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
/
Dynamics of marine bacterial communities from surface to bottom and the factors controling them
(USC Thesis Other)
Dynamics of marine bacterial communities from surface to bottom and the factors controling them
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
Dynamics of marine bacterial communities
from surface to bottom and the factors
controlling them
Jacob A. Cram
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)
December 2014
Copyright 2014 Jacob Adrian Cram
Cram
2
Acknowledgements
All of the work described here was collaborative, and would not have been possible without the
support of many individuals. My advisor, Jed Fuhrman’s leadership, mentorship, support and
good nature has been invaluable as I carried out this research. My dissertation committee
Douglas Capone, David Caron, Fengzhu Sun and Najmedin Meshkati provided substantial help
and advice. Jed’s research group consisted of a number of fantastic individuals alongside of
whom it was a great privilege to work. I especially thank Cheryl Chow, Anand Patel, Rohan
Sachdeva, Mahira Kakajiwala, David Needham, Alma Parada, Elizabeth Teel, Laura Gómez-
Consarnau, Catherine Roney, Ella Sieradzki, Nathan Ahlgren, Lyria Berdjeb and Erin Fichot.
This time series would not have been possible without the help of numerous individuals. I
especially thank John Heilderberg, Dale Kiefer, Roberta Marinelli, Troy Gunderson, Diane Kim,
Adriane Jones, Allie Lie, Sarah Hu, Victoria Campbell, Michael Morando, Bridget Seegers, Xiao
Liu, Lin Chang, Tu-My To, Henry Ho, Kieran Bartholow, Vartis Tsontos, Tim Lam and the
crews of the RVs Seawatch and Yellowfin. I thank William Berelson, Burt Jones, John
Heidelberg and Sergio Sañudo-Wilhelmy, for advice. I thank the Wrigley Institute for
Environmental Studies and its personnel for help and support in the field including Donal
Manahan, Robeta Marinelli, Linda Duguay, Sean Conner, Gerry Smith, Gordon Boivin, Lauren
Oudin, Kellie Spafford, Karla Heildelberg and J. Heidelberg. This work was supported by NSF
grant numbers 0703159, 1136818 and a grant from the Gordon and Betty Moore Foundation
Marine Microbiology Initiative and the Wrigley Institute for Environmental Studies.
Cram
3
Contents
Acknowledgements ......................................................................................................................... 2
List of Figures ................................................................................................................................. 4
List of Tables ................................................................................................................................... 5
Abstract: .......................................................................................................................................... 6
Chapter 1 : Temporal variability of marine bacteria throughout the world’s oceans ...................... 8
Temporal variability is a key component of biogeography....................................................... 10
A review of time series, what they tell us about variable throughout the water column, and
how that variability shapes ecosystem processes. ..................................................................... 11
Temporal variability of specific species.................................................................................... 20
Research Aims........................................................................................................................... 21
References ................................................................................................................................. 21
Chapter 2 : Seasonal and interannual variability of the marine bacterioplankton community
throughout the water column over ten years ................................................................................. 32
Abstract: .................................................................................................................................... 32
Introduction: .............................................................................................................................. 33
Methods: .................................................................................................................................... 38
Results: ...................................................................................................................................... 43
Discussion ................................................................................................................................. 59
Conclusion ................................................................................................................................. 70
Supplement ................................................................................................................................ 72
Supplemental Methods .............................................................................................................. 88
References ................................................................................................................................. 96
Chapter 3 : Network analysis of planktonic marine bacteria throughout the entire water
column......................................................................................................................................... 106
Abstract ................................................................................................................................... 106
Introduction ............................................................................................................................. 107
Methods ................................................................................................................................... 111
Results ..................................................................................................................................... 115
Discussion ............................................................................................................................... 128
Conclusion:.............................................................................................................................. 132
Supplement .............................................................................................................................. 134
References ............................................................................................................................... 141
Chapter 4 : Dilution reveals how viral lysis and grazing shape microbial communities ............ 145
Abstract ................................................................................................................................... 145
Introduction ............................................................................................................................. 146
Methods ................................................................................................................................... 150
Results ..................................................................................................................................... 158
Discussion ............................................................................................................................... 171
Conclusions ............................................................................................................................. 179
Supplement .............................................................................................................................. 180
References ............................................................................................................................... 183
Bibliography ............................................................................................................................... 188
Cram
4
List of Figures
Figure 2-1 Median values of environmental parameters .............................................................. 45
Figure 2-2 Mean Bray-Curtis similarities of all pairs of samples ................................................. 48
Figure 2-3 A. Mean biodiversity index scores by depth ............................................................... 51
Figure 2-4 Heat map of relative abundances of bacterial taxa ...................................................... 56
Figure 3-1 Association networks for each depth..........................................................................119
Figure 3-2 Association network between depths ........................................................................ 122
Figure 3-3 Association network Deltaproteobacteria OTUs at 890m ......................................... 125
Figure 3-4 Association network Marine Group A phylum at 150m, 500m and 890m................ 126
Figure 4-1 Outline of bacterial dilution setup ............................................................................. 152
Figure 4-2 Total abundances of bacteria, virus like particles, protists ........................................ 161
Figure 4-3 Non metric multidimensional scaling plots ............................................................... 164
Figure 4-4 Abundances of selected OTUs over time .................................................................. 167
Figure 4-5 Abundances of selected OTU over time .................................................................... 168
Supplemental Figure 2-1 Plots of median environmental parameters .......................................... 73
Supplemental Figure 2-2 Richness statistics................................................................................. 74
Supplemental Figure 2-3 Heat map of relative abundances of bacterial groups by sample ......... 75
Supplemental Figure 2-4 Heat maps of relative abundances of taxa by month ............................ 76
Supplemental Figure 2-5 Heat maps of the relative abundance of the taxonomic groups given in
Supplemental Figure 2-4, by sample ............................................................................................ 78
Supplemental Figure 2-6 The abundance of the five most abundant species at each depth across
time ............................................................................................................................................... 80
Supplemental Figure 3-1 Association network positive bacterial correllations.......................... 134
Supplemental Figure 3-2 Association network positive and negative bacterial correllations .... 135
Supplemental Figure 3-3 Association network,Deltaproteobacteria at 150, 500 and 890m ....... 136
Supplemental Figure 3-4 Association network SAR11 Surface-1 clade .................................... 137
Supplemental Figure 3-5 Association network AEGEAN-169 clade ......................................... 138
Supplemental Figure 4-1 Distribution of groups, at SPOT which separated control group and
0.2μm dilution treatments on day 3 ............................................................................................ 180
Supplemental Figure 4-2 Distribution of groups that separate the control group and 0.2um
dilution treatment on day six....................................................................................................... 181
Supplemental Figure 4-3 Distribution of groups that separate the 0.2um dilution treatment and
0.02um dilution treatment on day six ......................................................................................... 181
Cram
5
List of Tables
Table 2-1: Summary of metrics of community seasonality at each depth .................................... 49
Table 2-2 Rho values of Mantel tests, community similarity vs environmental parameters ........ 53
Table 2-3 The abundance and temporal characteristics of taxonomic groups .............................. 57
Table 3-1 Topological statistics for networks of bacteria at each depth ..................................... 120
Table 3-2 Summary statistics for networks, correlations between bacteria at different depths .. 123
Table 3-3 Description of nodes seen in networks ....................................................................... 128
Table 4-1 Results of analysis of similarities ............................................................................... 165
Table 4-2 Results of similarity percentages tests ........................................................................ 170
Supplemental Table 2-1 GAMM environmental parameters ........................................................ 83
Supplemental Table 2-2 GAMM Class and Phylum ..................................................................... 84
Supplemental Table 2-3 GAMM Order and Family ..................................................................... 85
Supplemental Table 2-4 GAMM, groups from Supplemental Figure 2-4 .................................... 87
Supplemental Table 3-1 Topological statistics for networks of bacteria at each depth .............. 139
Supplemental Table 3-2 Summary cross-network statistics ....................................................... 140
Supplemental Table 4-1 Results of analysis of similarities ........................................................ 182
Cram
6
Abstract:
Marine bacteria throughout the water column are an important part of the marine ecosystem and
drive global geochemical processes, yet their long term ecological dynamics have been studied
largely in the euphotic zone and adjacent seasonally mixed depths. Exploring microbial
dynamics at monthly and interannual time scales throughout the water column provides an
opportunity to explore how microorganisms are shaped by their dynamic environment. This
environment includes not only the physical and chemical characteristics of the water in which the
microbes live but also the other organisms present in the water column. Predator-prey
relationships and infection processes are likely of particular importance as they preferentially
remove some organisms while leaving others, thereby shaping microbial communities. The San
Pedro Channel, off the coast of Los Angeles is an ideal model system for exploring microbial
dynamics, as its physics, chemistry and biology have been regularly sampled through the San
Pedro Ocean Time-series (SPOT). Previous analysis of data from SPOT has identified seasonal
and interannual dynamics of chemistry, biology and especially the structure of microbial
communities in surface water. The community structure of the deep water column had, before
now, been studied little; this work provides a first synthesis of temporal patterns over the entire
water column. This study examines temporal patterns in the community structure monthly at five
depths throughout the water column: 5m, the deep chlorophyll maximum (~15-40m), 150m,
500m and 890m (maximum depth 900m). Seasonal and long term variability of microbial
community structure, as well as the microbial community’s relationship to environmental factors
were examined at each depth using a variety of statistical approaches. Network association
analysis, a technique for visualizing patterns in statistical associations between many pairs of
variables simultaneously was applied to identify novel patterns in community structure and
Cram
7
dynamics. Network analysis also illuminated relationships between community structures at
different depths. In addition to analysis of observational data, we used experimental methods to
elucidate the way that predation and infection shape microbial communities.
Our analyses suggested that microbial communities varied seasonally in the euphotic zone and at
the bottom of the water column much more strongly than in the middle of the water column.
Furthermore changes in the abundance of many OTUs throughout the water column followed
changes in community structure and environmental factors in shallower depths. These
observations suggest that sinking particles and migrating zooplankton likely act to link different
depths of the water column. Experimental approaches artificially decreasing predator-prey
interactions and viral infection illustrated that these “top down” factors play an important role in
shaping microbial communities by favoring particular OTUs. All of our analyses focus on
identifying which specific OTUs vary over time, associate statistically with environmental
parameters and other microbes, and are removed by grazing and infection. Together our analysis
methods provide valuable insight about the processes, microbial interactions and key bacterial
groups that contribute to microbial dynamics in the ocean.
Cram
8
Chapter 1 : Temporal variability of marine
bacteria throughout the world’s oceans
Marine microscopic organisms drive many global geochemical processes, thus playing a central
role in maintaining the function of earth's ecosystems (Falkowski et al. 2008; Strom 2008). The
smallest and most abundant organisms are prokaryotes (Ducklow & Carlson 1992; Whitman et
al. 1998), single celled organisms in the genetically and physiologically distinct domains of
Bacteria and Archaea (Woese et al. 1990). Despite their small size, Bacteria and Archaea often
dominate marine biomass (Fuhrman et al. 1989) and productivity (Fuhrman & Azam 1982;
Azam et al. 1983).
Marine bacteria include heterotrophic members, which take up dissolved organic matter as well
as autotrophic members which fix inorganic carbon dioxide. (Sherr & Sherr 2000; Kirchman
2008). Heterotrophic bacteria are a vital part of marine food web dynamics because they recycle
dissolved organic nutrients through the microbial loop (Azam et al. 1983), a process that occurs
to different degrees throughout the entire water column (Aristegui et al. 2009). Autotrophic
bacteria play key roles in a number of nutrient cycles: Photoautotrophic bacteria along with
phytoplankton fix carbon in surface waters, while chemoautotrophic bacteria oxidize and reduce
nitrogen species especially in the deep ocean (Zehr & Ward 2002; Spieck & Bock 2005). Many
heterotrophic bacteria appear to use light for energy, while predominantly autotrophic bacteria
such as Synechococcus take up reduced carbon compounds complicating traditional distinctions
between these categories (Béjà & Suzuki 2008). Microbial taxa are diverse; surface waters have
been estimated to have between 100 and 500 species level groups of microorganims, depending
on location and season (Ladau et al. 2013). In addition, diversity in the Sargasso Sea has been
Cram
9
shown to vary with depth (Vergin et al. 2013b). While early biogeochemistry literature was able
to make great progress by generalizing the roles of microbes as a whole when discussing
ecological processes, microbes have different effects on the environment and interact with each
other (Fuhrman et al. 1994) necessitating understanding of community structure when
investigating ecosystem processes. For instance, dissolved organic matter (DOM) pools are
complex (Nagata 2008) and are made of many different dissolved organic compounds. Abundant
microbial groups often metabolize only certain classes of compounds (e.g. Dupont et al. 2012;
Tripp 2013) with different microbes metabolizing these diverse compounds. Thus, one would
expect that the relative abundances and activities of different microbial groups might lead to
different interactions with the DOM pool.
The mesopelagic ocean is defined as the region of the ocean below the euphotic zone where light
is available for photosynthesis, but above 1000m (Aristegui et al. 2009). While the primary
source of energy in the euphotic zone is sunlight, bacteria in the mesopelegic rely on dissolved
organic matter for energy, often transported from surface waters, or else produce energy through
chemoautotrophy (Aristegui et al. 2009). While bacteria are more abundant in cells per unit
volume in surface waters than in the mesopelagic, the mesopelagic has greater total volume and
as a result there as many prokaryotic cells in the mesopelagic ocean as the euphotic zone
(Reinthaler et al. 2006; Herndl et al. 2008; Aristegui et al. 2009). It is no surprise, given the
difference between surface and deep waters that the structure both of microbial communities
(Treusch et al. 2009) and microbial genes (DeLong et al. 2006) vary with depth.
Of the two domains of prokaryotes, bacteria are the most diverse and abundant, in most observed
marine environments, (Aller & Kemp 2008). Because bacterial community composition varies
geographically at a number of scales (Hewson et al. 2006b; Rusch et al. 2007; Pommier et al.
Cram
10
2007; Fuhrman et al. 2008; Fortunato et al. 2011) and across time (Morris et al. 2005; Fuhrman
& Steele 2008; Gilbert et al. 2009), the overall effect of bacteria on ocean chemistry and marine
food webs likely varies regionally and depends on interactions within that community.
Understanding the interactions between bacteria and their environment necessitates first
understanding this spatio-temporal distribution of microbes. From there, it is logical to look for
patterns that inform why a set of bacteria live in a given location. The study of which microbes
are found in different locations, depths and times in the ocean is called biogeography. While
biogeography has been most easily and thoroughly studied in surface waters, microbial
communities do have spatiotemporal patterning in deep water (Agogué et al. 2011), and
understanding these communities throughout the water column is critical.
Temporal variability is a key component of biogeography
In their review of bacterial biogeography literature, Martiny et al. (2006) explain that the
microbial community in a given location and time is shaped by on two factors: habitat and
provinces. A habitat is the set environmental conditions presently influencing the microbial
community. A province is the historical circumstances, which include geographic separation, that
shape that microbial community. Habitats are defined by a combination of bottom up, top down
and sideways controls (Fuhrman & Hagstrom 2008). Bottom up controls are factors in the
chemical and physical environment that effect bacterial growth. These include energy inputs like
light and dissolved carbon, nutrients such as nitrogen, phosphorus, or trace elements. They also
include other factors that can affect cellular processes, such as temperature, salinity (Church
2008) and vitamins (Ayers 1960; Rodionov et al. 2003). Top down controls are those factors that
cause removal of bacteria and generally include predation by protists (Fuhrman & Hagstrom
2008; Jurgens & Massana 2008) and viral infection (Fuhrman 1999; Suttle 2007; Breitbart 2012).
Cram
11
Examples of sideways controls are competitive and symbiotic interactions between bacteria.
These interactions include resource competition, allelopathy and predation by bacteria (Fuhrman
& Hagstrom 2008).
Provinces are the historical events that led up to the community having a given set of
characteristics, and are a combination of dispersal limitation and past environmental conditions
(Martiny et al. 2006). It has been suggested that microbes face no barriers to dispersal, but rather
are shaped only by environmental processes, a contention known as the “everything is
everywhere” hypothesis (Fenchel et al. 1997; Finlay 2002). However, because inhospitable
environments may separate habitats for which a microbe is adapted, it is likely that at least some
microbes have barriers to dispersal (Martiny et al. 2006; Medlin 2007). It has been recently
suggested that ocean environments contain, in some concentration, every microbe that has ever
been abundant in that environment; a suggestion known as the “everything is everywhen”
hypothesis (Gibbons et al. 2013). Beyond dispersal, the histories of environmental factors play
clearer roles in shaping microbial communities. For instance, microbial communities after a
“boom” event may experience a similar environment to those that exist at other times of year, but
the residual effects of blooms have been shown to shape microbial communities for days to
weeks (Larsen et al. 2004; Teeling et al. 2012; Tada et al. 2012; Paver et al. 2013)
A review of time series, what they tell us about variable throughout the water
column, and how that variability shapes ecosystem processes.
Examining microbial time-series provides insight into how both present environment and
historical factors shape environments. Time series distributed globally together illustrate not only
how microbial communities change across time, but also how temporal variability relates to
geography. A number of short term (less than one year) time series have suggested seasonal
Cram
12
patterns in a number of environments, that vary in scale depending on region (Rich et al. 2011;
Teeling et al. 2012; Sintes et al. 2013; Ottesen et al. 2013). Furthermore, a number of long term
(greater than one year) time series have identified long term patterns in total numbers of bacteria,
chlorophyll abundance and other microbially relevant parameters (Jeandel et al. 1998; Harrison
2002; Wiltshire et al. 2010). However, there are only a handful of sites that have been well
studied over multiple years. These include the Sargasso Sea, the North Pacific Gyre, the North
West Mediterranean Sea, the Western English Channel and the San Pedro ocean time-series,
which is the primary subject of this dissertation. In this section, we focus on patterns in
community structure identified at different time scales. We are particularly interested in how
microbial community structure varies in deep water and how that variability relates to surface
environments.
Large time-series and other data sets present challenges in identifying trends and patterns due to
the data's complexity. A variety of tools are available for investigating large multivariate datasets
(Ramette 2007). One statistical tool, association network analysis, warrants special note, as it has
been particularly fruitful in identifying novel patterns. Association networks allow statistical
comparisons of many pairs of parameters in complex data sets, and allow these comparisons to
be visualized in such a way that one can identify patterns within those associations. In microbial
data sets, networks provide an overview that allows researchers to observe overlying patterns in
statistical associations that describe the data set and to generate hypotheses about ecologically
meaningful relationships (Fuhrman & Steele 2008; Faust et al. 2012; Cram et al. 2014a).
In two of the environments, the Western English Chanel and the San Pedro Chanel, in addition to
variability across time, the authors applied network association analysis using a local similarity
analysis technique (Ruan et al. 2006) to identify statistical associations between species level
Cram
13
OTUs with the aim of identifying groups of bacterial, protistan and viral OTUs that had similar
abundance patterns (Fuhrman & Steele 2008; Gilbert et al. 2012). These related patterns suggest
symbioses, predation events, or shared resources (Fuhrman & Steele 2008). In addition to marine
environments, association network analysis has been used to evaluate associations between
microbial taxa and between microbial genes in various environments including lake systems
(Eiler et al. 2012; Kara et al. 2012), soil (Zhou et al. 2010; Barberán et al. 2011), the human
microbiome (Arumugam et al. 2011; Faust et al. 2012), and globally through meta-analysis
across diverse sampling sites (Chaffron et al. 2010; Freilich et al. 2010). What follows is a
description of the major long term microbial time series. We describe patterns that are found at
different time scales throughout the water column. In sites where network association analysis
was applied we indicate the novel patterns that it identified.
Bermuda
The Bermuda Atlantic Time Series (BATS) is located southeast of Bermuda, on the northern
edge of the Sargasso Sea at a depth of 4800m (Michaels et al. 1994; Steinberg et al. 2001).
Temperature profiles at BATS vary seasonally with significant winter convective mixing down
to depths of 160-200m, followed by re-stratification. This mixing is reflected by seasonal
variability of nutrient and chlorophyll concentrations in the top 200m of the water column. Less
physical variability is seen below these depths.
These seasonal patterns in the physics and chemistry of BATS are reflected in the microbial
community structure, which varies seasonally in response to this vertical mixing event and which
is driven by seasonal variability of a several of taxa (Morris et al. 2005, 2012; Treusch et al.
2009; Giovannoni & Vergin 2012; Vergin et al. 2013a, 2013b). This seasonal variability at
BATS is primarily evident in the euphotic zone and is associated primarily with the spring bloom
Cram
14
(Treusch et al. 2009). Environmental variability at BATS has been shown to be shaped by the
movement of ~200km mesoscale eddies (Sweeney et al. 2003; Ewart et al. 2008). Mesoscale
eddies come in three types, cyclonic, anticyclonic and mode water eddies, each of which have
been shown to have unique impacts on surface and underlying waters (Sweeney et al. 2003;
Ewart et al. 2008). Microbial community structure at BATS was carefully investigated across a
mode water eddy (Nelson et al. 2014), a feature in which isopycnals are lensed, rotation is
clockwise and upwelling occurs (Sweeney et al. 2003). The mode water eddy appeared to enrich
the abundance of some taxa, deplete other taxa and uplift taxa from the mesopelagic zone into
shallower waters (Nelson et al. 2014). It was also considered a bloom forming eddy in that it
appeared to uplift nutrients and increase the abundance of phytoplankton and bacterial
productivity (McGillicuddy et al. 2007; Ewart et al. 2008; Nelson et al. 2014).
Hawaii
The Hawaii Ocean Time-series (HOT) is located in the North Pacific Gyre, 100km north of the
Hawaiian island of Oahu. Like BATS, it is located over 4800m deep water (Karl & Lukas 1996).
Compared to BATS, HOT’s water column has been characterized less saline, lower oxygen and
higher in nitrogen. The sea surface at HOT has less variable temperature, mixed layer depth and
nutrient ratios (Church et al. 2013).
Microbial communities at HOT, as measured by terminal restriction fragment length
polymorphism (TRFLP), have been shown to stratify by depths, and show less seasonal
variability than BATS in the euphotic zone (Giovannoni & Vergin 2012). Some seasonality
however, is evident at HOT, both in surface waters and in some, but not all layers of the water
column above 200m (Eiler et al. 2011). It has been suggested that HOT’s year round surface
communities are similar to the summer surface community at BATS (Giovannoni & Vergin
Cram
15
2012). At diel time scales, microbial communities at HOT have been shown to have differing
gene expression between day and night with genes related to photosynthesis and related energy
acquisition processes more active during the day (Poretsky et al. 2009).
Northwest Mediterranean
The Mediterranean Sea, unlike other marine environments with well characterized time series, is
a mostly enclosed body of water, and is characterized by warmer saltier waters, high surface
irradiance and relatively low nutrient levels. The surface environment shows seasonal variability
particularly in temperature, light levels, rainfall and stratification (Duarte et al. 1999). Winter
mixing in the North West Mediterranean has been shown to extend down to 100m (D’Ortenzio
2005). There is also notable long term variability, much of it linked to anthropogenic effects
(Duarte et al. 1999).
There are two microbial observatories in the North West Mediterranean Sea, located just over
100km from each other. In the North, about four km off the coast of Banyuls France is the
Banyuls observatory. In the south about 1km off coast of Blanes Spain is the Blanes Bay
Observatory.
Archaeal communities have been investigated over multiple years at both of these time series.
Archaea were monitored using a combination of qPCR and clone libraries over a 4.5 year period
at Blanes Bay. The total abundance of Archaea, and several groups within Archaea including
Group II Euryarchaeota, Group I Crenarchaeota and AmoA gene were shown to vary seasonally
by qPCR. Meanwhile clone library data suggested that the community composition of Archaeal
subgroups shifted seasonally as well (Galand et al. 2010). 16s pyrosequencing of Archaeal rRNA
and rDNA at Banules showed seasonal variability of not only abundant Archaea, but also of less
Cram
16
abundant Archaeal groups. These less abundant groups showed several prevailing patterns. Some
groups were occasionally abundant (local seed bank); some were never abundant at that site
(nonlocal seed bank); others were never abundant, but actively metabolizing (active-but rare)
(Hugoni et al. 2013). At both sites, some year to year variability in the abundances of certain
microbial groups was evident, but in neither case was a long term trend discussed (Galand et al.
2010; Hugoni et al. 2013).
Bacteria were investigated over two disconnected one year time periods in two separate studies
at Blanes Bay. In the first case (1997-1998) DGGE fingerprinting was used (Schauer et al. 2003),
while in the second year (2003-2004) combination of clone libraries, DGGE and FISH (Alonso-
Sáez et al. 2007) were applied. In both years, seasonal patterns in bacterial community structure
were identified. Seasonal groups of bacteria in both studies included taxa, such as SAR11 that
were also seasonal in other environments, such as BATS (Schauer et al. 2003; Alonso-Sáez et al.
2007).
Western English Channel
Station L4 in the Western English Channel is located 10km off shore, in water that is only 55m
deep (Pingree & Griffiths 1978; Southward et al. 2004). Situated at a boundary region between
oceanic and coastal environments (Southward et al. 2004) L4’s environment is subject to
estuarine outflow from Plymouth sound and seasonal mixing (Southward et al. 2004) in winter to
depths of 35m (Gilbert et al. 2012). This seasonal mixing has been shown to influence dynamics,
physical and chemical parameters in the channel (Pingree & Griffiths 1978; Southward et al.
2004).
Cram
17
The Plymouth Marine Lab (PML) time series samples monthly at station L4 and examined
microbes at the sea surface, along with many classically identified phytoplankton and
zooplankton (Gilbert et al. 2009, 2012). Microbial populations have been shown to be strongly
seasonal in their composition in this environment (Gilbert et al. 2009, 2012). Network analysis
examined the 300 most abundant microbes, identified by high throughput sequencing. Overall
network structure was characterized by a few features. First, a large interconnected group of
bacteria in which nodes were connected to most of the other nodes in the group. This group was
statistically associated with winter months which coincide with high dissolved nitrate and nitrite
concentrations. Second, there was a more loosely-connected group of Eukaryotes and
Prokaryotes that appeared to be most abundant during the spring. Generally, it appeared that
most Eukaryotic OTUs were correlated with other Eukaryotes while bacterial OTUs correlated
with other bacteria, while there were also cases in which particular taxa were shown to correlate
with a number of taxa from the other domain (Gilbert et al. 2012).
Southern California
The San Pedro Ocean Time-Series (SPOT) is the site on which this dissertation will focus. It is a
coastal time series situated in the San Pedro Channel off the coast of Southern California. SPOT
is directly above the bottom of the San Pedro Basin at a depth of 890m. Seasonal factors at
SPOT involve both seasonal mixing (Chow et al. 2013) as well as seasonal variability in current
structure (Hickey 1992). A ten year time-series of the basin's temperature, nutrient and oxygen
profiles was run about a decade prior to the establishment of the microbial observatory and
suggested that water is generally contained in the basin. Periodically basin water is replaced in
basin flushing events where cooler high nutrient high oxygen water replaces warmer low nutrient
low oxygen water (Berelson 1991). Furthermore, particle flux, which has been shown to be an
Cram
18
important factor in driving variability in deep water processes throughout the ocean (Herndl et al.
2008) has been shown to vary seasonally both in magnitude and chemical composition at SPOT
(Collins et al. 2011).
At SPOT, Bacterial (Fuhrman & Steele 2008; Chow et al. 2013) and Protistan (Kim et al. 2013)
communities have both been examined, along with the abundance of some Archaeal groups
(Beman et al. 2010, 2011) from the surface and deep chlorophyll maximum layer over a period
of ten years. These studies used molecular approaches including ARISA and TRFLP
fingerprinting and qPCR methods (Brown et al. 2005; Beman et al. 2010; Kim et al. 2012).
Microbial communities in SPOT surface waters have been shown to vary seasonally over the ten
year data set (Fuhrman et al. 2006; Fuhrman & Steele 2008; Hatosy et al. 2013; Kim et al. 2013;
Chow et al. 2013). They also show a number of patterns of variability at short time scales, with
some species level operational taxonomic units (OTU) showing stable abundance over time,
others showing steady or sudden increases and decreases in abundance (Needham et al. 2013).
Microbial communities in deeper water varied little at 500m over a short section of the time
series (Hewson et al. 2006a) while seasonal variability was tested for but not identified at 890m
(Hatosy et al. 2013).
A focus of SPOT has been in identifying co-occurrence patterns between different
microorganisms using association network analysis. Complex patterns of interactions have been
identified between species level groups in the surface mixed layer and Deep Chlorophyll
Maximum (Steele et al. 2011; Chow et al. 2014). At short time scales, microbial communities
have been shown to have different interaction patterns than at long time scales. For instance,
over the period of one month, closely related SAR11 OTUs show similar temporal dynamics;
meanwhile over seasonal and interannual time scales those same OTUs show different dynamics
Cram
19
to each other (Fuhrman & Steele 2008; Steele et al. 2011; Needham et al. 2013). In deeper
environments co-occurrence patterns have been identified between a few key taxa (Beman et al.
2011).
Interactions between bacteria, protists and some viruses were examined and non-random patterns
of association were identified (Steele et al. 2011; Needham et al. 2013; Chow et al. 2014). For
instance it appeared that that bacteria are more likely to statistically associate with other bacteria
and protists with other protists, but there were also many examples of associations between the
two domains (Chow et al. 2014). Of particular note was one cluster of strongly co-occurring
Eukaryotes (Steele et al. 2011). The authors hypothesized that such a pattern could be due to
symbiosis, endoparasatism, or several genes within the same organism. Analysis of overall
network topology (mathematical descriptions of network patters) suggested that the network
demonstrated “small world properties.” This meant that there were several particular OTUs that
were highly connected with many other OTUs, which suggest the presence of keystone species
(Steele et al. 2011).
General trends
Microbial communities at all sites show depth stratification (if samples are taken below the
surface) and more seasonality in surface waters than in deep waters. There appears to be a
general pattern in which seasonal variability is generally a strong factor at all sites except HOT.
A likely scenario is that, because environmental factors have the strongest seasonal variability
closest to the poles and weakest towards the equator, this environmental variability drives
differences in microbial community variability. It is challenging to compare these environments
quantitatively because many of them of them have been analyzed using different methods.
Approaches have emerged to cross compare such differently analyzed datasets and have
Cram
20
suggested, for instance, that the SPOT and PML datasets have similar patterns of short term,
seasonal and interannual variability (Hatosy et al. 2013). These investigations have revealed that
both SPOT and the western English Channel show greater seasonal variability than interannual
variability, but that interannual variability does exist at both sites (Hatosy et al. 2013; Chow et al.
2013).
Temporal variability of specific species
Microbial communities are made of many different groups of organisms, each of which have
important ecological roles. The aforementioned time-series have provided valuable information
about the spatiotemporal distributions of many specific groups of micro-organisms. For instance,
closely related groups of SAR11 have been shown to have different temporal distributions over
time and depth in the Sargasso Sea (Carlson et al. 2009; Vergin et al. 2013a). The abundance
patterns of rare microbes (relative abundance < 0.1 of the community) showed different patterns
that also differed from more abundant organisms. For instance, some rare organisms showed
spikes in abundance either at specific depths or throughout the water column (Vergin et al.
2013b). The dynamics of classically deep water micro-organisms are less well understood. The
Marine Group A phylum is known to be an important component of deep, especially low oxygen
waters and is known to have variability across both depth and space (Allers et al. 2013), though
less is known about its temporal dynamics. At SPOT, the temporal distributions of Nitrospina
and Archaea have been examined in deep waters. It was shown that Nitrospina OTUs followed
similar patterns over time, but did not covary with the abundance of various archaeal markers
(Beman et al. 2010).
Cram
21
Research Aims
The goal of this research is to shed light on seasonal and long term patterns of microbial
variability throughout the water column. We will identify how environmental variability, along
with interactions between microorganisms works to shape microbial communities. The
remainder of this dissertation is structured as follows: Chapter 2 examines microbial
communities and their chemical and physical environment at each depth within the water
column. This section focuses on identifying seasonal and interannual trends in microbial
community structure, trends in environmental and biotic factors, and on how these environmental
factors relate to community structure. It also identifies how environmental variability shapes
microbial communities structure. Chapter 3 focuses on statistical associations between bacteria
with the aim not only of asking which groups of microorganisms are associated within a depth
but how microbial communities at different depths are related to the environment and
community structure at other depths. Chapter 4 takes an experimental approach to understanding
how interactions among grazers, viruses and microbes shape microbial communities by
artificially decreasing interactions between microorganisms. All of these approaches focus not
only on how microbial communities vary as a whole, but on how specific groups of organisms
vary over time and what these patterns likely tell us about ecosystem function.
References
Agogué H, Lamy D, Neal PR, Sogin ML, Herndl GJ. (2011). Water mass-specificity of bacterial
communities in the North Atlantic revealed by massively parallel sequencing. Mol. Ecol.
20:258–274.
Aller JY , Kemp PF. (2008). Are Archaea inherently less diverse than Bacteria in the same
environments? FEMS Microbiol. Ecol. 65:74–87.
Cram
22
Allers E, Wright JJ, Konwar KM, Howes CG, Beneze E, Hallam SJ, et al. (2013). Diversity and
population structure of Marine Group A bacteria in the Northeast subarctic Pacific Ocean. ISME
J. 7:256–268.
Alonso-Sáez L, Balagué V , Sà EL, Sánchez O, González JM, Pinhassi J, et al. (2007).
Seasonality in bacterial diversity in north-west Mediterranean coastal waters: assessment through
clone libraries, fingerprinting and FISH. FEMS Microbiol. Ecol. 60:98–112.
Aristegui J, Gasol JM, Duarte CM, Herndl GJ. (2009). Microbial oceanography of the dark
ocean’s pelagic realm. Limnol. Oceanogr. 54:1501–1529.
Arumugam M, Raes J, Pelletier E, Le Paslier D, Yamada T, Mende DR, et al. (2011). Enterotypes
of the human gut microbiome. Nature 473:174–180.
Ayers WA. (1960). Specificity of the vitimin B12 requirement in certain marine bacteria. J.
Bacteriol. 80:744.
Azam F, Fenchel T, Field J, Gray J, Meyrriel L, Thingstad F. (1983). The ecological role of
water-column microbes in the sea. Mar. Ecol. -Prog. Ser. 10:257–263.
Barberán A, Bates ST, Casamayor EO, Fierer N. (2011). Using network analysis to explore co-
occurrence patterns in soil microbial communities. ISME J. 6:343–351.
Béjà O, Suzuki MT. (2008). Photoheterotrophic Marine Prokaryotes. In:Microbial Ecology of the
Oceans, Kirchman, DL, ed (ed)., John Wiley & Sons, Inc., pp. 131–157.
http://onlinelibrary.wiley.com/doi/10.1002/9780470281840.ch5/summary (Accessed June 2,
2014).
Beman JM, Sachdeva R, Fuhrman JA. (2010). Population ecology of nitrifying Archaea and
Bacteria in the Southern California Bight. Environ. Microbiol. 12:1282–1292.
Beman JM, Steele JA, Fuhrman JA. (2011). Co-occurrence patterns for abundant marine
archaeal and bacterial lineages in the deep chlorophyll maximum of coastal California. ISME J.
5:1077–1085.
Berelson WM. (1991). The Flushing of Two Deep-Sea Basins, Southern California Borderland.
Limnol. Oceanogr. 36:1150–1166.
Breitbart M. (2012). Marine Viruses: Truth or Dare. Annu. Rev. Mar. Sci. 4:425–448.
Cram
23
Brown MV , Schwalbach MS, Hewson I, Fuhrman JA. (2005). Coupling 16S-ITS rDNA clone
libraries and automated ribosomal intergenic spacer analysis to show marine microbial diversity:
development and application to a time series. Environ. Microbiol. 7:1466–1479.
Carlson C, Morris R, Parsons R, Treusch A, Giovannoni S, Vergin K. (2009). Seasonal dynamics
of SAR11 populations in the euphotic and mesopelagic zones of the northwestern Sargasso Sea.
ISME J. 3:283–295.
Chaffron S, Rehrauer H, Pernthaler J, von Mering C. (2010). A global network of coexisting
microbes from environmental and whole-genome sequence data. Genome Res. 20:947–959.
Chow C-ET, Kim DY , Sachdeva R, Caron DA, Fuhrman JA. (2014). Top-down controls on
bacterial community structure: microbial network analysis of bacteria, T4-like viruses and
protists. ISME J. 8:816–829.
Chow C-ET, Sachdeva R, Cram JA, Steele JA, Needham DM, Patel A, et al. (2013). Temporal
variability and coherence of euphotic zone bacterial communities over a decade in the Southern
California Bight. ISME J. 7:2259–2273.
Church MJ. (2008). Resource control of bacterial dynamics in the sea. In:Microbial Ecology of
the Oceans, Hoboken, N.J., pp. 335–382.
Church MJ, Lomas MW, Muller-Karger F. (2013). Sea change: Charting the course for
biogeochemical ocean time-series research in a new millennium. Deep Sea Res. Part II Top.
Stud. Oceanogr. 93:2–15.
Collins LE, Berelson W, Hammond DE, Knapp A, Schwartz R, Capone D. (2011). Particle fluxes
in San Pedro Basin, California: A four-year record of sedimentation and physical forcing. Deep
Sea Res. Part Oceanogr. Res. Pap. 58:898–914.
Cram JA, Sun F, Fuhrman JA. (2014). Marine bacterial, archaeal and protistan association
networks. In:Encyclopedia of Metagenomics, SpringerReference.
D’Ortenzio F. (2005). Seasonal variability of the mixed layer depth in the Mediterranean Sea as
derived from in situ profiles. Geophys. Res. Lett. 32.
http://doi.wiley.com/10.1029/2005GL022463 (Accessed May 12, 2014).
Cram
24
DeLong EF, Preston CM, Mincer T, Rich V , Hallam SJ, Frigaard N-U, et al. (2006). Community
Genomics Among Stratified Microbial Assemblages in the Ocean’s Interior. Science 311:496–
503.
Duarte CM, Agustı
́ S, Kennedy H, Vaqué D. (1999). The Mediterranean climate as a template for
Mediterranean marine ecosystems: the example of the northeast Spanish littoral. Prog. Oceanogr.
44:245–270.
Ducklow HW, Carlson CA. (1992). Oceanic Bacterial Production. In:Advances in Microbial
Ecology, Marshall, KC, ed (ed). Advances in Microbial Ecology, Springer US, pp. 113–181.
http://link.springer.com.libproxy.usc.edu/chapter/10.1007/978-1-4684-7609-5_3 (Accessed May
31, 2014).
Dupont CL, Rusch DB, Yooseph S, Lombardo M-J, Richter RA, Valas R, et al. (2012). Genomic
insights to SAR86, an abundant and uncultivated marine bacterial lineage. ISME J. 6:1186–1199.
Eiler A, Hayakawa DH, Rappé MS. (2011). Non-random assembly of bacterioplankton
communities in the subtropical north pacific ocean. Front. Microbiol. 2:140.
Eiler A, Heinrich F, Bertilsson S. (2012). Coherent dynamics and association networks among
lake bacterioplankton taxa. ISME J. 6:330–342.
Ewart CS, Meyers MK, Wallner ER, McGillicuddy Jr. DJ, Carlson CA. (2008). Microbial
dynamics in cyclonic and anticyclonic mode-water eddies in the northwestern Sargasso Sea.
Deep Sea Res. Part II Top. Stud. Oceanogr. 55:1334–1347.
Falkowski PG, Fenchel T, Delong EF. (2008). The Microbial Engines That Drive Earth’s
Biogeochemical Cycles. Science 320:1034–1039.
Faust K, Sathirapongsasuti JF, Izard J, Segata N, Gevers D, Raes J, et al. (2012). Microbial co-
occurrence relationships in the human microbiome. PLoS Comput. Biol. 8:e1002606.
Fenchel T, Esteban GF, Finlay BJ. (1997). Local versus Global Diversity of Microorganisms:
Cryptic Diversity of Ciliated Protozoa. Oikos 80:220–225.
Finlay BJ. (2002). Global dispersal of free-living microbial eukaryote species. SCIENCE
296:1061–1063.
Cram
25
Fortunato CS, Herfort L, Zuber P, Baptista AM, Crump BC. (2011). Spatial variability
overwhelms seasonal patterns in bacterioplankton communities across a river to ocean gradient.
ISME J. 6:554–563.
Freilich S, Kreimer A, Meilijson I, Gophna U, Sharan R, Ruppin E. (2010). The large-scale
organization of the bacterial network of ecological co-occurrence interactions. Nucleic Acids
Res. 38:3857–3868.
Fuhrman J, Azam F. (1982). Thymidine incorporation as a measure of heterotrophic
bacterioplankton production in marine surface waters - evaluation and field results. Mar. Biol.
66:109–120.
Fuhrman J, Sleeter T, Carlson C, Proctor L. (1989). dominance of bacterial biomass in the
sargasso sea and its ecological implications. Mar. Ecol.-Prog. Ser. 57:207–217.
Fuhrman J, Steele J. (2008). Community structure of marine bacterioplankton: patterns,
networks, and relationships to function. Aquat. Microb. Ecol. 53:69–81.
Fuhrman JA. (1999). Marine viruses and their biogeochemical and ecological effects. Nature
399:541–548.
Fuhrman JA, Hagstrom A. (2008). Bacterial and archaeal community structure and its patterns.
In:Microbial Ecology of the Oceans, Wiley: Hoboken, N.J., pp. 45–90.
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.
103:13104 –13109.
Fuhrman JA, Lee SH, Masuchi Y , Davis AA, Wilcox RM. (1994). Characterization of Marine
Prokaryotic Communities via DNA and RNA. Microb. Ecol. 28:133–145.
Fuhrman JA, Steele JA, Hewson I, Schwalbach MS, Brown MV , Green JL, et al. (2008). A
latitudinal diversity gradient in planktonic marine bacteria. Proc. Natl. Acad. Sci. 105:7774–
7778.
Galand PE, Gutiérrez-Provecho C, Massana R, Gasol JM, Casamayor EO. (2010). Inter-annual
recurrence of archaeal assemblages in the coastal NW Mediterranean Sea (Blanes Bay Microbial
Observatory). Limnol. Oceanogr. 55:2117–2125.
Cram
26
Gibbons SM, Caporaso JG, Pirrung M, Field D, Knight R, Gilbert JA. (2013). Evidence for a
persistent microbial seed bank throughout the global ocean. Proc. Natl. Acad. Sci. 110:4651–
4655.
Gilbert JA, Field D, Swift P, Newbold L, Oliver A, Smyth T, et al. (2009). The seasonal structure
of microbial communities in the Western English Channel. Environ. Microbiol. 11:3132–3139.
Gilbert JA, Steele JA, Caporaso JG, Steinbrück L, Reeder J, Temperton B, et al. (2012). Defining
seasonal marine microbial community dynamics. ISME J. 6:298–308.
Giovannoni SJ, Vergin KL. (2012). Seasonality in Ocean Microbial Communities. Science
335:671–676.
Harrison PJ. (2002). Station papa time series: Insights into ecosystem dynamics. J. Oceanogr.
58:259–264.
Hatosy SM, Martiny JBH, Sachdeva R, Steele J, Fuhrman JA, Martiny AC. (2013). Beta
diversity of marine bacteria depends on temporal scale. Ecology 94:1898–1904.
Herndl G, Agogué H, Baltar F, Reinthaler T, Sintes E, Varela M. (2008). Regulation of aquatic
microbial processes: the ‘microbial loop’ of the sunlit surface waters and the dark ocean
dissected. Aquat. Microb. Ecol. 53:59–68.
Hewson I, Steele J, Capone D, Fuhrman J. (2006a). Remarkable heterogeneity in meso- and
bathypelagic bacterioplankton assemblage composition. Limnol. Oceanogr. 51:1274–1283.
Hewson I, Steele JA, Capone DG, Fuhrman JA. (2006b). Temporal and spatial scales of variation
in bacterioplankton assemblages of oligotrophic surface waters. Mar. Ecol.-Prog. Ser. 311:67–77.
Hickey BM. (1992). Circulation over the Santa Monica-San Pedro Basin and Shelf. Prog.
Oceanogr. 30:37–115.
Hugoni M, Taib N, Debroas D, Domaizon I, Dufournel IJ, Bronner G, et al. (2013). Structure of
the rare archaeal biosphere and seasonal dynamics of active ecotypes in surface coastal waters.
Proc. Natl. Acad. Sci. 110:6004–6009.
Jeandel C, Ruiz-Pino D, Gjata E, Poisson A, Brunet C, Charriaud E, et al. (1998). KERFIX, a
time-series station in the Southern Ocean: a presentation. J. Mar. Syst. 17:555–569.
Cram
27
Jurgens K, Massana R. (2008). Protistan grazing on marine bacterioplankton. In:Microbial
Ecology of the Oceans, Hoboken, N.J., pp. 383–441.
Kara EL, Hanson PC, Hu YH, Winslow L, McMahon KD. (2012). A decade of seasonal
dynamics and co-occurrences within freshwater bacterioplankton communities from eutrophic
Lake Mendota, WI, USA. ISME J. 7:680–684.
Karl DM, Lukas R. (1996). The Hawaii Ocean Time-series (HOT) program: Background,
rationale and field implementation. Deep Sea Res. Part II Top. Stud. Oceanogr. 43:129–156.
Kim DY , Countway PD, Jones AC, Schnetzer A, Yamashita W, Tung C, et al. (2013). Monthly to
interannual variability of microbial eukaryote assemblages at four depths in the eastern North
Pacific. ISME J. http://www.nature.com/ismej/journal/vaop/ncurrent/full/ismej2013173a.html
(Accessed December 10, 2013).
Kim DY , Countway PD, Yamashita 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.
Methods 91:527–536.
Kirchman DL. (2008). Introduction and Overview. In:Microbial Ecology of the Oceans,
Kirchman, DL, ed (ed)., John Wiley & Sons, Inc., pp. 1–26.
http://onlinelibrary.wiley.com/doi/10.1002/9780470281840.ch1/summary (Accessed June 2,
2014).
Ladau J, Sharpton TJ, Finucane MM, Jospin G, Kembel SW, O’Dwyer J, et al. (2013). Global
marine bacterial diversity peaks at high latitudes in winter. ISME J. 7:1669–1677.
Larsen A, Flaten GAF, Sandaa R-A, Castberg T, Thyrhaug R, Erga SR, et al. (2004). Spring
Phytoplankton Bloom Dynamics in Norwegian Coastal Waters: Microbial Community
Succession and Diversity. Limnol. Oceanogr. 49:180–190.
Martiny JBH, Bohannan BJM, Brown JH, Colwell RK, Fuhrman JA, Green JL, et al. (2006).
Microbial biogeography: putting microorganisms on the map. Nat. Rev. Microbiol. 4:102–112.
McGillicuddy DJ, Anderson LA, Bates NR, Bibby T, Buesseler KO, Carlson CA, et al. (2007).
Eddy/Wind Interactions Stimulate Extraordinary Mid-Ocean Plankton Blooms. Science
316:1021–1026.
Cram
28
Medlin LK. (2007). If everything is everywhere, do they share a common gene pool? Gene
406:180–183.
Michaels AF, Knap AH, Dow RL, Gundersen K, Johnson RJ, Sorensen J, et al. (1994). Seasonal
patterns of ocean biogeochemistry at the U.S. JGOFS Bermuda Atlantic time-series study site.
Deep Sea Res. Part Oceanogr. Res. Pap. 41:1013–1038.
Morris RM, Frazar CD, Carlson CA. (2012). Basin-scale patterns in the abundance of SAR11
subclades, marine Actinobacteria (OM1), members of the Roseobacter clade and OCS116 in the
South Atlantic. Environ. Microbiol. 14:1133–1144.
Morris RM, Vergin KL, Cho JC, Rappe MS, Carlson CA, Giovannoni SJ. (2005). Temporal and
spatial response of bacterioplankton lineages to annual convective overturn at the Bermuda
Atlantic Time-series Study site. Limnol. Oceanogr. 50:1687–1696.
Nagata T. (2008). Organic Matter–Bacteria Interactions in Seawater. In:Microbial Ecology of the
Oceans, Kirchman, DL, ed (ed)., John Wiley & Sons, Inc., pp. 207–241.
http://onlinelibrary.wiley.com/doi/10.1002/9780470281840.ch7/summary (Accessed June 1,
2014).
Needham DM, Chow C-ET, Cram JA, Sachdeva R, Parada A, Fuhrman JA. (2013). Short-term
observations of marine bacterial and viral communities: patterns, connections and resilience.
ISME J. 7:1274–1285.
Nelson CE, Carlson CA, Ewart CS, Halewood ER. (2014). Community differentiation and
population enrichment of Sargasso Sea bacterioplankton in the euphotic zone of a mesoscale
mode-water eddy. Environ. Microbiol. 16:871–887.
Ottesen EA, Young CR, Eppley JM, Ryan JP, Chavez FP, Scholin CA, et al. (2013). Pattern and
synchrony of gene expression among sympatric marine microbial populations. Proc. Natl. Acad.
Sci. 110:E488–E497.
Paver SF, Hayek KR, Gano KA, Fagen JR, Brown CT, Davis-Richardson AG, et al. (2013).
Interactions between specific phytoplankton and bacteria affect lake bacterial community
succession. Environ. Microbiol. 15:2489–2504.
Pingree RD, Griffiths DK. (1978). Tidal fronts on the shelf seas around the British Isles. J.
Geophys. Res. Oceans 83:4615–4622.
Cram
29
Pommier T, CanbäCk B, Riemann L, BoströM KH, Simu K, Lundberg P, et al. (2007). Global
patterns of diversity and community structure in marine bacterioplankton. Mol. Ecol. 16:867–
880.
Poretsky RS, Hewson I, Sun S, Allen AE, Zehr JP, Moran MA. (2009). Comparative day/night
metatranscriptomic analysis of microbial communities in the North Pacific subtropical gyre.
Environ. Microbiol. 11:1358–1375.
Ramette A. (2007). Multivariate analyses in microbial ecology. FEMS Microbiol. Ecol. 62:142–
160.
Reinthaler T, van Aken H, Veth C, Arístegui J, Robinson C, Williams PJ le B, et al. (2006).
Prokaryotic respiration and production in the meso- and bathypelagic realm of the eastern and
western North Atlantic basin. Limnol. Oceanogr. 51:1262–1273.
Rich VI, Pham VD, Eppley J, Shi Y , DeLong EF. (2011). Time-series analyses of Monterey Bay
coastal microbial picoplankton using a ‘genome proxy’ microarray. Environ. Microbiol. 13:116–
134.
Rodionov DA, Vitreschak AG, Mironov AA, Gelfand MS. (2003). Comparative Genomics of the
Vitamin B12 Metabolism and Regulation in Prokaryotes. J. Biol. Chem. 278:41148 –41159.
Ruan Q, Dutta D, Schwalbach MS, Steele JA, Fuhrman JA, Sun F. (2006). Local similarity
analysis reveals unique associations among marine bacterioplankton species and environmental
factors. Bioinformatics 22:2532–2538.
Rusch DB, Halpern AL, Sutton G, Heidelberg KB, Williamson S, Yooseph S, et al. (2007). The
Sorcerer II Global Ocean Sampling Expedition: Northwest Atlantic through Eastern Tropical
Pacific. PLoS Biol. 5:e77.
Schauer M, Balagué V , Pedrós-Alió C, Massana R. (2003). Seasonal changes in the taxonomic
composition of bacterioplankton in a coastal oligotrophic system. Aquat. Microb. Ecol. 31:163–
174.
Sherr EB, Sherr BF. (2000). Marine Microbes: An overview. In:Microbial Ecology of the
Oceans, Kirchman, D., ed (ed)., Wiley-Liss, pp. 351–386.
Cram
30
Sintes E, Witte H, Stodderegger K, Steiner P, Herndl GJ. (2013). Temporal dynamics in the free-
living bacterial community composition in the coastal North Sea. FEMS Microbiol. Ecol.
83:413–424.
Southward AJ, Langmead O, Hardman-Mountford NJ, Aiken J, Boalch GT, Dando PR, et al.
(2004). Long-Term Oceanographic and Ecological Research in the Western English Channel.
In:Advances in Marine Biology, V ol. V olume 47, Academic Press, pp. 1–105.
http://www.sciencedirect.com/science/article/pii/S0065288104470011 (Accessed April 2, 2014).
Spieck E, Bock E. (2005). The Lithoautotrophic Nitrite-Oxidizing Bacteria. In:Bergey’s
Manual® of Systematic Bacteriology, Brenner, DJ, Krieg, NR, Staley, JT, & Garrity, GM, eds
(ed)., Springer US, pp. 149–153.
http://www.springerlink.com/content/l67712075q6r051p/abstract/ (Accessed November 15,
2012).
Steele JA, Countway PD, Xia L, Vigil PD, Beman JM, Kim DY , et al. (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. Part Ii-Top. Stud. Oceanogr. 48:1405–1447.
Strom SL. (2008). Microbial Ecology of Ocean Biogeochemistry: A Community Perspective.
Science 320:1043–1045.
Suttle CA. (2007). Marine viruses — major players in the global ecosystem. Nat. Rev. Microbiol.
5:801–812.
Sweeney EN, McGillicuddy Jr. DJ, Buesseler KO. (2003). Biogeochemical impacts due to
mesoscale eddy activity in the Sargasso Sea as measured at the Bermuda Atlantic Time-series
Study (BATS). Deep Sea Res. Part II Top. Stud. Oceanogr. 50:3017–3039.
Tada Y , Taniguchi A, Sato-Takabe Y , Hamasaki K. (2012). Growth and succession patterns of
major phylogenetic groups of marine bacteria during a mesocosm diatom bloom. J. Oceanogr.
68:509–519.
Cram
31
Teeling H, Fuchs BM, Becher D, Klockow C, Gardebrecht A, Bennke CM, et al. (2012).
Substrate-Controlled Succession of Marine Bacterioplankton Populations Induced by a
Phytoplankton Bloom. Science 336:608–611.
Treusch AH, Vergin KL, Finlay LA, Donatz MG, Burton RM, Carlson CA, et al. (2009).
Seasonality and vertical structure of microbial communities in an ocean gyre. ISME J. 3:1148–
1163.
Tripp HJ. (2013). The unique metabolism of SAR11 aquatic bacteria. J. Microbiol. 51:147–153.
Vergin KL, Beszteri B, Monier A, Thrash JC, Temperton B, Treusch AH, et al. (2013a). High-
resolution SAR11 ecotype dynamics at the Bermuda Atlantic Time-series Study site by
phylogenetic placement of pyrosequences. ISME J. 7:1322–1332.
Vergin KL, Done B, Carlson CA, Giovannoni SJ. (2013b). Spatiotemporal distributions of rare
bacterioplankton populations indicate adaptive strategies in the oligotrophic ocean. Aquat.
Microb. Ecol. 71:1–13.
Whitman WB, Coleman DC, Wiebe WJ. (1998). Prokaryotes: The unseen majority. Proc. Natl.
Acad. Sci. 95:6578–6583.
Wiltshire KH, Kraberg A, Bartsch I, Boersma M, Franke H-D, Freund J, et al. (2010). Helgoland
Roads, North Sea: 45 Years of Change. Estuaries Coasts 33:295–310.
Woese CR, Kandler O, Wheelis ML. (1990). Towards a natural system of organisms: proposal
for the domains Archaea, Bacteria, and Eucarya. Proc. Natl. Acad. Sci. 87:4576–4579.
Zehr JP, Ward BB. (2002). Nitrogen Cycling in the Ocean: New Perspectives on Processes and
Paradigms. Appl. Environ. Microbiol. 68:1015–1024.
Zhou J, Deng Y , Luo F, He Z, Tu Q, Zhi X. (2010). Functional Molecular Ecological Networks.
mBio 1:e00169–10.
Cram
32
Chapter 2 : Seasonal and interannual variability
of the marine bacterioplankton community
throughout the water column over ten years
Abstract:
Microbial activities that affect global oceanographic and atmospheric processes happen
throughout the water column, yet the long-term ecological dynamics of microbes have been
studied largely in the euphotic zone and adjacent seasonally mixed depths. We investigated
temporal patterns in the community structure of free-living bacteria, by sampling approximately
monthly from 5m, the deep chlorophyll maximum (~15-40m), 150m, 500m and 890m, in San
Pedro Channel (maximum depth 900m, hypoxic below ~500m), off the coast of Southern
California. Community structure and biodiversity (inverse Simpson index) showed seasonal
patterns near the surface and bottom of the water column, but not at intermediate depths. Inverse
Simpson’s index was highest in the winter in surface waters and in the spring at 890m, and
varied interannually at all depths. Biodiversity appeared to be driven partially by exchange of
microbes between depths and was highest when communities were changing slowly over time.
Meanwhile, communities from the surface through 500m varied interannually. After accounting
for seasonality, several environmental parameters co-varied with community structure at the
surface and 890m, but not the intermediate depths. Abundant and seasonally variable groups
included, at 890m, Nitrospina, Flavobacteria and Marine Group A. Seasonality at 890m is likely
driven by variability in sinking particles, which originate in surface waters, pass transiently
through the middle water column, and accumulate on the seafloor where they alter the chemical
environment. Seasonal sub-euphotic groups are likely those whose ecology is strongly influenced
by these particles. This surface-to-bottom, decade-long, study identifies seasonality and
Cram
33
interannual variability not only of overall community structure, but also of numerous taxonomic
groups and near-species level operational taxonomic units.
Introduction:
Exploration of long-term diversity and temporal dynamics of marine microbial communities
living near the sea surface has revealed seasonality, connectivity to ecosystem parameters, and
co-occurrence patterns (Fuhrman et al. 2006; Steele et al. 2011; Gilbert et al. 2012). However,
relatively little is known about whether these characteristics are consistent with or influence sub-
surface microbial communities, especially those below the euphotic zone. The mesopelagic
region is of great biogeochemical interest, as it is the site of much of the Earth's carbon
remineralization (Aristegui et al. 2009; Herndl & Reinthaler 2013). The mesopelagic is also the
site of key parts of the nitrogen cycle (Zehr & Ward 2002) and sulfur cycle (Canfield et al.
2010), particularly in oxygen minimum zones (OMZs) that are found in association with coastal
upwelling (Wright et al. 2012). The San Pedro Ocean Time Series (SPOT) provides an
opportunity to study the entire water column, from surface to mesopelagic, in a coastal seasonal
upwelling system that includes an oxygen minimum at the bottom of the water column. SPOT is
located over the San Pedro Basin, in the San Pedro Channel, off the coast of Southern California.
The San Pedro Basin has well characterized topography and currents (Hickey 1991, 1992). Of
particular relevance to this study is the seasonally variable structure of subsurface currents
characterized by the surfacing of the California countercurrent during winter months (Hickey
1992) and a seasonally variable mixed layer shallower than 40m (Chow et al. 2013). SPOT's
surface shows repeating seasonal environmental and biological patterns, relationships between
microbial communities and biotic and abiotic parameters, and a prevalence of putative
interactions between microbes, all of which likely influence community structure (Fuhrman et al.
Cram
34
2006; Fuhrman & Steele 2008; Steele et al. 2011; Needham et al. 2013; Chow et al. 2013, 2014).
The physical and chemical environment at SPOT is regularly sampled and previous work has
characterized variability in particle flux (Collins et al. 2011), nutrient availability and
biogeochemistry (Hamersley et al. 2011) and turnover of water within the San Pedro Basin
(Berelson 1991). This physiochemical and surface-biological information provides a useful
background for understanding the temporal dynamics of bacteria throughout the water column.
Most surface microbial communities share common features, including, at a broad level,
dominance by Alphaproteobacteria, especially the SAR11 clade, and presence of other common
groups, such as Cyanobacteria, Actinobacteria and Gammaproteobacteria (especially the SAR86
clade) (Pommier et al. 2007; Fuhrman & Hagstrom 2008). Marine bacterial communities at the
ocean’s surface have been shown to vary globally (Rusch et al. 2007; Pommier et al. 2007), at
mesoscales (10-100km), ) across ocean fronts (Pinhassi et al. 2003; Hewson et al. 2006b), at
smaller scales between river plumes, bays and estuaries and their surrounding environments
(Casamayor et al. 2002; Crump et al. 2004; Fortunato et al. 2011; Yeo et al. 2013) and even
down to the micrometer scale (Long & Azam 2001); however, within a given water mass, at least
up to several km wide, communities are coherent (Hewson et al. 2006b).
Microbes share common community structure but vary around these common features
temporally, as they do on spatial scales. At the San Pedro Ocean Time-Series Station (SPOT),
pairs of samples taken days, weeks, months or years apart share many of the same near-species
level operational taxonomic units (OTUs), often at similar abundances (Needham et al. 2013;
Chow et al. 2013). In fact, most pairs of samples, even those taken at opposite times of year,
many years apart, have on average at least 36% similarity (Chow et al. 2013). Overlying this
stability, surface assemblages the San Pedro Channel vary seasonally, with some taxa more
Cram
35
abundant in surface waters in particular seasons. Seasonal patterns have also been observed in
the surface waters of the Bermuda Atlantic Time Series (BATS) (Steinberg et al. 2001; Morris et
al. 2005; Treusch et al. 2009; Giovannoni & Vergin 2012; Vergin et al. 2013b), the Plymouth
Marine Lab (PML) Western English Channel time series (Gilbert et al. 2012), as well as at other
locations (Kan et al. 2007; Rich et al. 2011). The Hawaii Ocean Time Series (HOT) is located in
a region that has less seasonal variation in surface conditions than other time series and appears
to have a less defined, but still detectable, seasonal patterning of its surface microbial community
(Eiler et al. 2011; Giovannoni & Vergin 2012). In the SPOT dataset, at time scales of one to four
years, samples that are taken more years apart generally have more dissimilar community
structure. Meanwhile samples collected more than four years apart have comparable
dissimilarity. Seasonal and longer-term changes in community composition were reduced in the
deep chlorophyll maximum (DCM) layer compared to surface waters (Chow et al. 2013).
Generally, seasonal patterns appear to be ubiquitous in surface waters, though they appear to
vary in magnitude between locations. Thus, seasonality at SPOT is representative of seasonality
elsewhere, making it a good environment to investigate the less explored topic of sub-surface
seasonality.
Below the euphotic zone microbial communities vary between depths (e.g. Garcia-Martinez &
Rodriguez-Valera 2000; DeLong et al. 2006) and vary at large scales between water masses
(Hamilton et al. 2008; Galand et al. 2009a, 2009b; Agogué et al. 2011). The microbial
communities of the mesopelagic ocean, (150m-1000m) are generally believed to be less
seasonally variable than those found at the surface. Microbial communities at BATS have been
shown to vary seasonally from the surface down to depths of 300m (the deepest depth studied).
These patterns are driven by the winter mixing of the top 200m-250m of water, and subsequent
Cram
36
summer stratification (Morris et al. 2005; Carlson et al. 2009; Treusch et al. 2009). In contrast,
the mesopelagic at HOT, like its surface waters, appears to have less distinct seasonal patterning
overall, although sub-surface layers within or just below the deep HOT euphotic zone, for
instance the 75m and 150m depths, show seasonal trends (Eiler et al. 2011). Because HOT shows
little subsurface seasonality and studies at BATS have not shown seasonal patterns in
communities more than 300m deep, (which is only 50m below the layer of deepest mixing)
(Treusch et al. 2009), neither time series has identified seasonality of microbial community
structure below the deepest seasonal mixing. Furthermore, neither of these projects has
investigated community variability near the sea floor (~4800m at each site). Over a shorter (~1
year) time series, microbial communities at 300m in the Northwest Mediterranean Sea vary in
abundance, activity and structure, and reportedly relate to water stratification, dissolved organic
carbon, and transparent exopolymeric particles (Weinbauer et al. 2013). Previous mid-water
work at SPOT demonstrated repeating patterns, particularly of nitrifying organisms including
Archaeal groups and Nitrospina OTUs (Beman et al. 2010). Community structure measured over
four years at SPOT at 500m has been shown to be quite stable relative to spatial data sets
spanning large regions of the ocean (Hewson et al. 2006a). Microbial communities in immediate
near shore regions like the Columbia River estuary (Fortunato et al. 2011) or Kaneohe Bay,
Hawaii (Yeo et al. 2013) with strong spatial environmental gradients, have been found to have
microbial community structure more affected by such gradients than by purely seasonal
influences. Contrastingly, temporal trends have been shown to prevail over spatial gradients off
of the coast of New Jersey (Nelson et al. 2008). SPOT, as a relatively deep coastal zone with
hypoxic bottom waters, is representative of a biogeochemically important marine ecosystem, and
Cram
37
extending observations to the dynamics of the whole water column’s bacterial community allows
more complete understanding of this system’s ecology.
Alpha-diversity or biodiversity in this study is defined as the number of near-species level taxa
detected above an abundance threshold, and/or the evenness of those taxa. The degree of
biodiversity’s environmental relevance is an ongoing subject of debate, though it is generally
agreed that in many systems more diverse environments are more stable and better able to
respond to environmental perturbation (McGrady-Steed et al. 1997; Hooper et al. 2005).
Biodiversity in the Western English Channel (Gilbert et al. 2012) and BATS (Vergin et al.
2013b) is reported to be highest in surface waters in the winter. Ladau et al. (2013) suggested
that biodiversity is highest in winter in most high latitude locations. While, at BATS,
biodiversity below 150m is consistently high (Vergin et al. 2013b), globally less is known about
biodiversity below the surface and it is with this aim that we quantify biodiversity throughout the
year at multiple depths. To investigate the natural history of the microbial community at SPOT,
between depths, years and seasons, at multiple taxonomic levels, we ask the following questions:
1. What is the degree of change of the (surface to 890m) microbial community at multiple
time-scales?
2. At what depths and seasons do the highest and lowest levels of biodiversity occur, in
terms of both richness and evenness?
3. Which environmental and biological parameters relate to community structure?
4. Which individual OTUs and taxonomic groups contribute to the seasonal and long term
variability of the microbial community?
Cram
38
Answering these questions will allow us a novel perspective of how microbes throughout the
water column change over time and how individual groups contribute to overall shifts in
community structure at these depths.
Methods:
Sampling
Samples were collected by Niskin bottle from five depths (5m, deep chlorophyll maximum
(DCM), 150m, 500m and 890m) at the San Pedro Ocean Time Series (SPOT) Microbial
Observatory site (33° 33’N, 118° 24’W) approximately monthly from August 2000 to January
2011. Sampling at 890m began in 2003 and samples from below 5m were not collected in 2007.
Environmental measurements
An in situ sensor (Sea-Bird Electronics) measured temperature and salinity. Oxygen was
measured using Winkler titrations from 2000-2006 and with an in situ oxygen electrode (Seabird,
model 13) over the entire data set. The electrode oxygen values were linearly related to Winkler
values (R
2
= 0.93). Winkler data were always used in our analyses when available for the
appropriate depths. If only electrode data were available, oxygen values were estimated from the
electrode values using a linear fit of electrode vs Winkler values.
Nitrate, nitrite and phosphate samples were stored at -20C and measured by standard
colorimetric techniques (Parsons 1984). Total bacterial and viral abundances were determined by
SYBR green epifluorescence microscopy (Noble & Fuhrman 1998; Fuhrman et al. 2006; Patel et
al. 2007). Bacterial heterotrophic productivity was measured in triplicate 10ml seawater samples
by [
3
H]leucine incorporation with a conversion factor of 1.5 * 10
-17
cells/mol of leucine
(Kirchman et al. 1985; Fuhrman et al. 2006; Chow et al. 2013).
Cram
39
We calculated values for cell turnover time (leucine), and excess phosphate concentration (P*) as
follows:
Cell Turnover Time (days) = Bacteria Concentration (cells/ml) / Productivity (cells/ml/day).
P* = [Phosphate] Concentration – [Nitrate + Nitrite] Concentration/16 (as in Deutsch et al.
2007).
Monthly and eight-day average estimates of surface chlorophyll a concentration and integrated
primary productivity, eight day average estimates of photosynthetically active radiation, colored
dissolved organic matter, and particulate organic carbon concentrations were downloaded as
satellite data from NASA. Meteorological data including daily air temperature, precipitation
(measured at Avalon airport, Santa Catalina Island, 16km away), wave period and height
(measured by a buoy in Santa Monica basin, 40km away), upwelling, and multivariate El Niño
Southern Oscillation index scores (MEI) (compiled by NOAA) were downloaded and
synthesized (supplement).
Bacterial community structure
For each depth, seawater (5m- 10L, deep chlorophyll maximum- 10L, 150m- 15L, 500m- 20L,
890m- 15L) was filtered, sequentially through a, ~1µm pore size, A/E filter, and a 0.2µm pore
size Durapore filter (both 142mm in diameter). Bacterial DNA was isolated as described
previously (Fuhrman et al. 1988; Chow et al. 2013) from the frozen (-80° C) 0.2 µm Durapore
filter by hot SDS lysis followed by phenol-chloroform purification (Fuhrman et al. 1988) and
stored in TE at -80°C. Automated Ribosomal Intergenic Spacer Analysis (ARISA) was
performed by amplifying 2ng of bacterial DNA and performing fragment analysis on 10ng of the
amplified product as initially described by Brown et al. (2005) and with modifications described
Cram
40
by Chow et al. (2013). As in both of these articles, technical replicates were generated by
running each sample in two lanes of the ARISA gel. Peaks were called in DAx (Chow et al.
2013) and ARISA fragments were normalized to relative abundances, calculated as proportions
of total abundance, and dynamically binned (Ruan et al. 2006). Fragments were classified, using
clone libraries, based on Greengenes (DeSantis et al. 2006), SILVA (Quast et al. 2012) and RDP
(Maidak et al. 2001) taxonomies as outlined previously, (Brown et al. 2005; Needham et al.
2013; Chow et al. 2013). In brief, we used existing sequenced clone libraries from SPOT and
elsewhere, as well as data sets that contained both 16s ribosomal subunit and intergenic spacer
(ITS) DNA sequences to map ARISA fragment sizes to taxonomic identity. Because we were
investigating communities throughout the water column, we modified earlier protocols assigning
identity to peaks by prioritizing clones from the depths where each ARISA fragment was most
abundant on average, rather than always prioritizing surface clones (supplement). In order to
establish maximum and minimum possible values of community similarity, we determined the
mean similarity of machine replicates (maximum) and the statistical difference between 1000
arbitrarily chosen pairs of samples in which the order of the OTUs had been randomized
(minimum).
Seasonality and inter-annual variability of environmental parameters
Each measured environmental parameter was tested for seasonal and interannual variability by
way of a generalized additive mixed effects model (GAMM) (Wood 2004, 2006) in which data
were fit with two splines, one seasonally cyclic and another that fit the overall dataset (see Wood
2006, 321–324; Ferguson et al. 2008). From this model we report estimated degrees of freedom
(EDF) and p-values for both spline functions and identify the month and year of highest and
lowest abundance in the data set. Seasonal and interannual parameters are defined as those
Cram
41
whose GAMM function’s seasonal or interannual component had a p-value of less than 0.05,
respectively. We compared this model to another model in which we attempted to fit the variable
using the same cyclic seasonal spline and a long term spline based on the multivariate El-Niño
Southern Oscillation index (MEI) rather than year (supplement).
Seasonality and inter-annual variability of overall community structure
Seasonal and interannual patterns were investigated by determining whether samples taken in
similar times of year (seasonal) or similar years (interannual) had similar microbial community
structure, using both visual and statistical approaches. Visually, we compared temporal distance
between pairs of samples (lag) to the Bray-Curtis similarity of those samples. For each month of
possible lag, we calculated mean and 95% confidence intervals of Bray-Curtis similarities
(supplement). More quantitatively, we compared the Bray-Curtis similarity of each pair of
samples to whether the pair of samples was taken in similar or different times of year (seasonal)
and whether the samples were taken in similar or different dates (interannual). The relationship
between these temporal and community similarity were evaluated statistically using a Mantel test
(supplement). Depths with seasonally or interannually variable communities were defined as
those with corresponding Mantel p-values less than 0.05.
Alpha diversity
For each time point and depth, Shannon index and inverse Simpson index (ISI) were calculated
using the “diversity” function in the “vegan” R-package (Oksanen et al. 2013). Sample richness
was estimated by quantifying, at each depth, the number of OTUs with a relative abundance of
more than 0.01%, 0.1%, and 1% of the total community. For subsequent analysis we used 0.1%
as our richness cutoff because it was well above our ARISA detection threshold of 0.01%, giving
us good confidence that the OTUs were indeed present in our data set. Pielou's evenness was
Cram
42
calculated as “H/ln(S)”, where H is the Shannon Index, S was the number of OTUs found with
greater than 0.1% relative abundance. For each depth, we calculated mean values and 95%
confidence intervals of each of these metrics. Analysis of variance was applied to determine
whether there was statistically significant difference in richness and ISI between depths
(supplement). We also determined whether richness and ISI related to season, long term trends,
the degree of change of community structure, or the similarity of community structure between
depths (supplement).
Environmental parameters and community structure
Partial Mantel tests were applied to determine which environmental parameters relate to
community structure at each depth. To identify which environmental variables could predict
community structure beyond seasonal effects, we statistically removed the effects of seasonality
and long term variability from this environmental variability analysis (supplement).
Temporal dynamics of microbial taxa over time
Relative abundances for each taxon (at multiple taxonomic levels) were determined in each
sample by summing the relative abundance of all ARISA OTUs within that group. Generalized
additive mixed effects models (GAMMs), were used to model the abundance of taxonomic
groups’ relative abundances using cyclic seasonal splines and interannually variable splines. This
GAMM approach was similar in structure to that applied to environmental variables (see above).
These GAMMs determined which taxonomic groups, and which of the 100 most abundant OTUs
at each depth showed seasonality and long term variability. They also identified the months and
years in which these groups were most and least abundant (supplement).
Cram
43
Results:
Variability of environmental and biological parameters, between seasons, years and depths.
Temperature and salinity varied seasonally at several depths throughout the water column, but
showed greater variability in surface waters than deeper water (Figure 2-1, Supplemental Figure
2-1, Supplemental Table 2-1). Satellite measurements of surface chlorophyll a, primary
productivity, particulate organic carbon concentration, colored dissolved organic matter
concentration, sea surface height ,were all seasonally variable with the highest concentrations or
rates in the spring. Mixed layer depth, upwelling, wave height, average wave period, dominant
wave period, precipitation, daily maximum and minimum air temperatures, and wind speeds all
varied seasonally.
Nitrate concentration was consistently high in deep waters (25-50uM at 150m, 500m, 890m), and
was seasonally variable in the deep chlorophyll maximum and 150m where highest
concentrations occurred in May and June, respectively. Nitrate was low in surface waters (<1µM
at 5m). Nitrite concentration was low and variable at all depths yet statistically significantly
seasonal in only at 890m (Supplemental Figure 2-1, Supplemental Table 2-1). Phosphate, like
nitrate, was low at 5m and the DCM (varied between undetectable and 1µM), was generally
around 3uM at 150m and ranged from 3-5uM at 500m and 890m. Phosphate showed seasonal
variability at the DCM and 150m where it followed the concentration of nitrate and at 500m
where it was most abundant in February. P*, the excess phosphate over nitrate concentration
relative to predicted Redfield ratios, was generally higher at deeper depths (ranging from 0uM at
5m to 2µM at 890m) and showed seasonal variability at 5m where it was highest in January.
Oxygen concentrations ranged from saturated (>200µM) in surface waters to strongly hypoxic at
Cram
44
the bottom of the water column (<10µM) and were seasonally variable at 5m and 150m, but not
so at other depths.
Bacterial abundances were about three fold higher in the euphotic zone (5m and DCM, ~1 x 10
6
cells/ml), where they varied seasonally, than in deeper waters (150m, 500m, 890m, ~3 x 10
5
cells/ml) where they did not show seasonal variability. Bacterial growth rates, as measured by
leucine uptake, varied seasonally at all depths except 500m. Estimated average doubling times of
bacteria were around five days in the euphotic zone, considerably faster than the estimated
average doubling time of around 100 days in the deeper depths. At all depths, many
environmental parameters showed long term variability over the course of the data set at most
depths, showing increasing, decreasing or nonlinear trends (Supplemental Table 2-1). There were
very few cases in which a secondary model, such as using MEI rather than year to predict long
term variability, improved the model and produced a statistically meaningful result. This result
suggests the El Niño Southern Oscillation had minimal effect on environmental variables in our
dataset, likely because there were no strong El-Niño years during the time these data were
collected.
Cram
45
Figure 2-1 Median values of environmental parameters at each depth for each month across the ten year
SPOT data set. Symbols represent depth at which the parameter was measured. Confidence bars are plus
or minus one median adjusted deviation.
Bacterial community data overview
332 distinct OTUs were detected over the entire dataset, where OTU were considered only if
their peak in the electropherogram constituted at least 0.01% of the total ARISA peak area in at
least one month. We assigned taxonomic identities to 131 OTUs. The most abundant OTUs tend
to be the most easily identified via our clone libraries and sequence databases, and we assigned
identity to OTUs that cumulatively comprised ~90% of the bacterial community, (5m, 93%;
DCM, 91%; 150m, 86%; 500m, 90%; 890m, 85%).
Seasonal variability in microbial community structure
At all depths, samples taken one month apart were between 50% and 60% similar. In contrast,
machine replicates (same sample) averaged 78% similar, while simulated random unrelated
Cram
46
samples averaged 6.5% similar. As has been reported previously at the surface (Hatosy et al.
2013; Chow et al. 2013), when temporal difference between samples was plotted against Bray-
Curtis dissimilarity, we observed a sinusoidal pattern with maximum Bray-Curtis similarities at
multiples of one year, suggesting a seasonal repetition. We observed this sinusoidal pattern at
890m but not at the intermediate depths (Figure 2-2). In contrast, at the deep chlorophyll
maximum and 150m, samples taken six months apart were not statistically dissimilar from
samples taken one month apart (Supplemental Table 2-1). A Mantel test comparing the Bray-
Curtis dissimilarities to seasonal difference confirmed these seasonal patterns at the surface and
890m, and suggested weaker but still statistically significant seasonality at the deep chlorophyll
maximum. Meanwhile, the Mantel test did not suggest seasonality at 150m or 500m
(Supplemental Table 2-1).
Inter-annual variability in microbial community structure
In surface waters, communities sampled less than a year apart averaged 50% Bray-Curtis
similarity while samples taken one to two years apart had only 45% similarity, a statistically
significant decrease (ANOVA F=12.84, DF = 11, P < 10
-15
; Tukey corrected t-test P < 10
-6
).
Samples taken farther apart in time appeared to be less similar up through four years apart, as
was found previously (Chow et al. 2013). All pairs of samples four or more years apart tended to
be ~37% similar on average, while oscillating seasonally. In contrast, at the DCM, 150m and
500m depths, communities one year apart were roughly as similar as communities sampled two
and six years apart. Communities sampled more than seven years apart were less similar than
those taken fewer than seven years apart at these intermediate depths (ANOVA F = 6.11DCM,
4.59 150m, 14.37 500m, DF = 11 ,P < 10
-6
; Tukey corrected t-test P < 0.05). At 890m there was
a weaker but still statistically detectable decrease in similarity between samples taken one year
Cram
47
apart and samples taken four or five years apart (ANOVA F = 2.95, DF = 8, P < 0.01, Tukey
corrected t-test P <0.05) (Figure 2-2). At all depths except 890m, Mantel tests suggested that
there was a statistically significant relationship between the temporal distance of samples and
their Bray-Curtis distance, further indicating the presence of a long term trend (Table 2-1).
Cram
48
Figure 2-2 Mean Bray-Curtis similarities of all pairs of samples (y-axis) separated by different intervals
of time (x-axis). Circles represent the mean similarity of all pairs of samples taken a given number of
months apart (Intermonthly; left y-axis). Thus, the first circle is the mean Bray-Curtis similarity of all
Cram
49
pairs of samples taken 15-45 days (~1 month) apart, the second circle is the mean similarity of all pairs of
samples taken 46-76 days (~2 months), the twelfth circle (aligned with the one year lag tick mark)
represents samples taken twelve months apart and so on. Squares represent mean similarity of samples
taken a given number of years apart (Interannual; right y-axis). Accordingly, the first square represents the
mean Bray-Curtis similarity of all pairs of samples taken less than one year apart, the second is all pairs of
samples taken between one and two years apart and so on. Error bars, for both types of data points,
represent 95% confidence intervals of the mean similarity. Points with non-overlapping error bars suggest
statistically significant differences in mean similarity between samples taken different distances apart in
time. For instance, samples taken six months apart in the surface (the sixth circular data point) are
statistically less similar than samples taken one month apart (first circular point), while samples taken
twelve months apart (twelfth circular point) are not less similar than samples taken one month apart.
ΔSim
1month
-6mon
P
Seasonal
Mantel
P
% Seas
OTUs
(R<0.1)
% Seas
OTUs
(R<0.2)
Temporal
Mantel
P
% Year
OTUs
(R<0.1)
%# Year
OTUs
(R<0.2)
5m 9.8% <0.001 0.123±.020 <0.001 37 13 0.139±0.059 <0.001 33 12
DCM 5.3% 0.058 0.038±0.031 0.026 23 7 0.113±0.070 0.007 26 10
150m 4.1% 0.129 0.020±0.036 0.156 9 3 0.100±.008 0.019 18 5
500m 6.9% 0.028 -0.006±.003 0.584 7 4 0.150±0.078 .001 26 11
890m 9.9% <.01 0.079±.038 0.002 20 9 0.054±0.075 .115 27 10
X
2
37.6 9.7 9 3.4
DF 4 4 4 4
P <.001 0.046 0.062 0.5
Table 2-1: Summary of metrics of community seasonality at each depth. ΔSim_1mon-6mon is the decrease
in mean similarity between all pairs of samples taken one month apart and all pairs of samples taken six
months apart. P-value is for an associated independent t-test investigating whether the groups of pairs
have statistically significantly different means. Seasonal Mantel reports the R value ± 95% confidence
intervals of the Mantel test correlating seasonal distance (distance apart in time of year of two samples)
against the Bray-Curtis distance of the community structure of all pairs of samples. Temporal Mantel
reports the r value ± 95% confidence intervals of the Mantel test correlating the number of days apart
pairs of samples were collected against the Bray-Curtis distance of the community structure of all pairs of
samples. The corresponding p-values are from permutation tests of the Mantel statistics. %SeasOTUs are
the number, out of the top 100 most abundant OTUs that are shown to be predictable from a
nonparametric regression model with R values above a threshold (0.1 or 0.2 depending on column) and
whose seasonal spline functions differed statistically significantly from the null model
(P<0.05). %YearOTUs are the number, out of the top 100 most abundant OTUs that are shown to be
predictable from a nonparametric regression model with R values above a threshold (0.1 or 0.2 depending
on column) and whose spline functions following year to year variability differed statistically
significantly from the null model (P<0.05). X
2
, DF and P rows specify the chi-squared statistic, degrees of
freedom and p-value of the test of whether the differences in numbers of seasonal or long term variable
taxa are more different between depths than would be predicted by chance alone.
Cram
50
Alpha diversity relates to season and certain community structure patterns
Both average richness (number of OTUs with > 0.1% relative abundance) and ISI varied among
depths (Richness: ANOVA F=3.81, DF = 4, P<0.01; ISI: ANOVA F = 37.37, DF = 4, P<0.001)
(Figure 2-3A). The 150m depth had the highest mean richness and ISI, which was statistically
significantly higher than the richness at 5m only (Tukey corrected P < 0.01) and ISI at all depths
except 890m (Tukey P < 0.01) (Figure 2-3A). At 5m and the DCM, ISI (but not richness) was
highest in the winter, while at 890m both richness and ISI values were highest in the spring
(Figure 2-1, Supplemental Figure 2-1, Supplemental Table 2-1). In the deep chlorophyll
maximum layer, richness was highest in winter while ISI did not vary seasonally. At 150m and
500m neither richness nor ISI varied seasonally.
We observed a positive correlation at all depths between either richness and ISI and the
similarity of that sample to communities at most other depths. As an example, in months when
community structure at150m and 500m were similar, the 150m community had a high ISI
(Figure 2-3B). This finding is also true of most other pairs of depths (Figure 2-3C, Supplemental
Figure 2-2). We further observed that, at 150m and 500m, months in which community structure
had changed little in the past month also had higher richness and ISI than months that had
undergone larger changes in the past month (Figure 2-3C, Supplemental Figure 2-2).
Cram
51
Figure 2-3 A. Mean biodiversity index scores by depth. X axis labels are the biodiversity metrics under
evaluation. Richness are mean number of species in a given sample with greater than 1% 0.1% and
0.01%. Inverse Simpson (ISI) and Shannon H are biodiversity indexes and Pelou J measures evenness.. Y
axes are the values for each metric (OTUs for richness and inverse Simpson’s index, unitless for Shannon
Biodiversity and Peilou’s evenness. Symbols represent depths and are described in the legend. Bars
represent 95% confidence intervals from the mean.
B. Similarity between communities at 150m and 500m (x axis) compared to the ISI of the community at
150m (y-axis). Each point represents the relationship between ISI and similarity between depths for a
single month. The solid line is a trend line and the dashed lines represent 95% confidence intervals of this
trend. The R value of this correlation is 0.49 with associated confidence intervals from of 0.30 to 0.64.
This R value and its confidence interval correspond to the darkened symbol and bar in figure 3c. This
figure is provided as an example of correlations seen between depths and ISI shown in Figure C. Other
pairs of depths, shown in Figure 3c, show similar trends.
C. R-values of correlations (y-axis) between inverse Simpson index (ISI) values for each depth given as a
symbol, and its similarity to other depths or to the previous month (x-axis). Symbol shapes signify depth
at which the biodiversity was measured. The first five parameters signify the Bray-Curtis similarity
between that the depth measured (indicated by symbol) and each other depth (vs5m, vsDCM, vs150m,
Cram
52
vs500m, vs890m). Higher R values represent a stronger correlation between the biodiversity at the
measured depth and its similarity to the depth given by the x label. For instance, the y-value of the
darkened diamond shape is 0.32, corresponding to slope of the relationship between biodiversity at 150m
and inter-depth similarity between 150m and 500m depicted in Figure 3B.The “vsLastMonth” column
represents the relationship between ISI and the Bray-Curtis similarity between that sample and a sample
taken at that same depth collected in the previous month/ Bars represent 95% confidence intervals for the
R value. Confidence intervals not overlapping the x axis indicate statistically significant correlations.
Environmental parameters relate to community structure
After seasonal and long term variability were factored out, similar surface communities generally
were characterized by similar monthly mean chlorophyll a concentrations, monthly mean
primary productivity, particulate organic carbon concentration, virus abundance and leucine
incorporation rates (P < 0.01) (Table 2-2). At the DCM, 150m and 500m, no environmental
parameter related significantly to community structure with P < 0.01. At 890m, the depth of the
mixed layer and the amount of precipitation on the day of sampling related to community
structure (P < 0.01). For these observations in which P < 0.01, the false discovery rate, or
fraction of patterns likely to be false positives, (Q-value) was 3% in the surface and 9% at 890m,
providing us with reasonable confidence in the result, despite the multiple comparisons.
Conversely, use of P < 0.05 had higher false discovery rates (variable between depths but below
25% in all depths except 500m) suggesting a proportion (though still a less than a quarter) of the
weaker correlations are due to random chance. Notable statistically weaker (P < 0.05) factors
relating to community structure included the abundance of bacteria and the dominant (but not
average) wave period at 5m, the DCM and 150m; surface chlorophyll a and surface primary
productivity as measured by satellites to the DCM community; and the mixed layer depth to the
150m community. The false discovery rate of 65% at 500m suggests that the one positive
relationship observed at that depth is probably due to random chance and that no measured
environmental parameters explained community variability at 500m.
Cram
53
5 CMAX 150 500 890
Physical/
Chemical
Temperature 0.100 * 0.109 0.066 0.019 -0.082
Salinity -0.094 -0.045 -0.102 0.201 * -0.108
NO2 0.08 0.022 -0.036 -0.075 -0.033
NO3 0.034 0.034 -0.033 0.054 0.093
PO4 0.016 0.02 -0.066 -0.127 -0.107
P* 0.0161 0.0099 -0.0657 -0.1688 -0.079
O2 0.115 * 0.068 0.039 0.059 -0.049
Satellite Chl_A 0.203 ** 0.112 * -0.058 -0.058 -0.185
Prim_Prod 0.175 ** 0.151 * 0.028 -0.029 -0.11
Chl_A8 0.086 0.032 -0.079 -0.059 -0.104
Prim_Prod8 0.066 0.091 -0.033 -0.069 -0.031
POC 0.1682 ** 0.0719 -0.0064 -0.0276 -0.0523
CDOM 0.052 0.041 0.098 0.028 0.037
PAR -0.1028 -0.1498 0.0323 -0.0071 0.0274
SSHD_Sat 0.06 0.095 0.073 -0.042 -0.015
Surface MLD 0.024 -0.02 0.115 * 0.061 0.249 **
Cmax_Depth 0.042 0.035 0.033 0.12 -0.03
Svd -0.0549 -0.0069 -0.0674 0.0452 0.0039
Upwelling 0.0022 0.0056 0.0665 -0.0849 -0.0292
WaveHeight -0.012 0.027 -0.079 -0.075 -0.016
AvgWavePd -0.072 -0.074 0.04 0.063 0.08
DomWavePd 0.087 * 0.120 * 0.101 * 0.019 0.036
PRCP 0.019 0.173 * 0.117 0.118 0.319 **
TempMax 0.032 0.03 0.06 0.074 0.088
TempMin 0.018 0.058 0.073 0.048 0.029
AvgWind 0.025 -0.076 -0.042 -0.065 -0.072
WindGust 0.0033 -0.0928 -0.011 0.0822 -0.0601
Global MEI 0.0011 -0.0641 -0.063 -0.0704 0.0903
Biotic Bact 0.16245 * 0.13242 * 0.22517 * -0.00091 -0.05753
Vir 0.162 ** 0.082 0.166 * 0.024 -0.04
VBR 0.015 -0.033 0.08 -0.077 -0.03
Leu 0.250 *** 0.088 -0.083 0.139 0.131
TurnoverLeu 0.0636 0.1099 -0.0051 0.0365 0.0896
False
Discovery
Q: P<0.01 3.45% NA NA NA 9.1%
Q: P<0.05 9.6% 24.8% 21.4% 65.0% 9.1%
Table 2-2 Rho values of Mantel tests relating community similarity to environmental and biotic
parameters, after factoring out the effects of seasonality and depth. Higher values indicate that samples
with similar values for a given parameter generally have similar community structures (as measured by
bray Curtis similarity), and that samples with different values for that parameter have different
community structures. Surface Satellite and Global values are measured for the entire water column,
Chemical and Biotic values are measured at each depth. Stars indicate p-value * = 0.05, ** = 0.01.
Abbreviations: NO2, nitrite; NO3, nitrate, PO4, phosphate, P* excess phosphate; O2, oxygen
concentration; Chl_A_Sat, chlorophyll A, monthly average; Chl_A_Sat8, ibid eight day average;
Cram
54
Prim_Prod, primary productivity estimate, monthly average; Prim_Prod8, ibid eight day average; POC,
particulate organic carbon; CDOM, colored dissolved organic matter to chlorophyll ratio; SSHD, sea
surface height differential; PAR, photosynthetically active radiation; MLD, mixed layer depth;
Cmax_Depth, depth of the chlorophyll maximum; Svd, Sverdrup transport;AvgWavePd and
DomWavePd, average and dominant wave periods; PRCP, precipitation; TempMax and TempMin,
maximum and minimum daily air temperature; AvgWind, average wind speed; WindGust, maximum two
minute wind gust speed; MEI, Multivariate El Niño Southern Oscellation Index; Bact, bacterial
abundance; Vir, virus abundance; VBR virus to bacterial abundance ratio; Leu, growth rate as measured
by leucine incorporation; TurnoverLeu, cell turnover time as measured by leucine incorporation and
bacterial abundance data.
Dynamics of dominant marine bacteria clades
The relative abundances of several taxonomic groups were examined both at phylum to class and
at family levels at each depth (Figure 2-4, Table 2-3, Supplemental Figure 2-3). Furthermore, the
subset of taxa that showed seasonal variability differed between depths.
Alphaproteobacteria were the most abundant class at every depth, and varied seasonally only at
890m. The Alphaproteobacteria were dominated by the SAR11 clade, which was present
throughout the water column and seasonal in surface waters with highest abundance in summer.
The AEGEAN-169 clade, which is closely related to SAR11 (Alonso-Sáez et al. 2007), was most
abundant in deeper depths where SAR11 was less abundant (Figure 2-4B, Supplemental Figure
2-3B, Table 2-3).
The Surface 1 sub-clade was the dominant sub-group of SAR11 at all depths (Supplemental
Figure 2-5A), with highest abundance in surface waters. It was seasonal only at 890m, where it is
least abundant (Supplemental Table 2-4). Meanwhile the less abundant SAR11 subclades
Surface 2 and Surface 4 were seasonally variable at several depths. SAR11 Deep 1 was abundant
in 150m and 500m waters, but never seasonally variable. It was seasonally variable only at 890m
(Supplemental Table 2-4). SAR11 Surface 1 was primarily comprised of three OTUs:
OTU_666.4, OTU_686.9 and OTU_670.5 (in order of abundance and where the number
Cram
55
indicates ARISA fragment length) (Supplemental Figure 2-4B, Supplemental Figure 2-5B) and
several other less abundant OTUs. Each Surface 1 OTUs seemed to exhibit different seasonal
and interannual patterns from each other.
Gammaproteobacteria were also abundant at all depths (Figure 2-4A, Table 2-3). They were
dominated by SAR86 in the surface, and other OTUs such as SUP05 in deeper waters (Figure
2-4B, Table 2-3). Actinobacteria, dominated by the OCS155 clade, Cyanobacteria, which were
predominantly Prochorococcus, and Chloroplasts, from eukaryotic picophytoplankton, were all
more abundant in surface waters than deep waters (Figure 2-4). Cyanobacteria, most abundant in
late fall, and Chloroplasts, abundant in late winter and early spring, were both seasonal in the
surface waters (Supplemental Table 2-2).
More abundant in deeper waters were Deltaproteobacteria, Flavobacteria and the Marine Group
A phylum (MGA) (Figure 2-4A, Table 2-3), all of which were seasonally variable at 890m with
relative abundance maxima in the month of March. Within these classes, bacteria from the NS9
(Flavobacteria), Nitrospina (Deltaproteobacteria) and Arctic 96B-7 (MGA) groups appeared to
drive the seasonal variability of the broader taxonomic groups (Figure 2-4, Table 2-3).
Arctic96B-7 also showed seasonal variability at 150m and 500m, while SAR324 appeared to
drive seasonality of Deltaproteobacteria at 500m. Investigation of genus level groups of Marine
Group A and clades Flavobacteria showed diverse seasonal and interannual patterns
(Supplemental Table 2-4).
In addition to seasonal variability, many taxa showed long term increases, decreases or non-
linear changes in abundance.
Cram
56
Figure 2-4 Heat map of relative abundances of bacteria from class and phylum level taxonomic groups
(A) as well as family and order level groups (B). Each panel represents a different taxonomic grouping.
The x axis indicates months and y axis indicates sampled depths (not to scale). Colors correspond to the
summed relative abundance of all OTUs identified as falling within that taxonomic group, averaged by
month. Relative abundances scores for each color are given in the scale bar at right; note that abundances
are on a log scale.
Cram
57
Table 2-3 The abundance and temporal characteristics of taxonomic groups seen in Figure 2-4. Bold cells
have relative abundance greater than 5%, while italicized cells have relative abundance greater than 2%.
Abundance scores are given as a medians ± one median adjusted deviation. Month abbreviations (eg Dec)
correspond to the month of the year that a taxon has greatest relative abundance according as fit by a
Taxon Resol. 5m DCM 150m 500m 890m
Actinobacteria Phylum
17.8±(10.5)%
19.5±(10.4)%
3.3±(1.9)%
1.7±(0.9)%
0.9±(0.7)%
>OCS155 Clade
17.4±(10.7)%
19.4+(10.6)%
2.4±(1.6)%
0.3±(.3)%
0.3±(0.3)
Flavobacteria Class
4.7±(3.6)%
Feb
7.5±(4.3)%
10.4±(3.1)%
8.8±(2.7)%
13.9±(3)%
Aug
>NS9 Clade
0.6±(0.7)%
0.7±(0.8)%
4.2±(2.2)%
D
3.8±(1.6)%
N
3.5±(1.3)%
Aug
Cyanobacteria Phylum
2.8±(3.1)%
Nov:I
1.4±(1.4)%
1.1±(0.6)%
I
1.1±(0.7)%
I
0.8±(0.5)%
>Prochloro. Genus
1.6±(2.0)
Nov:I
0.8±(1.0)%
Nov:N
1.0±(0.5)%
1.0±0.6
0.7±0.5
Chloroplast Organelle
3.8±(4.4)%
Feb:I
3.4±(4.1)%
Feb
0.8±(0.7)%
Apr
1.2±(0.6)%
0.3±(0.4)%
Jan
α-proteobacteria Class 38.7±(7.2)%
35.8±(6.4)%
42.2±(5.4)%
43.4±(6.7)%
25.2±(9.5)%
Jan
>SAR11 SilvaTag
27.5±(7.4)%
Jun:N
24.0±(7.3)%
N
24.2±(6.1)%
20.3±(6.1)%
13.9±(5.2)%
>>Surface_1 Ecotype 18.8±(7.4)%
18.5±(5.8)%
12.5±(5.6)%
7.1±(4.5)%
D
5.8±(2.7)%
Feb
>AEGEAN-169 Clade 7.5± (2.9)%
6.5±(2.3)%
12.6±(5)%
N
17.8±(6.1)%
7.9±(3.3)%
γ-proteobacteria Class
9.4±(3)%
Aug
9.0±(3)%
8.5±(2.7)%
May:N
7.5±(2.6)%
N
8.4±(2.9)%
>SAR86 SilvaTag 6.9±(2.6)%
6.4±(3.3)%
1.6±(1.1)%
0.5±(0.5)%
0.7±(0.7)%
>SUP05 Clade 0.03±(0.04)%
0.1±(0.2)%
2.0±(1.4)%
1.5±(0.9)
Jun:N
1.7±(1.6)
δ-proteobacteria Class 2.4±(1.8)%
3.9±(3.4)%
8.1±(2.4)%
9.9±(3.8)%
Dec:D
22.3±(9.1)%
Aug
>Nitrospina Genus
1.3±(1)%
Oct:D
1.4±(1.3)%
4.9±(1.7)%
5.2±(2.3)%
14.4±(8)%
Aug:N
>SAR324 Clade
0.3±(0.4)%
Mar:D
0.9±(1.1)%
Apr
3.2±(1.6)%
D
4.2±(2.3)%
Dec:D
7.5±(3.1)%
Marine Group A Phylum
2.1±(1.4)%
2.7±(1.7)%
Nov:D
8.2±(3)%
Oct:D
6.3±(2.4)%
12.2±(3.5)%
Aug
>Arctic96B-7 Order
0.9±(0.8)%
Nov:D
1.9±(1.6)%
5.9±(3)%
Oct:D
2.4±(1.1)%
Jan:D
8.4±(3.6)%
Aug
>ZA3648c Order
1.1±(0.9)%
0.9±(0.8)%
2.2±(0.8)%
3.6±(2.1)%
3.1±(2)%
Cram
58
cyclic spline function in a general additive model. A letter code (D - Decreasing: I - Increasing, N - Non-
linear) identifies interannual trends in the data. Codes are only provided for splines that that differ
statistically significantly from a flat line (P>0.05) and are part of an overall general additive model with
overall R^2> 0.10. Further statistics for the general additive models are shown in Supplemental Table 2-2,
Supplemental Table 2-3, Supplemental Table 2-4.
Dynamics of abundant marine bacterial operational taxonomic units
Nonparametric regression models, using seasonal and long term splines to fit the data, showed
that for the 100 most abundant OTUs at each depth, 5m, the DCM and 890m had the highest
fraction of seasonal OTUs, while 150m and 500m had fewer seasonal OTUs (Table 2-1,
Supplemental Table 2-4). The abundances of the five most abundant bacteria at each depth
(Supplemental Figure 2-5) reflect the patterns seen for the 100 most abundant OTUs, with
seasonal bacteria present at the surface and bottom of the water column but not the middle water
column. The five most abundant bacteria at each depth show differences in their distribution
patterns across depths, with some abundant at multiple depths (e.g. most abundant AEGEAN-
169 OTUs), and others only abundant at one depth (e.g. SAR406 with ITS length of 709.4). Chi-
squared analysis confirmed that the depths did have statistically significant differences in the
number of seasonal bacteria, and not in the number of bacteria that showed long term variability.
The subset of the 100 most abundant bacteria that were determined to be seasonal or
interannually variable included OTUs from many abundant phylogenetic groups, including
groups that taken as a whole did not vary seasonally or interannually (Table S5).
Cram
59
Discussion
Seasonality is strongest at the surface and bottom of the water column
Seasonal patterns appeared to be strongest at the surface (5m), weak in the middle water column
(DCM, 150m and 500m), and strong again just above the sea floor (890m) according to a number
of metrics:
1. Community Structure: Samples taken in multiples of six months apart from each other
tended to be most different at 5m and 890m (Figure 2-2).
2. Biodiversity: Richness and inverse Simpson index were highest in the winter at 5m and in
the spring at 890m (Figure 2-3).
3. Taxonomic Groups: Several broad taxonomic groups of organisms were seasonal,
especially at the surface (such as SAR11 and Prochlorococcus) and at 890m (such as
Nitrospina and NS9 Flavobacteria) (Figure 2-4, Table 2-3, Supplemental Table 2-2,
Supplemental Table 2-3).
4. Individual OTUs: There were more seasonal individual OTUs at the surface, DCM and
890m than at 150m or 500m (Table 2-1, Supplemental Table 2-4).
Seasonality of surface communities is consistent with previous findings, both at this site
(Fuhrman et al. 2006; Chow et al. 2013) and elsewhere (Morris et al. 2005; Carlson et al. 2009;
Treusch et al. 2009; Gilbert et al. 2012; Vergin et al. 2013b), and is likely governed by strong
seasonality in many environmental factors. Photosynthetically available radiation, day length,
nutrient concentrations, seawater temperature and mixing of the water column all vary seasonally
in the surface (Figure 2-1, Supplemental Figure 2-1, Supplemental Table 2-1) and many or all of
these factors likely influence the bacterial community. However, covariance of these variables
precludes identifying their individual relative contributions to seasonality. Primary productivity
and concentration of chlorophyll a are also seasonally variable, and appear to relate to
Cram
60
community structure even beyond this seasonal effect, i.e. after seasonality is factored out (Table
2-2).
At intermediate depths, the decrease in seasonality below the photic zone is consistent with
findings at HOT and BATS (as reviewed in Giovannoni & Vergin 2012). Light levels,
temperature, and turbulence are less seasonally variable at SPOT’s deeper depths. At 500m,
temperature is stable and light intensity is insufficient for photosynthesis. Unlike BATS, the
mixed layer at SPOT is almost always above 40m, so the 150m, 500m or 890m depths likely
never mix with surface water directly (even though mixed layer depth does appear to relate to
community structure at 890m (Table 2-2).
Seasonality of the deepest depths, has not been seen previously, even in a prior analysis of a
shorter subset of the SPOT 890m data (Hatosy et al. 2013), either because the additional samples
gave this study more statistical power, or else because the statistical tests employed in this study
are more sensitive. Seasonality in surface waters through sinking particles, or perhaps vertically
migrating meso-to-mega fauna, contribute to the seasonality near the seafloor. Particle flux,
which varies in magnitude and composition (Collins et al. 2011) over the course of the year
likely contributes to the observed seasonality at 890m, which is at the bottom of the water
column. Particles are believed to be the source of key substrates for bacteria in the deep ocean
and likely transport organic nutrients from the surface to the bottom, where they may become
entrained in the nepheloid layer or land on the sea floor, degrade, and affect nearby overlying
waters (Francois et al. 2002; Herndl & Reinthaler 2013). Therefore, the longer residence time of
particles and their associated organic matter at or near the sea floor, compared to midwaters,
likely leads to stronger seasonality there. Bear in mind that our study assessed the free-living
bacterial communities and not those attached to larger particles at the time of sampling.
Cram
61
Furthermore, many particles are heavily colonized by bacteria (Simon et al. 2002) and may
transport bacteria to bottom waters where they may be released (Sohrin et al. 2011). The
presence of cyanobacteria (presumably photosynthetic) at 890m reflect similar findings in other
systems (Sohrin et al. 2011) and further suggests bacterial transport. Particles have been shown
to sink at a rate averaging about 83m/day (Collins et al. 2011), which suggests that it should take
about 11 days to travel from the surface to the bottom of the water column. Differences in sizes
and densities between particles cause different compounds or organisms to be transported from
the surface to the deep at different rates. Currents in the region have been shown to vary with
both depth and season, and are generally fastest in surface waters decreasing with depth, and
generally move from south to north parallel to the coast (Hickey 1992). These currents likely
cause spatial separation between the locations where particles are generated and where they
settle. Particle flux variability across space (Buesseler et al. 2009) may interact with seasonal
currents, resulting in seasonally variable particle delivery to the sea floor at SPOT.
In addition to sinking particles, crustaceans and other zooplankton likely influence the sub-
surface communities by consuming organisms and marine aggregates at the surface and egesting
rapidly sinking fecal pellets, in conjunction with diel vertical migration (Steinberg et al. 2000,
2002; Wilson & Steinberg 2010). Other organisms that may transport nutrients or bacteria
include cnidarians, which traverse the deep water column at SPOT (Schnetzer et al. 2011), giant
larvaceans which discard rapidly sinking, mucus houses (Robison et al. 2005), and smaller
larvaceans whose houses become part of the marine snow flux (Hansen, et al. 1996). These
pelagic organisms together link the entire water column such that features of any one depth may
affect other depths. It is important to note that, at all depths, despite seasonal variability, there is
Cram
62
stability as well. Most taxa that are more abundant in some months than others are still present,
but at lower abundance, in other months.
Ocean currents are seasonally variable, but show the most variability in surface waters (Hickey
1992), while water below the sill of the San Pedro basin at 750m is primarily trapped (Berelson
1991), suggesting that currents alone seem like an unlikely explanation for the greater
seasonality at 890m than at intermediate depths.
Inter-annual variability is apparent throughout the water column
While communities at all depths varied on time scales of years or longer, the scale of interannual
variability differed between depths. Surface samples taken one year apart tended to be most
similar, samples taken two years apart were less similar, and so forth on up through samples
taken four years apart, after which average similarity remained constant (as seen by Chow et al.
2013). This pattern may reflect seasonal trends and environmental factors that vary from year to
year around an average community. In contrast to the surface, at the DCM, 150m and 500m
depths, samples taken fewer than eight years apart tended to have about the same level of
dissimilarity, while samples taken eight or more years apart tended to be more dissimilar (Figure
2-2). At these intermediate depths, it is probable that some factor that we have not analyzed
varies on longer than annual time scales in the mid water column. At all depths, long term
processes may be related to inter-decadal processes such as the El Niño Southern Oscillation
Index or Pacific Decadal Oscillation index, which have been shown to affect the spatiotemporal
distributions of fish communities, often by way of predator prey interactions (Lehodey et al.
2006) and zooplankton communities (Marinovic et al. 2002). We were not able to detect a
relationship between the MEI and community structure in this study (Table 2-2), but because this
study is only one decade long and does not encompass any major El Niño events we would not
Cram
63
expect to isolate the effects of these long-scale processes. One caveat to consider is that only a
few pairs of years are eight or nine years apart in this dataset. Thus 8-9 year differences likely
reflect particularly strong differences between the early and late years of the data set, rather than
some consistent pattern across many years. As time series studies continue, the longer data sets
will provide information to better resolve these long time-scale trends.
Alpha-diversity patterns differ between depths
The seasonality of biodiversity at 5m (Table 2-1, Supplemental Table 2-1) parallels trends seen
in the Western English Channel (Gilbert et al. 2012) and at BATS (Vergin et al. 2013b), while
the biodiversity’s seasonality at 890m is an original observation, to our knowledge. The
biodiversity maximum in the mid water column at 150m (Figure 2-3A) reflects patterns seen
elsewhere in the ocean for other organisms including copepods, ostracods (Angel 1993; Lindsay
& Hunt 2005), and fish (Badcock & Merrett 1976). Biodiversity is often highest at ecotones, the
interface between different environments, (Angel 1993; Barton et al. 2010; Ribalet et al. 2010)
and the 150m depth is such an interface between the euphotic and disphotic zones.
The correlation, at most pairs of depths, between high similarity and high biodiversity at those
depths (Figure 2-3B and C, Supplemental Figure 2-2) suggests mixing of communities between
depths or immigration of microbes from one depth to another may drive increased diversity at
those depths. For instance, diversity may be highest in the surface in the winter (when storms
may mix water to approx. 40 m) because the deeper, mixed in or upwelled, water layer brings
microbes from what had been in the stratified waters below the mixed layer into the surface,
thereby increasing the number and evenness of species in the surface. Higher biodiversity at
150m and 500m, among samples that changed little from month to month (Figure 2-3B and C),
Cram
64
suggests that particularly low disturbance allows for more niche differentiation and thus higher
diversity.
Environmental variability relates to community structure
At the Surface
The relationship between microbial community structure and chlorophyll a concentrations,
particulate organic carbon concentrations, primary productivity, heterotrophic productivity and
viral abundance at the surface (Table 2-2), extend Chow et al.’s (2013) observations of particular
environmental variables relating to the abundance of certain OTUs by identifying variables that
relate to community structure beyond the effects of seasonality. For instance, while Chow et al.
(2013) observed that nitrate was related to community structure, both nitrate and community
structure are seasonally variable and in this study we were not able to differentiate nitrate effects
away from seasonal effects statistically. Thus, we do not know whether nitrate itself or some
other seasonal factor shapes community structure. Conversely, chlorophyll a concentrations and
heterotrophic productivity (leucine incorporation) are related to the community when de-
convoluted from seasonality, suggesting links between these parameters beyond season, even
though these parameters also show seasonality.
Chlorophyll a concentrations likely relate to community structure because bacteria consume
phytoplankton exudates (Obernosterer & Herndl 1995; Fouilland et al. 2014), live symbiotically
with phytoplankton (Caron 2000; Aota & Nakajima 2001) and/or respond to unmeasured
variable(s) along with phytoplankton (Steele et al. 2011; Chow et al. 2014). Community structure
may relate to leucine incorporation because some OTUs are more metabolically active than
others. Bacterial abundance and viral abundance, but not virus-to-bacteria ratios, relate to
community structure, suggesting that denser communities contain different members than less
Cram
65
dense communities. This difference may result from greater numbers of interactions between
bacteria, grazers, and viruses in more dense communities. Alternatively, particular sets of
environmental conditions may favor both dense microbial communities and particular microbial
groups. Particulate organic carbon has been implicated as an important substrate for many
marine bacteria (Azam 1998; Simon et al. 2002) and it seems likely that the bacteria that co-
occur with particles at the surface may be directly associated with (living on or eating) these
particles.
At 890m
It appeared that, at 890m, both the depth of the mixed layer and precipitation (rainfall) rates,
measured on the day of sampling, related to community structure, beyond the effect of season
(Table 2-2). Both of these measurements are surface parameters, further supporting our
hypothesis that conditions at the top of the water column influence variability at the bottom of
the water column. Mixed layer depth is known to influence the range of depths at which
organisms can produce biomass and the types of dominant phytoplankton and zooplankton
(Sverdrup 1953; Bissett et al. 1994; Arrigo et al. 1999; Eslinger et al. 2001) and this variation in
surface productivity and plankton community structure likely in turn influences the rate of
particle production, particle type, and delivery to the sea floor. Collins et al. (2011) found that
there was no direct correlation between rainfall and mid-water particle flux nor between rainfall
and particulate carbon to nitrogen ratios, suggesting that links between rainfall and deep water
community structure are by way of some process other than surface to bottom particle flux.
Precipitation may trigger turbidity flows in which sediment influx from land after rainfall rapidly
moves down submarine canyons, such as nearby Redondo Canyon (see Sholkovitz & Soutar
1975). These turbidity flows likely transfer sediment to bottom waters and/or distribute it into the
Cram
66
lower water column by entrainment (Drake & Gorsline 1973; Drake 1974). Correlations to
rainfall on the day of sampling may relate to the observation that major rainstorms last multiple
days.
Mid Water Column
The absence of observed parameters strongly relating to community structure at the DCM, 150m,
and 500m depths, in combination with lack of seasonality at those depths, suggests that mid-
water community structure variability is due to unmeasured influences. These influences could
include unmeasured environmental factors, interactions between microorganisms, neutral
processes (see Chave 2004), or interactions between multiple environmental influences.
Meanwhile the weaker statistical association between bacterial abundance and composition at
DCM and 150m suggests that, as at the surface, denser communities are comprised of different
sets of organisms than sparser communities, rather than simply being comprised of more of the
same organisms. The community’s relationship, at these depths, to dominant wave period
suggests that mixing, physical processes caused by wave action, or some unmeasured factor
correlated with wave height (such as light penetration), affects community structure down to
150m.
Variability of specific taxonomic groups drives overall community variability
We saw different seasonally variable taxonomic groups at the surface, middle and bottom of the
SPOT water column (Table 2-3, Supplemental Table 2-2, Supplemental Table 2-3, Supplemental
Table 2-4). The seasonal surface groups, like the surface community as a whole, likely respond
to seasonality in light levels, mixing and nutrient levels. Prochlorococcus are adapted to
oligotrophic stratified warm water conditions (Partensky et al. 1999) and tend to occur during the
summer and fall (Figure 2-4), when Chlorophyll a levels are lowest (Supplemental Figure 2-1),
Cram
67
as a result of this adaptation. We were surprised to see that Prochlorococcus appeared to be
present at around 1% relative abundance in deep waters. This may reflect transport of this
organism from surface waters, as has been seen for Synechococcus in other environments (Sohrin
et al. 2011). Alternatively, this deep water cyanobacteria signal could be the result of other
bacteria with the same ITS length as cyanobacteria inhabiting the deep ocean. The latter
explanation seems particularly probable as no Cyanobacteria were found in our 890m SPOT
clone libraries. SAR11 were dominant members of the community at all depths and, as at BATS
(Carlson et al. 2009, 11; Vergin et al. 2013a), appeared to structure by depth. Furthermore, as at
BATS, SAR11 abundance, and the abundance of several SAR11 subclades, appeared to vary
seasonally (Supplemental Figure 2-4A, Supplemental Table 2-4) suggesting either different
environmental factors shaping this broad group’s abundances or differences in the taxonomic
identification schemes used in these two studies. ARISA, compared to tag sequencing and other
fragment analysis approaches, is better able to identify operational taxonomic units within
SAR11 clades. While the cumulative abundance of the SAR11 Surface 1 clade did not show
seasonal variability at any depth, individual OTUs within this clade showed differing seasonal
patterns (Supplemental Table 2-4), suggesting that this finer resolution is important in
understanding the dynamics of this abundant group.
Marine Group A (MGA) as a whole, and the taxonomic groups therein appeared to be abundant
seasonal components of community structure at 150m and 890m, with highest abundance in the
fall (150m) and summer (890m). The MGAs have previously been found in water with low
oxygen, especially oxygen minimum zones (Allers et al. 2013) and has been implicated in sulfur
cycling (Wright et al. 2013). While oxygen varies seasonally at 150m (Supplemental Figure 2-1,
Supplemental Table 2-1), this depth is high in oxygen (~100µM) year round, while oxygen levels
Cram
68
at 890m do not appear to vary seasonally, suggesting that the seasonality of these bacteria at our
site is driven by some factor other than bulk oxygen availability.
Seasonal taxa at 890m are likely related indirectly to seasonally variable surface parameters.
Flavobacteria remineralize particles and complex organic molecules (Kirchman 2002) and are
likely seasonal because the flux and content of particles or other food sources are also seasonally
variable. Nitrospina are likely nitrite oxidizers (Fuessel et al. 2012) and have been implicated as
drivers of nitrite oxidation in subsurface layers in much of the ocean (Mincer et al. 2007; Beman
et al. 2013). At SPOT they have previously been shown to fluctuate in abundance and respond to
variability of nitrite and ammonia oxidizing Archaea (Beman et al. 2010). The present study
supports this finding by identifying that the concentration of their substrate, nitrite, is seasonally
variable in concentration at 890m (Supplemental Figure 2-1, Supplemental Table 2-1). SAR324
was abundant and seasonally variable at 500m, and some strains are reported to oxidize reduced
sulfur or methane (Swan et al. 2011). Because Nitrospina and SAR324 are thought to be
chemoautotrophic, seasonality in these organisms’ abundances likely reflect variable carbon
fixation, in addition to the previously described processes.
Despite these seasonal patterns, one remarkable feature of broad taxonomic groups is their
stability. While communities, evaluated on an OTU level, vary significantly between months,
seasons and years, communities within a given depth appear to be dominated consistently by a
common set of broader taxa at similar relative abundances.
Individual OTUs drive community variability
All depths, regardless of overall seasonality, contain some OTUs that are seasonally variable and
others that are not. The prevalence of non-seasonal OTUs at depths with overall community
Cram
69
structure seasonality suggests that there are many OTUs that interact with their environment in
ways not affected by this variability. Conversely, at depths in which most OTUs do not vary
seasonally, there are always at least a few OTUs, making up a smaller subset of community
variability, that have a seasonally variable niche. These seasonal OTUs at non-seasonal depths
indicate that at least some environmental properties vary over time, particularly those properties
relating to transient particles sinking towards the bottom. During phytoplankton blooms, Teeling
et al. (2012) observed that finer level taxonomic level groups of bacteria adapted to different
ecological niches by expressing different metabolic processes. The observation that non-seasonal
taxa contain seasonal OTUs (Table 2-3, Supplemental Table 2-2, Supplemental Table 2-3,
Supplemental Table 2-4) supports this idea of niche differentiation within the broader taxa and is
a compelling reason to focus future ecological analysis of community structure on fine, rather
than coarse taxonomic groups, in order to discernmechanisms.
Advantages and considerations of using ARISA
The ARISA technique is well-suited to analysis of this data set, because its reliance on the
finely-resolved length of the hypervariable 16S-23S ITS region allows resolution of closely
related OTUs (Brown et al. 2005), often better than current next generation 16S tag sequencing
approaches (Chow et al. 2013). It is nevertheless important to consider that any PCR based
technique, including this one, has biases, such as over-representing some DNA fragments (e.g.
shorter ones), meaning that our relative abundance estimates may have biases towards some taxa
and against others. Fortunately, any such bias is likely consistent from sample to sample,
suggesting the conclusions drawn from the seasonal and interannual dynamics, the main focus of
this report, are justified. All of the statistics used to find patterns in this study are generally
insensitive to such biases. It is also important to consider that this version of ARISA detects
Cram
70
neither Archaea nor bacteria from the Planctomycetes and SAR202 phyla all of which are
members of many marine microbial communities, especially deep ones (Woebken et al. 2008;
Treusch et al. 2009). It is also inevitable that some OTUs will contain members of more than one
unrelated group, so the dynamics of any OTU may be the combined dynamics of any groups that
share a common intergenic spacer length. Although some ARISA OTUs (~10% of the
community, on average) are currently unknown, as they do not have corresponding clones in our
clone libraries, they represent a minority of our community fingerprint data. Despite being a
nearly decade old, this ARISA and clone library combined approach, as shown by (Brown et al.
2005) and subsequent studies, is still an effective method capable of generating high quality data
suitable for this type of study.
Conclusion
Bacterial communities at 5m and 890m vary seasonally. After seasonality had been factored out,
the residual variability was related to surface environmental parameters at both depths. In
contrast, the mid water column community as a whole did not appear to co-vary with seasonal or
any other measured environmental parameters. This pattern suggests links between the surface
environment and deep water communities, possibly driven by rapidly sinking particles and
migrating plankton. The community varies interannually throughout the top 500m of the water
column, suggesting a contribution of slower-than-seasonal variability in the environment of the
entire water column that affects microbial communities. This variability in community structure
appears to be driven by particular OTUs and sometimes broader taxonomic groups, and likely
reflects OTU-specific or group-specific responses to environmental variability. In contrast to
overall community structure, the Marine Group A phylum and a few other taxa are seasonally
variable throughout the aphotic zone; these organisms’ consistent deep water seasonality
Cram
71
suggests a potential tie between their ecology and seasonally variable particle flux, zooplankton
migration, or other seasonal factors to which the rest of the community does not respond. The
patterns reported here present a useful initial set of expectations for similar sites, including
coastal basins and oxygen minimum zones.
Cram
72
Supplement
Cram
73
Supplemental Figure 2-1 Plots of median environmental parameters for each month across the ten year
SPOT data set. This figure complements Figure 2-1 and shows additional environmental parameters.
Confidence bars are plus or minus one median adjusted deviation (MAD).
Cram
74
Supplemental Figure 2-2 R-values of correlations (y-axis) between Richness (0.1% cutoff) values for each
depth, given as a symbol, and its similarity to other depths or to the previous month (x-axis). This figure
complements Figure 3C and shows the same relationships, but for richness. Symbol shapes signify depth
at which the richness was measured. The first five parameters signify the Bray-Curtis similarity between
that the depth measured (indicated by symbol) and each other depth (vs5m, vsDCM, vs150m, vs500m,
vs890m). Higher R values represent a stronger correlation between the biodiversity at the measured depth
and its similarity to the depth given by the x label. The “vsLastMonth” column represents the relationship
between richness and the Bray-Curtis similarity between that sample and a sample taken at that same
depth collected in the previous month. Bars represent 95% confidence intervals for the R value.
Confidence intervals not overlapping the x axis indicate statistically significant correlations.
Cram
75
Supplemental Figure 2-3 Heat map of relative abundances of bacteria from of phylum and class level
groups (A) and family and order level groups (B). This figure complements Figure 2-4 and shows the
values at each time point, rather than monthly averages. The x-axis represents the date on which the
sample was taken, the y-axis represents depth and the colors indicate relative abundance of all organisms
within that group.
Cram
76
Supplemental Figure 2-4 Heat maps of relative abundances of various taxonomic groups averaged by
month. Each panel represents a different taxonomic group. The x-axis indicates months and y-axis
indicates sampled depths (not to scale). Colors correspond to the summed relative abundance of all OTUs
identified as falling within that taxonomic group, averaged by month. Relative abundances scores for each
color are given in the scale bar at right; note that abundances are on a log scale. A. Abundance of known
subclades of the SAR11clade. B. Abundance of OTUs within the SAR11 surface 1 subclade. C.
Abundance of finest level Silva taxonomy tags that constitute the Flavobacteria Class. D. Abundance of
genera of the Marine Group A Phylum.
Cram
77
Cram
78
Supplemental Figure 2-5 Heat maps of the relative abundance of the taxonomic groups given in
Supplemental Figure 2-4, by sample. The x-axis indicates time points; the y-axis indicates sampled
depths. Colors correspond to the summed relative abundance of all OTUs identified as falling within that
taxonomic group. A. Abundance of known subclades of the SAR11 clade. B. Abundance of OTUs within
the SAR11 Surface 1 subclade. C. Abundance of finest level Silva taxonomy tags that constitute the
Flavobacteria Class. D. Abundance of genera of the Marine Group A Phylum.
Cram
79
Cram
80
Supplemental Figure 2-6 The abundance of the five most abundant species at each depth across time. The
x axis of each plot is day of the year. Below the x-axis, the “target depth” is stated; this is the depth from
which the taxa plotted are the five most abundant for that depth. The y-axis is the relative abundance of
that operational taxonomic unit. The positions of the plotted numbers represent the relative abundance of
samples collected at the “target depth”. The value of the numbers indicates the year from which the data
point was taken, minus 2000. The curved line is a spline fitting these points.
Cram
81
Month Year
Depth Variable Trans Max Min EDF P Max Min EDF P R
2
5 Temperature
Sep Feb 5.44 0.000 2000 2011 0.03 0.381 0.83
CMAX Temperature
Dec May 3.42 0.000 2010 2000 0.97 0.000 0.39
150 Temperature
Dec May 4.09 0.000 2000 2011 0.00 0.605 0.49
500 Temperature
Aug Apr 4.09 0.000 2000 2011 0.00 0.870 0.42
890 Temperature
Mar Sep 0.41 0.323 2011 2003 0.40 0.213 0.02
5 Salinity
Jul Feb 3.64 0.000 2000 2004 5.90 0.000 0.64
CMAX Salinity
May Nov 2.62 0.001 2000 2010 0.93 0.009 0.20
150 Salinity
May Dec 2.80 0.000 2000 2011 0.38 0.213 0.20
500 Salinity
May Nov 0.00 0.575 2000 2011 0.01 0.327 0.00
890 Salinity
Apr Sep 0.00 0.762 2004 2011 0.99 0.000 0.23
5 Nitrite log
May Oct 0.00 0.741 2000 2011 0.01 0.421 0.00
CMAX Nitrite log
Jun Nov 1.42 0.130 2000 2010 0.52 0.161 0.05
150 Nitrite log
Apr Sep 0.01 0.523 2000 2010 0.96 0.001 0.13
500 Nitrite log
Dec Jul 0.00 0.766 2000 2011 0.86 0.016 0.07
890 Nitrite log
Jun Dec 1.84 0.049 2003 2011 0.53 0.164 0.12
5 Nitrate log
May Nov 0.60 0.263 2011 2000 0.01 0.785 0.01
CMAX Nitrate log
Jun Dec 2.18 0.008 2000 2010 0.94 0.007 0.17
150 Nitrate log
Jun Nov 2.50 0.001 2010 2000 1.05 0.000 0.34
500 Nitrate log
Mar Aug 0.00 0.749 2011 2000 0.97 0.001 0.15
890 Nitrate log
Dec Aug 2.29 0.064 2011 2003 0.68 0.093 0.14
5 Phosphate log
Dec Aug 0.00 0.891 2000 2011 1.01 0.000 0.18
CMAX Phosphate log
May Oct 2.61 0.005 2003 2010 2.37 0.000 0.42
150 Phosphate log
Apr Oct 1.96 0.012 2000 2010 3.10 0.000 0.72
500 Phosphate log
Feb Aug 1.90 0.017 2000 2009 4.51 0.000 0.74
890 Phosphate log
Jan Aug 1.41 0.115 2003 2011 1.19 0.000 0.59
5 P* log
Jan Sep 3.62 0.001 2000 2011 1.12 0.000 0.42
CMAX P* log
Jan Aug 0.00 0.634 2000 2010 1.05 0.000 0.26
150 P* log
Mar Oct 0.81 0.221 2000 2010 2.37 0.000 0.74
500 P* log
Feb Aug 1.29 0.112 2000 2009 3.81 0.000 0.62
890 P* log
Jan Jul 0.43 0.318 2003 2011 1.15 0.000 0.55
5 Oxygen
Apr Oct 2.21 0.002 2000 2011 1.05 0.000 0.27
CMAX Oxygen
Aug Feb 0.00 0.653 2000 2010 1.08 0.000 0.28
150 Oxygen
Oct Apr 1.84 0.015 2000 2011 1.08 0.000 0.31
500 Oxygen
Sep Mar 1.57 0.076 2000 2011 0.01 0.555 0.05
890 Oxygen
Nov Mar 0.01 0.783 2011 2003 0.00 0.984 0.00
5 Chlorophyll (Month) log
Mar Oct 4.33 0.000 2011 2000 0.98 0.001 0.41
5 Prim. Prod. (Month) log
Apr Dec 4.07 0.000 2010 2000 0.93 0.005 0.27
5 Chlorophyll (8 Day) log
Mar Sep 3.37 0.000 2011 2000 0.96 0.001 0.30
5 Prim. Prod. (8 Day) log
Apr Nov 2.64 0.002 2010 2000 0.90 0.010 0.15
Cram
82
Month Year
Depth Variable Trans Max Min EDF P Max Min EDF P R
2
5 POC log
Apr Oct 4.17 0.000 2011 2000 0.98 0.000 0.35
5 CDOM sqrt
Apr Aug 4.34 0.000 2011 2000 0.47 0.179 0.31
5 PAR sqrt
Jul Dec 6.31 0.000 2000 2011 0.01 0.515 0.90
5 SSHD
Sep Apr 4.82 0.000 2001 2010 0.00 0.882 0.58
5 Mixed Layer Depth log
Jan Jul 4.00 0.000 2000 2011 0.00 0.755 0.56
5 CMAX Depth Cmax
Sep Apr 2.20 0.006 2000 2010 0.95 0.002 0.21
5 Sverdrup Trans.
Jan Jul 2.00 0.011 2000 2011 0.84 0.023 0.10
5 Upwelling
May Dec 3.47 0.000 2000 2011 0.01 0.647 0.31
5 Wave Height sqrt
Feb Sep 2.71 0.000 2000 2011 0.42 0.198 0.14
5 Avg. Wave Pd. sqrt
Jan Aug 1.95 0.033 2000 2011 0.67 0.091 0.07
5 Dom. Wave Pd.
Oct May 0.01 0.622 2000 2011 0.00 0.841 0.00
5 Precipitation log
May Oct 1.87 0.022 2011 2000 0.87 0.011 0.10
5 Max. Air Temp sqrt
Aug Jan 3.36 0.000 2000 2011 0.96 0.001 0.36
5 Min. Air Temp sqrt
Aug Feb 3.39 0.000 2000 2011 0.94 0.001 0.37
5 Avg. Wind Speed
Jan Jul 2.23 0.009 2000 2011 0.01 0.639 0.08
5 Wind Gust Speed sqrt
Dec Jul 2.17 0.007 2000 2011 0.01 0.796 0.08
5 Bacterial Abundance sqrt
May Nov 2.34 0.002 2011 2000 0.94 0.003 0.17
CMAX Bacterial Abundance sqrt
Jun Dec 2.06 0.011 2010 2000 0.97 0.001 0.20
150 Bacterial Abundance sqrt
Apr Oct 1.51 0.056 2011 2000 0.87 0.013 0.11
500 Bacterial Abundance sqrt
Nov May 0.67 0.245 2011 2000 0.91 0.005 0.10
890 Bacterial Abundance sqrt
Oct Apr 1.06 0.156 2003 2011 0.55 0.150 0.05
5 Viral Abundance sqrt
Aug Feb 0.53 0.277 2004 2000 3.45 0.000 0.17
CMAX Viral Abundance sqrt
Aug Jan 0.03 0.536 2010 2000 1.02 0.000 0.15
150 Viral Abundance sqrt
Feb Aug 1.19 0.142 2011 2000 0.66 0.100 0.05
500 Viral Abundance sqrt
Feb Oct 0.01 0.943 2011 2000 0.87 0.012 0.07
890 Viral Abundance sqrt
Jul Jan 0.00 0.866 2011 2003 0.01 0.490 0.00
5 Virus:Bacteria Ratio sqrt
Oct Apr 1.82 0.019 2000 2011 0.00 0.924 0.06
CMAX Virus:Bacteria Ratio sqrt
Nov May 1.15 0.164 2010 2000 0.00 0.438 0.02
150 Virus:Bacteria Ratio sqrt
Jan Jun 0.00 0.396 2011 2000 0.00 0.886 0.00
500 Virus:Bacteria Ratio sqrt
Apr Oct 1.04 0.164 2000 2011 0.10 0.305 0.02
890 Virus:Bacteria Ratio sqrt
May Nov 0.58 0.288 2011 2003 0.55 0.155 0.03
5 Growth (Leucine) log
Apr Nov 2.71 0.000 2011 2000 0.00 0.691 0.18
CMAX
Growth (Leucine)
log
May Oct 3.19 0.000 2010 2000 0.00 0.551 0.32
150
Growth (Leucine)
log
May Jan 3.26 0.018 2000 2011 0.01 0.349 0.11
500
Growth (Leucine)
log
Jul Jan 0.23 0.350 2000 2011 0.00 0.472 0.00
890
Growth (Leucine)
log
Aug Feb 2.51 0.006 2003 2011 0.31 0.245 0.17
5 Turnover Time (Leu) log
Nov Jun 2.03 0.015 2011 2000 0.82 0.029 0.11
CMAX Turnover Time (Leu) log
Oct Apr 2.19 0.002 2010 2000 0.69 0.086 0.18
150 Turnover Time (Leu) log
Jan Aug 1.51 0.139 2011 2000 0.86 0.017 0.10
500 Turnover Time (Leu) log
Dec Jun 1.00 0.189 2011 2000 0.84 0.024 0.08
Cram
83
Month Year
Depth Variable Trans Max Min EDF P Max Min EDF P R
2
890 Turnover Time (Leu) log
Jan Jul 1.84 0.053 2011 2003 0.00 0.822 0.09
5 Richness (0.1%)
Oc)(t May 1.37 0.132 2011 2000 0.01 0.467 0.03
CMAX Richness (0.1%)
Jun Dec 0.00 0.698 2000 2010 0.70 0.079 0.03
150 Richness (0.1%)
Mar Oct 0.02 0.428 2000 2011 0.56 0.145 0.02
500 Richness (0.1%)
Jul Dec 0.00 0.926 2001 2011 0.86 0.019 0.08
890 Richness (0.1%)
Mar Aug 2.66 0.007 2003 2011 0.00 0.324 0.20
5 1/Simpson Index
Jan May 4.35 0.000 2011 2000 0.95 0.001 0.24
CMAX 1/Simpson Index
Jan Jul 1.99 0.014 2010 2003 1.87 0.000 0.27
150 1/Simpson Index
Apr Oct 0.03 0.527 2011 2000 1.06 0.000 0.20
500 1/Simpson Index
Mar Oct 0.00 0.903 2011 2000 0.94 0.001 0.13
890 1/Simpson Index
Mar Aug 2.51 0.000 2011 2003 0.87 0.013 0.27
Supplemental Table 2-1 . Generalized additive mixed model results, modified to treat residuals as a
stochastic time series process for all environmental parameters. Log-odds transformed abundance data
were fit a cyclic spline with one year period to measure seasonality and a time dependent spline to model
interannual variability. “Depth” indicates the depth from which the sample was taken. “Variable”
indicates the environmental parameter modelled. “Trans” indicates the transformation function applied to
the environmental data before the model was fit. “log” represents log(x+a) transformations where “a” is a
constant. Variables under the “Month” describe the seasonal spline while “Year” variables describe the
interannual spline. For both splines “Max” and “Min” represent the months and years with highest or
lowest abundance of that parameter. “EDF” is estimated degrees of freedom for the spline and represents
the complexity of the spline. EDF terms of one represent perfectly sinusoidal seasonal splines or perfectly
linear interannual splines. “P” represents the p-value for the fit of that spline. R
2
indicates the r squared
value for the entire model, only taxa with R
2
fits of higher than 0.10 are shown. Abbreviated categories
are as follows: Particulate Organic Carbon (POC), Colored Dissolved Organic Matter (CDOM),
Photosynthetically Active Radiation (PAR), Sea Surface Height Differential (SSHD), Chlorophyll
Maximum Layer Depth (CMAX Depth).
Cram
84
Month Year
Depth Taxon Max Min EDF P Max Min EDF P R
2
5m
Cyanobacteria Nov Apr 2.89 0.000 2000 2011 0.98 0.000 0.27
Chloroplast Feb Sep 2.44 0.001 2000 2011 0.85 0.017 0.16
Gammaproteobacteria Aug May 3.98 0.000 2000 2011 0.01 0.508 0.16
Flavobacteria Feb Oct 3.31 0.000 2011 2000 0.49 0.169 0.17
CMAX
Chloroplast Feb Sep 2.71 0.003 2010 2000 0.00 0.943 0.12
Marine Group A Nov Apr 2.55 0.016 2010 2000 0.88 0.009 0.14
150m
Cyanobacteria Apr Oct 1.23 0.114 2000 2011 0.92 0.004 0.10
Chloroplast Apr Oct 2.67 0.000 2000 2011 0.07 0.335 0.20
Gammaproteobacteria May Nov 2.03 0.007 2011 2000 0.78 0.046 0.12
Marine Group A Oct Feb 2.50 0.018 2011 2000 0.83 0.023 0.12
500m
Cyanobacteria May Nov 0.01 0.806 2000 2011 0.89 0.009 0.10
Deltaproteobacteria Dec Jul 2.17 0.006 2011 2000 0.87 0.011 0.13
890m
Chloroplast Jan Jul 2.13 0.009 2011 2003 0.20 0.278 0.13
Alphaproteobacteria Jan Aug 2.31 0.004 2011 2003 0.22 0.270 0.16
Deltaproteobacteria Aug Feb 1.68 0.037 2003 2011 0.69 0.086 0.11
Flavobacteria Aug Feb 2.87 0.000 2003 2011 0.00 0.362 0.31
Marine Group A Aug Jan 1.95 0.013 2003 2011 0.25 0.262 0.12
Supplemental Table 2-2 Generalized additive mixed model results, modified to treat residuals as a
stochastic time series process, for Class and Phylum level taxonomic groups, that are found in Figure
2-4A, Supplemental Figure 2-3A. Log-odds transformed abundance data were fit a cyclic spline with one
year period to measure seasonality and a time dependent spline to model interannual variability. Depth
indicates the depth from which the sample was taken. Taxon indicates the class or phylum level group
modelled. Month variables describe the seasonal spline and Year variables describe the interannual spline.
For both “Max” and “Min” represent the months and years with highest or lowest abundance of that taxa.
EDF is estimated degrees of freedom for the spline and represents the complexity of the spline. EDF
terms of one represent perfectly sinusoidal seasonal splines or perfectly linear interannual splines. “P”
represents the p-value for the fit of that spline. P-values of less than 0.05 correspond to false discovery
rates of 0.040 for the month spline and 0.018 for the year spline. R
2
indicates the r squared value for the
entire model, only taxa with R
2
fits of higher than 0.10 are shown.
Cram
85
Month Year
Depth Taxon Max Min EDF P Max Min EDF P R
2
5m
Prochlorococcus Nov Apr 3.08 0.000 2000 2011 0.94 0.002 0.27
Nitrospina Oct Apr 1.83 0.020 2011 2000 0.91 0.006 0.11
SAR324 Mar Aug 3.05 0.000 2011 2000 0.92 0.004 0.23
Arctic96B-7 Nov May 2.34 0.003 2011 2000 1.00 0.000 0.21
CMAX
Prochlorococcus Nov May 1.67 0.040 2000 2010 0.88 0.011 0.11
SAR324 Apr Sep 2.44 0.000 2010 2000 0.00 0.614 0.15
150m
NS9 Nov May 1.27 0.107 2011 2000 0.95 0.002 0.12
Arctic96B-7 Oct Mar 2.09 0.011 2011 2000 0.83 0.024 0.12
SAR324 Dec Jul 1.53 0.079 2011 2000 0.80 0.034 0.18
500m
NS9 Nov May 1.03 0.183 2006 2000 1.98 0.000 0.20
Arctic96B-7 Jan Jul 1.80 0.045 2011 2000 0.93 0.002 0.14
SUP05 Jun Dec 2.01 0.014 2003 2011 0.84 0.023 0.18
890m
Nitrospina Aug Jan 1.89 0.021 2003 2011 0.80 0.039 0.15
NS9 Aug Feb 2.46 0.000 2003 2011 0.74 0.062 0.24
Arctic96B-7 Aug Feb 2.25 0.003 2003 2011 0.06 0.320 0.16
Supplemental Table 2-3 Generalized additive mixed model results, modified to treat residuals as a
stochastic time series process, for the selected Order and Family level taxonomic groups, that are found in
Figure 2-4B and Supplemental Figure 2-3B. Log-odds transformed abundance data were fit a cyclic
spline with one year period to measure seasonality and a time dependent spline to model interannual
variability. Only taxa with R
2
values greater than 0.10 are shown. Column headings are the same as for
Supplemental Table 2-2. P-values of less than 0.05 correspond to false discovery rates of 0.047 for
months and 0.015 for years.
Cram
86
Month Year
Depth Taxon Max Min EDF P Max Min EDF P R
2
Sar11 Clades Q(0.05) =0.116 Q(0.05) = 0.022
5 SAR11 Surface_2 Oct May 3.25 0.000 2011 2000 0.27 0.256 0.28
5 SAR11 Surface_4 Jul Mar 2.82 0.021 2011 2005 2.27 0.001 0.19
CMAX SAR11 Surface_2 Dec Jun 2.68 0.000 2010 2000 0.63 0.109 0.18
CMAX SAR11 Surface_4 Aug Apr 4.20 0.000 2010 2000 0.85 0.026 0.22
500 SAR11 Surface_1 May Nov 0.01 0.940 2011 2000 1.01 0.000 0.15
890 SAR11 Surface_1 Feb Aug 2.05 0.006 2011 2003 0.01 0.360 0.13
890 SAR11 Surface_4 Jan Aug 2.86 0.020 2011 2003 0.75 0.065 0.17
Sar11 Surface 1 OTUs Q(0.05) =0.090 Q(0.05) =0.171
5 SAR11_657.6 Nov May 0.01 0.442 2011 2000 1.03 0.000 0.16
5 SAR11_663.4 Dec Jun 1.36 0.086 2011 2000 6.23 0.000 0.49
5 SAR11_683.9 Nov May 0.62 0.252 2011 2000 1.00 0.000 0.13
5 SAR11_686.9 Jun Oct 2.75 0.001 2000 2011 0.01 0.506 0.12
5 SAR11_692.2 Jan Jul 2.56 0.000 2011 2000 0.93 0.003 0.20
CMAX SAR11_657.6 Dec Jun 1.59 0.057 2010 2000 0.89 0.007 0.13
CMAX SAR11_692.2 Jan Jun 1.65 0.058 2010 2000 0.88 0.011 0.10
500 SAR11_584 Feb Aug 0.01 0.568 2006 2000 2.33 0.003 0.12
500 SAR11_657.6 Oct Apr 0.00 0.657 2011 2000 0.97 0.000 0.13
500 SAR11_663.4 Apr Oct 0.60 0.258 2011 2000 1.01 0.001 0.13
500 SAR11_683.9 Dec May 0.01 0.799 2011 2000 0.99 0.000 0.22
500 SAR11_686.9 Nov May 0.00 0.905 2011 2000 3.24 0.000 0.32
890 SAR11_663.4 Mar Sep 0.01 0.497 2011 2003 0.93 0.003 0.13
890 SAR11_670.5 Feb Aug 1.98 0.019 2011 2003 0.00 0.855 0.11
890 SAR11_683.9 Mar Nov 3.17 0.001 2011 2003 0.85 0.021 0.25
Flavobacteria Q(0.05) =0.100 Q(0.05) =0.025
5 Flavobacteriaceae Mar Sep 1.89 0.033 2000 2011 0.91 0.005 0.11
5 NS4 Dec May 2.39 0.001 2000 2011 0.01 0.719 0.11
5 NS7 Apr Sep 3.85 0.000 2000 2011 0.00 0.556 0.24
5 Owenweeksia Mar Aug 3.06 0.000 2011 2000 0.89 0.008 0.22
CMAX Flavobacteriaceae Mar Sep 0.38 0.310 2000 2010 0.99 0.000 0.14
CMAX Owenweeksia Apr Aug 3.34 0.000 2010 2000 0.85 0.017 0.21
150 NS9 Nov May 1.27 0.107 2011 2000 0.95 0.002 0.12
500 Formosa Nov May 2.86 0.000 2011 2000 0.91 0.005 0.28
500 NS5 May Nov 1.26 0.111 2000 2011 0.92 0.004 0.11
500 NS9 Nov May 1.03 0.183 2006 2000 1.98 0.000 0.20
890 Fluviicola Jul Jan 0.00 0.931 2011 2003 0.98 0.000 0.18
890 NS9 Aug Feb 2.46 0.000 2003 2011 0.74 0.062 0.24
Cram
87
Month Year
Depth Taxon Max Min EDF P Max Min EDF P R
2
Marine Group A Q(0.05) =0.002 Q(0.05) =0.011
5 Arctic95A-2 Feb Sep 0.00 0.766 2011 2000 0.98 0.000 0.12
5 SargSea-WGS Sep Mar 2.11 0.005 2011 2000 0.92 0.004 0.14
5 SGSH944 Dec May 2.44 0.000 2011 2000 0.49 0.172 0.14
150 SargSea-WGS Sep Apr 2.35 0.004 2011 2000 0.91 0.004 0.17
150 SGSH944 Feb Jul 1.26 0.160 2011 2000 0.95 0.001 0.12
150 ZA3312c Oct Mar 2.14 0.003 2011 2000 0.01 0.886 0.11
500 ZA3312c Mar Aug 0.01 0.607 2011 2000 1.00 0.000 0.16
890 Arctic95A-2 May Nov 1.72 0.055 2011 2003 0.92 0.004 0.17
890 SargSea-WGS Sep Feb 2.06 0.006 2003 2011 0.00 0.608 0.13
890 SGSH944 Mar Oct 2.02 0.024 2011 2003 0.91 0.015 0.17
Supplemental Table 2-4 Generalized additive mixed model results, modified to treat residuals as a
stochastic time series process, for the taxonomic groups seen in Supplemental Figure 2-4. Log-odds
transformed abundance data were fit a cyclic spline with one year period to measure seasonality and a
time dependent spline to model interannual variability. Only taxa with R^2 values greater than 0.10 are
shown. Column headings are the same as for Supplemental Table 2-2. False discovery rates, associated
with p-values of 0.05 “Q(0.05)” for both months and years are reported in the header lines.
Cram
88
Supplemental Methods
Satellite measurements
Monthly and eight day average estimates for surface chlorophyll a concentration and surface
productivity were downloaded from the Coastwatch browser website. The chlorophyll-a
measurements were from the data categories “Chlorophyll a, SeaWiFS, 0.04167 degrees, West
US Science Quality + Chlorophyll-a” and “Chlorophyll-a, Aqua MODIS NPP, 0.05 DEGREES,
Global, Science Quality”. Productivity was extracted from the categories “Primary Productivity,
SeaWiFS and Pathfinder, 0.1degrees, Global, EXPERIMENTAL” and “Primary Productivity,
NASA Aqua MODIS and Pathfinder, 0.1 degrees, Global, EXPERIMENTAL” (Behrenfeld &
Falkowski 1997). Data from SeaWiFS and MODIS together spanned our data set, with overlap in
the middle. In cases where the data overlapped, we gave priority to the MODIS data.
Surface eight day averages of photosynthetically active radiation (PAR) (Frouin et al. 2003),
colored dissolved organic matter to chlorophyll a ratio (CDOM) (Mannino et al. 2008) and
particulate organic carbon concentrations (Morel & Gentili 2009) were downloaded from the
ocean color data site. A 3x3 grid of pixels was extracted from around SPOT using the SeaDas
program (Fu et al. 1998), and the weighted average (using weights from the SeaDas output) of
this grid was used in downstream analysis. Monthly sea surface height differential was
downloaded from Coastwatch as “Sea Surface Height Deviation, Aviso, 0.25 degrees, Global,
Science Quality” (Ducet et al. 2000).
We obtained meteorological data, including minimum and maximum daily air temperatures and
precipitation data from the weather station at nearby Avalon airport (33.405°N 118.415°W).
Wave height, average wave period, and dominant wave period from a buoy in nearby Santa
Monica Bay (33.749°N, 119.053°W) were downloaded from the National Buoy Data Center.
Pacific Fisheries and Environmental Laboratory (PFEL) estimates of coastal upwelling, and
Sverdrup transport at (33°N, 119°W), along with Multivariate ENSO Index scores were
downloaded from the National Oceanographic and Atmospheric Administration (NOAA).
Assigning Taxonomic Identities to ARISA peaks
We assigned taxonomic identity to each ARISA fragment size by identifying which clones from
our clone libraries had fragment sizes that fell within the range of peak sizes that were assigned
Cram
89
to an ARISA OTU bin. In cases in which an ARISA OTU corresponded in size to more than one
clone in our clone library database, we prioritized our clones based on where they were isolated.
For fragments that were more abundant in surface waters than at 890m, we prioritized fragments
according to the first number in parentheses, while fragments that were more abundant at 890m
than 5m we prioritized according to the second number in parentheses:
(1;3) observed ARISA length of SPOT clones from 5m across all seasons (2;2) SPOT clones
from 150m (3;1) SPOT clones from 890m (4;5) Clones from the Pacific Ocean (near Hawaii)
and Atlantic Ocean (near the Amazon river outfall) from 5m. (5;7) published cyanobacterial
intergenic spacer (ITS) sequences (6;6) observed ARISA lengths from 16S-ITS clones from
surface waters of the Indian Ocean: (7;9) in silico amplification of marine isolate genomes from
the photic zone(8;4) Clones from the Pacific and Atlantic oceans from 500m and below (9;8) in
silico amplification of marine isolate genomes from below the photic zone. Chow et al (2013)
provide a full description about these datasets and how they were used to assign identity to
ARISA OTUs.
In cases in which more than one clone from the highest priority category fell into a given bin, we
selected the clone that had the highest number of instances in our clone libraries.
Environmental parameter variability
We tested for seasonal variability of each measured environmental parameter by applying
generalized additive mixed effects models (Wood 2004, 2006). Each variable was modelled
according to the equation
y = μ + m1(time) + m2(DoY) + ε
In this equation “y” is the transformed (for normalization purposes) value of the environmental
parameter. m1(time) is a univariate smooth thin plate regression spline modelling long term
variability as a function of the number of days that had elapsed since the beginning of the study.
m2(DoY) (Day of Year) is a cyclic penalized cubic regression smooth spline of one year period.
μ is essentially the mean of y, and the m1 and m2 functions describe how y deviates from this
Cram
90
mean over time. ε is the error term. The model was set up to allow for the data to have a
continuous lag-1 autoregressive structure. This model reflects equation 1 in Ferguson et al.
(2008) as well as approaches demonstrated elsewhere (Wood 2006, 321–324) and identifies
seasonal variability that is not perfectly sinusoidal as well as long term trends that are non-linear.
The model was run using the “gamm” function of the “mgcv” R-package (Wood 2011).
For both the seasonal and long term spline function we determined the model’s estimated
degrees of freedom (EDF) which is essentially a measure of the complexity of the spline. For
instance, seasonal splines of EDF of 1 are perfectly sinusoidal while higher EDF relate to
unevenly shaped seasonal peaks or local maxima. Long term splines with EDF of 1 are linear,
while higher EDF correspond to curved long term splines which may have maximum and
minimum values at years within (rather than at the extremes of) the dataset. We also determined
p-values for both the seasonal and long term splines, where P is the probability of the null
hypothesis that the EDF of the smooth term is actually zero (no prediction by that spline). R2
values for the entire model were also determined.
For each fitted nonparametric regression model, we interrogated the cyclic seasonal spline to
determine the month in which that factor had the highest value and the month in which that
factor had the lowest value. We interrogated the long-term spline function to determine whether
there appeared to be a linear on non-linear change over time and identified the years that
appeared to have the highest and lowest values.
To test whether each variable appeared to relate to the Multivariate El-Niño Southern Oscillation
Index (MEI) a second GAMM model using MEI instead of year as a predictor variable for the
long term spline was fit to each variable. Thus this model was of the form y = μ + m1(MEI) +
Cram
91
m2(DoY) + ε. We determined the Akaike information criterion (AIC) of both the original (Year)
and modified (MEI) GAMM models. In cases where the second model had a lower AIC than the
former, and in which p-values suggested the m1(MEI) model had good fit we would say that the
variable seemed to be driven by variability in MEI.
Seasonal variability of microbial community structure
Graphical approach
We used the “vegdist” function in the “vegan” (Oksanen et al. 2011) package to estimate Bray-
Curtis dissimilarity in community structure between all pairs of samples in the dataset, thereby
generating a dissimilarity matrix; we calculated similarity matrix by subtracting the dissimilarity
scores from one. We determined upper and lower bounds for these similarity scores by
examining similarity scores between machine replicates (upper bounds) and randomized samples
(lower bounds). Machine replicates were identical samples run on different fragment analysis gel
lanes. We determined the machine replicate similarity for every sample in the data set and
calculated average machine replicate similarity. Similarities between randomized samples were
determined by arbitrarily picking pairs of samples and then shuffling the orders of the
abundances of each OTU. This process was repeated 1000 times and the average value of
similarity between randomized samples was recorded.
We determined the temporal difference or lags, in days, between all pairs of samples and the
Bray-Curtis similarity between those same pairs of samples. Pairs of samples were binned by
their lags in by 30.416 day (the average number of days in a month) intervals and average
similarity for pairs of samples in each monthly bin was determined. Accordingly our first bin
returned the average Bray-Curtis similarity value for all samples collected between 15 and 45
days apart, the next bin returned the average for all samples between 46 and 76 days apart and so
Cram
92
forth. Bray-Curtis similarity scores that oscillated with a period of one year were considered
seasonal. A t-test was applied to ask whether samples that were taken one month (15 to 45 days)
apart were statistically more similar than samples taken six months apart.
Mantel test approach
Mantel tests were applied to look for seasonality using a ‘seasonal difference matrix’ (S). S was
calculated as follows: “D”, a matrix of the difference in serial days between each pair of samples
was calculated. 2) “DM”, a matrix containing the 365.25 day modulus of each value in D was
calculated. “S” was calculated from each value of DM such that if the value was less than
180.625 it was kept and if greater the value was subtracted from 365.25 and the difference was
kept. These ‘seasonal difference matrixes’ were compared to the community’s Bray-Curtis
dissimilarity matrix using the “mantel” function in the “ecodist” package for R (Goslee & Urban
2007). Depths where the seasonal matrix significantly correlated to the Bray-Curtis dissimilarity
matrix were said to be seasonal.
Interannual variability of microbial community structure
We binned samples by 365.25 day (the average number of days in a year) intervals and applied
the same analysis described previously. Thus the first bin would contain the mean of all samples
taken between 1 month and 12 months apart, the second all samples taken between 13 and 24
months apart and so on. For each depth we performed ANOVA to ask whether the mean
similarity between samples within each bin differed between those bins. In cases in which the
ANOVA suggested statistically significant differences, we performed a Tukey corrected t-test for
each pair of bin categories to determine which bins had statistically different mean similarity
scores. As for the seasonality comparison, we performed Mantel tests to examine the relationship
between difference in serial day (“D” as calculated above) and community dissimilarity. Depths
Cram
93
in which samples that were more temporally distant had higher Bray-Curtis dissimilarity scores
would be said to show long term change in community structure.
Alpha diversity
Variability between depths
Mean values of Richness of species with greater than 1% 0.1% and 0.01% relative abundance,
inverse Simpson index (ISI), Shannon indexes of biodiversity and Pelou’s index of evenness
were determined at each depth along with 95% confidence intervals of those means. We
investigated whether richness at 0.1% and ISI differed between depths using analysis of variance
(“AOV” function in the R's “stats” package). A Tukey corrected t-test compared all pairs of
depths in order to determine which pairs of depths have different mean richness and ISI.
Relation to season
Richness and ISI were investigated with the same nonparametric regression model used to
investigate seasonality. Seasonal and long term spline functions were investigated and depths
with seasonal and long term trends were noted. We identified months and years of highest and
lowest biodiversity and parameters for the splines used to fit these data.
Relation to community similarity between depths
We examined, for each depth, whether richness and/or ISI was correlated with the similarity of
that depths community structure to the community structure at each other depth. Our goal was to
identify whether biodiversity at each depth was driven by influence of OTUs from other depths.
Inter-depth community similarity was determined for each pair of depths as the Bray-Curtis
similarity between those depths’ communities in a given month. Scatterplots of richness vs inter-
depth similarity and ISI vs inter-depth similarity were visually investigated to determine whether
Cram
94
simple correlations were sufficient to describe relationships between the factors. After
determining that no non-linear relationships were present, Pearson correlation tests were applied
to determine whether richness and/or ISI at each depth was statistically correlated with the inter-
depth community similarity between that depth and each other depth. R values of the Pearson
correlations and the 95% confidence intervals of those R values were identified for each
comparison. Relationships whose and confidence intervals did not overlap zero were identified
as having a statistically significant relation between inter-depth community similarity and
biodiversity.
Relation to community change
We queried whether alpha diversity was higher for communities that were changing the most
rapidly. To calculate the rate of community change, we compared the Bray Curtis dissimilarity of
each month to the community in the previously sampled month and refer to this dissimilarity
score henceforth as “Bray-Curtis shift (BCS)”. As in the inter-depth similarity analysis,
scatterplots of BCS vs Richness and BCS vs ISI at each depth were investigated for non-linear
relationships. After determining that no parabolic relationships were present, Pearson correlation
tests were applied to determine whether BCS was statistically significantly related to richness
and Simpson’s index at each depth. R values of the Pearson correlations and the 95% confidence
intervals of those R values were identified and depths with confidence intervals not overlapping
zero were identified as having a statistically significant relation between community change and
biodiversity.
Environmental parameters and community structure: Mantel tests
We applied partial Mantel tests that examine the model Y = a +bS + cD +dX where Y is the Bray
Curtis similarity matrix of the community structure, S is the seasonal distance matrix, D is the
Cram
95
serial day distance matrix (S and D are described above), and X is the similarity matrix for the
variable of interest. Our environmental data set was missing values for a few environmental
variables. Because Mantel tests are not able to handle missing data, we filled in our data set
using multiple imputation, a method which fills in missing values with numbers that are
reasonable estimates but reflect the uncertainty of the data (King et al. 2001). We generated 25
imputed data sets using the “Amelia” R-package (Honaker et al. 2006; Honaker & King 2010)
and performed the Mantel test, using the “ecodist” R-package (Goslee & Urban 2007), on each
of these imputed data sets. We then report the median rho score of the 25 Mantel tests, and the p-
value corresponding to this median rho score. Because we ran many tests in parallel, in addition
to calculating p-values for each environmental parameter, we also estimated the false discovery
rate “Q” from the p-values at each depth using the “qvalue” R-package (Dabney et al. 2004).
Temporal dynamics of microbial taxa over time
Transformations
Taxonomic groups and individual OTUs were log of odds transformed using the “logit” function,
in the “car” R-package (Fox and Weisberg, 2011), with an adjustment factor of 0.001.
Taxonomic Groups
We examined seasonality and long term temporal variability of class level taxonomic groups,
more abundant order and family level taxonomic groups, each of the sub clades of SAR11, all of
the OTUs of SAR11 Surface group 1, clades of Flavobacteria and genera of Marine group A. To
investigate these temporal dynamics we attempted to fit the group’s log-odds transformed
abundance (Y’) using the same nonparametric model used to fit the environmental variables. For
each OTU, months and years of both maximal and minimal abundance, estimated degrees of
freedom of each spline term and p-values for each spline term were generated. We report all
Cram
96
taxonomic groups which were fit by this model with an R^2 value greater than 0.10. False
discovery rates (Q) were calculated from the p-values of each category of taxa investigated.
OTUs
The previously described nonparametric regression model was also applied to each of the 100
most abundant OTUs (where Y’ is the transformed relative abundance of the OTU under
investigation). We recorded the number of bacteria, out of 100 that were fit with R^2 values of
0.1 and 0.2. Of these bacteria, we determined which were fit by the seasonal spline with a p-
value of less than 0.05 and which were fit by the long-term spline with a p-value less than 0.05.
To determine if the fraction of seasonally variable and interannually variable bacteria (seasonal
term P < 0.05, R2 > 0.2) differed between depths, we applied the “chisq.test” function in R. We
recorded the taxonomic identity, ARISA fragment length and statistics for each temporally
variable OTU that was fit by the model with an R^2 value of greater than 0.2.
References
Agogué H, Lamy D, Neal PR, Sogin ML, Herndl GJ. (2011). Water mass-specificity of bacterial
communities in the North Atlantic revealed by massively parallel sequencing. Molecular Ecology
20:258–274.
Allers E, Wright JJ, Konwar KM, Howes CG, Beneze E, Hallam SJ, et al. (2013). Diversity and
population structure of Marine Group A bacteria in the Northeast subarctic Pacific Ocean. ISME
J 7:256–268.
Alonso-Sáez L, Balagué V , Sà EL, Sánchez O, González JM, Pinhassi J, et al. (2007).
Seasonality in bacterial diversity in north-west Mediterranean coastal waters: assessment through
clone libraries, fingerprinting and FISH. FEMS Microbiology Ecology 60:98–112.
Angel MV . (1993). Biodiversity of the Pelagic Ocean. Conservation Biology 7:760–772.
Aota Y , Nakajima H. (2001). Mutualistic relationships between phytoplankton and bacteria
caused by carbon excretion from phytoplankton. Ecol. Res. 16:289–299.
Aristegui J, Gasol JM, Duarte CM, Herndl GJ. (2009). Microbial oceanography of the dark
ocean’s pelagic realm. Limnol. Oceanogr. 54:1501–1529.
Cram
97
Arrigo KR, Robinson DH, Worthen DL, Dunbar RB, DiTullio GR, VanWoert M, et al. (1999).
Phytoplankton Community Structure and the Drawdown of Nutrients and CO2 in the Southern
Ocean. Science 283:365–367.
Azam F. (1998). Microbial Control of Oceanic Carbon Flux: The Plot Thickens. Science 280:694
–696.
Badcock J, Merrett NR. (1976). Midwater fishes in the eastern North Atlantic—I. Vertical
distribution and associated biology in 30°N, 23°W, with developmental notes on certain
myctophids. Progress in Oceanography 7:3–58.
Barton AD, Dutkiewicz S, Flierl G, Bragg J, Follows MJ. (2010). Patterns of Diversity in Marine
Phytoplankton. Science 327:1509–1511.
Behrenfeld MJ, Falkowski PG. (1997). A Consumer’s Guide to Phytoplankton Primary
Productivity Models. Limnol. Oceanogr. 42:1479–1491.
Beman JM, Leilei Shih J, Popp BN. (2013). Nitrite oxidation in the upper water column and
oxygen minimum zone of the eastern tropical North Pacific Ocean. ISME J 7:2192–2205.
Beman JM, Sachdeva R, Fuhrman JA. (2010). Population ecology of nitrifying Archaea and
Bacteria in the Southern California Bight. Environmental Microbiology 12:1282–1292.
Berelson WM. (1991). The Flushing of Two Deep-Sea Basins, Southern California Borderland.
Limnol. Oceanogr. 36:1150–1166.
Bissett W, Meyers M, Walsh J, Mullerkarger F. (1994). The Effects of Temporal Variability of
Mixed-Layer Depth on Primary Productivity Around Bermuda. J. Geophys. Res.-Oceans
99:7539–7553.
Brown MV , Schwalbach MS, Hewson I, Fuhrman JA. (2005). Coupling 16S-ITS rDNA clone
libraries and automated ribosomal intergenic spacer analysis to show marine microbial diversity:
development and application to a time series. Environ Microbiol 7:1466–1479.
Buesseler KO, Pike S, Maiti K, Lamborg CH, Siegel DA, Trull TW. (2009). Thorium-234 as a
tracer of spatial, temporal and vertical variability in particle flux in the North Pacific. Deep Sea
Research Part I: Oceanographic Research Papers 56:1143–1167.
Canfield DE, Stewart FJ, Thamdrup B, Brabandere LD, Dalsgaard T, Delong EF, et al. (2010). A
Cryptic Sulfur Cycle in Oxygen-Minimum–Zone Waters off the Chilean Coast. Science
330:1375–1378.
Carlson C, Morris R, Parsons R, Treusch A, Giovannoni S, Vergin K. (2009). Seasonal dynamics
of SAR11 populations in the euphotic and mesopelagic zones of the northwestern Sargasso Sea.
ISME J 3:283–295.
Caron DA. (2000). Symbiosis and mixotrophy among pelagic microorganisms. In:Microbial
Ecology of the Oceans, Kirchman, DL, ed (ed)., Wiley-Liss: New York, NY , pp. 495–523.
Cram
98
Casamayor EO, Massana R, Benlloch S, Ovreas L, Diez B, Goddard VJ, et al. (2002). Changes
in archaeal, bacterial and eukaryal assemblages along a salinity gradient by comparison of
genetic fingerprinting methods in a multipond solar saltern. Environ Microbiol 4:338–348.
Chave J. (2004). Neutral theory and community ecology. Ecology Letters 7:241–253.
Chow C-ET, Kim DY , Sachdeva R, Caron DA, Fuhrman JA. (2014). Top-down controls on
bacterial community structure: microbial network analysis of bacteria, T4-like viruses and
protists. ISME J 8:816–829.
Chow C-ET, Sachdeva R, Cram JA, Steele JA, Needham DM, Patel A, et al. (2013). Temporal
variability and coherence of euphotic zone bacterial communities over a decade in the Southern
California Bight. ISME J 7:2259–2273.
Collins LE, Berelson W, Hammond DE, Knapp A, Schwartz R, Capone D. (2011). Particle fluxes
in San Pedro Basin, California: A four-year record of sedimentation and physical forcing. Deep
Sea Research Part I: Oceanographic Research Papers 58:898–914.
Crump BC, Hopkinson CS, Sogin ML, Hobbie JE. (2004). Microbial Biogeography along an
Estuarine Salinity Gradient: Combined Influences of Bacterial Growth and Residence Time.
Appl. Environ. Microbiol. 70:1494–1505.
Dabney A, Storey JD, qvalue Tutorial PRS, Install R, Packages IB, Packages UB, et al. (2004).
Q-value estimation for false discovery rate control. Medicine 344:539–548.
DeLong EF, Preston CM, Mincer T, Rich V , Hallam SJ, Frigaard N-U, et al. (2006). Community
Genomics Among Stratified Microbial Assemblages in the Ocean’s Interior. Science 311:496–
503.
DeSantis TZ, Hugenholtz P, Larsen N, Rojas M, Brodie EL, Keller K, et al. (2006). Greengenes,
a chimera-checked 16S rRNA gene database and workbench compatible with ARB. Appl.
Environ. Microbiol. 72:5069–5072.
Deutsch C, Sarmiento JL, Sigman DM, Gruber N, Dunne JP. (2007). Spatial coupling of nitrogen
inputs and losses in the ocean. Nature 445:163–167.
Drake DE. (1974). Distribution and Transport of Suspended Particulate Matter in Submarine
Canyons Off Southern California. In:Suspended Solids in Water, Gibbs, RJ, ed (ed). Marine
Science, Springer US, pp. 133–153. http://link.springer.com/chapter/10.1007/978-1-4684-8529-
5_9 (Accessed November 26, 2013).
Drake DE, Gorsline DS. (1973). Distribution and Transport of Suspended Particulate Matter in
Hueneme, Redondo, Newport, and La Jolla Submarine Canyons, California. Geological Society
of America Bulletin 84:3949–3968.
Ducet N, Le Traon PY , Reverdin G. (2000). Global high-resolution mapping of ocean circulation
from TOPEX/Poseidon and ERS-1 and -2. J. Geophys. Res. 105:19477–19498.
Cram
99
Eiler A, Hayakawa DH, Rappé MS. (2011). Non-random assembly of bacterioplankton
communities in the subtropical north pacific ocean. Frontiers in microbiology 2:140.
Eslinger DL, Cooney RT, McRoy CP, Ward A, Kline TC, Simpson EP, et al. (2001). Plankton
dynamics: observed and modelled responses to physical conditions in Prince William Sound,
Alaska. Fish Oceanogr. 10:81–96.
Ferguson CA, Carvalho L, Scott EM, Bowman AW, Kirika A. (2008). Assessing ecological
responses to environmental change using statistical models. Journal of Applied Ecology 45:193–
203.
Fortunato CS, Herfort L, Zuber P, Baptista AM, Crump BC. (2011). Spatial variability
overwhelms seasonal patterns in bacterioplankton communities across a river to ocean gradient.
The ISME journal 6:554–563.
Fouilland E, Tolosa I, Bonnet D, Bouvier C, Bouvier T, Bouvy M, et al. (2014). Bacterial carbon
dependence on freshly produced phytoplankton exudates under different nutrient availability and
grazing pressure conditions in coastal marine waters. FEMS Microbiol Ecol 87:757–769.
Francois R, Honjo S, Krishfield R, Manganini S. (2002). Factors controlling the flux of organic
carbon to the bathypelagic zone of the ocean. Global Biogeochemical Cycles 16:34–1–34–20.
Frouin R, Franz BA, Werdell PJ. (2003). The SeaWiFS PAR product. Algorithm updates for the
fourth SeaWiFS data reprocessing 46–50.
Fu G, Baith K, McClain C. (1998). SeaDAS: The SeaWiFS data analysis system. In:Proceedings
of the 4th Pacific Ocean Remote Sensing Conference, Qingdao, China, pp. 28–31.
Fuessel J, Lam P, Lavik G, Jensen MM, Holtappels M, Guenter M, et al. (2012). Nitrite oxidation
in the Namibian oxygen minimum zone. ISME J 6:1200–1209.
Fuhrman J, Steele J. (2008). Community structure of marine bacterioplankton: patterns,
networks, and relationships to function. Aquat Microb Ecol 53:69–81.
Fuhrman JA, Comeau DE, Hagström Å, Chan AM. (1988). Extraction from Natural Planktonic
Microorganisms of DNA Suitable for Molecular Biological Studies. Appl Environ Microbiol
54:1426–1429.
Fuhrman JA, Hagstrom A. (2008). Bacterial and archaeal community structure and its patterns.
In:Microbial Ecology of the Oceans, Wiley: Hoboken, N.J., pp. 45–90.
Fuhrman JA, Hewson I, Schwalbach MS, Steele JA, Brown MV , Naeem S. (2006). Annually
reoccurring bacterial communities are predictable from ocean conditions. Proceedings of the
National Academy of Sciences 103:13104 –13109.
Galand PE, Casamayor EO, Kirchman DL, Potvin M, Lovejoy C. (2009a). Unique archaeal
assemblages in the Arctic Ocean unveiled by massively parallel tag sequencing. ISME J 3:860–
869.
Cram
100
Galand PE, Lovejoy C, Hamilton AK, Ingram RG, Pedneault E, Carmack EC. (2009b). Archaeal
diversity and a gene for ammonia oxidation are coupled to oceanic circulation. Environmental
Microbiology 11:971–980.
Garcia-Martinez J, Rodriguez-Valera F. (2000). Microdiversity of uncultured marine
prokaryotes: the SAR11 cluster and the marine Archaea of Group I. Mol Ecol 9:935–948.
Gilbert JA, Steele JA, Caporaso JG, Steinbrück L, Reeder J, Temperton B, et al. (2012). Defining
seasonal marine microbial community dynamics. ISME J 6:298–308.
Giovannoni SJ, Vergin KL. (2012). Seasonality in Ocean Microbial Communities. Science
335:671–676.
Goslee SC, Urban DL. (2007). The ecodist package for dissimilarity-based analysis of ecological
data. Journal of Statistical Software 22:1–19.
Hamersley MR, Turk KA, Leinweber A, Gruber N, Zehr JP, Gunderson T, et al. (2011). Nitrogen
fixation within the water column associated with two hypoxic basins in the Southern California
Bight. Aquat Microb Ecol 63:193–205.
Hamilton AK, Lovejoy C, Galand PE, Ingram RG. (2008). Water masses and biogeography of
picoeukaryote assemblages in a cold hydrographically complex system. Limnol. Oceanogr.
53:922–935.
Hansen, J, Kirboe T, Alldredge A. (1996). Marine snow derived from abandoned larvacean
houses: sinking rates, particle content and mechanisms of aggregate formation. Mar Ecol Prog
Ser 141:205–215.
Hatosy SM, Martiny JBH, Sachdeva R, Steele J, Fuhrman JA, Martiny AC. (2013). Beta
diversity of marine bacteria depends on temporal scale. Ecology 94:1898–1904.
Herndl GJ, Reinthaler T. (2013). Microbial control of the dark end of the biological pump.
Nature Geosci 6:718–724.
Hewson I, Steele J, Capone D, Fuhrman J. (2006a). Remarkable heterogeneity in meso- and
bathypelagic bacterioplankton assemblage composition. Limnol. Oceanogr. 51:1274–1283.
Hewson I, Steele JA, Capone DG, Fuhrman JA. (2006b). Temporal and spatial scales of variation
in bacterioplankton assemblages of oligotrophic surface waters. Marine Ecology-Progress Series
311:67–77.
Hickey BM. (1992). Circulation over the Santa Monica-San Pedro Basin and Shelf. Progress In
Oceanography 30:37–115.
Hickey BM. (1991). Variability in two deep coastal basins (Santa Monica and San Pedro) off
southern California. Journal of Geophysical Research: Oceans 96:16689–16708.
Cram
101
Honaker J, King G. (2010). What to do about missing values in time-series cross-section data.
American Journal of Political Science 54:561–581.
Honaker J, King G, Blackwell M. (2006). Amelia II: A program for missing data.
http://www.icesi.edu.co/CRAN/web/packages/Amelia/vignettes/amelia.pdf (Accessed July 8,
2013).
Hooper DU, Chapin FS III, Ewel JJ, Hector A, Inchausti P, Lavorel S, et al. (2005). Effects of
Biodiversity on Ecosystem Functioning: A Consensus of Current Knowledge. Ecological
Monographs 75:3–35.
Kan J, Suzuki MT, Wang K, Evans SE, Chen F. (2007). High Temporal but Low Spatial
Heterogeneity of Bacterioplankton in the Chesapeake Bay. Appl. Environ. Microbiol. 73:6776–
6789.
King G, Honaker J, Joseph A, Scheve K. (2001). Analyzing Incomplete Political Science Data:
An Alternative Algorithm for Multiple Imputation. The American Political Science Review
95:49–69.
Kirchman D, K’nees E, Hodson R. (1985). Leucine incorporation and its potential as a measure
of protein synthesis by bacteria in natural aquatic systems. Appl Environ Microbiol 49:599–607.
Kirchman DL. (2002). The ecology of Cytophaga–Flavobacteria in aquatic environments. FEMS
Microbiology Ecology 39:91–100.
Ladau J, Sharpton TJ, Finucane MM, Jospin G, Kembel SW, O’Dwyer J, et al. (2013). Global
marine bacterial diversity peaks at high latitudes in winter. ISME J 7:1669–1677.
Lehodey P, Alheit J, Barange M, Baumgartner T, Beaugrand G, Drinkwater K, et al. (2006).
Climate Variability, Fish, and Fisheries. Journal of Climate 19:5009–5030.
Lindsay DJ, Hunt JC. (2005). Biodiversity in midwater cnidarians and ctenophores: submersible-
based results from deep-water bays in the Japan Sea and north-western Pacific. J. Mar. Biol.
Assoc. U.K. 85:503–517.
Long RA, Azam F. (2001). Microscale patchiness of bacterioplankton assemblage richness in
seawater. Aquat Microb Ecol 26:103–113.
Maidak BL, Cole JR, Lilburn TG, Jr CTP, Saxman PR, Farris RJ, et al. (2001). The RDP-II
(Ribosomal Database Project). Nucl. Acids Res. 29:173–174.
Mannino A, Russ ME, Hooker SB. (2008). Algorithm development and validation for satellite-
derived distributions of DOC and CDOM in the U.S. Middle Atlantic Bight. Journal of
Geophysical Research: Oceans 113:n/a–n/a.
Marinovic BB, Croll DA, Gong N, Benson SR, Chavez FP. (2002). Effects of the 1997-1999 El
Nino and La Nina events on zooplankton abundance and euphausiid community composition
within the Monterey Bay coastal upwelling system. Prog. Oceanogr. 54:265–277.
Cram
102
McGrady-Steed J, Harris PM, Morin PJ. (1997). Biodiversity regulates ecosystem predictability.
Nature 390:162–165.
Mincer TJ, Church MJ, Taylor LT, Preston C, Karl DM, DeLong EF. (2007). Quantitative
distribution of presumptive archaeal and bacterial nitrifiers in Monterey Bay and the North
Pacific Subtropical Gyre. Environmental Microbiology 9:1162–1175.
Morel A, Gentili B. (2009). A simple band ratio technique to quantify the colored dissolved and
detrital organic material from ocean color remotely sensed data. Remote Sensing of Environment
113:998–1011.
Morris RM, Vergin KL, Cho JC, Rappe MS, Carlson CA, Giovannoni SJ. (2005). Temporal and
spatial response of bacterioplankton lineages to annual convective overturn at the Bermuda
Atlantic Time-series Study site. Limnol. Oceanogr. 50:1687–1696.
Needham DM, Chow C-ET, Cram JA, Sachdeva R, Parada A, Fuhrman JA. (2013). Short-term
observations of marine bacterial and viral communities: patterns, connections and resilience.
ISME J 7:1274–1285.
Nelson JD, Boehme SE, Reimers CE, Sherrell RM, Kerkhof LJ. (2008). Temporal patterns of
microbial community structure in the Mid-Atlantic Bight. FEMS Microbiology Ecology 65:484–
493.
Noble RT, Fuhrman JA. (1998). Use of SYBR Green I for rapid epifluorescence counts of marine
viruses and bacteria. Aquat Microb Ecol 14:113–118.
Obernosterer I, Herndl GJ. (1995). Phytoplankton extracellular release and bacterial growth:
dependence on the inorganic N: P ratio. Marine Ecology Progress Series 116:247–257.
Oksanen J, Blanchet FG, Kindt R, Legendre P, Minchin PR, O’Hara RB, et al. (2013). vegan:
Community Ecology Package. http://CRAN.R-project.org/package=vegan.
Oksanen J, Blanchet FG, Kindt R, Legendre P, O’Hara RB, Simpson GL, et al. (2011). vegan:
Community Ecology Package. http://CRAN.R-project.org/package=vegan.
Parsons TR. (1984). A manual of chemical and biological methods for seawater analysis. 1st ed.
Pergamon Press: Oxford [Oxfordshire] ; New York.
Partensky F, Hess WR, Vaulot D. (1999). Prochlorococcus, a Marine Photosynthetic Prokaryote
of Global Significance. Microbiol. Mol. Biol. Rev. 63:106–127.
Patel A, Noble RT, Steele JA, Schwalbach MS, Hewson I, Fuhrman JA. (2007). Virus and
prokaryote enumeration from planktonic aquatic environments by epifluorescence microscopy
with SYBR Green I. Nat Protoc 2:269–276.
Pinhassi J, Winding A, Binnerup SJ, Zweifel UL, Riemann B, Hagstrom A. (2003). Spatial
variability in bacterioplankton community composition at the Skagerrak-Kattegat Front. Marine
Ecology-Progress Series 255:1–13.
Cram
103
Pommier T, CanbäCk B, Riemann L, BoströM KH, Simu K, Lundberg P, et al. (2007). Global
patterns of diversity and community structure in marine bacterioplankton. Mol Ecol 16:867–880.
Quast C, Pruesse E, Yilmaz P, Gerken J, Schweer T, Yarza P, et al. (2012). The SILVA ribosomal
RNA gene database project: improved data processing and web-based tools. Nucleic Acids
Research 41:D590–D596.
Ribalet F, Marchetti A, Hubbard KA, Brown K, Durkin CA, Morales R, et al. (2010). Unveiling
a phytoplankton hotspot at a narrow boundary between coastal and offshore waters. PNAS
107:16571–16576.
Rich VI, Pham VD, Eppley J, Shi Y , DeLong EF. (2011). Time-series analyses of Monterey Bay
coastal microbial picoplankton using a ‘genome proxy’ microarray. Environmental Microbiology
13:116–134.
Robison BH, Reisenbichler KR, Sherlock RE. (2005). Giant Larvacean Houses: Rapid Carbon
Transport to the Deep Sea Floor. Science 308:1609–1611.
Ruan Q, Steele JA, Schwalbach MS, Fuhrman JA, Sun F. (2006). A dynamic programming
algorithm for binning microbial community profiles. Bioinformatics 22:1508–1514.
Rusch DB, Halpern AL, Sutton G, Heidelberg KB, Williamson S, Yooseph S, et al. (2007). The
Sorcerer II Global Ocean Sampling Expedition: Northwest Atlantic through Eastern Tropical
Pacific. Plos Biol 5:e77.
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 Research Part I: Oceanographic Research Papers
58:16–26.
Sholkovitz E, Soutar A. (1975). Changes in Composition of Bottom Water of Santa-Barbara
Basin - Effect. Deep-Sea Research 22:13–&.
Simon M, Grossart H-P, Schweitzer B, Ploug H. (2002). Microbial ecology of organic aggregates
in aquatic ecosystems. Aquat Microb Ecol 28:175–211.
Sohrin R, Isaji M, Obara Y , Agostini S, Suzuki Y , Hiroe Y , et al. (2011). Distribution of
Synechococcus in the dark ocean. Aquat Microb Ecol 64:1–14.
Steele JA, Countway PD, Xia L, Vigil PD, Beman JM, Kim DY , et al. (2011). Marine bacterial,
archaeal and protistan association networks reveal ecological linkages. ISME J 5:1414–1425.
Steinberg DK, Carlson CA, Bates NR, Goldthwait SA, Madin LP, Michaels AF. (2000).
Zooplankton vertical migration and the active transport of dissolved organic and inorganic
carbon in the Sargasso Sea. Deep Sea Research Part I: Oceanographic Research Papers 47:137–
158.
Cram
104
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 Research Part Ii-Topical Studies in Oceanography
48:1405–1447.
Steinberg DK, Goldthwait SA, Hansell DA. (2002). Zooplankton vertical migration and the
active transport of dissolved organic and inorganic nitrogen in the Sargasso Sea. Deep Sea
Research Part I: Oceanographic Research Papers 49:1445–1461.
Sverdrup HU. (1953). On conditions for the vernal blooming of phytoplankton. Journal du
Conseil 18:287–295.
Swan BK, Martinez-Garcia M, Preston CM, Sczyrba A, Woyke T, Lamy D, et al. (2011).
Potential for Chemolithoautotrophy Among Ubiquitous Bacteria Lineages in the Dark Ocean.
Science 333:1296–1300.
Teeling H, Fuchs BM, Becher D, Klockow C, Gardebrecht A, Bennke CM, et al. (2012).
Substrate-Controlled Succession of Marine Bacterioplankton Populations Induced by a
Phytoplankton Bloom. Science 336:608–611.
Treusch AH, Vergin KL, Finlay LA, Donatz MG, Burton RM, Carlson CA, et al. (2009).
Seasonality and vertical structure of microbial communities in an ocean gyre. ISME J 3:1148–
1163.
Vergin KL, Beszteri B, Monier A, Thrash JC, Temperton B, Treusch AH, et al. (2013a). High-
resolution SAR11 ecotype dynamics at the Bermuda Atlantic Time-series Study site by
phylogenetic placement of pyrosequences. ISME J 7:1322–1332.
Vergin KL, Done B, Carlson CA, Giovannoni SJ. (2013b). Spatiotemporal distributions of rare
bacterioplankton populations indicate adaptive strategies in the oligotrophic ocean. Aquat
Microb Ecol 71:1–13.
Weinbauer MG, Liu J, Motegi C, Maier C, Pedrotti ML, Dai M, et al. (2013). Seasonal
variability of microbial respiration and bacterial and archaeal community composition in the
upper twilight zone. Aquat Microb Ecol 71:99–115.
Wilson SE, Steinberg DK. (2010). Autotrophic picoplankton in mesozooplankton guts: evidence
of aggregate feeding in the mesopelagic zone and export of small phytoplankton. Mar Ecol Prog
Ser 412:11–27.
Woebken D, Lam P, Kuypers MMM, Naqvi SWA, Kartal B, Strous M, et al. (2008). A
microdiversity study of anammox bacteria reveals a novel Candidatus Scalindua phylotype in
marine oxygen minimum zones. Environmental Microbiology 10:3106–3119.
Wood SN. (2011). Fast stable restricted maximum likelihood and marginal likelihood estimation
of semiparametric generalized linear models. Journal of the Royal Statistical Society: Series B
(Statistical Methodology) 73:3–36.
Cram
105
Wood SN. (2006). Generalized additive models: an introduction with R. Chapman & Hall/CRC:
Boca Raton, FL.
Wood SN. (2004). Stable and Efficient Multiple Smoothing Parameter Estimation for
Generalized Additive Models. Journal of the American Statistical Association 99:673–686.
Wright JJ, Konwar KM, Hallam SJ. (2012). Microbial ecology of expanding oxygen minimum
zones. Nat Rev Micro 10:381–394.
Wright JJ, Mewis K, Hanson NW, Konwar KM, Maas KR, Hallam SJ. (2013). Genomic
properties of Marine Group A bacteria indicate a role in the marine sulfur cycle. ISME J.
http://www.nature.com/ismej/journal/vaop/ncurrent/abs/ismej2013152a.html (Accessed
September 12, 2013).
Yeo SK, Huggett MJ, Eiler A, Rappé MS. (2013). Coastal Bacterioplankton Community
Dynamics in Response to a Natural Disturbance. PLoS ONE 8:e56207.
Zehr JP, Ward BB. (2002). Nitrogen Cycling in the Ocean: New Perspectives on Processes and
Paradigms. Applied and Environmental Microbiology 68:1015–1024
Cram
106
Chapter 3 : Network analysis of planktonic
marine bacteria throughout the entire water
column
Abstract
Interactions between microbes and stratification across depths are both believed to be important
drivers of microbial communities, though little is known about how these associations differ
between and across depths. We have monitored the free-living microbial community at the San
Pedro Ocean Time Series station (SPOT) monthly for 10 years at five different depths: 5 meters
(m), the chlorophyll maximum layer, 150m, 500m, and 890m. Association network analysis
revealed surprising complexity of time lagged patterns among microorganisms and
environmental parameters throughout the water column. The euphotic zone, DCM and 890m
depth each contain two somewhat mutually exclusive modules of inter-correlated bacteria and
environmental conditions, suggesting mutually exclusive environmental states. We observed
pairs of organisms that co-occurred with zero or one month time lags at different depths, with
two thirds of correlations between depths lagged such that changes in the abundance of deeper
organisms followed changes in shallower organisms. Taken in conjunction with the analysis of
seasonality, these trends suggest that the planktonic microbial communities throughout the water
column are linked to the environmental conditions and/or the microbial communities in
overlying waters. Poorly understood groups including Marine Group A, Nitrospina, and
AEGEAN-169 clades contained OTUs that showed diverse association patterns, suggesting these
groups contain multiple ecological species, each shaped by different factors. These observations
support the hypothesis that sinking particles and vertical migrating animals transport materials
that significantly shape the time-varying patterns of microbial community composition.
Cram
107
Introduction
Microbial communities throughout the water column show long term, seasonal and short term
dynamics that relate to various biological and nonbiological environmental parameters (Cram et
al. 2014b; Gilbert et al. 2012; Giovannoni & Vergin 2012; Hatosy et al. 2013; Needham et al.
2013; Chow et al. 2013). Microbial interactions have been observed experimentally (Jurgens et
al. 1999; Miller & Bassler 2001; Jürgens & Matz 2002; Bonilla-Findji et al. 2009), through
physical attachment (Malfatti et al. 2009; Malfatti & Azam 2009) and inferred from statistical
associations (Steele et al. 2011; Chow et al. 2014), . These interactions appear to be major
drivers of the ecological dynamics of many marine bacterioplankton, and they include predator-
prey, mutualisms, parasitism, and competition (see Strom 2008). Pairwise network association
analysis techniques (reviewed in Faust & Raes 2012; Cram et al. 2014a) have proven a valuable
tool in looking at trends in the many statistical associations in microbial communities in a variety
of environments including lake systems(Eiler et al. 2012; Kara et al. 2012), soil (Zhou et al.
2010; Barberán et al. 2011), the human microbiome (Arumugam et al. 2011; Faust et al. 2012),
globally through meta-analysis across diverse sampling sites (Chaffron et al. 2010; Freilich et al.
2010). In marine surface waters, studies employing pairwise association analysis over time have
suggested that the abundance of particular bacteria tend to be best predicted by the abundance of
other microorganisms, rather than environmental variability (at least of parameters that were
measured) (Fuhrman & Steele 2008; Steele et al. 2011; Needham et al. 2013; Chow et al. 2013,
2014). Network analysis has identified many associations between organisms that are driven by
seasonal variability, especially at locations where seasonality is strong (Gilbert et al. 2012), and
this seasonal pattern has been de-convoluted in lake environments to show different inter-
organismal associations between seasons (Kara et al. 2012). Time scale also appears to be an
Cram
108
important determinant of microbial interactions, with certain closely related groups of organisms,
e.g. the SAR11 cluster, co-occurring more closely at daily time scales than at monthly ones
(Needham et al. 2013). The dynamics of microbial communities throughout an entire deep (900
m) water column were recently described, and one of the most surprising discoveries was that of
seasonality at the bottom of the water column (Cram et al. 2014b). This seasonality at 890m
suggests surface influences on deeper depths by way of sinking particle flux and/or migrating
zooplankton (see Cram et al. 2014b).
Network association analysis provides an opportunity to more closely examine links between the
environments and communities at each depth. In this study, we examined both the interaction
patterns at each depth throughout the water column, and the interactions between different
bacterial operational taxonomic units (OTU) and between bacterial OTUs and environmental
parameters at different depths.
Chow et al. (2013) examined similarities between association networks in the surface and DCM
and observed some associations that were present at both depths, specifically associations
between OTUs from the SAR11 clade and other OTUs. They observed other associations that
differed between depths such as the relationships between temperature, salinity and nutrient
concentrations to the abundances of many OTUs. Previous studies have examined patterns
within surface and chlorophyll maximum depth, and a common feature of these networks in the
presence of modules, highly interconnected groups of organisms (Steele et al. 2011; Chow et al.
2013). Modules in other studies have been shown to represent groups of organisms that have
similar seasonal patterns (Gilbert et al. 2012) that sometimes occur independently of
environmental parameters and may represent organisms that are associated with each other
Cram
109
through symbiosis, common niches, or through other means (Steele et al. 2011; Chow et al.
2013).
Here, we extend these prior analyses of euphotic zone depths to the rest of the water column, and
investigate networks at every depth visually and statistically. Visually, we identify modules or
groups of co-occurring organisms and determine whether they show similar or different patterns
at different depths. Statistically, we quantify whether OTUs show similar or different patterns
between depths though analysis of network topology. Network topology analysis is the
quantitative characterization of overall network properties, such density and clustering. Density
is defined as the number of edges in a network divided by the number of possible edges (edges
that could be in the network if every node was connected to every other possible node) (Deng et
al. 2012) and can be used to quantify how highly connected the nodes are in a given network.
Average clustering describes whether the network associates into clumps of highly
interconnected organisms and higher clustering coefficients represent greater “modularity” in a
network, i.e., more clustered networks show multiple groups of co-occurring organisms (Deng et
al. 2012). Path length, the average number of nodes in a network, is also used to quantify the
structure of networks. Previous ecological networks have shown that bacteria tend to have higher
average clustering coefficients than would be expected for a similarly sized random network
(Steele et al. 2011; Faust et al. 2012; Deng et al. 2012; Chow et al. 2014), but have path lengths
that were not much longer than for random networks. Networks that meet these characteristics
are defined as having “small world” properties, which characterize many networks, including
biological ones (Watts & Strogatz 1998; Humphries & Gurney 2008; Telesford et al. 2011; Deng
et al. 2012).
Cram
110
In lake systems, networks of bacterial interactions in different seasons were all shown to have
small world properties, but clustering coefficients differed between seasons in a way that was
related to biodiversity, with highest clustering and shortest path length in the Spring (the season
of lowest Richness and Shannon Biodiversity Index), and lowest clustering in the Fall (highest
Richness and Biodiversity) (Kara et al. 2012). In this study, we use topological statistics to
determine the emergent properties in networks of communities at each depth with to determine
how patterns of interactions differ across the water column.
Microbial communities throughout the oceans show seasonal patterns, and vary between depths
(Cram et al. 2014b; Treusch et al. 2009; Giovannoni & Vergin 2012). Closely related organisms,
such as different members of the SAR11cluster appear to show different spatio-temporal patterns
(Carlson et al. 2009; Brown et al. 2012) indicating the value of comparing interactions between
related organisms and their surrounding environments. Seasonal variability at SPOT has been
shown to be strongest in surface waters, weaker deeper in the water column and strong again
near the sea-floor. In the middle water column, some species, but not others are seasonal. It has
been suggested that seasonality of some species in the mid water column, and many species near
the bottom of the water column is driven by sinking particles, which originate in surface layers
and migrate to deeper layers where they are alter the ecology (Cram et al. 2014b). We aim to
shed light on how organisms interact with and are shaped by the flux of these particles in
different water column depths by investigating statistical associations between organisms at each
depth and organisms at other water column depths. Particles take time to sink: At SPOT they
have been shown to sink at a mean rate of 83m/day (Collins et al. 2011), which suggests that it
should take about 11 days to travel from the surface to the bottom of the water column. Since
this is an average, it is likely that smaller less dense particles sink more slowly, while denser
Cram
111
larger particles sink more quickly. We would also expect that particles take time to decompose or
be consumed by microorganisms, and it has been suggested that particles take longer to
decompose in deeper waters than in the mixed layer (Kiorboe 2001). As microbes are digesting
something, it takes time for their numbers to respond in a meaningful way. The aims of this
paper are to identify patterns of associations between microbes within and between depths and to
elucidate the ecology of less studied groups of marine organisms, especially those that are
abundant in the deep water column. We hypothesize that relationships between organisms at
different depths are 1) less common than relationships between organisms at the same depth and
2) likely to be time lagged such that surface changes likely precede changes in deeper depths.
Methods
Sample Collection:
Samples were collected monthly from the San Pedro Ocean Time Series between August 2000
and January 2011 from five depths 5m, the Deep Chlorophyll Maximum Layer (between 5m and
40m), 150m, 500m. Starting from 2003 samples from and 890m were collected as well.
Collection, processing and analysis procedures are as described previously (Cram et al. 2014b)
(see also Brown et al. 2005; Chow et al. 2013). We investigated biological and environmental
parameters, the collection of which were described previously (Cram et al. 2014b; Beman et al.
2008), which are summarized in Table 3-3.
Network Analysis:
Networks were generated using the extended local similarity analysis (eLSA) program (Ruan et
al. 2006; Xia et al. 2013). We used eLSA to identify global, time lagged, Spearman correlations
between bacterial and environmental nodes at all depths. Our network analysis takes a slightly
different approach to several previous network analyses at SPOT. Previous analysis have
Cram
112
employed “local” similarity analysis, which identifies time lagged associations as well as “local”
associations, which are those that only occur over a portion of the data set. In this study we were
most interested in relationships that were consistent throughout the study period, so we applied
“global” Spearman correlations in order to look for correlations that are present throughout the
entire data set.
In order to have good cross comparability between depths, and in consideration that not all
depths were collected on every sampling day, we selected only dates for our analysis where
samples had been collected at most depths. For instance, we excluded all samples from before
2003 when 890m was not sampled and from most of 2007 where only surface data were
available. In cases where a few depths were missing, we selected sampling dates such that each
depth was sampled a similar number of times (5m, 64; DCM 62; 150m, 62; 500m, 61; 890m, 62
samples). We also considered only those bacterial OTUs that were present at least 5 times in the
data set. The eLSA metric that we used allowed a possible one month time lag, and interpolated
missing values by filling in with zeros after normalization (analogous to interpolating with
median values). LSA output data was preliminarily filtered by identifying and retaining only
those associations that had P values of less than 1%, (which corresponded to Q values of less
than 1%) and Spearman R values greater than 0.5 or less than -0.5. For each depth we eliminated
bacteria that could not be detected at least 25 times (corresponding with 20% of the sampling
dates) in the data set.
We imported our LSA networks as well as metadata about variables including variable names,
and for bacterial nodes, mean abundance and taxonomic data into Cytoscape 2.8.3 (Smoot et al.
2011). Our network from the data described above will be referred to as the “all vs all” network
because it contains comparisons of all nodes at all depths vs each other depth. In order to focus
Cram
113
on patterns in the bacterial community, we generated a filtered network containing only bacterial
nodes but not environmental parameters from the “all vs all” network; this network will be
referred to henceforth as the “bacterial all vs all” network. From the “bacterial all vs all network”
we also created a child network in which we kept only edges representing positive correlations
and nodes connected by those edges, which we call the “positive bacteria all vs all” network.
From the “bacterial all vs all” network and from the “positive bacterial all vs all” network, we
generated 10 child networks (child networks have all of the properties of the parents networks
but only some nodes and edges are kept), containing bacterial nodes from all possible pairs of
depths. These networks that contain nodes from more than one depth will be referred to as
“between depth networks.” We thus had a total of 20 between depth networks, 10 each from the
“bacterial all vs all” and “positive bacterial all vs all” networks.
In order to look at patterns within depths, we generated five “within depth networks”, each of
which only contained nodes from one depth. For each of those five networks, we also created
secondary child networks contain only bacterial nodes “within depth bacterial networks”, for
statistical purposes. From the within depth bacterial networks we also generated “positive within
depth bacterial networks” containing only edges representing positive correlations and the nodes
connected by those edges.
Network Statistics:
Density:
The “density” parameter tells us how interconnected nodes are within a depth. Density can be
defined as the number of edges in a network divided by the possible number of edges that could
exist in that network (if all nodes were connected to all other nodes) (Deng et al. 2012). For the
all depths bacterial network and all depths positive bacterial network, we determined three edge
Cram
114
density attributes (1) the total density (2) the density of edges that connected pairs of nodes from
the same depth and (3) the density of edges that connected pairs of nodes from different depths.
For each of the within depth bacterial networks and within depth positive bacterial networks we
determined (1) the density of edges connecting nodes within that depth and (2) the density of
edges connecting that depth to nodes at other depths. Then for each between depth network we
determined (1) the density of edges connecting pairs of nodes of at the same depth and (2) the
density of edges connecting pairs of nodes from each of the two depths.
The “density ratio” for each depth is the density of the edges connecting nodes within a depth to
each other divided by the density of edges connecting nodes from that depth to nodes at other
depths. Density ratios for each pair of depths is the density of the edges connecting nodes within
each depth divided by the density of edges connecting nodes from one depth to nodes at the other
depth.
Clustering:
Network analysis used the Network Analyzer Plugin (Assenov et al. 2008; Steele et al. 2011;
Chow et al. 2014) on each of the within depth networks to determine “Clustering” and “Path
Length We also generated random networks (needed for statistical comparisons) of the same
number of nodes and edges as the original networks using the random network plugin in
Cytoscape 2.6.0 (the random network plugin is not compatible with Cytoscape 2.8.3 which was
used for all other analysis). We loaded these random networks back into Cytoscape 2.8.3 and
compared the ratios of clustering coefficients in our networks to those of random networks.
Cram
115
Directionality:
“Directionality” quantifies whether changes in shallower depths generally happened before
correlated changes in deeper depths. We investigated the “bacterial all vs all”, “positive bacterial
all vs all”, all “bacterial between depth networks” and “positive bacterial between depth
networks”. In all cases we examined the edges that connected nodes from different depths and
asked what fraction of those edges were time lagged. Of the time lagged edges, we asked what
fraction of those edges were time lagged such that the changes in the node that was at the
shallower depth in the water column preceded correlated changes in the node that was deeper in
the water column. These edges were referred to as “downhill” lagged. Conversely, “uphill”
lagged edges are those in which changes in the deeper node lead changes in the shallower node.
Hub and Spoke Networks:
We created hub and spoke networks (as seen in Fuhrman & Steele 2008; Steele et al. 2011;
Chow et al. 2014) around nodes from particular taxonomic groups. To generate these networks
from the “all v all” network, we first filtered this network to include only bacterial nodes that
occurred at least 36 times in the dataset and had a mean relative abundance of at least 1%. We
then selected OTUs of particular interest and created sub-networks from those nodes and their
nearest neighbors and immediate adjoining edges.
Results
Association networks identified modular structures in networks in the euphotic zone and bottom
of the water column (Figure 3-1, Supplemental Figure 3-1, Supplemental Figure 3-2) as well as a
prevalence “downhill” of time lagged interactions between nodes and parameters at different
depth, where changes in surface environments preceded changes in deeper environments (Figure
3-3, Figure 3-4, Supplemental Figure 3-3, Supplemental Figure 3-4, Supplemental Figure 3-5).
Cram
116
Patterns within individual depths:
Networks examining correlations (Spearman R > 0.5) between bacterial OTUs at each depth
(Supplemental Figure 3-1, Supplemental Figure 3-2) as well as networks showing correlations
among bacterial OTUs and environmental parameters (Figure 3-1) showed different overall
association patterns between depths. When we looked only at connections between bacterial
OTUs, several patterns were immediately apparent. At 890m, if we examined only positive
associations between bacteria, the network formed two “modules” or highly interconnected
groups of organisms (Figure 3-1E, Supplemental Figure 3-1E). If we also examined negative
associations, it was clear that there were many negative correlations between these modules
(Figure 3-1E, Supplemental Figure 3-2E) suggesting there were two groups of organisms, that
were often seen at the same time. The networks at 5m and the DCM showed two large modules,
but neither of these was as highly interconnected as the network at 890m. In all cases one module
had many interconnected nodes while the other had a few nodes that were connected in a more
diffuse (most nodes only connected to one or two other nodes) pattern. Both of these networks
had a number of pairs of organisms that appeared to associate statistically with each other but not
other OTUs. As at 890m, adding negative correlations to these networks showed that nodes in
both of the main two modules had negative associations with nodes in the other module. Nodes
at 150m and 500m, in contrast, appeared to lack a discernable structure, both when negative
correlations were included and when they were ignored.
Topological analysis of the bacterial networks showed that 890m had the highest edge density,
the 5m, the DCM and 890m had higher clustering than intermediate depths, and that path length
of all depths was similar to those of random networks (Table 3-1, Supplemental Table 3-1).
Cram
117
Density:
890m has a high edge density, relative to the other depths (Table 3-1), meaning that, for the
number of nodes present in this network, there was a higher probability of any two nodes being
correlated than at other depths. There were just as many bacteria at 890m that occurred at least
25 times as there were at other depths, meaning that while more than half of the bacteria detected
at 890m appeared to associate with several other OTUs, just under half of the bacteria at this
depth did not correlate in abundance with any other OTU at 890m. The other depths all had
similar edge densities suggesting that any pair of nodes was about equally likely to be connected
at one of these depths at any other. Edge density was lowest at 500m. When only positive
associations between nodes were considered (Supplemental Table 3-1) similar patterns emerged.
Clustering:
At all depths, networks were more clustered than would be expected in similar sized random
networks. DCM and 890m have higher clustering coefficients than other depths. However only
the DCM and not the 890m depth has a high clustering coefficient ratio (clustering coefficient
divided by the coefficient of a similarly sized random network) and clustering log response ratio.
This is because, due to its high edge density a random network with the same number of nodes
and edges as the 890m network would also be more clustered than other networks. When only
positive associations were considered, both the DCM and 890m had higher clustering
coefficients and clustering log response ratios, suggesting that the differences in clustering
patterns between these two depths were driven primarily by the negative associations.
Path Length:
Mean path lengths between nodes at 150m, 500m and 890m were slightly longer than in
randomly generated networks. In contrast path lengths in the 5m networks were shorter. In the
Cram
118
DCM networks path lengths were about the same when both positive and negative associations
were considered and were shorter when only positive associations were considered).
Environmental Parameters in Within Depth Network
Networks at each depth that also included environmental parameters (Figure 3-1) generally
showed that most OTUs associated statistically with other OTUs but not any of the
environmental parameters that we measured. Many environmental parameters, predictably,
associated with each other. Usually there were a few OTU nodes among highly interconnected
modules that associated with environmental parameters, and we suggest that these associations
generally suggest loose relationships between that module and the noted environmental
parameters. At 5m, DCM and 890m two modules were evident, in all cases one of these modules
contained several nodes that corresponded to increasing day length (DDL) and high levels of
surface chlorophyll, conditions indicating that that community was most abundant in the spring.
At each depth, we refer to this module as “Mod I”. The other module, which we call Mod II, was
comprised of OTUs that associated negatively with the modules in Mod I.
At 5m and DCM, Mod II had more nodes than Mod I. In the DCM several nodes in Mod II
appeared to associate positively with the abundance of the amoA gene (Archaeal ammonia mono
oxygenase), this variable was not measured at the other depths. At 5m ModII appeared to divide
into two sub modules, one of which had a number of OTUs that correlated positively with the
bacterial diversity measures Inverse Simpson Index and Pielou’s evenness index. At 890m Mod I
appeared to have more nodes than MOD II, in contrast to 5m and the DCM. Many nodes in Mod
I were associated with high Inverse Simpson, Richness and Pelou’s evenness, and associated
with a one month lag with the concentration of nitrite and nitrate.
Cram
119
Figure 3-1 Association network showing statistically significant, time lagged and non time-lagged
correlations between bacterial and environmental nodes at each depth (A, 5m; B, DCM; C, 150m; D,
500m; E, 890m). Nodes represent bacterial OTUs (circles) and environmental parameters (squares) at
Cram
120
each depth. Edges (lines) represent correlations, sometimes time lagged, between the bacterial OTUs.
Shown are bacteria that occur at least 25 times and edges that have lagged Spearman correlations such
that |R|>0.55, P<0.01, Q<0.05. Solid edges represent positive correlations, dashed edges represent
negative correlations. Arrows indicate time lagged associations. Node identities are indicated in Table
3-3. Modules, clusters of highly connected nodes, are circled. In all cases, Mod I corresponds to the
module with nodes connected to high surface Chlorophyll_A and increasing daylight (spring bloom) and
Mod II corresponds to nodes that are correlated positively to each other and negatively to nodes in Mod I.
5m DCM 150m 500m 890m AllvAll
Eligible Nodes 110 102 113 106 106 537
Nodes (N) 73 77 83 82 57 463
Edges (E) 112 112 141 124 154 2301
Cl 0.17 0.29 0.15 0.15 0.27 0.13
L 3.63 4.30 4.19 4.51 2.96 3.16
Cl_r 0.02 0.01 0.03 0.04 0.13 0.02
L_r 3.85 4.10 3.63 3.92 2.59 2.92
Cl/Cl_R 7.0 29.1 5.4 3.4 2.0 6.3
I_Cl 1.92 3.28 1.66 1.20 0.70 1.80
I_L -0.06 0.05 0.14 0.14 0.13 0.08
Density 4.3% 3.8% 4.1% 3.7% 9.6% 2.2%
Outgoing Density 1.6% 2.1% 1.7% 1.8% 2.8% NA
Density Ratio 2.71 1.82 2.47 2.12 3.43 NA
Table 3-1 Topological statistics for networks of bacteria at each depth (Figure 3-1), and for a network of
OTUs at all depths (Figure 3-2). This table complements Supplemental Figure 3-2 which visually depicts
the network described here. These networks include only nodes for bacteria that are present at >0.01%
abundance greater than 25 times (Eligible Nodes) and edges that have a possibly time lagged, global,
absolute Spearman R value of greater than 0.5 or less than -0.5. Nodes are the bacterial OTUs that are
connected by at least one edge to another node. Edges are the number of correlations between bacterial
OTUs. Density is the number of edges (E) divided by the number of possible edges {N*(N-1)/2} such
that {Density = E/(N*(N-1)/2)}. Cl is the clustering coefficient for the network. L is the mean path length
for the network. Cl_r and L_r are the clustering coefficients and path lengths for equivalently sized (same
number of nodes and edges) randomly distributed networks. Cl/Cl_R is the ratio of the clustering
coefficient to the random network's clustering coefficient. ICl and IL are the log response ratios of the
clustering and path length coefficients {I_Cl = log(Cl) -log(Cl_r)}. It is apparent that 890m has higher
density than the other depths. Clustering coefficient relative to random networks (Cl/Clr and ICl) is
highest in the DCM. 890m, like the DCM, has a high level of absolute clustering, which reflects as
several highly connected groups in Supplemental Figure 3-1E.
Cram
121
Patterns between OTUs at different depths:
Networks that examined multiple depths simultaneously indicated the presence of many
statistical connections between nodes from different depths. In many cases, nodes, especially
OTUs from different depths, were correlated in a time lagged fashion. An example is an OTU
from the Actinobacterial OCS115 clade with ITS length of 435.5 in the DCM which “led” an
OTU of the Nitrospina genus with ITS length of 729.4 at 890m. By led, we mean that the
OCS115’s abundance either increased or decreased and then a month later, the Nitrospina
abundance would often change in a similar direction (Figure 3-2).
The “all vs all bacterial network” (Figure 3-3), shows pairwise associations between bacteria at
every depth and bacteria at that same depth and every other depth. These associations include
positive and negative, time lagged and unlagged correlations. In this network, more than half of
the correlations between bacteria at different depths are time lagged by one month such that
changes in the OTU in the shallower depth leads changes in the abundance of OTU in the deeper
by one month (Figure 3-3, Table 3-2). Some pairs of depths had particularly high fractions of
downhill lagged edges (e.g. DCM and 890 and150m and 890m both had 76% of their edges
downhill lagged) while the surface and DCM had a high fraction of unlagged edges.
Significantly, the overwhelming majority of the lagged connections between depths were
“downhill.”At all depths, nodes were about twice as likely to correlate with another node at that
same depth (internal density) as they were to connect to a node at another depth (outgoing
density), relative to the number of possible connections that could occur within or between
depths (Table 3-2). Compared to other depths, nodes at 890m had the highest density of
connections both to other nodes at 890m as well as to nodes at other depths. Meanwhile, both 5m
and 890m depths had high internal edge density relative to outgoing edge density, while the
Cram
122
DCM, 150m and especially 500m depths had the lower edge densities relative to internal edge
density (Table 3-1, Supplemental Table 3-1).
Figure 3-2 Association network showing significant, time lagged and non time-lagged correlation
between bacterial nodes both within and between depths. Nodes (circles) represent bacterial
OTUs at each depth. Edges (lines) represent correlations, sometimes time lagged, between the
bacterial OTUs. Shown are bacteria that occur at least 25 times and edges that have lagged
Spearman correlations such that |R|>0.5, P<0.01, Q<0.05. Black edges are instantaneous
Cram
123
correlations between depth as well as correlations (both time lagged and not) that that connect
nodes from the same depth. Blue edges represent “downward” time lagged correlations in which
the two nodes are from different depths and changes in the shallower node precede changes in
the deeper node by one month. Green edges represent “upward” time-lagged correlations such
that changes in the deeper node precede changes in the shallower node. Large arrows show
numbers of edges connecting depths (including both positive and negative correlations) that are
not time-lagged (center, black), upward lagged (top, green) and downward lagged (bottom, blue).
The legend applies to Figure 3-2, Figure 3-3, and Figure 3-4 and describes the meanings
assigned to node and edge characteristics.
Variable/
Formula AllvAll
5m-
DCM
5m-
150m
5m-
500m
5m-
890m
DCM-
150m
DCM-
500m
DCM-
890m
150m-
500m
150m-
890m
500m-
890m
Nodes N 463 159 163 162 157 182 179 173 182 179 179
Edges E 2301 489 466 525 390 436 396 327 391 396 441
Within Depth Nodes Ni 342 128 129 125 122 149 145 142 146 143 139
Within Depth Edges Ei 445 189 193 178 179 186 171 172 175 176 161
Cross Depth Edges Eo 894 122 81 136 67 75 82 53 64 83 131
Within Depth Density Ei/Ei_Max 3.8% 4.6% 4.6% 4.5% 4.8% 3.4% 3.3% 3.5% 3.3% 3.5% 3.4%
Cross Depth Density Eo/Eo_max 1.1% 1.4% 0.9% 1.6% 1.0% 0.8% 0.9% 0.8% 0.7% 1.2% 2.1%
Density Ratio
(Ei/EiM)/
(Eo/EoM) 3.38 3.38 5.16 2.77 4.62 4.31 3.47 4.36 4.53 2.84 1.58
Uphill U
83
(5.0%)
7
(3.2%)
10
(7.0%)
8
(8.8%)
12
(9.7%)
4
(2.9%)
3
(2.3%)
11
(4.2%)
6
(3.5%)
3
(1.8%)
19
(9.0%)
Unlagged F
478
(28.8%)
104
(47.9%)
33
(23.1%)
23
(25.3%)
29
(23.4%)
40
(29.0%)
28
(21.1%)
60
(23.2%)
35
(20.5%)
57
(33.3%)
69
(32.7%)
Downhill D
1097
(66.2%)
106
(48.8%)
100
(69.9%)
60
(65.9%)
83
(66.9%)
94
(68.1%)
102
(76.7%)
188
(72.6%)
130
(76.0%)
111
(64.9%)
123
(58.3%)
Table 3-2 Summary Statistics for networks that examine cross correlations between bacteria at different
depths. The AllvAll column reflects the network shown in Figure 3-2. The other columns reflect the
nodes within and connecting pairs of depths. Nodes (N) are the number of bacterial OTUs at a specific
depth that occur at least 25 times in the data set connected by at least one edge (correlation) in the
specified network. Edges (E) are the total number of Spearman correlations with |R| > 0.5 in that
network. Within Depth Nodes (Ni) are the nodes in these networks that are connected by at least one edge
to another node at that same depth. Within depth edges (Ei) are those edges that connect pairs of nodes
from the same depth. Cross Depth Edges (Eo) are those edges that connect two nodes from different
depths. Within Depth Density, reflects the probability that a pair of nodes from the same depth are
connected by an edge; it is the quotient of the number of edges found connecting nodes that are from the
same depth, divided by the maximum number of edges that could possibly connect nodes of the same
depth in that network. Cross Depth Density (Eo/Eo_max) reflects the probability that two nodes from
different depths are connected by an edge; it is the total number of edges connecting nodes from different
depths, compared to the number of possible edges that could connect nodes from those depths. The
density ratio (ei/eim)/(eo/eom), reflects how much likelier two nodes are to be connected if they are from
the same depth, than if they are from different depths; it is the quotient of the Within Depth Density and
the Cross Depth Density of a network. Uphill (U) is the number of time lagged edges that connect pairs of
nodes from different depths such that changes in the node at the deeper depth precede changes in the done
at the shallower depth. Unlagged (F) is the number of non time-lagged edges that connect pairs of nodes
Cram
124
from different depths. Downhill is the number of time lagged edges that connect pairs of odes from
different depths such that the changes in the shallower node precede correlated changes in the deeper
node.
Example Networks:
Hub and spoke networks, focusing on particular bacterial taxonomic groups reflected the trends
seen in Figure 3-3, but illustrated that different related OTUs appeared to associate with different
factors. As an example of how these figures suggest interdepth associations, Figure 3-3 shows
the four OTUs, from the Deltaproteobacteria class, that were found at 890m, which were present
in the dataset at least 36 times, and associated statistically with other nodes. It is apparent that the
Nitrospina OTUs with ITS length 650.2 associates both positively and negatively, with a time
lag of one month, with a number of OTUs and parameters from 5m, the DCM and 890m. This
Nitrospina also correlates without lag to a number of OTUs and parameters at 500m and 890m.
In contrast, a SAR324 OTU (770.5bp) associates only with two OTU from the DCM and not
from other depths. Two other Nitrospina OTUs associate each with a number of parameters, only
some of which they share with the more connected Nitrospina 650.2 OTU. It is evident, when we
also consider Deltaproteobacteria from 500m and 150m, that these OTUs show different patterns
at different depths and that there is a general pattern where changes in various parameters in
shallower depths precede changes in deeper depths, cascading through the water column
(Supplemental Figure 3-3).
A similar pattern can be seen for the Marine Group A OTUs and the parameters and OTUs that
associate with the (Figure 3-4). For instance, the cascading pattern of changes at shallower
depths leading changes at deeper depths is apparent, as are differences in associations between
different OTUs in the same depth and the same OTUs at different depths. For instance,
MGA_653.2 correlates to parameters at 890m and 500m both when it is investigated at 890m or
Cram
125
500m, but those parameters are different depending on where it is evaluated. For instance it
responds with a time delay to the abundance of phosphate and nitrite at 500m at 890m, but
relates negatively and with delay to nitrite when evaluated at 500m. It correlates more with
parameters from 150m and above when evaluated at 150m.
Similar patterns can be seen among the different OTUs of the AEGEAN-169 and SAR11 Surface
1 clades throughout the water column (Supplemental Figure 3-4, Supplemental Figure 3-5).
Figure 3-3 Association network of correlations between Deltaproteobacteria OTUs at 890m and other
bacteria and environmental parameters found at every depth throughout the water column. All bacteria
shown have a mean abundance of greater than 1% and are found with a relative abundance of greater than
0.01% in more than 36 samples. Lines represent statistically significant, potentially time lagged
correlations (Spearman |R| > 0.57, P<0.01, Q < 0.05). Nodes represent individual measurements of
bacterial abundances (circles) or environmental parameters (squares). Abbreviated names are followed by
OTU fragment size. Node Sizes of bacterial nodes represent the average abundance of OTUs. Solid lines
show a positive correlation, dashed lines show a negative correlation, arrows indicate a 1-month shift in
Cram
126
the correlation. Abbreviations for the nodes are translated in Table 3-3. Edge labels represent Spearman R
values. Edge colors are the same as in Figure 3-2. Legends are present in Figure 3-1 and Figure 3-3.
Figure 3-4 Association network of correlations between bacteria from the Marine Group A phylum at
150m, 500m and 890m and other bacteria and environmental parameters found at every depth throughout
the water column. Nodes and edges have the same thresholds (P, Q, and R values for Edges, mean
abundance and occurrence thresholds for bacterial nodes) as in Figure 3-3.
nodeType Code Parameter Taxonomy Note
ARISA AEG169 AEGEAN-169 SAR11 Clustered
ARISA Altero Alteromonas Gammaprotobacteria
ARISA Bacter Bacteriovoraceae Deltaproteobacteria
ARISA Plastid Chloroplast
ARISA Cronob Cronobacter Gammaproteobacteria
ARISA Cyanob Cyanobacteria
ARISA E01-9C E01-9C-26 Gammaprotobacteria
ARISA Flavob Flavobacteriaceae
Cram
127
nodeType Code Parameter Taxonomy Note
ARISA Fluvii Fluviicola Flavobacteria
ARISA Formos Formosa Flavobacteria
ARISA Hyd24- Hyd24-01 Gammaprotobacteria
ARISA Marino Marinoscillum Sphinobacteria
ARISA Microb Microbacteriaceae Actinobacteria
ARISA Nitros Nitrospina Deltaproteobacteria
ARISA NS2b NS2b Flavobacteria
ARISA NS5 NS5 Flavobacteria
ARISA NS9 NS9 Flavobacteria
ARISA OCS116 OCS116 Alphaproteobacteria
ARISA OCS155 OCS155 Actinobacteria
ARISA OM43 OM43 Betaproteobacteria
ARISA OTU Unidentified Operational Taxonomic Unit
ARISA Owenwe Owenweeksia Flavobacteria
ARISA PAUC34 PAUC34f or Marine Group A
ARISA Piscir Piscirickettsiaceae Gammaproteobacteria
ARISA Pro Prochlorococcus Cyanobacteria
ARISA Pro_HL(I) Prochlorococcus_High Light Strain (I)
ARISA Pro_LL(IV) Prochlorococcus_Low Ligth Strain (IV)
ARISA Rhodob Rhodobacteraceae Alphaproteobacteria
ARISA Roseob Roseobacter Alphaproteobacteria
ARISA SAR11 SAR11 Alphaproteobacteria
ARISA SAR116 SAR116 Alphaproteobacteria
ARISA SAR324 SAR324 Gammaprotobacteria
ARISA MGA Marine Group A/SAR406
ARISA SAR86 SAR86 Gammaproteobacteria
ARISA Shewan Shewanella Gammaproteobacteria
ARISA Sva099 Sva0996 Actinobacteria
ARISA Thioba Thiobacillus Betaproteobacteria
ARISA Thioth Thiothrix Gammaproteobacteria
ARISA ZD0405 ZD0405 Gammaprotobacteria
ARISA ZD0417 ZD0417 Gammaprotobacteria
Env Bact Bacterial Abundance
Env bc.shift Bray Curtis Rate of Community Change
Env CDOM Colored Dissolved Organic Matter
Env Chl_A_Sat Chlorophyll A Satelite Estemate
Env CHL-A Chlorophyll A Bottle Concentraton
Env Cmax_Depth Chlorophyll Maximum Depth
Env DL Day Length
Env DDL Day Length Rate of Change
Env Depth Depth
Env Elapsed_Days Elapsed days since initiation of study
Env InvSimpson Inverse Simpson biodiversity ndex
Env Leu Bacterial Productivity (Leucine)
Env MEI Multivaritate El-Niño Southern Oscillation Index
Env MLD Mixed Layer Depth
Env NO2 Nitrite Concentration
Env NO3 Nitrate Concentration
Cram
128
nodeType Code Parameter Taxonomy Note
Env Oxygen Oxygen Concentration
Env PAR Photo synthetically active radiation
Env PielouJ Pielou's evenness index
Env PO4 Phosphate Concentration
Env Prim_Prod Primary Productivity Satellite Estemate
Env P* Excess Phosphorous
Env qAmoA AmoA Gene Abundance
Env qArch Archaeal Abundance
Env qGroup1 Archaea Group I Abundance
Env RDate Date
Env Richness1000 Richness (OTUs with > 0.1% Abundance)
Env Salinity Salinity
Env st.shift Change in Salinity or Temperature (Euclidean Distance)
Env ShannonH Shannon Biodiversity Index
Env SiO3 Silicate Concentration
Env SSHD Sea Surface Height Differential (Measured by Satellite)
Env Temperature Temperature
Env Thy Bacterial Productivity (Thymidine)
Env TurnoverLeu Bacterial Turnover Time (Leucine)
Env TurnoverThy Bacterial Turnover Time (Thymidine)
Env Upwelling Upwelling
Env VBR Virus to Bacteria Ratio
Env Vir Viral Abundance
Env WaveHeight Mean Wave Height (day of sampling)
Env σθ Seawater density
Table 3-3 Description of nodes seen in networks. Node type indicates whether nodes are bacterial OTUs
(ARISA) or environmental or biotic parameters (Env). Code is the abbreviation seen in the networks
themselves. ARISA OTU codes are always followed in the networks by ITS length. Parameters describe
taxonomy of OTUs and environmental parameters. Taxonomy note provides phylum information to OTUs
that are defined at a finer than phylum taxonomic resolution.
Discussion
Interaction patterns vary between depths
Our results suggest that microbial interactions at each depth show patterns that are common
across depths, with some key differences between depths. Specifically all depths show modular
patterns in which there are co-occurring groups of organisms that tend to be negatively correlated
to each other, reminiscent of alternate states. At all depths, networks show small world
properties, with average clustering coefficients higher than would be expected by random chance
alone, and path lengths similar to random networks (Watts & Strogatz 1998). This “small world”
Cram
129
topology is a common feature of microbial association in time series between bacteria, protists
and viruses in marine systems (Steele et al. 2011; Chow et al. 2014) lake environments in all
seasons (Kara et al. 2012) and many other biological and non-biological systems (Humphries &
Gurney 2008). The small world nature of the networks suggests that microbes form co-occurring
groups and that there are several OTUs that associate statistically with many other nodes, while
most nodes have fewer associations. These highly connected groups have been likened to
“keystone” species (Steele et al. 2011), and may have particular environmental relevance in the
depths of the water column in which they are found. The higher clustering ratios at the DCM
than in other depths may suggest a greater importance of keystone-like species at this depth than
elsewhere.
Many, but not all of the groups of co-occurring bacteria relate to one or more environmental
parameters suggesting that community variability is likely driven by varying environmental
conditions, which may affect the community by cascading inter-microbial interactions. Also
apparent are differences in the overall network structures at different depths. 5m, DCM and
especially 890m appear to be highly modular, perhaps with seasonality in part driving this
modularity, since some nodes in each network are connected to the rate of change of day length.
This pattern reflects findings that the surface depths and 890m are the most seasonal depths
(Cram et al. 2014b) and expands these observations to show that only a few species are highly
correlated with season itself, while the remaining OTUs are highly correlated with other species,
only some of which are particularly seasonal. Thus, while Cram et al. (2014b) observed several
species that were statistically significantly seasonal, our findings here expand that observation to
suggest that these seasonal organisms may be extending the effect of seasonality to much of the
rest of the microbial community.
Cram
130
5m, the deep chlorophyll maximum and 890m show defined modules of bacteria, many of which
appear to be driven by environmental parameters (Growth rates and dissolved nutrients at DCM,
and Temperature and dissolved nutrients at 890m). This two module structure suggests that
communities at these depths tend to exist in one of two states. Some of the OTUs in these
modules appear to co-occur with environmental parameters, suggesting the state of the
environment is likely related to environmental variability and seasonality. However, since most
of the connections in these modules are between OTUs it is likely that ecological interactions
between organisms, or else between organisms and unmeasured parameters, play an important
role in determining community state. 150m and 500m show looser structures with fewer
connections between bacterial and environmental nodes. There does not appear to be the two
mode structure identified at other depths, suggesting these environments, instead of having a few
defined states have a less defined structure different OTUs each responding to different patterns.
These findings dovetail with previous findings that there is pronounced seasonality at 5m, the
DCM and 890mm but not at 150 or 500m, suggesting that Seasonal patterns may be the ultimate
drivers of the bimodal pattern at the euphotic and bottom communities, while more subtle
differences shape community structure in the mid water column depths.
Statistical associations suggest links between depths
Bacterial OTUs correlate with each other, both within and across depths. The presence of
correlations between depths implies one or more mechanisms linking different depths. Due to the
absence of mixing between depths, it is most likely that sinking particles (Collins et al. 2011)
and/or migrating organisms (Steinberg et al. 2000, 2002; Wilson & Steinberg 2010; Schnetzer et
al. 2011) link microbial communities at different depths. Given that the vast majority of lagged
correlations between depths are “downhill,” a likely mechanism would be that particles transport
Cram
131
nutrients from the surface, which are in turn utilized by the communities at depth. Note that since
our data refer only to the free-living bacteria, the “recipient” organisms either use dissolved
materials released from the particles or in some cases may be shed directly from particles
themselves. This is consistent with the long-standing paradigm in biological oceanography that
productivity in the euphotic zone drives most of what happens at depth, and in this case it is the
deep microbial community composition being driven indirectly by the surface environments and
communities.
We hypothesized that we should find higher edge density within depths than between depths
since we would expect OTUs that are in the same location to have more direct associations with
each-other than we would with OTUs at a remote location, and our data support this. In fact,
while OTUs at the same location can be related by way of symbiosis, shared resources,
competition or other direct means (Fuhrman and Steele 2008; Steele et al. 2011), OTUs at
different depths can only be related by way of some linking environmental parameter, such as the
particle flux and its decomposition described above.
Two of our hub and spoke type networks focus on Gammaproteobacteria, and Marine Group A,
both of which have previously been shown to be seasonal in the mid water column (Cram et al.
2014b) and biogeochemically important (Swan et al. 2011; Allers et al. 2013; Wright et al.
2013)). Further networks examine AEGEAN-169 and SAR11, which are particularly abundant
throughout the water column (Cram et al. 2014b; Alonso-Sáez et al. 2007; Carlson et al. 2009;
Brown et al. 2012). These networks illustrated that previously poorly characterized groups like
Marine Group A divide into multiple OTUs that associate very differently from each other,
suggesting they are each made up of many different “ecological species” (Fuhrman 2009)
(Figure 3-4). For instance the observation that at 890m one OTU “MGA_653.1” was positively
Cram
132
correlated, with a one month lag to nitrogen species and phosphate, suggests it might be adapted
to take advantage of nutrient rich conditions, but slow growing enough that it takes some time to
respond to these conditions. “MGA_762.8 correlated with one month lag to biodiversity in the
DCM, implying these organisms might respond to flux of a certain type of particle from surface
environments that for some reason is highest when diversity is highest. It is generally evident
that most MGA found at 150m and 500m and below lag changes in parameters shallower in the
water column. The exception to this is MGA_712.4 which, at 150m, is positively correlated to
nitrogen species at 500 and 890m and temperature at 890m. This OTU’s relationship to the deep
water column’s nitrogen might suggest that its abundance lags some unmeasured factor that
contributes to high nitrogen abundance deeper in the water column. Similarly, the Nitrospina at
890m (Figure 3-2), ostensibly within the same genus, have multiple disperate conections to
nitrogen species concentration and surface communities, many of them lagged.
Alphaproteobacterial groups such as SAR11 Surface-1 and AEGEAN-169 likewise show
complex ecological interactions with very different patterns shown between different OTUs. The
finding that the abundance of SAR11 Surface 1 OTUs from the surface tend to be related to
fewer environmental factors than OTUs from deeper in the water column where they are less
abundant (Supplemental Figure 3-4), suggests that while they are able to consistently dominate
surface waters, they are more dependent on environmental and community structure variability
in deeper waters.
Conclusion:
Our results are the first, to our knowledge, to show concurrent changes in community structure
between depths in a time series, throughout the water column. In agreement with our previous
recent results on seasonality (Cram et al. 2014), our data support the inference that depths are not
Cram
133
isolated from each-other but rather are linked, even the free-living bacteria within each depth.
Because vertical profiles suggest that the depth of deepest mixing is 40m (Chow et al. 2013),
mixing is likely not a viable mechanism for uniting these depths. We hypothesize that sinking
particles and/or migrating organisms link the environments by transporting nutrients through the
otherwise stratified water column. We furthermore show many associations among bacteria and
environmental parameters that help delineate potential interactions and niche characteristics of
many previously ecologically undefined though very abundant organisms.
Cram
134
Supplement
Supplemental Figure 3-1 Association network showing statistically significant, positive time
lagged and non time-lagged correlation between bacterial nodes within each depth (A, 5m; B,
DCM; C, 150m; D, 500m; E, 890m). Nodes represent bacterial OTUs (circles) at each depth.
Edges (lines) represent correlations, sometimes time lagged, between the bacterial OTUs.
Shown are bacteria that occur at least 25 times and edges that have lagged Spearman correlations
such that R>0.55, P<0.01, Q<0.05. Solid edges represent positive correlations, dashed edges
represent negative correlations. Arrows indicate time lagged associations. Node identities are
indicated in Table 3-3 and there is a legend in Figure 3-1.
Cram
135
Supplemental Figure 3-2 Association network showing statistically significant, positive and negative time
lagged and non time-lagged correlation between bacterial nodes within each depth (A, 5m; B, DCM; C,
150m; D, 500m; E, 890m). Nodes and edges are as in Supplemental Figure 3-1, though negative
associations, shown as dashed lines, are also present.
Cram
136
Supplemental Figure 3-3 Association network of correlations between Deltaproteobacteria OTUs at 150,
500 and 890m and other bacteria and environmental parameters found at every depth throughout the
water column. All bacteria shown have a mean abundance of greater than 1% and are found with a
relative abundance of greater than 0.01% in more than 36 samples. Lines represent statistically
significant, potentially time lagged correlations (Spearman |R| > 0.57, P<0.01, Q < 0.05). Nodes represent
individual measurements of bacterial abundances (circles) or environmental parameters (squares).
Abbreviated names are followed by OTU fragment size. Node Sizes of bacterial nodes represent the
average abundance of OTUs. Solid lines show a positive correlation, dashed lines show a negative
correlation, arrows indicate a 1-month shift in the correlation. Abbreviations for the nodes are translated
in Table 3-3. Edge labels represent Spearman R values. Edge colors are the same as in Figure 3-3.
Legends are present in Figure 3-1 and Figure 3-2.
Cram
137
Supplemental Figure 3-4 Association network of correlations between bacteria from the SAR11 Surface-1
clade and other bacteria and environmental parameters found at every depth throughout the water column.
Nodes and edges have the same thresholds (P, Q, and R values for Edges, mean abundance and
occurrence thresholds for bacterial nodes) as in Supplemental Figure 3-3.
Cram
138
Supplemental Figure 3-5 Association network of correlations between bacteria from the AEGEAN-169
clade and other bacteria and environmental parameters found at every depth throughout the water column.
Nodes and edges have the same thresholds (P, Q, and R values for Edges, mean abundance and
occurrence thresholds for bacterial nodes) as in Supplemental Figure 3-3.
Cram
139
5m DCM 150m 500m 890m AllvAll
Nodes (N) 68 71 75 74 54 446
Edges (E) 81 80 95 91 98 1339
Cl 0.13 0.23 0.18 0.18 0.29 0.15
L 3.51 4.03 5.37 6.49 4.38 4.07
Cl_r 0.03 0.02 0.07 0.02 0.02 0.01
L_r 4.66 4.75 3.87 4.38 3.11 3.59
Cl/Cl_R 3.3 9.4 2.4 6.4 9.6 6.7
I_Cl 1.36 2.65 0.95 2.19 2.58 2.32
I_L -0.28 -0.16 0.33 0.39 0.34 0.12
Density 3.6% 3.2% 3.4% 3.4% 6.8% 1.3%
Outgoing Density 1.6% 1.1% 0.9% 1.2% 1.0% NA
Density Ratio 2.19 2.98 3.85 2.84 7.09 NA
Supplemental Table 3-1 Topological statistics for networks of bacteria at each depth (Figure 3-1), and for
a network of OTUs at all depths (Figure 3-2) in which only positive correlations were generated. This
complements Table 3-1 which shows statistics for networks connected by both positive and negative
nodes and Supplemental Figure 3-2 which visually depicts the described networks. These networks
include only nodes for bacteria that are present at >0.01% abundance greater than 25 times and edges that
have a possibly time lagged, global, absolute Spearman R value of greater than 0.5. Nodes are the
bacterial OTUs that are connected by at least one edge to another node. Edges are the number of
correlations between bacterial OTUs. Density is the number of edges (E) divided by the number of
possible edges {N*(N-1)/2} such that {Density = E/(N*(N-1)/2)}. Cl is the clustering coefficient for the
network. L is the mean path length for the network. Cl_r and L_r are the clustering coefficients and path
lengths for equivalently sized (same number of nodes and edges) randomly distributed networks.
Cl/Cl_R is the ratio of the clustering coefficient to the random network's clustering coefficient. ICl and IL
are the log response ratios of the clustering and path length coefficients {ICl = log(Cl) -log(ClR)}. Similar
patterns can be seen as in Table 3-1, though the density ratio is much higher at 890m when only positive
edges are considered, than in Table 3-1 where negative edges are also considered.
Cram
140
5m DCM 150m 500m 890m AllvAll
Nodes (N) 68 71 75 74 54 446
Edges (E) 81 80 95 91 98 1339
Cl 0.13 0.23 0.18 0.18 0.29 0.15
L 3.51 4.03 5.37 6.49 4.38 4.07
Cl_r 0.03 0.02 0.07 0.02 0.02 0.01
L_r 4.66 4.75 3.87 4.38 3.11 3.59
Cl/Cl_R 3.3 9.4 2.4 6.4 9.6 6.7
I_Cl 1.36 2.65 0.95 2.19 2.58 2.32
I_L -0.28 -0.16 0.33 0.39 0.34 0.12
Density 3.6% 3.2% 3.4% 3.4% 6.8% 1.3%
Outgoing Density 1.6% 1.1% 0.9% 1.2% 1.0% NA
Density Ratio 2.19 2.98 3.85 2.84 7.09 NA
Supplemental Table 3-2 Summary Statistics for networks that examine cross correlations between bacteria
at different depths. This table complements Table 3-2 and treats only positive interactions between
variables, rather than both positive and negative interactions. Note that 58% of the edges in the AllvAll
network represent positive correlations, which are addressed here. The AllvAll column reflects the
network shown in Figure 3-2. The other columns reflect the nodes within and connecting pairs of depths.
Nodes (N) are the number of bacterial OTUs at a specific depth that occur at least 25 times in the data set
connected by at least one edge (correlation) in the specified network. Edges (E) are the total number of
Spearman correlations with R > 0.55 in that network. Within Depth Nodes (Ni) are the nodes in these
networks that are connected by at least one edge to another node at that same depth. Within depth edges
(Ei) are those edges that connect pairs of nodes from the same depth. Cross Depth Edges (Eo) are those
edges that connect two nodes from different depths. Within Depth Density, reflects the probability that a
pair of nodes from the same depth are connected by an edge; it is the quotient of the number of edges
found connecting nodes that are from the same depth, divided by the maximum number of edges that
could possibly connect nodes of the same depth in that network. Cross Depth Density (Eo/Eo_max)
reflects the probability that two nodes from different depths are connected by an edge; it is the total
number of edges connecting nodes from different depths, compared to the number of possible edges that
could connect nodes from those depths. The density ratio (ei/eim)/(eo/eom), reflects how much likelier
two nodes are to be connected if they are from the same depth, than if they are from different depths; it is
the quotient of the Within Depth Density and the Cross Depth Density of a network. Uphill (U) is the
number of time lagged edges that connect pairs of nodes from different depths such that changes in the
node at the deeper depth precede changes in the done at the shallower depth. Unlagged (F) is the number
of non time lagged edges that connect pairs of nodes from different depths. Downhill is the number of
time lagged edges that connect pairs of odes from different depths such that the changes in the shallower
node precede correlated changes in the deeper node.
Cram
141
References
Allers E, Wright JJ, Konwar KM, Howes CG, Beneze E, Hallam SJ, et al. (2013). Diversity and
population structure of Marine Group A bacteria in the Northeast subarctic Pacific Ocean. ISME
J. 7:256–268.
Alonso-Sáez L, Balagué V , Sà EL, Sánchez O, González JM, Pinhassi J, et al. (2007).
Seasonality in bacterial diversity in north-west Mediterranean coastal waters: assessment through
clone libraries, fingerprinting and FISH. FEMS Microbiol. Ecol. 60:98–112.
Arumugam M, Raes J, Pelletier E, Le Paslier D, Yamada T, Mende DR, et al. (2011). Enterotypes
of the human gut microbiome. Nature 473:174–180.
Assenov Y , Ramírez F, Schelhorn S-E, Lengauer T, Albrecht M. (2008). Computing topological
parameters of biological networks. Bioinformatics 24:282–284.
Barberán A, Bates ST, Casamayor EO, Fierer N. (2011). Using network analysis to explore co-
occurrence patterns in soil microbial communities. ISME J. 6:343–351.
Beman JM, Popp BN, Francis CA. (2008). Molecular and biogeochemical evidence for ammonia
oxidation by marine Crenarchaeota in the Gulf of California. ISME J. 2:429–441.
Bonilla-Findji O, Herndl GJ, Gattuso J-P, Weinbauer MG. (2009). Viral and Flagellate Control of
Prokaryotic Production and Community Structure in Offshore Mediterranean Waters. Appl.
Environ. Microbiol. 75:4801–4812.
Brown MV , Lauro FM, DeMaere MZ, Muir L, Wilkins D, Thomas T, et al. (2012). Global
biogeography of SAR11 marine bacteria. Mol. Syst. Biol. 8.
http://www.nature.com/msb/journal/v8/n1/full/msb201228.html (Accessed December 2, 2012).
Brown MV , Schwalbach MS, Hewson I, Fuhrman JA. (2005). Coupling 16S-ITS rDNA clone
libraries and automated ribosomal intergenic spacer analysis to show marine microbial diversity:
development and application to a time series. Environ. Microbiol. 7:1466–1479.
Carlson C, Morris R, Parsons R, Treusch A, Giovannoni S, Vergin K. (2009). Seasonal dynamics
of SAR11 populations in the euphotic and mesopelagic zones of the northwestern Sargasso Sea.
ISME J. 3:283–295.
Chaffron S, Rehrauer H, Pernthaler J, von Mering C. (2010). A global network of coexisting
microbes from environmental and whole-genome sequence data. Genome Res. 20:947–959.
Chow C-ET, Kim DY , Sachdeva R, Caron DA, Fuhrman JA. (2014). Top-down controls on
bacterial community structure: microbial network analysis of bacteria, T4-like viruses and
protists. ISME J. 8:816–829.
Chow C-ET, Sachdeva R, Cram JA, Steele JA, Needham DM, Patel A, et al. (2013). Temporal
variability and coherence of euphotic zone bacterial communities over a decade in the Southern
California Bight. ISME J. 7:2259–2273.
Cram
142
Collins LE, Berelson W, Hammond DE, Knapp A, Schwartz R, Capone D. (2011). Particle fluxes
in San Pedro Basin, California: A four-year record of sedimentation and physical forcing. Deep
Sea Res. Part Oceanogr. Res. Pap. 58:898–914.
Cram JA, Chow C-ET, Sachdeva R, Needham DM, Parada AE, Steele JA, et al. (2014a).
Seasonal and interannual variability of the marine bacterioplankton community throughout the
water column over ten years. ISME J.
Cram JA, Sun F, Fuhrman JA. (2014b). Marine bacterial, archaeal and protistan association
networks. In:Encyclopedia of Metagenomics, SpringerReference.
Deng Y , Jiang Y-H, Yang Y , He Z, Luo F, Zhou J. (2012). Molecular ecological network analyses.
BMC Bioinformatics 13:113.
Eiler A, Heinrich F, Bertilsson S. (2012). Coherent dynamics and association networks among
lake bacterioplankton taxa. ISME J. 6:330–342.
Faust K, Raes J. (2012). Microbial interactions: from networks to models. Nat. Rev. Microbiol.
10:538–550.
Faust K, Sathirapongsasuti JF, Izard J, Segata N, Gevers D, Raes J, et al. (2012). Microbial co-
occurrence relationships in the human microbiome. PLoS Comput. Biol. 8:e1002606.
Freilich S, Kreimer A, Meilijson I, Gophna U, Sharan R, Ruppin E. (2010). The large-scale
organization of the bacterial network of ecological co-occurrence interactions. Nucleic Acids
Res. 38:3857–3868.
Fuhrman J, Steele J. (2008). Community structure of marine bacterioplankton: patterns,
networks, and relationships to function. Aquat. Microb. Ecol. 53:69–81.
Fuhrman JA. (2009). Microbial community structure and its functional implications. Nature
459:193–199.
Gilbert JA, Steele JA, Caporaso JG, Steinbrück L, Reeder J, Temperton B, et al. (2012). Defining
seasonal marine microbial community dynamics. ISME J. 6:298–308.
Giovannoni SJ, Vergin KL. (2012). Seasonality in Ocean Microbial Communities. Science
335:671–676.
Hatosy SM, Martiny JBH, Sachdeva R, Steele J, Fuhrman JA, Martiny AC. (2013). Beta
diversity of marine bacteria depends on temporal scale. Ecology 94:1898–1904.
Humphries MD, Gurney K. (2008). Network ‘Small-World-Ness’: A Quantitative Method for
Determining Canonical Network Equivalence. PLoS ONE 3:e0002051.
Jürgens K, Matz C. (2002). Predation as a shaping force for the phenotypic and genotypic
composition of planktonic bacteria. Antonie Van Leeuwenhoek 81:413–434.
Cram
143
Jurgens K, Pernthaler J, Schalla S, Amann R. (1999). Morphological and compositional changes
in a planktonic bacterial community in response to enhanced protozoan grazing. Appl. Environ.
Microbiol. 65:1241–1250.
Kara EL, Hanson PC, Hu YH, Winslow L, McMahon KD. (2012). A decade of seasonal
dynamics and co-occurrences within freshwater bacterioplankton communities from eutrophic
Lake Mendota, WI, USA. ISME J. 7:680–684.
Kiorboe T. (2001). Formation and fate of marine snow: small-scale processes with large-scale
implications. Sci. Mar. 65:57–71.
Malfatti F, Azam F. (2009). Atomic force microscopy reveals microscale networks and possible
symbioses among pelagic marine bacteria. Aquat. Microb. Ecol. 58:1–14.
Malfatti F, Samo TJ, Azam F. (2009). High-resolution imaging of pelagic bacteria by Atomic
Force Microscopy and implications for carbon cycling. ISME J. 4:427–439.
Miller MB, Bassler BL. (2001). Quorum Sensing in Bacteria. Annu. Rev. Microbiol. 55:165–
199.
Needham DM, Chow C-ET, Cram JA, Sachdeva R, Parada A, Fuhrman JA. (2013). Short-term
observations of marine bacterial and viral communities: patterns, connections and resilience.
ISME J. 7:1274–1285.
Ruan Q, Dutta D, Schwalbach MS, Steele JA, Fuhrman JA, Sun F. (2006). Local similarity
analysis reveals unique associations among marine bacterioplankton species and environmental
factors. Bioinformatics 22:2532–2538.
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. Part Oceanogr. Res. Pap. 58:16–26.
Smoot ME, Ono K, Ruscheinski J, Wang PL, Ideker T. (2011). Cytoscape 2.8: new features for
data integration and network visualization. Bioinformatics 27:431–432.
Steele JA, Countway PD, Xia L, Vigil PD, Beman JM, Kim DY , et al. (2011). Marine bacterial,
archaeal and protistan association networks reveal ecological linkages. ISME J. 5:1414–1425.
Steinberg DK, Carlson CA, Bates NR, Goldthwait SA, Madin LP, Michaels AF. (2000).
Zooplankton vertical migration and the active transport of dissolved organic and inorganic
carbon in the Sargasso Sea. Deep Sea Res. Part Oceanogr. Res. Pap. 47:137–158.
Steinberg DK, Goldthwait SA, Hansell DA. (2002). Zooplankton vertical migration and the
active transport of dissolved organic and inorganic nitrogen in the Sargasso Sea. Deep Sea Res.
Part Oceanogr. Res. Pap. 49:1445–1461.
Strom SL. (2008). Microbial Ecology of Ocean Biogeochemistry: A Community Perspective.
Science 320:1043–1045.
Cram
144
Swan BK, Martinez-Garcia M, Preston CM, Sczyrba A, Woyke T, Lamy D, et al. (2011).
Potential for Chemolithoautotrophy Among Ubiquitous Bacteria Lineages in the Dark Ocean.
Science 333:1296–1300.
Telesford QK, Joyce KE, Hayasaka S, Burdette JH, Laurienti PJ. (2011). The ubiquity of small-
world networks. ArXiv11095454 Phys. http://arxiv.org/abs/1109.5454 (Accessed May 17, 2014).
Treusch AH, Vergin KL, Finlay LA, Donatz MG, Burton RM, Carlson CA, et al. (2009).
Seasonality and vertical structure of microbial communities in an ocean gyre. ISME J. 3:1148–
1163.
Watts DJ, Strogatz SH. (1998). Collective dynamics of ‘small-world’ networks. Nature 393:440–
442.
Wilson SE, Steinberg DK. (2010). Autotrophic picoplankton in mesozooplankton guts: evidence
of aggregate feeding in the mesopelagic zone and export of small phytoplankton. Mar. Ecol.
Prog. Ser. 412:11–27.
Wright JJ, Mewis K, Hanson NW, Konwar KM, Maas KR, Hallam SJ. (2013). Genomic
properties of Marine Group A bacteria indicate a role in the marine sulfur cycle. ISME J.
http://www.nature.com/ismej/journal/vaop/ncurrent/abs/ismej2013152a.html (Accessed
September 12, 2013).
Xia LC, Ai D, Cram J, Fuhrman JA, Sun F. (2013). Efficient statistical significance
approximation for local similarity analysis of high-throughput time series data. Bioinforma. Oxf.
Engl. 29:230–237.
Zhou J, Deng Y , Luo F, He Z, Tu Q, Zhi X. (2010). Functional Molecular Ecological Networks.
mBio 1:e00169–10.
Cram
145
Chapter 4 : Dilution reveals how viral lysis and
grazing shape microbial communities
Abstract
Grazing by protists and viral infection are the main factors that remove marine bacteria and both
processes have been shown in numerous environments to shape the structure and ecology of
microbial communities. We diluted planktonic microbial communities to determine how
decreasing the encounter rate between bacteria, grazers and viruses affects community structure.
In experimental treatments, sea water was diluted 10 fold in water from which protists and
bacteria had been removed (0.2μm filtered), or from which protists bacteria and viruses had been
removed (0.02μm filtered). After dilution, bacterial and protistan communities recovered to
starting abundances but viruses did not. After regrowth, the microbial community structure of
bacteria in both diluted treatments was significantly different from control communities.
Furthermore the community structure of bacteria diluted with 0.02μm filtered water was
significantly different, after regrowth, from communities diluted with 0.2μm filtered water.
Diluted treatments were dominated by formerly rare operational taxonomic units (OTUs), and
OTUs that are more abundant in the deep ocean, while abundant surface taxa such as SAR11 did
not increase in abundance after dilution. These results suggest that microbes face a tradeoff
between fast growth and resistance to predation and infection, and that the microbes that are
most abundant in the California coastal surface waters are both more grazer resistant and slower
growing than some organisms that are rare in surface water.
Cram
146
Introduction
Studies of the dynamics of marine microbes have revealed temporal patterns including
seasonality, long term changes, and relationships to environmental variability (Cram et al. 2014;
Schauer et al. 2003; Fuhrman et al. 2006; Alonso-Sáez et al. 2007; Gilbert et al. 2012;
Giovannoni & Vergin 2012; Chow & Fuhrman 2012; Ladau et al. 2013). More recently,
examination of statistical associations has suggested interactions between specific
bacterioplankton and the abundance of particular protists and viruses (Gilbert et al. 2009; Steele
et al. 2011; Needham et al. 2013; Chow et al. 2014). These relationships suggest that these
interactions play important roles in determining which bacteria dominate marine surface waters
in a given location, depth and time of year.
It is generally accepted that protistan grazers partially shape microbial communities by
preferentially grazing certain microbes and that many bacteria have phenotypes that can protect
them from grazing (Jurgens & Gude 1994; Jürgens & Matz 2002; Pernthaler 2005; Matz &
Kjelleberg 2005; Jurgens & Massana 2008). These phenotypes include small size, motility, cell
wall structure, morphology and other factors (Pernthaler 2005). Defense mechanisms likely
require higher a metabolic requirement or otherwise have potential fitness trade-offs (i.e. slower
growth). Thus it follows that variability in microbial grazing pressure likely promotes variability
in microbial community structure (Jürgens & Matz 2002). Many studies have investigated ways
grazers shape microbial communities (see Hahn & Höfle 2001) though only a few have focused
on taxonomic structure in marine, rather than freshwater environments. In general it has been
shown that both increasing and decreasing grazing pressure results in shifts in microbial
community structure (Suzuki 1999; Jurgens et al. 1999; Simek et al. 1999, 2001). Interactions
between grazing and nutrient availability have also been shown in which bacteria employing
Cram
147
different anti-predation strategies tend to prevail under limitation by different nutrients (Matz &
Jürgens 2003).
Viruses are also important agents of bacterial removal and are believed to be responsible for 10%
to 50% of total bacterial mortality in surface waters (Fuhrman 1999). While it is generally
understood that viruses shape microbial communities (Fuhrman 1999; Suttle 2007; Breitbart
2012), the modes of viral selection are less well understood than those of protists on bacteria.
“Kill the winner” theory suggests that bacteria face tradeoffs between growth and susceptibility
to both grazers and viruses (Thingstad & Lignell 1997; Thingstad 2000; Winter et al. 2010). To
avoid confusion, “kill the winner” here does not imply “frequency dependent selection” where
viruses infect the most abundant organism, but rather that both viruses and protists preferentially
remove organisms that are most active, implying that viruses and protists may shape
communities even if they are in a steady state (Thingstad & Lignell 1997; Thingstad 2000;
Winter et al. 2010). Several experiments have shown how viruses by themselves (Hewson &
Fuhrman 2003; Schwalbach et al. 2004; Bouvier & del Giorgio 2007), or alongside of grazers
(Jardillier et al. 2005; Weinbauer et al. 2007; Bonilla-Findji et al. 2009) shape microbial
communities (see Winter et al. 2010 for summary). More recent work has determined that both
grazing and infection were more prevalent in surface water (Anderson et al. 2012).
In a study of particular relevance to this one, water containing viruses and bacteria was diluted in
a number of proportions with viral replete and viral free water to determine if there was a
minimum viral density at which infection could be sustained (Wilcox & Fuhrman 1994). The
authors found that if the product of bacterial and viral abundances fell much below 10
12
then
viral abundance would not recover to starting concentrations. Bacteria on the other hand always
grew back to starting concentrations, regardless of how diluted they were. This pattern suggested
Cram
148
that viruses in the ocean primarily follow a lytic rather than a lysogenic life cycle, as in the latter
case the bacterial community would have resumed viral production once the bacterial population
grew back to higher abundance. This means that in coastal surface waters, infection pressure is
dependent on viral concentration (likely alongside other factors such as nutrient concentration, as
well as viral and bacterial community structure). This is convenient from an experimental
standpoint since decreasing ambient virus concentrations should quickly decrease infection
pressure on microbial communities. In this study we look at differences in community structure
of bacteria incubated with viruses and protists at ambient concentrations and bacteria diluted in
water that is free of bacteria and protists, or free of bacteria, protists and viruses.
This study explores the premise that the abundance of a given group of microbes is a function of
its growth and removal rates. Microbial OTUs that have constant abundance must have growth
rates equal to their removal rates. In a stable microbial community, there may be faster growing
and slower growing members, as long as the faster growing members are being removed at
proportionally faster rates. Conversely, changes in abundance of OTUs over time in a contained
water mass are likely the result of growth exceeding removal (abundance increases), or removal
exceeding growth (abundance decreases). Our goal in this study is to identify faster growing (and
by extension more rapidly removed) organisms through “dilution”. In our interpretation of this
study, we make several assumptions, which we postulate are mostly true for marine surface
environments. We will address the impacts of violations to these assumptions in the discussion.
First, we assume that bacteria live on dissolved organic matter (DOM) only. Second, we assume
that DOM is not affected by our filtration method. Third, we assume the DOM pool is not
substantially changed by the community’s response to dilution. In summary, we count on
Cram
149
microbial species experiencing a constant “bottom up” environment, in which only “top down”
controls primarily shape community structure.
Diluting bacteria in water from which protists and viruses are absent should decrease removal
rates substantially, since microbial removal is likely a function of their encounter rates (see
Schmoker et al. 2013) with protists and viruses, which in turn are dependent on protistan and
viral abundance. If bacteria are diluted in water that contains neither viruses nor protists and
given our assumptions above, we presume bacterial growth would remain constant, while
removal rates due to infection and grazing should decrease. Thus the fastest growing microbes
with high removal rates should increase in abundance. Meanwhile slow growing microbes with
slow removal rates should increase in abundance much more slowly. If bacteria are diluted in
water which contains viruses, but not protists, changes in abundance will reflect growth rates
minus removal due to grazing, but not infection (since infection processes are still removing
bacteria). Thus comparing changes in abundance of bacteria diluted in virus free water to virus
replete water should illuminate removal processes due to infection, while comparing changes in
bacterial abundance in samples that have been diluted in water without protists to undiluted
samples should reveal removal processes due to grazing.
The steady state “kill the winner” model contends that bacteria face tradeoffs between predation
resistance, virus resistance and growth (Thingstad & Lignell 1997; Thingstad 2000; Winter et al.
2010). Given this model, we hypothesize that after grazing and/or infection pressure is reduced
through dilution, the abundance of bacteria that are generally faster growing but less resistant to
removal will increase in abundance. Meanwhile, microbial bacteria that are slower growing but
more resistant to removal processes will not increase in abundance, and will make up a
proportionately smaller fraction of total community structure. Our goals in this study include
Cram
150
identifying which members of the microbial community are relatively more grazer resistant,
which are relatively more grazer susceptible, which are relatively more virus resistant and which
are relatively more susceptible to infection. This study also leverages the ten year SPOT time
series, to allow investigation of the dynamics of these resistant and susceptible organisms
throughout the water column.
Methods
Setup
This study was run in two separate summers. Water for the first iteration was collected ~7:00
pacific daylight savings time (pdt) on 01 June 2010 and for the second ~7:00 pdt on 14 June
2011. In both years approximately 200L of surface sea water was collected, by bucket from
(33°28'N, 118°29' W), just off of Catalina Island and ~13 km from the San Pedro Ocean Time-
series (SPOT) station (33°33'N, 118°24'W) and transported back to the Wrigley Marine Science
Center (WMSC), on Catalina Island. Collected water was filtered through 80um mesh to remove
large grazers and homogenized in 200L acid washed culture vats. To set up our experiment,
some of this homogenized water was set aside as “whole sea water”. A second subset was run in
series through a 142mm A/E prefilter (Pall Life Sciences) and then through 0.22μm 142mm
Durapore filters (Millipore) using gentle pressure filtration to remove all prokaryotes and larger
organisms and is referred to henceforth as “0.2μm filtered water”. A third subset was filtered as
for the 0.2μm treatment, and then through a 30 kilodalton (~0.02μm pore size) tangential flow
filter (Millipore) to remove all virus like particles, now called “0.02μm filtered water”.
Three experimental treatments, with three replicates each and 20L total volume per replicate
were set up in low density polyethylene flexible walled “cubitainers”. The first treatment was
comprised of 100% whole seawater. The second was 10% whole seawater diluted in 90% 0.2μm
Cram
151
filtered water. The third treatment was 10% whole seawater diluted in 90% 0.02μm filtered water
(Figure 4-1). Each of the nine samples (3 treatments * 3 replicates) were sampled for microscopy
and DNA as described below, and incubated for the duration of the experiment at ambient
surface ocean temperature and light levels in water flow through tanks.
Cram
152
Figure 4-1 Outline of bacterial dilution setup. (i-iii): Square boxes are inverted black and white images of
SYBR green epifluorescence slides of whole (80um filtered) sea water (i), 0.2μm Filtered water from
which bacteria and protists are removed through filtration through first a 1μm glass fiber filter and then a
0.2μm Durapore filter (ii) and 30kd filtered water which from which bacteria and protists were first
removed through 0.2μm filtration followed by viral removal by tangential flow filtration (iii). Cylinders
represent treatment groups, each containing three replicates. These include a control group (iv), a group
with 10% unfiltered water and 90% water from which protists and bacteria have been removed (v), and a
group with 10% unfiltered water and 90% water from which viruses bacteria and protists had all been
removed (vi).
Cram
153
Sampling
In both cases setup took about one day, so samples taken on “day 0” in both studies correspond
to 5:00 pdt 02 June 2010 and 15 June 2011. In 2010 samples were also taken on “day 1” (02 June
2010 14:00) “day 3” (04 June 2010 10:00) and “day 6” (07 June 2014 10:00). At each of these
time points, 1L of water was collected from each container. Every time samples were taken,
bottles were gently inverted three times to homogenize the sample. After sampling the containers
were compressed to remove all head-space. Of the 1L collected 50ml of water was subsampled
for microscopy analysis and the rest was filtered for DNA. In 2011 50ml samples were taken for
bacterial microscopy at 7:00 and 19:00 every day (except for day 0, where samples were taken
during setup), for the duration of the experiment. Larger 1L samples were collected for protist
microscopy and bacterial DNA on day 1 (16 June 2011), day 2 (17 June), day 3 (18 June), day 5
(20 June), day 7 (22 June) and day 9 (24 June).
Microscopy
For all forms of microscopy, a subsample of water was collected and fixed with 4% 0.2μm
filtered formalin. All samples were prepared as described below on an Olympus epifluorescence
microscope and total cells or virus like particles per milliliter were calculated.
SYBR green Bacterial and Virus counts
Bacteria and viruses were both quantified using SYBR green microscopy at all time-points in
2010 as described previously (Noble & Fuhrman 1998; Patel et al. 2007). Briefly, 2ml to 5ml of
fixed sample, depending on concentration, was filtered, stained, counted, and viewed on an
epifluorescence microscope (Olympic) under 450-495nm excitation and 515-560nm emission.
Bacteria and virus like particles were counted and concentrations in the sample were back
calculated.
Cram
154
Whatman 25mm diameter Anodisk filters were not available in 2011 and so we quantified
viruses in the control group at day 0 and in all treatments on day 9 using a modified protocol
with 13mm Anodisk filters. The protocol was modified such that only 500μl of fixed sample was
filtered and 6μl of SYBR solution was added to the filters. Adjusted conversion factors were
used for the different volume of water and slide diameter. In 2011 bacteria and viruses were
quantified with SYBR green at the beginning and the end of the experiment. In both years SYBR
green was used to confirm the near absence of bacteria from the 0.2ul filtered seawater used to
set up the experiments, and the near absence of viruses in the 30kd filtered water.
Acridine Orange Direct Counts of Bacteria
In 2011 bacterial concentrations were measured every six hours using acridine orange direct
counts (AODC) (Hobbie et al. 1977). Briefly, 2-5μl (depending on expected bacterial
concentration) of fixed (4% 0.02μm formalin) sample was filtered down to 2ul and incubated
with 50μl of acridine orange solution (0.1% acridine orange in 5% formalin, 0.2μm filtered) for
30 seconds. It was then filtered through a 25mm 0.2μm pore size black polycarbonate filter. The
filter was mounted between a slide and cover slip with Olympus Type F immersion oil. The
sample was viewed under the same emission and fluorescence spectra as the SYBR samples.
DAPI counts of Protists
In 2011, protist concentrations were measured by DAPI fluorescence. 25ml or 50ml of fixed
sample (depending on concentration) was filtered through a 0.8um polycarbonate filter until
500μl remained. This concentrate was incubated with 50ul of 50ug/ul DAPI solution (Sigma) in
the dark. Samples were filtered the rest of the way and then mounted between a slide and cover
slip with Olympus Type F immersion oil. Samples were counted under 350-400 ultraviolet
Cram
155
excitation and blue 415-450 emission. Autofluorescent protists were counted under 480nm
excitation and 660nm emission.
DNA Collection
DNA was collected from each sample by filtration in series, through a 47mm A/E filter (effective
pore size ~1μm) and a 0.22μm Durapore GVWP filter (Millipore, Billerica, MA, USA). Filters
were immediately frozen on dry ice and stored at -80C. Only the 0.2μm filter was used in
subsequent analysis.
ARISA + Clone Libraries
Extraction
DNA was extracted from filters using phenol-chloroform (Fuhrman et al. 1988) with the
following several modifications: Samples were extracted from the crushed 47mm filter with two
washes of 500μl SDS lysis buffer. Second, DNA was precipitated with 1x volume of Isopropanol
(instead of 2X volumes of ethanol) and the pellet was washed with 250ml of ethanol prior to
resuspension. DNA was quantified using the Quant-iT PicoGreen dsDNA Assay Kit
(Invitrogen), and a subsample was diluted to 0.5ng/ul. Both the fully concentrated stock samples
and diluted samples were stored at -80C prior to further analysis.
Amplification and processing
0.5ng of DNA was PCR amplified using a modified version of ARISA protocols applied
previously (Brown et al. 2005; Chow et al. 2013). Samples were amplified 50ul solution
containing 1X ThermoPol PCR buffer (New England Biolabs, NEB, Ipswich, MA USA), 2%
Tween 20 (Sigma Aldrich), 0.2mg/ml BSA (Sigma-Aldrich), 2mM dNTPs (Promega), , 0.4uM
each of 78 Forward primer and 79 TET labeled reverse primer (described in Brown et al. 2005;
Chow et al. 2013), and 2.5 units of ThermoPol buffer (NEB). Reaction mixtures were held at
Cram
156
95°C for 3 minutes, followed by 32 cycles of amplification at 95°C for 40s, 56°C for 40s and
72°C for 90s with a final elongation step of 72°C for 7 min.
DNA was concentrated using the Zymo DNA Clean & Concentrator -5 kit into 10ul volume and
DNA concentrations of amplified product were measured with PICO green. 5ng/μl dilutions of
each sample were made for fragment analysis.
Fragment Analysis
In 2010 fragment analysis was run in house, in replicate lanes using slab gel electrophoresis as
described previously (Brown et al. 2005; Chow & Fuhrman 2012). In 2011 fragment analysis
was run by the University of Illinois core sequencing facility on an ABI 3730xl automated
sequencer using the same custom standards used in the 2010 run. In both cases, 5ng of DNA was
analyzed per lane.
In both cases peaks were identified using DaX (Van Mierlo). Peaks were binned so as to match
bin sizes used in earlier studies (Cram et al. 2014; Needham et al. 2013; Chow et al. 2013).
Putative identities were assigned to each length using clone libraries and other known sequences
as described in Cram et al. (2014). This analysis yielded estimated relative abundances of a
number of identified and unidentified OTUs.
Microbial Observatory Data
Data from the SPOT microbial observatory were utilized in this study and were processed as
described previously (Cram et al. 2014). It is worth noting that OTU bins and identities were
generated in a manner that was consistent with the microbial observatory samples allowing the
abundances of OTUs identified in this study to be interrogated over time in the San Pedro
Channel.
Cram
157
Analyses
Total abundance patterns
Estimated absolute abundances were calculated by multiplying the relative abundance of each
OTU in each sample by the total prokaryote count for that sample. Total abundances of bacteria
and viruses (both years), as well as autotrophic and heterotrophic protists (2011 only) were
visualized using the ggplot2 (Wickham 2009) in R (Team 2011). Patterns of growth were
examined by looking at the shapes of the bacterial abundance curve.
Non-metric multidimensional scaling
Non-metric multidimensional scaling (NMDS) plots, one for each study year, approximating
Bray-Curtis dissimilarity between all pairs of samples, each year of the study was generated
using the metaMDS function in the “vegan” (Oksanen et al. 2013) R-package. Results were
plotted so as to identify differences between treatments and time points in base R. A third NMDS
plot was generated combining data from both 2010 and 2011 into one plot.
Analysis of similarities and Permutational ANOVA
Analysis of similarities (Clarke 1993) was applied to the data using the “anosim” function in the
vegan R-package to determine whether, for each time point, within each year, there were
statistically significant differences between all samples and between each pair of samples.
We applied permutational multivariate analysis of variance (Anderson 2001) by applying the
“adonis” vegan R-package to samples from both years together to determine whether the year
(iteration of the study), the experimental treatment, and/or the interaction between these two
variables, appeared to drive differences between samples. The adonis test was applied, one day at
a time, to samples collected on days 0, 1, and 3 in both studies. Samples from day 6 of 2010 and
day 5 of 2011 were pooled for a final time point.
Cram
158
Abundance profiles of key operational taxonomic units
We applied the similarity percentages (Clarke 1993) test using the “simper” vegan package to
determine OTUs that contributed most to differences between the control group and the 0.2μm
dilution treatments on day 3 (both years), and day 5 (2011 only) and day 6 (2010 only). We
further identified OTUs that drove differences between the 0.2 and 0.02μm dilution treatments
on day 5 (2010) and day 6 (2011).
For each pair of treatments to which the similarity percentages test was applied, we identified ten
OTU that contributed the most to differences between the two treatments such that five OTU
were more abundant in one sample, and five OTU that were more abundant in the other sample.
The abundances of each of the chosen OTUs were plotted over the course of the study in which
they drove differences between treatments.
Comparison to Microbial Observatory
The relative abundances of each of the OTUs chosen by similarity percentages analysis were also
plotted at all five depths over the course of the SPOT microbial observatory dataset. Plots were
investigated to determine the depths, times of year, and years during which the OTUs were most
abundant, and to determine the abundance patterns of said OTUs at the different depths.
Results
Physical parameters of starting water
Water conditions during the 2010 experiment were 18.9°C with salinity of 33.1ppt (measured
June 18 2010) and during the initiation of the 2011 experiment were measured as 18.4°C with
salinity of 32.8 ppt.
Cram
159
Total cell counts
Bacteria
In 2010, bacterial abundances in the control group started at about 3*10
6
/ml and remained near
that level over the course of the experiment (Figure 4-2). In 2011, control abundances started at
2*10
6
/ml and decreased by about half over the first three days of the experiment, and then
increased in abundance again after day 8. In both 0.2μm and 0.02μm diluted experimental
treatments, bacterial abundance started tenfold lower than in the control group, which was
expected, since these samples had been diluted tenfold. In 2010 bacteria in two out of three
replicates of both treatments increased in abundance nearly to control concentrations, and
decreased slightly by day 6. The exception was one 0.02μm diluted bacterial replicate which
grew more slowly, and caught up with the other replicates by day 6. In 2011, the experimental
groups likewise grew, starting one day after the start of the experiment and reaching maximal
abundance around day 3 or 4, after which their abundance slowly decreased until they had
returned to their starting abundance by day 7.
Viruses
Viral abundances in both years stayed fairly constant, regardless of treatment (Figure 4-2). In
2010, both the control group and 0.2μm treatments started at and remained near 6*10
7
VLP/ml.
In the 0.02μm dilution treatment, where viruses were diluted tenfold along with bacteria and
protists, viruses started at around 6*10
6
/ml and maintained this abundance over the course of the
study, perhaps increasing slightly in abundance between days 3 and 6. In June 2011, the control
group was measured at the start and end of the experiment, and virus counts increased in this
treatment from 2*10
7
to 6*10
7
. Because we had some trouble adapting our existing methods to
the 13mm Anodisk filters (25mm diameter filters were not available in 2011), we have less
Cram
160
confidence in our total viral counts in this year. Rather, this apparent increase in virus abundance
was below our detection threshold for (for 2011 only) and that viruses remained at approximately
the same order of magnitude in the diluted treatment as in the control group. In the 0.2μm
dilution, viruses ended the study at similar levels to the control group. We estimate that the
viruses in the 0.02μm dilution treatment started the experiment tenfold lower in abundance than
the other treatments. We observed that viruses in this treatment also finished the experiment with
abundances at a concentration of 3*10
5
, about an order of magnitude less than we estimated at
the beginning of the experiment.
Protists
Both autotrophic and heterotrophic protists in 2011 were shown to start with lower abundance in
the experimental treatment than in the control group. After a few days the abundance of both
categories in the control group decreased to match the abundances of the protists in the
experimental treatments. In the experimental groups, the abundance of autotrophic protists
increased somewhat over the course of the experiment, while the abundance of heterotrophic
protists remained relatively constant over the course of the experiment. In both cases, by day
three of the experiment, there was more variability in protist counts between replicates than there
were between treatments.
Cram
161
Figure 4-2 Total abundances of bacteria, virus like particles, autotrophic and heterotrophic protists. For
each box, the x-axis indicates days since the beginning of the study, the y-axis is estimated cells or virus
like per ml. Shapes indicate treatments, and are indicated in the legend at right. Each of three replicates is
plotted as a separate symbol and individual bottles are tracked with a grey line. Bacteria were measured
by SYBR green microscopy in 2011 and acridine orange direct counts in 2011. Viruses were measured in
both years with SYBR green. Autotrophic and heterotrophic protists were measured with DAPI where
protists with autofluorescence were considered to be autotrophic.
Shifts in community structure were most evident in experimental treatments
Non metric multidimensional scaled plots, showing differences in bacterial community structure
between pairs of treatments (Figure 4-3), as measured by ARISA, showed that in 2010 all
samples in the control group, regardless of date, were similar to each other. This indicated that
the bacterial community of the control group changed little over time. In contrast, community
structures of samples from the experimental groups were similar to control group samples on
days 0 and 1. By day 3 samples from both dilution treatments diverged from both the control
Cram
162
group and from samples taken earlier in the experiment. Meanwhile both diluted treatments had
similar microbial communities to each other. This divergence indicated that the microbial
communities in both experimental treatments all changed in a similar fashion between days 1 and
3, when the bacterial counts increased. On day 6 the two experimental dilution treatments
diverged from each other. This divergence indicated that the communities in the experimental
treatments had a secondary change in structure after the total bacterial counts had leveled off.
The nature of this change depended on whether viruses were diluted along with the bacteria and
protists, or whether viruses were left at full concentration. It is worth noting that on day 6, while
the dilution treatments were most dissimilar from each other on day six, they were more similar
to the control group on day 6 than on day 3.
In 2011 the control group’s community structure changed slightly over the course of the
experiment, though less than changes in the experimental treatments. The diluted experimental
treatments diverged from the control group starting on day 2 and became even more different
from the control group by day 3. By day 5, samples in the two experimental treatments, as in
2011, appeared to have diverged from each other. By days 7 and 9, in contrast, community
structure in all but one of the replicates in both experimental dilution treatments began to
converge with the control group.
Analysis of similarities indicated that there was a statistically significant difference between the
three treatments on days 3 and 6 in 2010 and every day of the study except day 0 in 2011
(Supplemental Table 4-1). No statistically significant differences between any pair of treatments
were evident due to sample size considerations. However, non-significant R values from the
analysis of similarities supported the patterns described above.
Cram
163
Permutational ANOVA pooled results from both years and with this additional statistical power
found statistically significant differences, for each pair of treatments, between years and
treatments where the ANOSIM could not (Table 4-1). For all pairs of samples considered, the
Permutational ANOVA confirmed that, regardless of the subset of samples considered,
communities differed between years. When the control group was compared to either the 0.2μm
or 0.02μm dilution treatments, there were statistically significant differences between
communities in each of the two treatments. There was also a statistically significant
year*treatment interaction term suggesting the effect of the treatments on community structure
differed between years. When the 0.2μm and 0.02μm dilution treatments were compared, there
was a statistically significant difference between treatments when data from day 6 2010 was
pooled with data from day 5 in 2011, but not on days 1-3 of the study.
Cram
164
Figure 4-3 Non metric multidimensional scaling plots in which distance between points represents Bray-
Curtis dissimilarity between the microbial communities in pairs of samples. In all cases, symbols
represent treatment, and are the same as in Figure 4-2. Numbers indicate the day the sample was
collected. Thus a triangle with a three inside would represent a sample from the 0.02μm dilution treatment
that was collected on day 3 of the study. Grey lines connect individual bottles over time. A. All samples
taken from the experiment that was run in 2010. Note that the control group (circles) stays relatively
consistent over time, while both the 0.2μm (squares) and 0.02μm dilution treatments (triangles) deviate
from the control by day 3 and from each other on day 6. B. Samples from 2011. C. Samples taken in both
2010 and 2011. Notice that the community structures of the experiments differ at all time-points between
years but both change over time. Permutational ANOV A tests (Table 4-1) suggest that differences between
the control group and the experimental treatments are event by day 3, and evident between the
experimental treatments by day 6.
Cram
165
Groups
Day
2010
Day
2011
R
2
Year
R
2
Treatment
R
2
Year*
Treatment
P
Year
P
Treatment
P.
Year *
Treatment
All 0 0 0.85 0.04 0.01 0.000 0.198 0.781
All 1 1 0.83 0.04 0.03 0.000 0.209 0.238
All 3 3 0.35 0.30 0.19 0.000 0.000 0.000
All 6 5 0.36 0.23 0.16 0.000 0.000 0.001
Ctrl – 0.2um 0 0 0.83 0.04 0.01 0.001 0.207 0.644
Ctrl – 0.2um 1 1 0.85 0.02 0.04 0.001 0.506 0.095
Ctrl – 0.2um 3 3 0.41 0.29 0.17 0.000 0.000 0.000
Ctrl – 0.2um 6 5 0.51 0.16 0.12 0.000 0.002 0.005
Ctrl – 0.02um 0 0 0.86 0.02 0.00 0.001 0.310 0.757
Ctrl – 0.02um 1 1 0.85 0.04 0.02 0.000 0.126 0.273
Ctrl – 0.02um 3 3 0.33 0.32 0.21 0.000 0.000 0.000
Ctrl – 0.02um 6 5 0.34 0.26 0.18 0.000 0.000 0.001
0.2um-0.02um 0 0 0.88 0.04 0.00 0.001 0.143 0.800
0.2um-0.02um 1 1 0.83 0.03 0.02 0.001 0.247 0.284
0.2um-0.02um 3 3 0.61 0.09 0.06 0.000 0.089 0.182
0.2um-0.02um 6 5 0.45 0.12 0.09 0.000 0.043 0.088
Table 4-1 Results of analysis of similarities investigating whether there is greater Bray-Curtis
dissimilarity between groups than within groups for each pair of treatments at each day of each study
year. The year indicates the study year under analysis. Treatments indicates the pairs of treatments being
considered where “Ctrl” indicates the whole seawater control group, and 0.2μm and 0.02μm indicate the
dilution treatments. “All” indicates that all three treatments are considered together. “R” indicates the
analysis of similarities’ R statistic. “P” indicates the associated p-value. Note that p-values are only less
than 0.05 in the all samples comparison because sample sizes are too small in the other comparisons.
Identification of OTUs that drove community structure differences
Similarity percentages analysis identified OTUs that contributed most strongly to the
dissimilarities between the control treatment and 0.2um dilution treatment on days 3, 5 and 6. It
also identified OTUs that contributed to the dissimilarities between the 0.2um and 0.02um
dilution treatments on days 5 and 6. We report organisms that were the strongest contributors to
differences between these groups (Table 4-2). We examined the abundances of organisms that
drove differences between control group and 0.2um treatment (Table 4-2) on day 3 (both years)
(Figure 4-4) day 6 (2010) (Figure 4-5a) and day 5 (2011) (Figure 4-5b). Abundance profiles
suggested that the OTUs that were most abundant in the control group, tended to have started out
Cram
166
as abundant at day 0 and remained abundant over the course of the experiment. In the
experimental dilution treatment these same OTUs started out with one tenth their control group
abundance and changed little over the course of the experiment, even as other OTU grew. The
OTUs that were more abundant in the 0.2μm dilution treatment than the control group generally
started the experiment with undetectable abundance, and then grew rapidly between days 1 and 3
or between days 1 and 6, depending on organism. These organisms usually reached a peak
abundance after which their abundance decreased, particularly in 2011, where total bacterial
abundances decreased in the dilution treatments towards the end of the study.
Organisms that drove differences between the 0.2μm and 0.02μm dilution treatments (on days 5
and 6) generally showed less obvious patterns than the ones that separated the control group and
0.2μm dilution treatment, with differences in patterns found between replicates. Many of these
were the same organisms found to differentiate the diluted groups from controls. Generally, it
appeared that these organisms could show a number of different patterns. Some grew in both
dilution treatments, but grew the most if viruses were also diluted, while others grew more if
viruses were not diluted. Still other taxa were abundant in the control group, and did not grow
much in response to dilution but did slightly better in one of the dilution treatments than the
other.
Cram
167
Figure 4-4 Abundances of selected OTUs over time. A. Five organisms that were more abundant in the
whole seawater control group than the 0.2μm dilution treatment in summer 2010 that most strongly
contributed to the difference between the community structure of the control group and 0.2um
experimental group on day 3 of the study. B. Ibid, but for summer of 2011. C. Five organisms that were
more abundant in the 0.2um Dilution group than in the control treatment in the summer of 2010. C. Ibid,
but for summer of 2011.
Cram
168
Figure 4-5 Abundances of selected OTU over time. Each sub-figure contains the five OTUs that were
most abundant in one experimental treatment, on day 6 of the study in 2010 or day 5 of the study in 2011
that contributed the most to differences between the control group and the 0.2um dilution treatments. A.
Five organisms that were more abundant in the whole seawater control group than the 0.2um dilution
treatment in summer 2010. B. Ibid, but for summer of 2011. C. The five organisms that were more
abundant in the 0.2um dilution group than in the control treatment in the summer of 2010. C. Ibid, but for
summer of 2011.
Cram
169
Year Day
Comparison A –
B Drives
short-name_ITS-
length Identity contr sd ratio av.A av.B
2010 3 Ctrl – 0.2μm Ctrl SAR11_666.4 SAR11 0.173 0.032 5.340 0.387 0.041
2010 3 Ctrl – 0.2μm Ctrl SAR11_686.9 SAR11 0.069 0.026 2.630 0.187 0.049
2010 3 Ctrl – 0.2μm Ctrl OCS155_435.5 OCS155 0.031 0.005 6.340 0.076 0.015
2010 3 Ctrl – 0.2μm Ctrl SAR11_670.5 SAR11 0.030 0.020 1.520 0.069 0.012
2010 3 Ctrl – 0.2μm Ctrl SAR11_682.4 SAR11 0.020 0.011 1.820 0.050 0.010
2010 3 Ctrl – 0.2μm 0.2μm Marino_779.2 Marinoscillum 0.143 0.038 3.790 0.016 0.301
2010 3 Ctrl – 0.2μm 0.2μm OTU_808 Unidentified 0.051 0.007 7.300 0.004 0.107
2010 3 Ctrl – 0.2μm 0.2μm OTU_502.9 Unidentified 0.020 0.016 1.270 0.000 0.040
2010 3 Ctrl – 0.2μm 0.2μm Fluvii_646.9 Fluviicola 0.017 0.012 1.380 0.019 0.028
2010 3 Ctrl – 0.2μm 0.2μm SAR324_770.5 SAR324 0.016 0.010 1.660 0.003 0.036
2011 3 Ctrl – 0.2μm Ctrl SAR11_670.5 SAR11 0.046 0.009 4.828 0.113 0.021
2011 3 Ctrl – 0.2μm Ctrl SAR116_689.8 SAR116 0.030 0.011 2.585 0.089 0.030
2011 3 Ctrl – 0.2μm Ctrl OTU_618.6 Unidentified 0.025 0.007 3.765 0.058 0.008
2011 3 Ctrl – 0.2μm Ctrl AEGEAN_674.2 AEGEAN-169 0.021 0.009 2.438 0.055 0.013
2011 3 Ctrl – 0.2μm Ctrl AEGEAN_662 AEGEAN-169 0.016 0.004 4.223 0.037 0.004
2011 3 Ctrl – 0.2μm 0.2μm OTU_810.6 Unidentified 0.056 0.061 0.918 0.001 0.111
2011 3 Ctrl – 0.2μm 0.2μm SAR406_762.8 SAR406 0.055 0.012 4.686 0.001 0.110
2011 3 Ctrl – 0.2μm 0.2μm SAR324_767.5 SAR324 0.046 0.007 6.548 0.000 0.091
2011 3 Ctrl – 0.2μm 0.2μm OTU_811 Unidentified 0.041 0.044 0.928 0.001 0.081
2011 3 Ctrl – 0.2μm 0.2μm SAR324_770.5 SAR324 0.031 0.006 4.929 0.000 0.062
2010 6 Ctrl – 0.2μm Ctrl SAR11_666.4 SAR11 0.110 0.052 2.110 0.374 0.155
2010 6 Ctrl – 0.2μm Ctrl SAR11_686.9 SAR11 0.056 0.033 1.700 0.199 0.087
2010 6 Ctrl – 0.2μm Ctrl SAR11_670.5 SAR11 0.038 0.022 1.720 0.121 0.046
2010 6 Ctrl – 0.2μm Ctrl SAR11_682.4 SAR11 0.022 0.007 3.080 0.064 0.045
2010 6 Ctrl – 0.2μm Ctrl SAR116_689.8 SAR116 0.009 0.006 1.650 0.042 0.040
2010 6 Ctrl – 0.2μm 0.2μm OTU_808 Unidentified 0.066 0.009 7.410 0.020 0.152
2010 6 Ctrl – 0.2μm 0.2μm NS9_556.7 NS9 0.036 0.028 1.290 0.000 0.072
2010 6 Ctrl – 0.2μm 0.2μm OTU_758.3 Unidentified 0.022 0.016 1.350 0.001 0.044
2010 6 Ctrl – 0.2μm 0.2μm AEGEAN_676.9 AEGEAN-169 0.016 0.013 1.260 0.003 0.035
2010 6 Ctrl – 0.2μm 0.2μm Thioba_840 Thioba 0.012 0.005 2.300 0.016 0.039
2011 6 Ctrl – 0.2μm Ctrl OTU_618.6 Unidentified 0.024 0.009 2.710 0.055 0.006
2011 6 Ctrl – 0.2μm Ctrl SAR11_692.2 SAR11 0.023 0.018 1.290 0.065 0.060
2011 6 Ctrl – 0.2μm Ctrl SAR11_670.5 SAR11 0.021 0.015 1.370 0.101 0.060
2011 6 Ctrl – 0.2μm Ctrl OTU_480 Unidentified 0.016 0.002 9.230 0.031 0.000
2011 6 Ctrl – 0.2μm Ctrl AEGEAN_674.2 AEGEAN-169 0.014 0.009 1.620 0.061 0.036
2011 6 Ctrl – 0.2μm 0.2μm OTU_810.6 Unidentified 0.055 0.040 1.380 0.004 0.112
2011 6 Ctrl – 0.2μm 0.2μm SAR324_767.5 SAR324 0.036 0.012 3.150 0.000 0.073
2011 6 Ctrl – 0.2μm 0.2μm Pseudo_937.8 Pseudospirillum 0.025 0.010 2.450 0.010 0.060
2011 6 Ctrl – 0.2μm 0.2μm SAR406_762.8 SAR406 0.019 0.009 2.020 0.001 0.038
2011 6 Ctrl – 0.2μm 0.2μm SAR324_773.1 SAR324 0.016 0.010 1.600 0.008 0.034
2010 6 0.2μm – 0.02μm 0.2μm SAR11_666.4 SAR11 0.076 0.051 1.480 0.155 0.002
2010 6 0.2μm – 0.02μm 0.2μm OTU_808 Unidentified 0.057 0.023 2.480 0.152 0.039
2010 6 0.2μm – 0.02μm 0.2μm SAR11_686.9 SAR11 0.036 0.032 1.120 0.087 0.018
2010 6 0.2μm – 0.02μm 0.2μm NS9_556.7 NS9 0.036 0.028 1.290 0.071 0.000
2010 6 0.2μm – 0.02μm 0.2μm SAR11_670.5 SAR11 0.023 0.017 1.330 0.046 0.001
2010 6 0.2μm – 0.02μm 0.02μm Marino_779.2 Marinoscillum 0.063 0.049 1.280 0.019 0.135
2010 6 0.2μm – 0.02μm 0.02μm NS4_579.3 NS4 0.055 0.072 0.760 0.003 0.112
2010 6 0.2μm – 0.02μm 0.02μm SAR11_682.4 SAR11 0.046 0.032 1.440 0.045 0.107
2010 6 0.2μm – 0.02μm 0.02μm SAR324_770.5 SAR324 0.039 0.029 1.340 0.007 0.084
2010 6 0.2μm – 0.02μm 0.02μm SAR116_689.8 SAR116 0.034 0.020 1.730 0.040 0.090
2011 6 0.2μm – 0.02μm 0.2μm SAR11_670.5 SAR11 0.024 0.016 1.544 0.060 0.011
2011 6 0.2μm – 0.02μm 0.2μm SAR116_689.8 SAR116 0.015 0.009 1.573 0.045 0.020
2011 6 0.2μm – 0.02μm 0.2μm AEGEAN_674.2 AEGEAN-169 0.013 0.010 1.337 0.036 0.009
2011 6 0.2μm – 0.02μm 0.2μm Pseudo_937.8 Pseudospirillum 0.009 0.006 1.475 0.061 0.059
2011 6 0.2μm – 0.02μm 0.2μm SAR11_683.9 SAR11 0.009 0.010 0.923 0.043 0.030
2011 6 0.2μm – 0.02μm 0.02μm OTU_810.6 Unidentified 0.082 0.057 1.453 0.112 0.268
Cram
170
Table 4-2 Results of similarity percentages tests which identified, for pairs of treatments, on specific days,
five organisms each that were more abundant in each of the two treatments and that contributed the most
to differences in community structure between the two treatments. “Year”, indicates the year in which the
study was run. “Day” indicates the number of days elapsed since the beginning of the study. “Comparison
A-B” indicates the pair of treatments investigated. “Drives”, indicateds the treatment in which an
organism is most abundant. “Short-name_ITS-length” indicates an abbreviated taxonomic identity and
ITS length of the identified organism. “Identity” is the unabbreviated taxonomic identity of the organism.
“Contr” indicates the mean percent contribution of that organism to the Bray-Curtis dissimilarities
between the two groups. “SD” is the standard deviation of percent contributions. “Ratio” is the mean to
standard deviation ratio of the contribution. “Av.A” and “Av.B” indicate the relative abundance of the
organism in the category listed first and second, respectively, in the “Comparison A-B” column.
Abundance profiles in the San Pedro Chanel of species that drove differences between
treatments
Organisms that were most abundant in the control group were, in natural surface waters, usually
abundant year round, though all fluctuated in abundance over the course of the year
(Supplemental Figure 4-1A and B; Supplemental Figure 4-2A and B). Examples of these
organisms included several SAR11 OTUs and Actinobacterial OCS155 OTUs. Organisms that
grew in the 0.2μm diluted treatments showed one of three patterns (Supplemental Figure 4-1C
and D; Supplemental Figure 4-2C and D): Some OTUs were usually rare, with occasional spikes
in abundance in surface waters; examples include a Marinoscillum with ARISA fragment length
of 779.2bp and a Fluviicola with 646.9 fragment length (Supplemental Figure 4-1C). Some
OTUs spiked occasionally, often with highest abundance in one season, such as an unidentified
bacteria with 808.0bp fragment length (Supplemental Figure 4-1C) and a Pseudospirillum with
973.8 fragment length (Supplemental Figure 4-2D). Still other OTUs were usually more
abundant in deeper water layers. Some were putatively identified as related to the AEGEAN-169
and SAR324 and SAR406 groups (Supplemental Figure 4-1C and D; Supplemental Figure 4-2C
2011 6 0.2μm – 0.02μm 0.02μm SAR11_692.2 SAR11 0.020 0.012 1.745 0.060 0.070
2011 6 0.2μm – 0.02μm 0.02μm Prochl_812.4 Prochlorococcus 0.014 0.015 0.914 0.017 0.022
2011 6 0.2μm – 0.02μm 0.02μm SAR406_762.8 SAR406 0.013 0.009 1.502 0.039 0.058
2011 6 0.2μm – 0.02μm 0.02μm SAR324_773.1 SAR324 0.013 0.008 1.518 0.034 0.041
Cram
171
and D), though since the clones we used to identify these ARISA fragments were isolated from
deep water and not the surface, we recognize these could actually be other bacteria with similar
ITS length to these deep water groups.
OTUs that drove differences between the 0.2μm and 0.02μm dilution treatments showed a mix of
temporal abundance patterns, since they were comprised both of OTUs that grew under diluted
conditions and OTUs that were abundant in the control group.
Discussion
This is the first study, to our knowledge, to examine fine-scale taxonomic changes in community
structure in response to alleviation of predator/phage pressure. It allows an exciting new look
into how both of these top-down pressures affect community structure. Overall, different
portions of the community respond differently to reduce phage or grazer pressure, implying that
bacteria have a range of strategies with some bacteria growing faster and others growing slower
but having better defense against grazing and/or infection.
Bacteria, but not viruses or protists, recover from dilution
The recovery of bacteria to nearly starting concentrations, in both iterations of the experiment,
suggests that total bacterial concentrations are controlled, at least partially, by top down forces:
Since no nutrients were added to our bottles we expect that growth rates of organisms should
remain essentially the same after dilution, while the decreased predator and virus densities
should lead to decreased removal rates. Thus the increase in abundance of bacteria in the
experimental groups is likely a result of this decreased grazing pressure no longer balancing
growth rates. We then expect that the bacteria are most abundant after dilution and re-growth are
those that are both growing and being grazed at the highest rate in the wild. That bacteria
returned to the normal surface ocean concentration of 10
6
cells/ml and then did not grow could
Cram
172
be due to either “top down” or “bottom up” limitation, or else a combination of both factors. Top
down limitation would imply that grazers were able to re-establish high removal rates toward the
end of the study. Bottom up limitation would be the case in which bacterial growth rates slowed
down over the course of the experiment, which might be caused by the expanding bacterial
population removing labile dissolved organic matter or nutrients faster than they could be
replaced.
The finding that protists did not recover over the course of the 2011 experiment suggests that
some factor other than increased encounter rates with protists caused the bacteria to level off at
10
6
cells/ml. This could suggest that the bottom up factors described above were dominant.
Alternatively increased top down pressure could have manifested itself if the protistan
community changed such that there was a higher abundance of predators that are more voracious
or else are specialized to eat the faster growing bacteria. It will be valuable in future studies to
examine changes in protistan community structure in response to dilution.
The lack of recovery of virus counts is consistent with earlier findings that viruses do not recover
when they are diluted more than a certain threshold (Wilcox & Fuhrman 1994). In our 2010
experiment, our viral dilution treatments started with a bacterial concentration of ~3*10
5
cells/ml
and viral dilution treatments at ~6*10
6
cells/ml, which multiplied together give 1.8*10
12
cells*vlp/ml
2
. In 2011 this factor (2x10
5
bacteria/ml * 2x10
7
viruses/ml) was 4*10
11
cells*vlp/ml
2
. Both of these values were within range of or below the previously identified
recovery threshold, suggesting the non-recovery of viruses in our experiment were consistent
with previous findings. The observation that viruses decreased by about an order of magnitude
over the course of the nine day 2011 experiment suggested that most viruses at our site degraded
or were otherwise removed over the course of a week.
Cram
173
Likely trade-off between predation resistance and growth
Bacteria community composition in both experimental dilution treatments changed
coincidentally with increases in bacterial abundance over the first three days of the experiment,
while bacteria in the control groups stayed the same (Figure 4-2, Figure 4-3, Table 4-1). This
concurrence of growth with change indicates that the bacteria that drove the recovery in the
dilution treatments were different OTUs than the ones that were dominant at the beginning of the
experiment. We contend that dilution, by decreasing interactions between bacteria and grazers,
decreases the removal of grazing susceptible organisms and favors organisms that were already
growing quickly, but were previously being removed by grazers. Thus we argue that the
dominant bacteria in the dilution treatments on day 3 are more grazer susceptible than those in
the control group. Because there was no statistically detectable difference between the two
experimental dilution treatments on day 3 (Table 4-1), changes due to viral influence did not
have time to be revealed in the overall community structure at this time scale.
In contrast, the presence of statistically detectable, though weaker, differences between the two
experimental treatments by day 6 in 2010 and day 5 in 2011, suggested that decreasing the rates
of viral infection did alter community structure. This five to six day delay in the maximum effect
of virus amendment on community structure is slower than findings seen elsewhere (Winter et al.
2004; Weinbauer et al. 2007), which may reflect differences between environments. Since total
bacterial abundances were similar in both experimental groups by day 6, we contend that
differences between bacteria in the two treatments identify bacteria that were virus susceptible
(more abundant in the 0.02μm dilution treatment where viruses were sparse) as well as bacteria
that were virus resistant (more abundant in the 0.2μm dilution treatment where viruses were
undiluted).
Cram
174
Surface (but not deep) waters are likely dominated by grazer resistant bacteria
The abundance patterns of those OTUs that drove the differences in community structure
between the control group and 0.2μm diluted groups revealed that different OTUs responded in
different ways to decreased grazing pressure. Since most groups that dominated the control
treatments maintained constant abundance over time when they were diluted, it is likely that
these are slow growing bacteria. Their abundance in the control group is likely high compared to
the organisms that grew quickly in the experimental dilutions because they are able to avoid
removal by predation. In contrast, abundance patterns of organisms that grew to high abundance
in the diluted treatments suggest these organisms grow until they reach densities similar to the
control groups. Meanwhile these organisms are rare or undetectable in the control group. These
growth patterns suggest that the organisms found in the dilution treatments grow quickly, but are
continually removed in the control group by grazing. These results support the hypothesis
(Jürgens & Matz 2002; Winter et al. 2010) that there is a selection trade-off between fast growth
and predation resistance.
Most of the organisms that were more abundant in the control group than the dilution treatments
were common throughout the year in the surface waters (Supplemental Figure 4-1, Supplemental
Figure 4-2). Examples of these bacteria include SAR11s that have been postulated to have
predator resistance due to their small size (Malmstrom et al. 2004), and Actinobacteria which in
freshwater environments have not only been shown to resist grazing but also to take up
compounds released by protists grazing on other bacteria (Eckert et al. 2013). Meanwhile the
OTUs that grew in the dilution treatments were less common in surface water, suggesting that
the more grazing resistant organisms dominate marine surface waters throughout the year.
Organisms that grew in the 0.2μm dilution treatment were in some cases rare throughout the
Cram
175
water column with occasional spikes in abundance, suggesting that despite their fast growth rates
they are usually kept at low abundance by grazing (as in the control group of this experiment).
Spikes are likely results of temporary changes to food-web structure or bottom up factors that
decrease removal and/or increase growth of these organisms.
Meanwhile, many of the organisms that grew in the 0.2μm dilution treatments were found
throughout the year in the mesopelagic region of the ocean, suggesting that the lower abundance
of bacterial grazers in the deep ocean gives a competitive advantage to fast growing, rather than
grazer resistant organisms. This finding reflects observations in the Baltic Sea in which both
grazing and viral impacts on plankton are significantly lower in deep waters than at the surface
(Anderson et al. 2012).
Both grazing resistant and grazing susceptible organisms face a trade-off between infection
resistance and susceptibility
Comparison of the organisms that were more abundant in the 0.2μm and 0.02μm dilution
treatments identified OTUs that fared relatively better when virus abundance was decreased to
those that performed relatively better when virus abundance remained high. It is probable that
differences in the amount of bacterial infection by viruses drive differences between these two
diluted treatments. Organisms that fare better in the treatment with non-diluted viruses likely
have greater infection resistance while organisms that fare better when viruses are removed are
likely more susceptible to infection. There are likely trade-offs to being more predation and/or
virus resistant, namely slower growth, and OTUs appear to deploy a number of strategies. For
instance organisms that do best in the control group, but under dilution grow slightly faster in the
absence of viruses would be said to be more predation resistant, but infection susceptible (eg.
SAR11_682.4 in Supplemental Figure 4-3C), while those that are relatively more abundant in the
Cram
176
presence of viruses would be said to be more infection resistant (eg. SAR11_666.4 in
Supplemental Figure 4-3A). The observation that different SAR11 OTUs are identified as having
different infection resistance suggests variable strategies within this group, possibly promoting
the different dynamics of SAR11 over time(Carlson et al. 2009; Steele et al. 2011; Chow et al.
2013) and in relation to other organisms(Steele et al. 2011). Predation susceptible organisms face
similar tradeoffs with some organisms only growing fastest in the absence of viruses (e.g.
NS4_579.3 in Supplemental Figure 4-3C) and others doing relatively better when viruses remain
than when they are removed (e.g. Unknown OTU_808.0 in Supplemental Figure 4-3A). Thus the
different combinations of predation and infection resistance strategies may contribute in part to
why marine microbial communities are so diverse.
Grazing shaped communities more than infection
It appears that there are larger differences between the control group and both the 0.2μm and
0.02μm dilution treatments, than between the 0.2μm dilution treatment and the 0.02μm dilution
treatment (Figure 4-3, Table 4-1). Furthermore individual OTUs show clearer differences in their
abundance patterns between the control group and the dilution treatments than between the
0.2μm and 0.02μm dilution treatments (Figure 4-3, Figure 4-4, Figure 4-5). These patterns
suggest that the effect of diluting out bacteria and their grazers is greater than the added effect of
also removing viruses. This implies that grazing and perhaps inter-microbial associations have a
greater effect on community structure than infection, at least on the one week time scales
studied.
Successional patterns were evident later in the experiment
The observation that microbial communities in the experimental treatments were more similar to
microbial communities in the control group on day six than they were on day 3 in 2010, and that
Cram
177
experimental treatments became more like the control group on and after day 7 in 2011 (Figure
4-3), suggested successional patterns. It is likely that slower growing bacteria were able to catch
up in abundance with their faster growing counterparts, and perhaps possible that grazing
pressure increased (even if protist numbers did not, in 2010) due to changes in protistan
community structure in the experimental groups, selecting for similar organisms to the ones
found in the control group. Alternatively, in 2011 the collapse in total bacterial abundances later
in the study may have been due to the organisms that initially took over the experiment having
boom-bust dynamics typical of organisms in culture (Madigan & Martinko 2005) and that those
communities collapsed due to exhausting some resource or other factors.
Alternative interpretations and methodological considerations
We have emphasized predator prey interactions as the predominant factor shaping microbial
community structure differences between the 0.2μm and 0.02μm dilution treatments. Our
previous analysis relies on the assumptions that bacteria live on organic matter, that DOM is not
introduced by our filtration method, and that DOM is not affected by changes to bacterial
abundance, but it is likely that none of these assumptions are entirely valid. While dissolved
organic matter is generally accepted to be the main source of energy and carbon for most marine
bacteria (Kirchman 2008), bacteria are known to consume particulate organic matter and these
particles are presumably removed by our filtration process. Of course, a fraction of bacteria are
photo-autotrophic (namely Synecchococcus and Prochlorococcus) and these bacteria should face
similar growth efficiency tradeoffs to the heterotrophs. While, most organic matter in surface
waters is dissolved rather than particulate (Benner et al. 1997), only a small fraction of DOM is
labile (Nagata 2008). Most bacteria in marine surface waters are believed to be planktonic and
most heterotrophic activity is associated with the free living component in surface waters
Cram
178
(Iriberri et al. 1987), so we believe the not particle associated, presumably DOM consuming
bacteria likely dominate the patterns that we see. It is probable that our filtration method
somewhat increases the labile DOM pool at the beginning of the experiment because some
fraction of the bacterial and protistan cells caught by our filters likely lyse, releasing organic
matter into our filtrate. We minimized this effect by filtering our samples carefully and at low
pressure, but the amount of DOM released is still unknown. DOM enrichment would encourage
the growth of copeotrophic bacteria, and our bacteria that respond to the dilution treatments may
have responded to this effect. Another potentially confounding variable is that the rapidly
growing bacteria in the experimental dilution treatments remove labile DOM from the water,
decreasing the abundance of DOM. This factor would suggest the experimental dilution groups,
after a possible initial increase in DOM from the filtration process might have lower DOM
concentrations than the control group after some unknown number of days. This factor might
help to mitigate the effect of any DOM enrichment, but may also even select for more
oligotrophic bacteria later in the experiment.
It is worth considering that inter-microbial associations beyond predation may play a partial role
in shaping community structure and that decreasing density may also favor organisms that have
positive interactions with other microbes. For instance an organism that relies on compounds
released by other bacteria, such as B-Vitamins (Sanudo-Wilhelmy et al. 2014), would do poorly
in our dilution treatments. In marine environments, planktonic microbes are already quite dilute
and likely do not compete over space, but competition for some resource could also influence
differences between our treatments. Viruses likewise may have effects on community structure
beyond infection; they may for instance serve as a food source, and removing them may have
adverse effects on some organisms.
Cram
179
Protistan grazing has been shown to stimulate viral infection (Simek et al. 2001), and our
dilution method only identifies decreased virus pressure under conditions of decreased grazing. It
is possible that if viruses had been diluted while leaving bacteria and protists at a constant
concentration (not a trivial task), a greater effect of viral removal might be evident. Furthermore,
our dilution method decreases predation on bacteria but also likely decreases interactions
throughout the food-web, which may complicate the dynamics shown here.
Conclusions
Our results suggest that microbes that are abundant in surface waters throughout the San Pedro
Channel are likely more grazing resistant than less abundant organisms. Meanwhile there are
likely many actively growing organisms that are kept at relatively low abundance by protistan
grazing. Variability in protistan grazing and viral infection likely contributes to variability in the
surface community structure. In deep water, it is likely that predation is less of an important
factor in shaping microbial communities, since many organisms that are abundant in deep waters
grew in surface water when predation pressure was reduced. Our results support the steady state
of the “kill the winner” hypothesis (Thingstad & Lignell 1997; Thingstad 2000; Winter et al.
2010) in which tradeoffs between predation resistance, virus resistance and faster growth shape
microbial communities and promote diversity of surface communities.
Cram
180
Supplement
Supplemental Figure 4-1 Distribution of groups, in the San Pedro Ocean Time-Series (SPOT) over time,
which separated control group and 0.2μm dilution treatments on day 3 of the study. Each panel indicates
abundance of a particular OTU, whose abbreviated identity and ITS-length are given in the grey bar at
right. The Y axis of each panel indicates log-scaled relative abundance. The x-axis indicates the day of
the year each sample was taken. The black numbers represent samples taken at the surface (5m) and the
number indicates the year from which the sample was taken, minus two thousand, so a zero indicates a
sample taken in the year 2000 and a ten indicates a sample taken in 2010. Colored shapes indicate
samples taken at other depths than the surface and are described in the legend. A. OTUs that are more
abundant in the control group in 2010. B. Ibid for 2011. C. OTUs that are more abundant in the 0.2μm
dilution treatment in 2010. D. Ibid for 2011.
Cram
181
Supplemental Figure 4-2 Distribution of groups that separate the control group and 0.2um dilution
treatment on day six of the experiment. Layout as in Supplemental Figure 4-1.
Supplemental Figure 4-3 Distribution of groups that separate the 0.2um dilution treatment and 0.02um
dilution treatment on day six of the experiment. Layout as in Supplemental Figure 4-1.
Cram
182
Year Treatments Day R P
2010 All 0 0.02 0.463
1 0.07 0.332
3 0.88 0.004
6 0.60 0.002
Ctrl vs 0.2μm 0 -0.56 1.000
1 0.00 0.607
3 1.00 0.092
6 0.56 0.103
Ctrl vs 0.02μm 0 0.07 0.416
1 0.11 0.295
3 1.00 0.086
6 0.78 0.097
0.2μm vs 0.02μm 0 0.56 0.495
1 0.11 0.199
3 0.52 0.073
6 0.56 0.104
2011 All 0 0.22 0.145
1 0.48 0.032
2 0.83 0.003
3 0.84 0.026
5 0.76 0.008
7 0.28 0.034
9 0.61 0.021
Ctrl vs 0.2μm 0 0.44 0.106
1 0.50 0.354
2 0.93 0.111
3 1.00 0.096
5 0.93 0.090
7 0.44 0.096
9 1.00 0.097
Ctrl vs 0.02μm 0 -0.33 0.777
1 0.58 0.085
2 1.00 0.099
3 1.00 0.107
5 1.00 0.103
7 0.56 0.103
9 0.82 0.099
0.2μm vs 0.02μm 0 0.42 0.230
1 0.42 0.105
2 0.41 0.105
3 0.00 0.628
5 0.11 0.306
7 0.04 0.601
9 -0.15 0.893
Supplemental Table 4-1 Results of analysis of similarities investigating whether there is greater Bray-
Curtis dissimilarity between groups than within groups for each pair of treatments at each day of each
study year. The year indicates the study year under analysis. Treatments indicates the pairs of treatments
Cram
183
being considered where “Ctrl” indicates the whole seawater control group, and 0.2μm and 0.02μm
indicate the dilution treatments. “All” indicates that all three treatments are considered together. “R”
indicates the analysis of similarities’ R statistic. “P” indicates the associated p-value. Note that p-values
are only less than 0.05 in the all samples comparison because sample sizes are too small in the other
comparisons.
References
Anderson MJ. (2001). A new method for non-parametric multivariate analysis of variance.
Austral Ecol 26:32–46.
Anderson R, Winter C, Juergens K. (2012). Protist grazing and viral lysis as prokaryotic
mortality factors at Baltic Sea oxic-anoxic interfaces. Mar Ecol Prog Ser 467:1–14.
Bonilla-Findji O, Herndl GJ, Gattuso J-P, Weinbauer MG. (2009). Viral and Flagellate Control of
Prokaryotic Production and Community Structure in Offshore Mediterranean Waters. Appl
Environ Microbiol 75:4801–4812.
Bouvier T, del Giorgio PA. (2007). Key role of selective viral-induced mortality in determining
marine bacterial community composition. Environ Microbiol 9:287–297.
Breitbart M. (2012). Marine Viruses: Truth or Dare. Annu Rev Mar Sci 4:425–448.
Brown MV , Schwalbach MS, Hewson I, Fuhrman JA. (2005). Coupling 16S-ITS rDNA clone
libraries and automated ribosomal intergenic spacer analysis to show marine microbial diversity:
development and application to a time series. Environ Microbiol 7:1466–1479.
Chow C-ET, Fuhrman JA. (2012). Seasonality and monthly dynamics of marine myovirus
communities: Marine myovirus community dynamics at SPOT. Environ Microbiol 14:2171–
2183.
Chow C-ET, Kim DY , Sachdeva R, Caron DA, Fuhrman JA. (2014). Top-down controls on
bacterial community structure: microbial network analysis of bacteria, T4-like viruses and
protists. ISME J 8:816–829.
Chow C-ET, Sachdeva R, Cram JA, Steele JA, Needham DM, Patel A, et al. (2013). Temporal
variability and coherence of euphotic zone bacterial communities over a decade in the Southern
California Bight. ISME J 7:2259–2273.
Cram
184
Clarke KR. (1993). Non-parametric multivariate analyses of changes in community structure.
Aust J Ecol 18:117–143.
Cram JA, Chow C-ET, Sachdeva R, Needham DM, Parada AE, Steele JA, et al. (2014). Seasonal
and interannual variability of the marine bacterioplankton community throughout the water
column over ten years. ISME J.
Eckert EM, Baumgartner M, Huber IM, Pernthaler J. (2013). Grazing resistant freshwater
bacteria profit from chitin and cell-wall-derived organic carbon. Environ Microbiol 15:2019–
2030.
Fuhrman JA. (1999). Marine viruses and their biogeochemical and ecological effects. Nature
399:541–548.
Fuhrman JA, Comeau DE, Hagström Å, Chan AM. (1988). Extraction from Natural Planktonic
Microorganisms of DNA Suitable for Molecular Biological Studies. Appl Environ Microbiol
54:1426–1429.
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
103:13104 –13109.
Fuhrman JA, Steele JA, Hewson I, Schwalbach MS, Brown MV , Green JL, et al. (2008). A
latitudinal diversity gradient in planktonic marine bacteria. Proc Natl Acad Sci 105:7774–7778.
Gilbert JA, Field D, Swift P, Newbold L, Oliver A, Smyth T, et al. (2009). The seasonal structure
of microbial communities in the Western English Channel. Environ Microbiol 11:3132–3139.
Hahn MW, Höfle MG. (2001). Grazing of protozoa and its effect on populations of aquatic
bacteria. FEMS Microbiol Ecol 35:113–121.
Hewson I, Fuhrman JA. (2003). Viriobenthos production and virioplankton sorptive scavenging
by suspended sediment particles in coastal and pelagic waters. Microb Ecol 46:337–347.
Hobbie JE, Daley RJ, Jasper S. (1977). Use of nuclepore filters for counting bacteria by
fluorescence microscopy. Appl Environ Microbiol 33:1225–1228.
Cram
185
Jardillier L, Bettarel Y , Richardot M, Bardot C, Amblard C, Sime-Ngando T, et al. (2005).
Effects of Viruses and Predators on Prokaryotic Community Composition. Microb Ecol 50:557–
569.
Jürgens K, Güde H. (1994). The potential importance of grazing-resistant bacteria in planktonic
systems. Mar Ecol Progess Ser 112:169–188.
Jurgens K, Massana R. (2008). Protistan grazing on marine bacterioplankton. In:Microbial
Ecology of the Oceans, Hoboken, N.J., pp. 383–441.
Jürgens K, Matz C. (2002). Predation as a shaping force for the phenotypic and genotypic
composition of planktonic bacteria. Antonie Van Leeuwenhoek 81:413–434.
Jurgens K, Pernthaler J, Schalla S, Amann R. (1999). Morphological and compositional changes
in a planktonic bacterial community in response to enhanced protozoan grazing. Appl Environ
Microbiol 65:1241–1250.
Malmstrom RR, Kiene RP, Cottrell MT, Kirchman DL. (2004). Contribution of SAR11 Bacteria
to Dissolved Dimethylsulfoniopropionate and Amino Acid Uptake in the North Atlantic Ocean.
Appl Environ Microbiol 70:4129–4135.
Matz C, Jürgens K. (2003). Interaction of Nutrient Limitation and Protozoan Grazing Determines
the Phenotypic Structure of a Bacterial Community. Microb Ecol 45:384–398.
Matz C, Kjelleberg S. (2005). Off the hook – how bacteria survive protozoan grazing. Trends
Microbiol 13:302–307.
Needham DM, Chow C-ET, Cram JA, Sachdeva R, Parada A, Fuhrman JA. (2013). Short-term
observations of marine bacterial and viral communities: patterns, connections and resilience.
ISME J 7:1274–1285.
Oksanen J, Blanchet FG, Kindt R, Legendre P, Minchin PR, O’Hara RB, et al. (2013). vegan:
Community Ecology Package. http://CRAN.R-project.org/package=vegan.
Patel A, Noble RT, Steele JA, Schwalbach MS, Hewson I, Fuhrman JA. (2007). Virus and
prokaryote enumeration from planktonic aquatic environments by epifluorescence microscopy
with SYBR Green I. Nat Protoc 2:269–276.
Cram
186
Pernthaler J. (2005). Predation on prokaryotes in the water column and its ecological
implications. Nat Rev Microbiol 3:537–546.
Pommier T, CanbäCk B, Riemann L, BoströM KH, Simu K, Lundberg P, et al. (2007). Global
patterns of diversity and community structure in marine bacterioplankton. Mol Ecol 16:867–880.
Schwalbach MS, Hewson I, Fuhrman JA. (2004). Viral effects on bacterial community
composition in marine plankton microcosms. Aquat Microb Ecol 34:117–127.
Simek K, Kojecka P, Nedoma J, Hartman P, Vrba J, Dolan JR. (1999). Shifts in bacterial
community composition associated with different microzooplankton size fractions in a eutrophic
reservoir. Limnol Oceanogr 44:1634–1644.
Simek K, Pernthaler J, Weinbauer MG, Hornak K, Dolan JR, Nedoma J, et al. (2001). Changes
in bacterial community composition and dynamics and viral mortality rates associated with
enhanced flagellate grazing in a mesoeutrophic reservoir. Appl Environ Microbiol 67:2723–2733.
Steele JA, Countway PD, Xia L, Vigil PD, Beman JM, Kim DY , et al. (2011). Marine bacterial,
archaeal and protistan association networks reveal ecological linkages. ISME J 5:1414–1425.
Suttle CA. (2007). Marine viruses — major players in the global ecosystem. Nat Rev Microbiol
5:801–812.
Suzuki MT. (1999). Effect of protistan bacterivory on coastal bacterioplankton diversity. Aquat
Microb Ecol 20:261–272.
Team RDC. (2011). R: A Language and Environment for Statistical Computing. Vienna, Austria
http://www.R-project.org/.
Thingstad TF. (2000). Elements of a Theory for the Mechanisms Controlling Abundance,
Diversity, and Biogeochemical Role of Lytic Bacterial Viruses in Aquatic Systems. Limnol
Oceanogr 45:1320–1328.
Thingstad TF, Lignell R. (1997). Theoretical models for the control of bacterial growth rate,
abundance, diversity and carbon demand. Aquat Microb Ecol 13:19–27.
Weinbauer MG, Hornák K, Jezbera J, Nedoma J, Dolan JR, Šimek K. (2007). Synergistic and
antagonistic effects of viral lysis and protistan grazing on bacterial biomass, production and
diversity. Environ Microbiol 9:777–788.
Cram
187
Wickham H. (2009). ggplot2: elegant graphics for data analysis. Springer New York
http://had.co.nz/ggplot2/book.
Wilcox R, Fuhrman J. (1994). Bacterial viruses in coastal seawater: lytic rather than lysogenic
production. Mar Ecol-Prog Ser 114:35–45.
Winter C, Bouvier T, Weinbauer MG, Thingstad TF. (2010). Trade-Offs between Competition
and Defense Specialists among Unicellular Planktonic Organisms: the ‘Killing the Winner’
Hypothesis Revisited. Microbiol Mol Biol Rev 74:42–57.
Cram
188
Bibliography
Agogué H, Lamy D, Neal PR, Sogin ML, Herndl GJ. (2011). Water mass-specificity of bacterial
communities in the North Atlantic revealed by massively parallel sequencing. Mol. Ecol.
20:258–274.
Aller JY , Kemp PF. (2008). Are Archaea inherently less diverse than Bacteria in the same
environments? FEMS Microbiol. Ecol. 65:74–87.
Allers E, Wright JJ, Konwar KM, Howes CG, Beneze E, Hallam SJ, et al. (2013). Diversity and
population structure of Marine Group A bacteria in the Northeast subarctic Pacific Ocean. ISME
J. 7:256–268.
Alonso-Sáez L, Balagué V , Sà EL, Sánchez O, González JM, Pinhassi J, et al. (2007).
Seasonality in bacterial diversity in north-west Mediterranean coastal waters: assessment through
clone libraries, fingerprinting and FISH. FEMS Microbiol. Ecol. 60:98–112.
Anderson MJ. (2001). A new method for non-parametric multivariate analysis of variance.
Austral Ecol. 26:32–46.
Anderson R, Winter C, Juergens K. (2012). Protist grazing and viral lysis as prokaryotic
mortality factors at Baltic Sea oxic-anoxic interfaces. Mar. Ecol. Prog. Ser. 467:1–14.
Angel MV . (1993). Biodiversity of the Pelagic Ocean. Conserv. Biol. 7:760–772.
Aota Y , Nakajima H. (2001). Mutualistic relationships between phytoplankton and bacteria
caused by carbon excretion from phytoplankton. Ecol. Res. 16:289–299.
Aristegui J, Gasol JM, Duarte CM, Herndl GJ. (2009). Microbial oceanography of the dark
ocean’s pelagic realm. Limnol. Oceanogr. 54:1501–1529.
Arrigo KR, Robinson DH, Worthen DL, Dunbar RB, DiTullio GR, VanWoert M, et al. (1999).
Phytoplankton Community Structure and the Drawdown of Nutrients and CO2 in the Southern
Ocean. Science 283:365–367.
Arumugam M, Raes J, Pelletier E, Le Paslier D, Yamada T, Mende DR, et al. (2011). Enterotypes
of the human gut microbiome. Nature 473:174–180.
Cram
189
Assenov Y , Ramírez F, Schelhorn S-E, Lengauer T, Albrecht M. (2008). Computing topological
parameters of biological networks. Bioinformatics 24:282–284.
Ayers WA. (1960). Specificity of the vitimin B12 requirement in certain marine bacteria. J.
Bacteriol. 80:744.
Azam F. (1998). Microbial Control of Oceanic Carbon Flux: The Plot Thickens. Science 280:694
–696.
Azam F, Fenchel T, Field J, Gray J, Meyrriel L, Thingstad F. (1983). The ecological role of
water-column microbes in the sea. Mar. Ecol. -Prog. Ser. 10:257–263.
Badcock J, Merrett NR. (1976). Midwater fishes in the eastern North Atlantic—I. Vertical
distribution and associated biology in 30°N, 23°W, with developmental notes on certain
myctophids. Prog. Oceanogr. 7:3–58.
Barberán A, Bates ST, Casamayor EO, Fierer N. (2011). Using network analysis to explore co-
occurrence patterns in soil microbial communities. ISME J. 6:343–351.
Barton AD, Dutkiewicz S, Flierl G, Bragg J, Follows MJ. (2010). Patterns of Diversity in Marine
Phytoplankton. Science 327:1509–1511.
Behrenfeld MJ, Falkowski PG. (1997). A Consumer’s Guide to Phytoplankton Primary
Productivity Models. Limnol. Oceanogr. 42:1479–1491.
Béjà O, Suzuki MT. (2008). Photoheterotrophic Marine Prokaryotes. In:Microbial Ecology of the
Oceans, Kirchman, DL, ed (ed)., John Wiley & Sons, Inc., pp. 131–157.
http://onlinelibrary.wiley.com/doi/10.1002/9780470281840.ch5/summary (Accessed June 2,
2014).
Beman JM, Leilei Shih J, Popp BN. (2013). Nitrite oxidation in the upper water column and
oxygen minimum zone of the eastern tropical North Pacific Ocean. ISME J. 7:2192–2205.
Beman JM, Popp BN, Francis CA. (2008). Molecular and biogeochemical evidence for ammonia
oxidation by marine Crenarchaeota in the Gulf of California. ISME J. 2:429–441.
Beman JM, Sachdeva R, Fuhrman JA. (2010). Population ecology of nitrifying Archaea and
Bacteria in the Southern California Bight. Environ. Microbiol. 12:1282–1292.
Cram
190
Beman JM, Steele JA, Fuhrman JA. (2011). Co-occurrence patterns for abundant marine
archaeal and bacterial lineages in the deep chlorophyll maximum of coastal California. ISME J.
5:1077–1085.
Benner R, Biddanda B, Black B, McCarthy M. (1997). Abundance, size distribution, and stable
carbon and nitrogen isotopic compositions of marine organic matter isolated by tangential-flow
ultrafiltration. Mar. Chem. 57:243–263.
Berelson WM. (1991). The Flushing of Two Deep-Sea Basins, Southern California Borderland.
Limnol. Oceanogr. 36:1150–1166.
Bissett W, Meyers M, Walsh J, Mullerkarger F. (1994). The Effects of Temporal Variability of
Mixed-Layer Depth on Primary Productivity Around Bermuda. J. Geophys. Res.-Oceans
99:7539–7553.
Bonilla-Findji O, Herndl GJ, Gattuso J-P, Weinbauer MG. (2009). Viral and Flagellate Control of
Prokaryotic Production and Community Structure in Offshore Mediterranean Waters. Appl.
Environ. Microbiol. 75:4801–4812.
Bouvier T, del Giorgio PA. (2007). Key role of selective viral-induced mortality in determining
marine bacterial community composition. Environ. Microbiol. 9:287–297.
Breitbart M. (2012). Marine Viruses: Truth or Dare. Annu. Rev. Mar. Sci. 4:425–448.
Brown MV , Lauro FM, DeMaere MZ, Muir L, Wilkins D, Thomas T, et al. (2012). Global
biogeography of SAR11 marine bacteria. Mol. Syst. Biol. 8.
http://www.nature.com/msb/journal/v8/n1/full/msb201228.html (Accessed December 2, 2012).
Brown MV , Schwalbach MS, Hewson I, Fuhrman JA. (2005a). Coupling 16S-ITS rDNA clone
libraries and automated ribosomal intergenic spacer analysis to show marine microbial diversity:
development and application to a time series. Environ. Microbiol. 7:1466–1479.
Brown MV , Schwalbach MS, Hewson I, Fuhrman JA. (2005b). Coupling 16S-ITS rDNA clone
libraries and automated ribosomal intergenic spacer analysis to show marine microbial diversity:
development and application to a time series. Environ. Microbiol. 7:1466–1479.
Cram
191
Buesseler KO, Pike S, Maiti K, Lamborg CH, Siegel DA, Trull TW. (2009). Thorium-234 as a
tracer of spatial, temporal and vertical variability in particle flux in the North Pacific. Deep Sea
Res. Part Oceanogr. Res. Pap. 56:1143–1167.
Canfield DE, Stewart FJ, Thamdrup B, Brabandere LD, Dalsgaard T, Delong EF, et al. (2010). A
Cryptic Sulfur Cycle in Oxygen-Minimum–Zone Waters off the Chilean Coast. Science
330:1375–1378.
Carlson C, Morris R, Parsons R, Treusch A, Giovannoni S, Vergin K. (2009). Seasonal dynamics
of SAR11 populations in the euphotic and mesopelagic zones of the northwestern Sargasso Sea.
ISME J. 3:283–295.
Caron DA. (2000). Symbiosis and mixotrophy among pelagic microorganisms. In:Microbial
Ecology of the Oceans, Kirchman, DL, ed (ed)., Wiley-Liss: New York, NY , pp. 495–523.
Casamayor EO, Massana R, Benlloch S, Ovreas L, Diez B, Goddard VJ, et al. (2002). Changes
in archaeal, bacterial and eukaryal assemblages along a salinity gradient by comparison of
genetic fingerprinting methods in a multipond solar saltern. Environ. Microbiol. 4:338–348.
Chaffron S, Rehrauer H, Pernthaler J, von Mering C. (2010). A global network of coexisting
microbes from environmental and whole-genome sequence data. Genome Res. 20:947–959.
Chave J. (2004). Neutral theory and community ecology. Ecol. Lett. 7:241–253.
Chow C-ET, Fuhrman JA. (2012). Seasonality and monthly dynamics of marine myovirus
communities: Marine myovirus community dynamics at SPOT. Environ. Microbiol. 14:2171–
2183.
Chow C-ET, Kim DY , Sachdeva R, Caron DA, Fuhrman JA. (2014). Top-down controls on
bacterial community structure: microbial network analysis of bacteria, T4-like viruses and
protists. ISME J. 8:816–829.
Chow C-ET, Sachdeva R, Cram JA, Steele JA, Needham DM, Patel A, et al. (2013). Temporal
variability and coherence of euphotic zone bacterial communities over a decade in the Southern
California Bight. ISME J. 7:2259–2273.
Church MJ. (2008). Resource control of bacterial dynamics in the sea. In:Microbial Ecology of
the Oceans, Hoboken, N.J., pp. 335–382.
Cram
192
Church MJ, Lomas MW, Muller-Karger F. (2013). Sea change: Charting the course for
biogeochemical ocean time-series research in a new millennium. Deep Sea Res. Part II Top.
Stud. Oceanogr. 93:2–15.
Clarke KR. (1993). Non-parametric multivariate analyses of changes in community structure.
Aust. J. Ecol. 18:117–143.
Collins LE, Berelson W, Hammond DE, Knapp A, Schwartz R, Capone D. (2011). Particle fluxes
in San Pedro Basin, California: A four-year record of sedimentation and physical forcing. Deep
Sea Res. Part Oceanogr. Res. Pap. 58:898–914.
Cram JA, Chow C-ET, Sachdeva R, Needham DM, Parada AE, Steele JA, et al. (2014). Seasonal
and interannual variability of the marine bacterioplankton community throughout the water
column over ten years. ISME J.
Cram JA, Sun F, Fuhrman JA. (2014). Marine bacterial, archaeal and protistan association
networks. In:Encyclopedia of Metagenomics, SpringerReference.
Crump BC, Hopkinson CS, Sogin ML, Hobbie JE. (2004). Microbial Biogeography along an
Estuarine Salinity Gradient: Combined Influences of Bacterial Growth and Residence Time.
Appl Env. Microbiol 70:1494–1505.
D’Ortenzio F. (2005). Seasonal variability of the mixed layer depth in the Mediterranean Sea as
derived from in situ profiles. Geophys. Res. Lett. 32.
http://doi.wiley.com/10.1029/2005GL022463 (Accessed May 12, 2014).
Dabney A, Storey JD, qvalue Tutorial PRS, Install R, Packages IB, Packages UB, et al. (2004).
Q-value estimation for false discovery rate control. Medicine (Baltimore) 344:539–548.
DeLong EF, Preston CM, Mincer T, Rich V , Hallam SJ, Frigaard N-U, et al. (2006). Community
Genomics Among Stratified Microbial Assemblages in the Ocean’s Interior. Science 311:496–
503.
Deng Y , Jiang Y-H, Yang Y , He Z, Luo F, Zhou J. (2012). Molecular ecological network analyses.
BMC Bioinformatics 13:113.
Cram
193
DeSantis TZ, Hugenholtz P, Larsen N, Rojas M, Brodie EL, Keller K, et al. (2006). Greengenes,
a chimera-checked 16S rRNA gene database and workbench compatible with ARB. Appl.
Environ. Microbiol. 72:5069–5072.
Deutsch C, Sarmiento JL, Sigman DM, Gruber N, Dunne JP. (2007). Spatial coupling of nitrogen
inputs and losses in the ocean. Nature 445:163–167.
Drake DE. (1974). Distribution and Transport of Suspended Particulate Matter in Submarine
Canyons Off Southern California. In:Suspended Solids in Water, Gibbs, RJ, ed (ed). Marine
Science, Springer US, pp. 133–153. http://link.springer.com/chapter/10.1007/978-1-4684-8529-
5_9 (Accessed November 26, 2013).
Drake DE, Gorsline DS. (1973). Distribution and Transport of Suspended Particulate Matter in
Hueneme, Redondo, Newport, and La Jolla Submarine Canyons, California. Geol. Soc. Am.
Bull. 84:3949–3968.
Duarte CM, Agustı
́ S, Kennedy H, Vaqué D. (1999). The Mediterranean climate as a template for
Mediterranean marine ecosystems: the example of the northeast Spanish littoral. Prog. Oceanogr.
44:245–270.
Ducet N, Le Traon PY , Reverdin G. (2000). Global high-resolution mapping of ocean circulation
from TOPEX/Poseidon and ERS-1 and -2. J. Geophys. Res. Oceans 105:19477–19498.
Ducklow HW, Carlson CA. (1992). Oceanic Bacterial Production. In:Advances in Microbial
Ecology, Marshall, KC, ed (ed). Advances in Microbial Ecology, Springer US, pp. 113–181.
http://link.springer.com.libproxy.usc.edu/chapter/10.1007/978-1-4684-7609-5_3 (Accessed May
31, 2014).
Dupont CL, Rusch DB, Yooseph S, Lombardo M-J, Richter RA, Valas R, et al. (2012). Genomic
insights to SAR86, an abundant and uncultivated marine bacterial lineage. ISME J. 6:1186–1199.
Eckert EM, Baumgartner M, Huber IM, Pernthaler J. (2013). Grazing resistant freshwater
bacteria profit from chitin and cell-wall-derived organic carbon. Environ. Microbiol. n/a–n/a.
Eiler A, Hayakawa DH, Rappé MS. (2011). Non-random assembly of bacterioplankton
communities in the subtropical north pacific ocean. Front. Microbiol. 2:140.
Cram
194
Eiler A, Heinrich F, Bertilsson S. (2012). Coherent dynamics and association networks among
lake bacterioplankton taxa. ISME J. 6:330–342.
Eslinger DL, Cooney RT, McRoy CP, Ward A, Kline TC, Simpson EP, et al. (2001). Plankton
dynamics: observed and modelled responses to physical conditions in Prince William Sound,
Alaska. Fish. Oceanogr. 10:81–96.
Ewart CS, Meyers MK, Wallner ER, McGillicuddy Jr. DJ, Carlson CA. (2008). Microbial
dynamics in cyclonic and anticyclonic mode-water eddies in the northwestern Sargasso Sea.
Deep Sea Res. Part II Top. Stud. Oceanogr. 55:1334–1347.
Falkowski PG, Fenchel T, Delong EF. (2008). The Microbial Engines That Drive Earth’s
Biogeochemical Cycles. Science 320:1034–1039.
Faust K, Raes J. (2012). Microbial interactions: from networks to models. Nat. Rev. Microbiol.
10:538–550.
Faust K, Sathirapongsasuti JF, Izard J, Segata N, Gevers D, Raes J, et al. (2012). Microbial co-
occurrence relationships in the human microbiome. PLoS Comput. Biol. 8:e1002606.
Fenchel T, Esteban GF, Finlay BJ. (1997). Local versus Global Diversity of Microorganisms:
Cryptic Diversity of Ciliated Protozoa. Oikos 80:220–225.
Ferguson CA, Carvalho L, Scott EM, Bowman AW, Kirika A. (2008). Assessing ecological
responses to environmental change using statistical models. J. Appl. Ecol. 45:193–203.
Finlay BJ. (2002). Global dispersal of free-living microbial eukaryote species. SCIENCE
296:1061–1063.
Fortunato CS, Herfort L, Zuber P, Baptista AM, Crump BC. (2011). Spatial variability
overwhelms seasonal patterns in bacterioplankton communities across a river to ocean gradient.
ISME J. 6:554–563.
Fouilland E, Tolosa I, Bonnet D, Bouvier C, Bouvier T, Bouvy M, et al. (2014). Bacterial carbon
dependence on freshly produced phytoplankton exudates under different nutrient availability and
grazing pressure conditions in coastal marine waters. FEMS Microbiol. Ecol. 87:757–769.
Francois R, Honjo S, Krishfield R, Manganini S. (2002). Factors controlling the flux of organic
carbon to the bathypelagic zone of the ocean. Glob. Biogeochem. Cycles 16:34–1–34–20.
Cram
195
Freilich S, Kreimer A, Meilijson I, Gophna U, Sharan R, Ruppin E. (2010). The large-scale
organization of the bacterial network of ecological co-occurrence interactions. Nucleic Acids
Res. 38:3857–3868.
Frouin R, Franz BA, Werdell PJ. (2003). The SeaWiFS PAR product. Algorithm Updat. Fourth
SeaWiFS Data Reprocess. 46–50.
Fu G, Baith K, McClain C. (1998). SeaDAS: The SeaWiFS data analysis system. In:Proceedings
of the 4th Pacific Ocean Remote Sensing Conference, Qingdao, China, pp. 28–31.
Fuessel J, Lam P, Lavik G, Jensen MM, Holtappels M, Guenter M, et al. (2012). Nitrite oxidation
in the Namibian oxygen minimum zone. Isme J. 6:1200–1209.
Fuhrman J, Azam F. (1982). Thymidine incorporation as a measure of heterotrophic
bacterioplankton production in marine surface waters - evaluation and field results. Mar. Biol.
66:109–120.
Fuhrman J, Sleeter T, Carlson C, Proctor L. (1989). dominance of bacterial biomass in the
sargasso sea and its ecological implications. Mar. Ecol.-Prog. Ser. 57:207–217.
Fuhrman J, Steele J. (2008). Community structure of marine bacterioplankton: patterns,
networks, and relationships to function. Aquat. Microb. Ecol. 53:69–81.
Fuhrman JA. (1999). Marine viruses and their biogeochemical and ecological effects. Nature
399:541–548.
Fuhrman JA. (2009). Microbial community structure and its functional implications. Nature
459:193–199.
Fuhrman JA, Comeau DE, Hagström Å, Chan AM. (1988). Extraction from Natural Planktonic
Microorganisms of DNA Suitable for Molecular Biological Studies. Appl. Environ. Microbiol.
54:1426–1429.
Fuhrman JA, Hagstrom A. (2008). Bacterial and archaeal community structure and its patterns.
In:Microbial Ecology of the Oceans, Wiley: Hoboken, N.J., pp. 45–90.
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.
103:13104 –13109.
Cram
196
Fuhrman JA, Lee SH, Masuchi Y , Davis AA, Wilcox RM. (1994). Characterization of Marine
Prokaryotic Communities via DNA and RNA. Microb. Ecol. 28:133–145.
Fuhrman JA, Steele JA, Hewson I, Schwalbach MS, Brown MV , Green JL, et al. (2008). A
latitudinal diversity gradient in planktonic marine bacteria. Proc. Natl. Acad. Sci. 105:7774–
7778.
Galand PE, Casamayor EO, Kirchman DL, Potvin M, Lovejoy C. (2009a). Unique archaeal
assemblages in the Arctic Ocean unveiled by massively parallel tag sequencing. ISME J. 3:860–
869.
Galand PE, Gutiérrez-Provecho C, Massana R, Gasol JM, Casamayor EO. (2010). Inter-annual
recurrence of archaeal assemblages in the coastal NW Mediterranean Sea (Blanes Bay Microbial
Observatory). Limnol. Oceanogr. 55:2117–2125.
Galand PE, Lovejoy C, Hamilton AK, Ingram RG, Pedneault E, Carmack EC. (2009b). Archaeal
diversity and a gene for ammonia oxidation are coupled to oceanic circulation. Environ.
Microbiol. 11:971–980.
Garcia-Martinez J, Rodriguez-Valera F. (2000). Microdiversity of uncultured marine
prokaryotes: the SAR11 cluster and the marine Archaea of Group I. Mol. Ecol. 9:935–948.
Gibbons SM, Caporaso JG, Pirrung M, Field D, Knight R, Gilbert JA. (2013). Evidence for a
persistent microbial seed bank throughout the global ocean. Proc. Natl. Acad. Sci. 110:4651–
4655.
Gilbert JA, Field D, Swift P, Newbold L, Oliver A, Smyth T, et al. (2009). The seasonal structure
of microbial communities in the Western English Channel. Environ. Microbiol. 11:3132–3139.
Gilbert JA, Steele JA, Caporaso JG, Steinbrück L, Reeder J, Temperton B, et al. (2012). Defining
seasonal marine microbial community dynamics. ISME J. 6:298–308.
Giovannoni SJ, Vergin KL. (2012). Seasonality in Ocean Microbial Communities. Science
335:671–676.
Goslee SC, Urban DL. (2007). The ecodist package for dissimilarity-based analysis of ecological
data. J. Stat. Softw. 22:1–19.
Cram
197
Hahn MW, Höfle MG. (2001). Grazing of protozoa and its effect on populations of aquatic
bacteria. FEMS Microbiol. Ecol. 35:113–121.
Hamersley MR, Turk KA, Leinweber A, Gruber N, Zehr JP, Gunderson T, et al. (2011). Nitrogen
fixation within the water column associated with two hypoxic basins in the Southern California
Bight. Aquat. Microb. Ecol. 63:193–205.
Hamilton AK, Lovejoy C, Galand PE, Ingram RG. (2008). Water masses and biogeography of
picoeukaryote assemblages in a cold hydrographically complex system. Limnol. Oceanogr.
53:922–935.
Hansen, J, Kirboe T, Alldredge A. (1996). Marine snow derived from abandoned larvacean
houses: sinking rates, particle content and mechanisms of aggregate formation. Mar. Ecol. Prog.
Ser. 141:205–215.
Harrison PJ. (2002). Station papa time series: Insights into ecosystem dynamics. J. Oceanogr.
58:259–264.
Hatosy SM, Martiny JBH, Sachdeva R, Steele J, Fuhrman JA, Martiny AC. (2013). Beta
diversity of marine bacteria depends on temporal scale. Ecology 94:1898–1904.
Herndl G, Agogué H, Baltar F, Reinthaler T, Sintes E, Varela M. (2008). Regulation of aquatic
microbial processes: the ‘microbial loop’ of the sunlit surface waters and the dark ocean
dissected. Aquat. Microb. Ecol. 53:59–68.
Herndl GJ, Reinthaler T. (2013). Microbial control of the dark end of the biological pump. Nat.
Geosci. 6:718–724.
Hewson I, Fuhrman JA. (2003). Viriobenthos production and virioplankton sorptive scavenging
by suspended sediment particles in coastal and pelagic waters. Microb. Ecol. 46:337–347.
Hewson I, Steele J, Capone D, Fuhrman J. (2006a). Remarkable heterogeneity in meso- and
bathypelagic bacterioplankton assemblage composition. Limnol. Oceanogr. 51:1274–1283.
Hewson I, Steele JA, Capone DG, Fuhrman JA. (2006b). Temporal and spatial scales of variation
in bacterioplankton assemblages of oligotrophic surface waters. Mar. Ecol.-Prog. Ser. 311:67–77.
Hickey BM. (1992). Circulation over the Santa Monica-San Pedro Basin and Shelf. Prog.
Oceanogr. 30:37–115.
Cram
198
Hickey BM. (1991). Variability in two deep coastal basins (Santa Monica and San Pedro) off
southern California. J. Geophys. Res. Oceans 96:16689–16708.
Hobbie JE, Daley RJ, Jasper S. (1977). Use of nuclepore filters for counting bacteria by
fluorescence microscopy. Appl. Environ. Microbiol. 33:1225–1228.
Honaker J, King G. (2010). What to do about missing values in time-series cross-section data.
Am. J. Polit. Sci. 54:561–581.
Honaker J, King G, Blackwell M. (2006). Amelia II: A program for missing data.
http://www.icesi.edu.co/CRAN/web/packages/Amelia/vignettes/amelia.pdf (Accessed July 8,
2013).
Hooper DU, Chapin FS III, Ewel JJ, Hector A, Inchausti P, Lavorel S, et al. (2005). Effects of
Biodiversity on Ecosystem Functioning: A Consensus of Current Knowledge. Ecol. Monogr.
75:3–35.
Hugoni M, Taib N, Debroas D, Domaizon I, Dufournel IJ, Bronner G, et al. (2013). Structure of
the rare archaeal biosphere and seasonal dynamics of active ecotypes in surface coastal waters.
Proc. Natl. Acad. Sci. 110:6004–6009.
Humphries MD, Gurney K. (2008). Network ‘Small-World-Ness’: A Quantitative Method for
Determining Canonical Network Equivalence. PLoS ONE 3:e0002051.
Iriberri J, Unanue M, Barcina I, Egea L. (1987). Seasonal variation in population density and
heterotrophic activity of attached and free-living bacteria in coastal waters. Appl. Environ.
Microbiol. 53:2308–2314.
Jardillier L, Bettarel Y , Richardot M, Bardot C, Amblard C, Sime-Ngando T, et al. (2005).
Effects of Viruses and Predators on Prokaryotic Community Composition. Microb. Ecol.
50:557–569.
Jeandel C, Ruiz-Pino D, Gjata E, Poisson A, Brunet C, Charriaud E, et al. (1998). KERFIX, a
time-series station in the Southern Ocean: a presentation. J. Mar. Syst. 17:555–569.
Jurgens K, Gude H. (1994). The Potential Importance of Grazing-Resistant Bacteria in
Planktonic Systems. Mar. Ecol. Prog. Ser. 112:169–188.
Cram
199
Jurgens K, Massana R. (2008). Protistan grazing on marine bacterioplankton. In:Microbial
Ecology of the Oceans, Hoboken, N.J., pp. 383–441.
Jürgens K, Matz C. (2002). Predation as a shaping force for the phenotypic and genotypic
composition of planktonic bacteria. Antonie Van Leeuwenhoek 81:413–434.
Jurgens K, Pernthaler J, Schalla S, Amann R. (1999). Morphological and compositional changes
in a planktonic bacterial community in response to enhanced protozoan grazing. Appl. Environ.
Microbiol. 65:1241–1250.
Kan J, Suzuki MT, Wang K, Evans SE, Chen F. (2007). High Temporal but Low Spatial
Heterogeneity of Bacterioplankton in the Chesapeake Bay. Appl. Environ. Microbiol. 73:6776–
6789.
Kara EL, Hanson PC, Hu YH, Winslow L, McMahon KD. (2012). A decade of seasonal
dynamics and co-occurrences within freshwater bacterioplankton communities from eutrophic
Lake Mendota, WI, USA. ISME J. 7:680–684.
Karl DM, Lukas R. (1996). The Hawaii Ocean Time-series (HOT) program: Background,
rationale and field implementation. Deep Sea Res. Part II Top. Stud. Oceanogr. 43:129–156.
Kim DY , Countway PD, Jones AC, Schnetzer A, Yamashita W, Tung C, et al. (2013). Monthly to
interannual variability of microbial eukaryote assemblages at four depths in the eastern North
Pacific. ISME J. http://www.nature.com/ismej/journal/vaop/ncurrent/full/ismej2013173a.html
(Accessed December 10, 2013).
Kim DY , Countway PD, Yamashita 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.
Methods 91:527–536.
King G, Honaker J, Joseph A, Scheve K. (2001). Analyzing Incomplete Political Science Data:
An Alternative Algorithm for Multiple Imputation. Am. Polit. Sci. Rev. 95:49–69.
Kiorboe T. (2001). Formation and fate of marine snow: small-scale processes with large-scale
implications. Sci. Mar. 65:57–71.
Cram
200
Kirchman D, K’nees E, Hodson R. (1985). Leucine incorporation and its potential as a measure
of protein synthesis by bacteria in natural aquatic systems. Appl. Environ. Microbiol. 49:599–
607.
Kirchman DL. (2008). Introduction and Overview. In:Microbial Ecology of the Oceans,
Kirchman, DL, ed (ed)., John Wiley & Sons, Inc., pp. 1–26.
http://onlinelibrary.wiley.com/doi/10.1002/9780470281840.ch1/summary (Accessed June 2,
2014).
Kirchman DL. (2002). The ecology of Cytophaga–Flavobacteria in aquatic environments. FEMS
Microbiol. Ecol. 39:91–100.
Ladau J, Sharpton TJ, Finucane MM, Jospin G, Kembel SW, O’Dwyer J, et al. (2013). Global
marine bacterial diversity peaks at high latitudes in winter. ISME J. 7:1669–1677.
Larsen A, Flaten GAF, Sandaa R-A, Castberg T, Thyrhaug R, Erga SR, et al. (2004). Spring
Phytoplankton Bloom Dynamics in Norwegian Coastal Waters: Microbial Community
Succession and Diversity. Limnol. Oceanogr. 49:180–190.
Lehodey P, Alheit J, Barange M, Baumgartner T, Beaugrand G, Drinkwater K, et al. (2006).
Climate Variability, Fish, and Fisheries. J. Clim. 19:5009–5030.
Lindsay DJ, Hunt JC. (2005). Biodiversity in midwater cnidarians and ctenophores: submersible-
based results from deep-water bays in the Japan Sea and north-western Pacific. J. Mar. Biol.
Assoc. U. K. 85:503–517.
Long RA, Azam F. (2001). Microscale patchiness of bacterioplankton assemblage richness in
seawater. Aquat. Microb. Ecol. 26:103–113.
Madigan MT, Martinko JM. (2005). Brock Biology Of Microorganisms 11th EDITION. 11th ed.
Lebanon, Indiana, U.S.A.: Pren.
Maidak BL, Cole JR, Lilburn TG, Jr CTP, Saxman PR, Farris RJ, et al. (2001). The RDP-II
(Ribosomal Database Project). Nucleic Acids Res. 29:173–174.
Malfatti F, Azam F. (2009). Atomic force microscopy reveals microscale networks and possible
symbioses among pelagic marine bacteria. Aquat. Microb. Ecol. 58:1–14.
Cram
201
Malfatti F, Samo TJ, Azam F. (2009). High-resolution imaging of pelagic bacteria by Atomic
Force Microscopy and implications for carbon cycling. ISME J. 4:427–439.
Malmstrom RR, Kiene RP, Cottrell MT, Kirchman DL. (2004). Contribution of SAR11 Bacteria
to Dissolved Dimethylsulfoniopropionate and Amino Acid Uptake in the North Atlantic Ocean.
Appl. Environ. Microbiol. 70:4129–4135.
Mannino A, Russ ME, Hooker SB. (2008). Algorithm development and validation for satellite-
derived distributions of DOC and CDOM in the U.S. Middle Atlantic Bight. J. Geophys. Res.
Oceans 113:n/a–n/a.
Marinovic BB, Croll DA, Gong N, Benson SR, Chavez FP. (2002). Effects of the 1997-1999 El
Nino and La Nina events on zooplankton abundance and euphausiid community composition
within the Monterey Bay coastal upwelling system. Prog. Oceanogr. 54:265–277.
Martiny JBH, Bohannan BJM, Brown JH, Colwell RK, Fuhrman JA, Green JL, et al. (2006).
Microbial biogeography: putting microorganisms on the map. Nat. Rev. Microbiol. 4:102–112.
Matz C, Jürgens K. (2003). Interaction of Nutrient Limitation and Protozoan Grazing Determines
the Phenotypic Structure of a Bacterial Community. Microb. Ecol. 45:384–398.
Matz C, Kjelleberg S. (2005). Off the hook – how bacteria survive protozoan grazing. Trends
Microbiol. 13:302–307.
McGillicuddy DJ, Anderson LA, Bates NR, Bibby T, Buesseler KO, Carlson CA, et al. (2007).
Eddy/Wind Interactions Stimulate Extraordinary Mid-Ocean Plankton Blooms. Science
316:1021–1026.
McGrady-Steed J, Harris PM, Morin PJ. (1997). Biodiversity regulates ecosystem predictability.
Nature 390:162–165.
Medlin LK. (2007). If everything is everywhere, do they share a common gene pool? Gene
406:180–183.
Michaels AF, Knap AH, Dow RL, Gundersen K, Johnson RJ, Sorensen J, et al. (1994). Seasonal
patterns of ocean biogeochemistry at the U.S. JGOFS Bermuda Atlantic time-series study site.
Deep Sea Res. Part Oceanogr. Res. Pap. 41:1013–1038.
Cram
202
Miller MB, Bassler BL. (2001). Quorum Sensing in Bacteria. Annu. Rev. Microbiol. 55:165–
199.
Mincer TJ, Church MJ, Taylor LT, Preston C, Karl DM, DeLong EF. (2007). Quantitative
distribution of presumptive archaeal and bacterial nitrifiers in Monterey Bay and the North
Pacific Subtropical Gyre. Environ. Microbiol. 9:1162–1175.
Morel A, Gentili B. (2009). A simple band ratio technique to quantify the colored dissolved and
detrital organic material from ocean color remotely sensed data. Remote Sens. Environ.
113:998–1011.
Morris RM, Frazar CD, Carlson CA. (2012). Basin-scale patterns in the abundance of SAR11
subclades, marine Actinobacteria (OM1), members of the Roseobacter clade and OCS116 in the
South Atlantic. Environ. Microbiol. 14:1133–1144.
Morris RM, Vergin KL, Cho JC, Rappe MS, Carlson CA, Giovannoni SJ. (2005). Temporal and
spatial response of bacterioplankton lineages to annual convective overturn at the Bermuda
Atlantic Time-series Study site. Limnol. Oceanogr. 50:1687–1696.
Nagata T. (2008). Organic Matter–Bacteria Interactions in Seawater. In:Microbial Ecology of the
Oceans, Kirchman, DL, ed (ed)., John Wiley & Sons, Inc., pp. 207–241.
http://onlinelibrary.wiley.com/doi/10.1002/9780470281840.ch7/summary (Accessed June 1,
2014).
Needham DM, Chow C-ET, Cram JA, Sachdeva R, Parada A, Fuhrman JA. (2013). Short-term
observations of marine bacterial and viral communities: patterns, connections and resilience.
ISME J. 7:1274–1285.
Nelson CE, Carlson CA, Ewart CS, Halewood ER. (2014). Community differentiation and
population enrichment of Sargasso Sea bacterioplankton in the euphotic zone of a mesoscale
mode-water eddy. Environ. Microbiol. 16:871–887.
Nelson JD, Boehme SE, Reimers CE, Sherrell RM, Kerkhof LJ. (2008). Temporal patterns of
microbial community structure in the Mid-Atlantic Bight. FEMS Microbiol. Ecol. 65:484–493.
Noble RT, Fuhrman JA. (1998). Use of SYBR Green I for rapid epifluorescence counts of marine
viruses and bacteria. Aquat. Microb. Ecol. 14:113–118.
Cram
203
Obernosterer I, Herndl GJ. (1995). Phytoplankton extracellular release and bacterial growth:
dependence on the inorganic N: P ratio. Mar. Ecol. Prog. Ser. 116:247–257.
Oksanen J, Blanchet FG, Kindt R, Legendre P, Minchin PR, O’Hara RB, et al. (2013). vegan:
Community Ecology Package. http://CRAN.R-project.org/package=vegan.
Oksanen J, Blanchet FG, Kindt R, Legendre P, O’Hara RB, Simpson GL, et al. (2011). vegan:
Community Ecology Package. http://CRAN.R-project.org/package=vegan.
Ottesen EA, Young CR, Eppley JM, Ryan JP, Chavez FP, Scholin CA, et al. (2013). Pattern and
synchrony of gene expression among sympatric marine microbial populations. Proc. Natl. Acad.
Sci. 110:E488–E497.
Parsons TR. (1984). A manual of chemical and biological methods for seawater analysis. 1st ed.
Pergamon Press: Oxford [Oxfordshire] ; New York.
Partensky F, Hess WR, Vaulot D. (1999). Prochlorococcus, a Marine Photosynthetic Prokaryote
of Global Significance. Microbiol. Mol. Biol. Rev. 63:106–127.
Patel A, Noble RT, Steele JA, Schwalbach MS, Hewson I, Fuhrman JA. (2007). Virus and
prokaryote enumeration from planktonic aquatic environments by epifluorescence microscopy
with SYBR Green I. Nat. Protoc. 2:269–276.
Paver SF, Hayek KR, Gano KA, Fagen JR, Brown CT, Davis-Richardson AG, et al. (2013).
Interactions between specific phytoplankton and bacteria affect lake bacterial community
succession. Environ. Microbiol. 15:2489–2504.
Pernthaler J. (2005). Predation on prokaryotes in the water column and its ecological
implications. Nat. Rev. Microbiol. 3:537–546.
Pingree RD, Griffiths DK. (1978). Tidal fronts on the shelf seas around the British Isles. J.
Geophys. Res. Oceans 83:4615–4622.
Pinhassi J, Winding A, Binnerup SJ, Zweifel UL, Riemann B, Hagstrom A. (2003). Spatial
variability in bacterioplankton community composition at the Skagerrak-Kattegat Front. Mar.
Ecol.-Prog. Ser. 255:1–13.
Cram
204
Pommier T, CanbäCk B, Riemann L, BoströM KH, Simu K, Lundberg P, et al. (2007). Global
patterns of diversity and community structure in marine bacterioplankton. Mol. Ecol. 16:867–
880.
Poretsky RS, Hewson I, Sun S, Allen AE, Zehr JP, Moran MA. (2009). Comparative day/night
metatranscriptomic analysis of microbial communities in the North Pacific subtropical gyre.
Environ. Microbiol. 11:1358–1375.
Quast C, Pruesse E, Yilmaz P, Gerken J, Schweer T, Yarza P, et al. (2012). The SILVA ribosomal
RNA gene database project: improved data processing and web-based tools. Nucleic Acids Res.
41:D590–D596.
Ramette A. (2007). Multivariate analyses in microbial ecology. FEMS Microbiol. Ecol. 62:142–
160.
Reinthaler T, van Aken H, Veth C, Arístegui J, Robinson C, Williams PJ le B, et al. (2006).
Prokaryotic respiration and production in the meso- and bathypelagic realm of the eastern and
western North Atlantic basin. Limnol. Oceanogr. 51:1262–1273.
Ribalet F, Marchetti A, Hubbard KA, Brown K, Durkin CA, Morales R, et al. (2010). Unveiling
a phytoplankton hotspot at a narrow boundary between coastal and offshore waters. Proc. Natl.
Acad. Sci. 107:16571–16576.
Rich VI, Pham VD, Eppley J, Shi Y , DeLong EF. (2011). Time-series analyses of Monterey Bay
coastal microbial picoplankton using a ‘genome proxy’ microarray. Environ. Microbiol. 13:116–
134.
Robison BH, Reisenbichler KR, Sherlock RE. (2005). Giant Larvacean Houses: Rapid Carbon
Transport to the Deep Sea Floor. Science 308:1609–1611.
Rodionov DA, Vitreschak AG, Mironov AA, Gelfand MS. (2003). Comparative Genomics of the
Vitamin B12 Metabolism and Regulation in Prokaryotes. J. Biol. Chem. 278:41148 –41159.
Ruan Q, Dutta D, Schwalbach MS, Steele JA, Fuhrman JA, Sun F. (2006a). Local similarity
analysis reveals unique associations among marine bacterioplankton species and environmental
factors. Bioinformatics 22:2532–2538.
Cram
205
Ruan Q, Steele JA, Schwalbach MS, Fuhrman JA, Sun F. (2006b). A dynamic programming
algorithm for binning microbial community profiles. Bioinforma. Oxf. Engl. 22:1508–1514.
Rusch DB, Halpern AL, Sutton G, Heidelberg KB, Williamson S, Yooseph S, et al. (2007). The
Sorcerer II Global Ocean Sampling Expedition: Northwest Atlantic through Eastern Tropical
Pacific. PLoS Biol. 5:e77.
Sanudo-Wilhelmy SA, Gomez-Consarnau L, Suffridge C, Webb EA. (2014). The Role of B
Vitamins in Marine Biogeochemistry. In:Annual Review of Marine Science, V ol 6, Carlson, CA
& Giovannoni, SJ, eds (ed). V ol. 6, Annual Reviews: Palo Alto, pp. 339–367.
Schauer M, Balagué V , Pedrós-Alió C, Massana R. (2003). Seasonal changes in the taxonomic
composition of bacterioplankton in a coastal oligotrophic system. Aquat. Microb. Ecol. 31:163–
174.
Schmoker C, Hernández-León S, Calbet A. (2013). Microzooplankton grazing in the oceans:
impacts, data variability, knowledge gaps and future directions. J. Plankton Res. fbt023.
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. Part Oceanogr. Res. Pap. 58:16–26.
Schwalbach MS, Hewson I, Fuhrman JA. (2004). Viral effects on bacterial community
composition in marine plankton microcosms. Aquat. Microb. Ecol. 34:117–127.
Sherr EB, Sherr BF. (2000). Marine Microbes: An overview. In:Microbial Ecology of the
Oceans, Kirchman, D., ed (ed)., Wiley-Liss, pp. 351–386.
Sholkovitz E, Soutar A. (1975). Changes in Composition of Bottom Water of Santa-Barbara
Basin - Effect. Deep-Sea Res. 22:13–&.
Simek K, Kojecka P, Nedoma J, Hartman P, Vrba J, Dolan JR. (1999). Shifts in bacterial
community composition associated with different microzooplankton size fractions in a eutrophic
reservoir. Limnol. Oceanogr. 44:1634–1644.
Simek K, Pernthaler J, Weinbauer MG, Hornak K, Dolan JR, Nedoma J, et al. (2001). Changes
in bacterial community composition and dynamics and viral mortality rates associated with
Cram
206
enhanced flagellate grazing in a mesoeutrophic reservoir. Appl. Environ. Microbiol. 67:2723–
2733.
Simon M, Grossart H-P, Schweitzer B, Ploug H. (2002). Microbial ecology of organic aggregates
in aquatic ecosystems. Aquat. Microb. Ecol. 28:175–211.
Sintes E, Witte H, Stodderegger K, Steiner P, Herndl GJ. (2013). Temporal dynamics in the free-
living bacterial community composition in the coastal North Sea. FEMS Microbiol. Ecol.
83:413–424.
Smoot ME, Ono K, Ruscheinski J, Wang PL, Ideker T. (2011). Cytoscape 2.8: new features for
data integration and network visualization. Bioinformatics 27:431–432.
Sohrin R, Isaji M, Obara Y , Agostini S, Suzuki Y , Hiroe Y , et al. (2011). Distribution of
Synechococcus in the dark ocean. Aquat. Microb. Ecol. 64:1–14.
Southward AJ, Langmead O, Hardman-Mountford NJ, Aiken J, Boalch GT, Dando PR, et al.
(2004). Long-Term Oceanographic and Ecological Research in the Western English Channel.
In:Advances in Marine Biology, V ol. V olume 47, Academic Press, pp. 1–105.
http://www.sciencedirect.com/science/article/pii/S0065288104470011 (Accessed April 2, 2014).
Spieck E, Bock E. (2005). The Lithoautotrophic Nitrite-Oxidizing Bacteria. In:Bergey’s
Manual® of Systematic Bacteriology, Brenner, DJ, Krieg, NR, Staley, JT, & Garrity, GM, eds
(ed)., Springer US, pp. 149–153.
http://www.springerlink.com/content/l67712075q6r051p/abstract/ (Accessed November 15,
2012).
Steele JA, Countway PD, Xia L, Vigil PD, Beman JM, Kim DY , et al. (2011). Marine bacterial,
archaeal and protistan association networks reveal ecological linkages. ISME J. 5:1414–1425.
Steinberg DK, Carlson CA, Bates NR, Goldthwait SA, Madin LP, Michaels AF. (2000).
Zooplankton vertical migration and the active transport of dissolved organic and inorganic
carbon in the Sargasso Sea. Deep Sea Res. Part Oceanogr. Res. Pap. 47:137–158.
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. Part Ii-Top. Stud. Oceanogr. 48:1405–1447.
Cram
207
Steinberg DK, Goldthwait SA, Hansell DA. (2002). Zooplankton vertical migration and the
active transport of dissolved organic and inorganic nitrogen in the Sargasso Sea. Deep Sea Res.
Part Oceanogr. Res. Pap. 49:1445–1461.
Strom SL. (2008). Microbial Ecology of Ocean Biogeochemistry: A Community Perspective.
Science 320:1043–1045.
Suttle CA. (2007). Marine viruses — major players in the global ecosystem. Nat. Rev. Microbiol.
5:801–812.
Suzuki MT. (1999). Effect of protistan bacterivory on coastal bacterioplankton diversity. Aquat.
Microb. Ecol. 20:261–272.
Sverdrup HU. (1953). On conditions for the vernal blooming of phytoplankton. J. Cons. 18:287–
295.
Swan BK, Martinez-Garcia M, Preston CM, Sczyrba A, Woyke T, Lamy D, et al. (2011).
Potential for Chemolithoautotrophy Among Ubiquitous Bacteria Lineages in the Dark Ocean.
Science 333:1296–1300.
Sweeney EN, McGillicuddy Jr. DJ, Buesseler KO. (2003). Biogeochemical impacts due to
mesoscale eddy activity in the Sargasso Sea as measured at the Bermuda Atlantic Time-series
Study (BATS). Deep Sea Res. Part II Top. Stud. Oceanogr. 50:3017–3039.
Tada Y , Taniguchi A, Sato-Takabe Y , Hamasaki K. (2012). Growth and succession patterns of
major phylogenetic groups of marine bacteria during a mesocosm diatom bloom. J. Oceanogr.
68:509–519.
Team RDC. (2011). R: A Language and Environment for Statistical Computing. Vienna, Austria
http://www.R-project.org/.
Teeling H, Fuchs BM, Becher D, Klockow C, Gardebrecht A, Bennke CM, et al. (2012).
Substrate-Controlled Succession of Marine Bacterioplankton Populations Induced by a
Phytoplankton Bloom. Science 336:608–611.
Telesford QK, Joyce KE, Hayasaka S, Burdette JH, Laurienti PJ. (2011). The ubiquity of small-
world networks. ArXiv11095454 Phys. http://arxiv.org/abs/1109.5454 (Accessed May 17, 2014).
Cram
208
Thingstad TF. (2000). Elements of a Theory for the Mechanisms Controlling Abundance,
Diversity, and Biogeochemical Role of Lytic Bacterial Viruses in Aquatic Systems. Limnol.
Oceanogr. 45:1320–1328.
Thingstad TF, Lignell R. (1997). Theoretical models for the control of bacterial growth rate,
abundance, diversity and carbon demand. Aquat. Microb. Ecol. 13:19–27.
Treusch AH, Vergin KL, Finlay LA, Donatz MG, Burton RM, Carlson CA, et al. (2009).
Seasonality and vertical structure of microbial communities in an ocean gyre. ISME J. 3:1148–
1163.
Tripp HJ. (2013). The unique metabolism of SAR11 aquatic bacteria. J. Microbiol. 51:147–153.
Vergin KL, Beszteri B, Monier A, Thrash JC, Temperton B, Treusch AH, et al. (2013a). High-
resolution SAR11 ecotype dynamics at the Bermuda Atlantic Time-series Study site by
phylogenetic placement of pyrosequences. ISME J. 7:1322–1332.
Vergin KL, Done B, Carlson CA, Giovannoni SJ. (2013b). Spatiotemporal distributions of rare
bacterioplankton populations indicate adaptive strategies in the oligotrophic ocean. Aquat.
Microb. Ecol. 71:1–13.
Watts DJ, Strogatz SH. (1998). Collective dynamics of ‘small-world’ networks. Nature 393:440–
442.
Weinbauer MG, Hornák K, Jezbera J, Nedoma J, Dolan JR, Šimek K. (2007). Synergistic and
antagonistic effects of viral lysis and protistan grazing on bacterial biomass, production and
diversity. Environ. Microbiol. 9:777–788.
Weinbauer MG, Liu J, Motegi C, Maier C, Pedrotti ML, Dai M, et al. (2013). Seasonal
variability of microbial respiration and bacterial and archaeal community composition in the
upper twilight zone. Aquat. Microb. Ecol. 71:99–115.
Whitman WB, Coleman DC, Wiebe WJ. (1998). Prokaryotes: The unseen majority. Proc. Natl.
Acad. Sci. 95:6578–6583.
Wickham H. (2009). ggplot2: elegant graphics for data analysis. Springer New York
http://had.co.nz/ggplot2/book.
Cram
209
Wilcox R, Fuhrman J. (1994). Bacterial viruses in coastal seawater: lytic rather than lysogenic
production. Mar. Ecol.-Prog. Ser. 114:35–45.
Wilson SE, Steinberg DK. (2010). Autotrophic picoplankton in mesozooplankton guts: evidence
of aggregate feeding in the mesopelagic zone and export of small phytoplankton. Mar. Ecol.
Prog. Ser. 412:11–27.
Wiltshire KH, Kraberg A, Bartsch I, Boersma M, Franke H-D, Freund J, et al. (2010). Helgoland
Roads, North Sea: 45 Years of Change. Estuaries Coasts 33:295–310.
Winter C, Bouvier T, Weinbauer MG, Thingstad TF. (2010). Trade-Offs between Competition
and Defense Specialists among Unicellular Planktonic Organisms: the ‘Killing the Winner’
Hypothesis Revisited. Microbiol. Mol. Biol. Rev. 74:42–57.
Winter C, Smit A, Herndl GJ, Weinbauer MG. (2004). Impact of Virioplankton on Archaeal and
Bacterial Community Richness as Assessed in Seawater Batch Cultures. Appl. Environ.
Microbiol. 70:804–813.
Woebken D, Lam P, Kuypers MMM, Naqvi SWA, Kartal B, Strous M, et al. (2008). A
microdiversity study of anammox bacteria reveals a novel Candidatus Scalindua phylotype in
marine oxygen minimum zones. Environ. Microbiol. 10:3106–3119.
Woese CR, Kandler O, Wheelis ML. (1990). Towards a natural system of organisms: proposal
for the domains Archaea, Bacteria, and Eucarya. Proc. Natl. Acad. Sci. 87:4576–4579.
Wood SN. (2011). Fast stable restricted maximum likelihood and marginal likelihood estimation
of semiparametric generalized linear models. J. R. Stat. Soc. Ser. B Stat. Methodol. 73:3–36.
Wood SN. (2006). Generalized additive models: an introduction with R. Chapman & Hall/CRC:
Boca Raton, FL.
Wood SN. (2004). Stable and Efficient Multiple Smoothing Parameter Estimation for
Generalized Additive Models. J. Am. Stat. Assoc. 99:673–686.
Wright JJ, Konwar KM, Hallam SJ. (2012). Microbial ecology of expanding oxygen minimum
zones. Nat. Rev. Microbiol. 10:381–394.
Wright JJ, Mewis K, Hanson NW, Konwar KM, Maas KR, Hallam SJ. (2013). Genomic
properties of Marine Group A bacteria indicate a role in the marine sulfur cycle. ISME J.
Cram
210
http://www.nature.com/ismej/journal/vaop/ncurrent/abs/ismej2013152a.html (Accessed
September 12, 2013).
Xia LC, Ai D, Cram J, Fuhrman JA, Sun F. (2013). Efficient statistical significance
approximation for local similarity analysis of high-throughput time series data. Bioinforma. Oxf.
Engl. 29:230–237.
Yeo SK, Huggett MJ, Eiler A, Rappé MS. (2013). Coastal Bacterioplankton Community
Dynamics in Response to a Natural Disturbance. PLoS ONE 8:e56207.
Zehr JP, Ward BB. (2002). Nitrogen Cycling in the Ocean: New Perspectives on Processes and
Paradigms. Appl. Environ. Microbiol. 68:1015–1024.
Zhou J, Deng Y , Luo F, He Z, Tu Q, Zhi X. (2010). Functional Molecular Ecological Networks.
mBio 1:e00169–10.
Abstract (if available)
Abstract
Marine bacteria throughout the water column are an important part of the marine ecosystem and drive global geochemical processes, yet their long term ecological dynamics have been studied largely in the euphotic zone and adjacent seasonally mixed depths. Exploring microbial dynamics at monthly and interannual time scales throughout the water column provides an opportunity to explore how microorganisms are shaped by their dynamic environment. This environment includes not only the physical and chemical characteristics of the water in which the microbes live but also the other organisms present in the water column. Predator-prey relationships and infection processes are likely of particular importance as they preferentially remove some organisms while leaving others, thereby shaping microbial communities. The San Pedro Channel, off the coast of Los Angeles is an ideal model system for exploring microbial dynamics, as its physics, chemistry and biology have been regularly sampled through the San Pedro Ocean Time-series (SPOT). Previous analysis of data from SPOT has identified seasonal and interannual dynamics of chemistry, biology and especially the structure of microbial communities in surface water. The community structure of the deep water column had, before now, been studied little
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Changes in the community composition of marine microbial eukaryotes across multiple temporal scales of measurement
PDF
Microbe to microbe: monthly microbial community dynamics and interactions at the San Pedro Ocean Time-series
PDF
Spatial and temporal dynamics of marine microbial communities and their diazotrophs in the Southern California Bight
PDF
Temporal variability of marine archaea across the water column at SPOT
PDF
Patterns of molecular microbial activity across time and biomes
PDF
Annual pattern and response of the bacterial and microbial eukaryotic communities in an aquatic ecosystem restructured by disturbance
PDF
Spatial and temporal investigations of protistan grazing impact on microbial communities in marine ecosystems
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
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
Ecological implications of daily-to-weekly dynamics of marine bacteria, archaea, viruses, and phytoplankton
PDF
Marine protistan diversity, spatiotemporal dynamics, and physiological responses to environmental cues
PDF
Using sequencing techniques to explore the microbial communities associated with ferromanganese nodules and sediment from the South Pacific gyre
PDF
The impact of the concentration and distribution of dissolved and particulate B-vitamins and their congeners on marine microbial ecology
PDF
The distribution of B-vitamins in two contrasting aquatic systems, and implications for their ecological and biogeochemical roles
PDF
Disentangling the ecology of bacterial communities in cnidarian holobionts
PDF
Ecological recovery dynamics of the benthic and pelagic fauna in response to extreme temperature events and low oxygen environments developed during the early Triassic
PDF
Phytoplankton bloom initiation in the Southern California Bight: a multi-year local and regional analysis
PDF
Environmental controls on alkalinity generation from mineral dissolution: from the mineral surface to the global ocean
Asset Metadata
Creator
Cram, Jacob Adrian
(author)
Core Title
Dynamics of marine bacterial communities from surface to bottom and the factors controling them
School
College of Letters, Arts and Sciences
Degree
Doctor of Philosophy
Degree Program
Biology
Publication Date
10/03/2014
Defense Date
06/17/2014
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
bacteria,infection,marine,microbial ecology,network association analysis,OAI-PMH Harvest,plankton,predation,San Pedro Channel,time-series,water column
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Fuhrman, Jed A. (
committee chair
), Capone, Douglas G. (
committee member
), Caron, David A. (
committee member
), Meshkati, Najmedin (
committee member
), Sun, Fengzhu Z. (
committee member
)
Creator Email
cram@usc.edu,cramjaco@gmail.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c3-488149
Unique identifier
UC11286744
Identifier
etd-CramJacobA-2998.pdf (filename),usctheses-c3-488149 (legacy record id)
Legacy Identifier
etd-CramJacobA-2998.pdf
Dmrecord
488149
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Cram, Jacob Adrian
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
bacteria
infection
marine
microbial ecology
network association analysis
plankton
predation
San Pedro Channel
time-series
water column