University of Southern California Dissertations and Theses

Multi-dataset analysis of bacterial heterotrophic variability at the San Pedro Ocean Time-series (SPOT): an investigation into the necessity and feasibility of incorporating a dynamic bacterial c...

(USC Thesis Other)

Multi-dataset analysis of bacterial heterotrophic variability at the San Pedro Ocean Time-series (SPOT): an investigation into the necessity and feasibility of incorporating a dynamic bacterial c...

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Copyright 2017 Elizabeth Nelson Teel
Multi-dataset analysis of bacterial
heterotrophic variability at the San
Pedro Ocean Time Series (SPOT): an
investigation into the necessity and
feasibility of incorporating a dynamic
bacterial component into global
ecosystem NPZD models.
Elizabeth N. Teel

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)
August 2017
Teel
2
Acknowledgments
This research was funded through the National Science Foundation, the NSF
Graduate Research Fellowship Program, the University of Southern California, the
Wrigley Institute for Environmental Sciences, the National Oceanographic and
Atmospheric Administration, and the University of Southern California Sea Grant.
This dissertation includes research completed through collaboration with many
laboratories in southern California, but I would like to especially thank my advisor,
Naomi Levine, my committee members, Burton Jones, Jed Fuhrman, and Doug
Hammond, and fellow researchers from USC and abroad, Xiao Liu, Erin McParland,
Matthew Ragan, Carl Oberg, Xuening Wen, Sara Rivero-Calle, Jennifer Cannon, Tara
Kerin, Aya Hozumi, Grace Kim, Thomas Leeuw, Popi Gonzalez, Loïc Marnat, Johanna
Holm, Rohan Sachdeva, Ella Sierazki, Alma Parada, Pieter Share, Stephanie Ho,
Megan Hall, Sabrina Ho, Rosalynn Sylvan, Jason Sylvan, Erin Fichot, Lily Momper,
Ben Tully, Michael Morando, David Needham, Jacob Cram, Cheryl Chow, Dario Diehl,
William Haskell, Nick Rollins, Arvind Pereira, Bridget Seegers, Ivona Cetinić, Ioannis
Georgakakis, Violaine Mendez, Nikolaos Zarokanellos, Efrain Perez, Fernando Cagua,
Dale Kiefer, John Heidelberg, Emmanuel Boss, Mary Jane Perry, Brian Seegers, Laura
Gómez Consarnau, Diane Kim, and Troy Gunderson for their tremendous advice
throughout my time at USC. Lastly, I would like to thank my son, husband, parents,
in-laws, and friends; this research would not have been possible without their
guidance and unending support.
Teel
3
Table of Contents
Acknowledgments ................................................................................................................... 2
Table of Contents ..................................................................................................................... 3
List of Figures ........................................................................................................................... 5
List of Tables ............................................................................................................................. 6
Abstract ...................................................................................................................................... 7
Chapter 1. The need for improved bacterial heterotrophic remineralization in
marine ecosystem models .................................................................................................... 8
1.1 Ocean Ecosystem Models ........................................................................................................ 8
1.2 Oceanic DOC Cycling .............................................................................................................. 10
1.3 Bacterial Community Structure and Function .............................................................. 14
1.4 Extrapolation of Time-series Data .................................................................................... 17
1.5 Research Goals ......................................................................................................................... 19
References ........................................................................................................................................ 23
Chapter 2. Contextualizing time-series data: Using high-resolution in-situ
gliders to quantify short-term regional oceanographic variability in the San
Pedro Channel ........................................................................................................................ 35
Abstract ............................................................................................................................................. 35
2.1 Introduction .............................................................................................................................. 36
2.2 Methods ...................................................................................................................................... 39
2.2.1 Glider Deployments ......................................................................................................................... 39
2.2.2 Glider Analyses .................................................................................................................................. 40
2.2.3 Ancillary Satellite Data ................................................................................................................... 43
2.3 Results ........................................................................................................................................ 44
2.3.1 Cross-Channel Oceanographic Trends .................................................................................... 44
2.3.2 Regional Similarity to SPOT Station .......................................................................................... 47
2.3.3 Surface to Sub-surface Connectivity ......................................................................................... 51
2.4 Discussion ................................................................................................................................. 54
2.4.1 Regional Application of SPOT Data ........................................................................................... 54
2.4.2 Determining Regional Domains for Time-Series Sites ...................................................... 56
2.4.3 Incorporation of Time-Series Data into Future Modeling Efforts ................................ 58
2.5 Conclusions ............................................................................................................................... 59
Acknowledgments ......................................................................................................................... 61
References ........................................................................................................................................ 75
Chapter 3. From Networks to Models: Investigation and parameterization of
bacterial carbon demand at the San Pedro Ocean Time-series site (SPOT). .... 80
Abstract ............................................................................................................................................. 80
3.1 Introduction .............................................................................................................................. 81
3.2 Study Site ................................................................................................................................... 83
3.3 Data and Methods ................................................................................................................... 85
3.3.1 Microbial Observatory Data ......................................................................................................... 85
3.3.2 Satellite Chlorophyll a and Photosynthetically Active Radiation data ...................... 86
3.3.3 Wrigley Time Series Data .............................................................................................................. 86
3.3.4 Primary Production Estimates .................................................................................................... 87
Teel
4
3.3.5 Bacterial Growth and Growth Efficiency ................................................................................ 89
3.4. Results ....................................................................................................................................... 92
3.4.1 “Black Box” Algorithms .................................................................................................................. 92
3.4.2 Community Composition and Function .................................................................................. 96
3.4.3 BGE and viral dynamics ............................................................................................................... 102
3.4.4 Linear Regression of BGE at SPOT ........................................................................................... 103
3.5 Discussion ............................................................................................................................... 103
3.6 Conclusions ............................................................................................................................. 107
Acknowledgments ....................................................................................................................... 108
References ...................................................................................................................................... 129
Chapter 4. Modeling the effects of free-living marine bacterial community
composition on heterotrophic remineralization rates and biogeochemical
carbon cycling ..................................................................................................................... 135
Abstract ........................................................................................................................................... 135
4.1 Introduction ............................................................................................................................ 136
4.2 Methods .................................................................................................................................... 141
4.2.1 Base Ecological Model ................................................................................................................... 141
4.2.2 Bacterial Model ................................................................................................................................ 143
4.2.3 Physical Framework and Environmental Drivers ............................................................ 151
4.2.4 Validation Dataset .......................................................................................................................... 153
4.3 Results ...................................................................................................................................... 154
4.3.1 Phytoplankton Dynamics: ........................................................................................................... 154
4.3.2 Bacterial Dynamics ......................................................................................................................... 156
4.3.3 Sensitivity tests ................................................................................................................................ 158
4.4 Discussion ............................................................................................................................... 160
4.4.1 Impact of future changes ............................................................................................................. 160
4.4.2 Next Steps ........................................................................................................................................... 164
4.5 Conclusions ............................................................................................................................. 166
Acknowledgments ....................................................................................................................... 166
References ...................................................................................................................................... 187
Bibliography ........................................................................................................................ 193

Teel
5
List of Figures
Figure 1. 1 The Southern California Bight ..................................................................................... 22

Figure 2. 1 Glider Deployment Map and Idealized Glider Transect ................................... 62
Figure 2. 2 Temperature and Chlorophyll Profiles for Ideal Profile Types .................... 63
Figure 2. 3 Cross-channel Variation of Profile Characteristics ............................................ 64
Figure 2. 4 Principal Component Analysis of Glider Profiles ................................................ 67
Figure 2. 5 Temporal Variation at SPOT and Coastal Locations .......................................... 69
Figure 2. 6 Investigation into Whole-Channel Oceanographic Events ............................. 70
Figure 2. 7 Integrated Chlorophyll below the First Optical Depth ..................................... 71

Supplemental Figure 2. 1 Principal Component Analysis of Ideal Profiles .................... 73
Supplemental Figure 2. 2 Optical Depth versus Glider-Satellite Mismatch ................... 74

Figure 3. 1 Surface Modules, Modified from Cram, 2014 .................................................... 109
Figure 3. 2 Satellite PAR data, MODIS Aqua .............................................................................. 112
Figure 3. 3 Chlorophyll Calibration ............................................................................................... 113
Figure 3. 4 Depth Resolved Seasonal Primary Prediction .................................................. 114
Figure 3. 5 Leucine to Thymidine Incorporation Ratios ...................................................... 115
Figure 3. 6 White et al. (1991) Predicted Growth Rates at the SPOT Site .................... 116
Figure 3. 7 Linear Regressions of Bacterial Production and Growth ............................. 117
Figure 3. 8 Temperature versus Bacterial Production and Growth ............................... 118
Figure 3. 9 Seasonality of Bacterial Production ...................................................................... 119
Figure 3. 10 Correlation between Bacterial and Phytoplankton Measurements ..... 120
Figure 3. 11 BGE versus Diversity, Evenness, and OTU Richness ................................... 121
Figure 3. 12 Average Monthly Bacterial Abundance and Satellite Chlorophyll ......... 122
Figure 3. 13 Growth Rate and Abundance of Mod1 and Mod2 ......................................... 123
Figure 3. 14 Mod2 Subgroups and Cyanobacterial Abundance ........................................ 124
Figure 3. 15 BGE for High and Low Mod2 Communities ..................................................... 125
Figure 3. 16 Community Dependence of BGE versus Temperature ............................... 126
Figure 3. 17 Viral Abundance and Bacterial Growth Efficiency ....................................... 127
Figure 3. 18 Predicted BGE as fn(PPint, T, %Mod2) .............................................................. 128

Figure 4. 1 Base Ecosystem Model Schematic .......................................................................... 168
Figure 4. 2 Bacterial Ecosystem Model Schematic ................................................................. 169
Figure 4. 3 Seasonal Cycles of Environmental Drivers ......................................................... 170
Figure 4. 4 Correlation between NO3 and Temperature at SPOT ................................... 171
Figure 4. 5 Base Ecosystem Seasonal Biomass Results ........................................................ 173
Figure 4. 6 Base Model Phytoplankton versus SPOT Estimates ....................................... 174
Figure 4. 7 Base Model Picophytoplankton versus SPOT Estimates .............................. 175
Figure 4. 8 Base Model Phytoplankton Growth Rates .......................................................... 176
Figure 4. 9 Bacterial Model Seasonal Biomass Results ........................................................ 177
Figure 4. 10 Bacterial Model Phytoplankton Growth Rates ............................................... 178
Teel
6
Figure 4. 11 Ecological Lag within Bacterial Model and at SPOT .................................... 179
Figure 4. 12 Bacterial Oscillation within Bacterial Model .................................................. 180
Figure 4. 13 Bacterial Model Bacteria versus SPOT Measurements ............................... 181
Figure 4. 14 Bacterial Model without Lability Restriction .................................................. 182
Figure 4. 15 Bacterial Model without Variable BGE .............................................................. 183
Figure 4. 16 Bacterial Model with Single Bacterial Group .................................................. 184
Figure 4. 17 Flux Sensitivity Analysis: Large Phytoplankton Biomass ......................... 185
Figure 4. 18 Flux Sensitivity Analysis: Picophytoplankton Biomass ............................ 186

List of Tables
Table 2. 1 Distribution of Profile Types ......................................................................................... 68

Supplemental Table 2. 1 Ideal Profile Characteristics ............................................................. 72

Table 3. 1 Operational Taxonomic Unit Abbreviations, from Cram, 2014 ................... 110
Table 3. 2 Ancillary Abbreviations, from Cram, 2014 ........................................................... 111

Table 4. 1 Model Parameterization for Base and Bacterial Models ................................ 172

Teel
7
Abstract
The oceans have taken up approximately half of anthropogenic carbon emissions
thereby buffering the impact of climate change; however, the rate at which the
ocean will continue to take up carbon from the atmosphere is controlled by many
mechanisms that may be affected by increased global temperatures. Understanding
the sensitivity of the ocean carbon pump to changes in climate is an area of active
research. Prokaryotic heterotrophic organisms are central to marine carbon
dynamics, but dynamic heterotrophic bacteria have yet to be incorporated into
global ocean ecosystem models. In this research, I developed a pipeline for utilizing
network analyses generated from interdisciplinary data sets to parameterize
bacterial function and to incorporate a mechanistic representation of bacterial
heterotrophic dynamics into a mixed layer Nutrient-Phytoplankton-Zooplankton-
Detritus model. In addition, the application of network analyses from a single time-
series location to broader regions was validated through a high-resolution spatial
and temporal in-situ study of water profile variability around the time-series site.
The primary findings of this dissertation are (1) monthly time-series sampling at the
San Pedro Ocean Time-series site is not likely to be strongly biased by nearshore
dynamics, (2) bacterial growth efficiencies at the San Pedro Ocean Time-series site
are best predicted as a function of integrated primary production, temperature, and
bacterial community composition, and (3) the incorporation of explicit mechanistic
bacterial dynamics into global ocean ecosystem models alters both bottom-up and
top-down dynamics and allows for the decoupling of bacterial production from
primary production.
Teel
8
Chapter 1. The need for improved bacterial heterotrophic
remineralization in marine ecosystem models
1.1 Ocean Ecosystem Models
Climate change threatens to increase drought, reduce food supply, and
disturb global weather patterns. Global climate models are indispensable for
predicting the severity of these effects. Specifically, modeling oceanic carbon cycling
is particularly important because the oceans play a critical role in the global carbon
cycle and have taken up approximately one third of anthropogenic CO2 emissions
since the Industrial Revolution (Khatiwala, et al., 2013). Carbon uptake by the
oceans occurs as a result of both physics and biology; however, biological
representation within ocean ecosystem models has been oversimplified. In a recent
review of Earth System Models used in the International Pannel on Climate Change’s
5
th
assessment, only eight of the oceanic ecosystem models incorporated two or
more phytoplankton groups and one or more zooplankton functional group
(Laufkötter et al., 2015 and references therein). Only one of these contained any
explicit representation of bacterial dynamics (Vichi et al., 2007a; Vichi et al., 2007b;
Vichi et al., 2009; Laufkötter et al, 2015). As a result of inconsistent simplifications
of biological parameters and variation in physics between models, net primary
production (NPP) estimates had low inter-model agreement; furthermore, these
estimates underpredicted the global observations and were only poorly correlated
with those observations (0.18 to 0.69), indicating large error and uncertainty in the
Teel
9
mechanization of the ocean carbon cycle (Laufkötter et al., 2015). This variation
shows up in predicted results as well, which do not even agree whether or not NPP
will increase or decrease under future ocean conditions (Laufkötter et al., 2015).
Improving these models will require changes in both the physics and the
biology represented within them. While there are many aspects of global carbon
models that need improvement, one of the most overtly underrepresented
mechanisms is heterotrophic remineralization in the surface ocean. Heterotrophic
remineralization in the oceans is one of the major predictors for the fate of carbon
fixed through primary production. This remineralization is biologically driven and
has timescales that vary by orders of magnitude (Jiao et al., 2010); however, in
general remineralization is assumed to operate as a simple function within ocean
ecosystem models. For example, in the widely used model published by Moore et al.
(2002), remineralization is represented by a single rate constant, which is only
variable as a function of temperature. The equation for the remineralization of DOC
in this model is:
(eq 1.1)
!"#$
!"
=−𝑑𝑜𝑐
!"#$%
∗𝑇𝑓𝑢𝑛𝑐 ∗𝐷𝑂𝐶
where docremin is the remineralization rate which is a constant (typically 0.1 day
-1
),
and Tfunc is a Q10 temperature function:
(eq 1.2) 𝑇𝑓𝑢𝑛𝑐= exp (−4000∗
!
!!!"#.!"
−
!
!"!.!"
)
where T represents the ambient mixed layer temperature. An update of this model
further simplified remineralization of DOC by removing this temperature function
and assuming that all labile DOC is remineralized immediately (Moore et al., 2004).
Teel
10
The Moore et al. (2004) model is used as the basis for the ecosystem model within
the widely cited and utilized National Center for Atmospheric Research (NCAR)
Community Earth System Model (CESM) (Doney et al., 2009; Lindsay, 2012; Moore
et al., 2013). Estimates from the NCAR CESM have been validated world-wide with
in-situ data, potentially justifying the absence of explicit heterotrophs from Earth
System Models (Moore et al., 2004); however, this lack of mechanistic
remineralization limits the sensitivity of the model to global change, which in turn
reduces the validity of model predictions for a suite of remineralization-dependent
features of the global carbon cycle, including POC export flux to the deep ocean,
nutrient availability in the surface ocean, and total surface ocean pH.
1.2 Oceanic DOC Cycling
A strong argument for the need for mechanistic remineralization within
oceanic ecosystem models lies in the complexity of the oceanic DOC cycle. Estimates
of carbon storage in the ocean in 2010 showed that, while the largest reservoir is
dissolved inorganic carbon (DIC) (37,400 Gt C), the carbon stored in the organic
fractions (682 GtC) rivals the carbon storage of the entire atmosphere (750 GtC)
(Jiao et al., 2010). The reservoir of DOC alone accounts for approximately 97% of
that carbon storage and can be further divided into labile, semi-labile, and
refractory pools, which have dramatically different bioavailability and lifetimes
(Hansell et al., 2009, Jiao et al., 2010). Through measurements of bacterial
production (BP) and primary production (PP), it has been estimated that
approximately half of all primary production is readily consumed by heterotrophic
Teel
11
bacteria as labile DOC, though this ratio of BP:PP has also been found to vary from
5% to 50% between studies (Fuhrman and Azam, 1980, Fuhrman and Azam, 1982,
Azam et al., 1983, Pomeroy and Deibel, 1986, Azam et al., 1993, Ducklow, 1999,
Anderson and Ducklow, 2001, Morán et al., 2002, Hoppe et al., 2002, Jiao et al.,
2010). The lifetime of this labile DOC in the surface ocean is thought to be hours to
days, but refractory DOC, due to its limited bioavailability, accumulates in the
surface ocean, can be exported through oceanic circulation, can persist for
thousands of years, and is estimated to account for approximately 624 Gt C (Hansell
et al., 2009, Jiao et al., 2010; Hansell et al., 2012). For this reason, oceanic DOC can
be viewed as a potential long-term reservoir for drawdown of carbon from the
atmosphere.
The primary sources of oceanic DOC are direct extracellular release (ER) by
phytoplankton, release of microbial cytoplasm through sloppy feeding by
zooplankton or viral lysis, and partial hydrolysis of particulate organic carbon (POC)
by bacteria and archaea (Azam, 1998, Carlson et al., 1998, Jiao et al., 2010, Becker et
al., 2014). The composition and bioavailability of these molecules is related to the
DOC source, microbial metabolism, and UV irradiation (Carlson et al., 2004, Mou et
al., 2008, Hansell et al., 2009, Jiao et al., 2010, McCarren et al 2010, Halewood et al.,
2012, Sarmiento and Gasol, 2012, Becker et al., 2014). While our understanding of
the mechanisms that control DOC production has been significantly advanced in
recent years, these processes are still poorly understood particularly in contrast to
the wealth of information regarding oceanic PP, ER and the percent extracellular
release (PER), (Smith et al., 1977, Fogg, 1983, Carlson et al., 1998, Teira et al., 2001,
Teel
12
Halewood et al., 2012). Estimates of PER range greatly between laboratory and field
studies. Baines and Pace (1991) proposed a cross-system estimate of PER at 13% of
PP, but values reported from culture studies and field incubations range from 1% to
as high as 86% (Carlson et al., 1998, Nagata et al., 2000, Sintes et al., 2010, Becker et
al., 2014). To the first order, PER appears to be negatively correlated with PP and
DIN (Fogg, 1983, Teira et al., 2001, Morán et al., 2002, Halewood et al., 2012).
Heterotrophic bacteria are the primary consumers of DOC in the surface
ocean and have been shown to be extremely important in the creation of
recalcitrant DOC (Kirchman et al., 1991, Azam, 1998; del Giorgio and Duarte, 2002,
Cherrier and Bauer, 2004, Jiao et al., 2010). Remineralization by heterotrophic
bacteria preferentially removes labile compounds, especially those that are rich in
nitrogen (N) and phosphorus (P) (Azam 1998, Jiao et al., 2010). This results in a
decoupling of C, N, and P cycling and the buildup of semi-labile and recalcitrant DOC
with a much higher C:N:P ratio than labile DOC (Azam et al., 1983, Aluwihare et al.,
1997, Azam, 1998, Jiao et al., 2010). Additionally, a portion of the recalcitrant
marine DOC is of bacterial or archaeal origin, such as cell wall components, implying
that recalcitrant DOC may be exuded during BP or released during viral lysis of
bacterial cells (Stoderegger and Herndl, 1998, Suttle, 2007; Jiao et al., 2010).
Laboratory culture experiments have shown that bacterial extracellular release of
semi-labile and recalcitrant DOC may account for 25% of bacterial mortality that is
often attributed to bacterial respiration (BR) (Stoderegger and Herndl, 1998).
Consequently, models that do not account for dynamic marine bacterial
heterotrophy will misrepresent several key biogeochemical processes; the
Teel
13
remineralization of POC to DIC, the concentrations of regenerated inorganic
nutrients available for primary production (PP), the overall stoichiometry of organic
matter in the surface waters, and the percentage of fixed carbon that would remain
in the ocean as recalcitrant DOC.
Recent research has also shown that ranges in the quantity and quality of
DOC produced by phytoplankton may be a source of variability in marine bacterial
community function (Halewood et al., 2012; Becker et al., 2014; Wear et al., 2015a;
Wear et al., 2015b). Rates and capacities for DOC remineralization have been
previously been shown in mesocosm experiments in the northwestern Sargasso Sea
to vary between bacterial communities of different depths (Carlson et al., 2004). The
link between bacterial community and metabolic potential has also supported by
research finding strong correlations between bacterial and protistan OTUs, which
have suggested the potential for tight coupling between phytoplankton type and
bacterial carbon uptake (Chow et al., 2013; Chow et al., 2014; Kim et al., 2014;
Needham and Fuhrman, 2016). The ability for the bacterial community to respond
to a phytoplankton bloom has been hypothesized to be driven by the metabolic
potential of low levels of copiotrophic microbes in the rare biosphere (Campbell et
al., 2011; Teeling et al., 2012; Hunt et al., 2013; Williams et al., 2013; Galand et al.,
2015). For example, cyanobacteria can produce large amounts of DOC exudate
during photosynthesis the composition and lability of which can vary significantly
(Becker et al., 2014). This suggests that some heterotrophic bacterial OTUs may be
adapted to grow on specific phytoplankton exudates.
Teel
14
1.3 Bacterial Community Structure and Function
Resolving accurate bacterial dynamics is one of the largest opportunities for
improvement of current ecosystem models and has gained momentum in the past
few years. Various ecosystem models have begun to account for bacterial biomass in
order to better model microzooplankton dynamics (PISCES, Tagliabue et al., 2011),
stoichiometry and DOC cycling (Vichi et al., 2007a; Vichi et al., 2007b; Vichi et al.,
2009; Hasumi and Nagata, 2014), light absorption and scattering (CoSINE, Xiu and
Chai, 2014), and viral dynamics (Weitz et al., 2015). Though the inclusion of
bacterial biomass, viruses, and multiple DOM pools are major improvements in
model formulation, each of these models is limited by a lack of realistic bacterial
dynamics. For example, all but one of these models includes a fixed value for
bacterial growth efficiency (BGE). These fixed values range from 0.1 (10%) (Hasumi
and Nagata, 2014) to 0.5 (50%) (CoSINE, Xiu and Chai, 2014). In the case of Vichi et
al. (2007a; 2007b; 2009), BGE varies as a function of oxygen concentration,
resulting in a BGE that is still fixed to a specific value for the surface mixed-layer.
In contrast to these models, field measurements of BGE have shown that it
varies with trophic state and bacterial production (BP), ranging from 0.01 (1%) in
oligotrophic regions to a maximum of approximately 0.5 (50%) in eutrophic coastal
zones (del Giorgio and Cole, 1998). BGE is defined as:
(eq 1.3) 𝐵𝐺𝐸=
!"
!"#
=
!"
!"!!"

A value of 0.5 for BGE, therefore, indicates that half of the carbon taken up by
bacteria will be respired and half will be assimilated into bacterial biomass.
Teel
15
Compared to a model where BGE is 0.1, a value of 0.5 would substantially increase
the amount of carbon available to the grazing food chain as bacterial biomass, would
reduce the buildup of DIC in the surface mixed layer, and would reduce the
concentration of remineralized nutrients available for local PP.
BGE is not a commonly measured variable for microbial communities due to
the challenge of measuring BR and BP on the same time-scales, 24- to 48-hour
incubations versus one- to two-hour incubations, respectively (Williams, 1981,
Fuhrman and Azam, 1982, Kirchman et al., 1985, Simon and Azam, 1989, del Giorgio
et al., 2006, del Giorgio et al., 2011). BGE can also be calculated as the loss of
bioavailable carbon over the duration of an incubation divided by the increase in BA
(Halewood et al., 2012), however, this method does not account for the effect of viral
lysis and requires removal of grazers as well as the accurate measurements of DIC,
DOC, and POC. Through synthesis of 40 studies in oceanic, coastal, and limnic
waters, del Giorgio and Cole (1998) derived the following relationship between BGE
and BP:
(eq 1.4) 𝐵𝐺𝐸=
!.!"#!!.!"∗!"
!.!!!!

While a simplification of actual dynamics, this relationship allows for BP
measurements, in µg C liter
-1
hr
-1
, to be used to estimate total BCD, which is a critical
step for parameterization of DOC remineralization by bacterial communities.
The variation of BP with environmental parameters has been studied
extensively (Hagstrom et al., 1979, Fuhrman et al., 1980; Fuhrman and Azam, 1982,
Kirchman et al., 1985, Smith and Azam, 1992, Kirchman et al., 1995, López-Urrutia
and Morán, 2007). In a review of BP studies from the marine and freshwater
Teel
16
environments in 1991, White et al. found that BP was positively correlated with BA,
chlorophyll a, and temperature. These correlations contained a large amount of
scatter, however, which at the time was attributed to fluctuations in season, nutrient
quality, and grazing pressure (White et al., 1991). Since 1991, the development of
community fingerprinting and sequencing has allowed for broad study of the
variation in marine microbial communities, and there is now strong indication that
community-level fluctuation in BP within free-living bacterial populations may be
related to community composition in addition to physical and chemical drivers
(DeLong et al., 1993, Azam, 1998, Ouverney and Fuhrman, 1999, Riemann et al.,
2000, Carlson et al., 2004, Alonso-Saez et al., 2010, Jiao et al., 2010, Teeling et al.,
2012, Dupont et al., 2012, Campbell and Kirchman, 2013; Chow et al., 2013, Cram,
2014; Galand et al., 2015; Ward et al., 2017).
Heterotrophic bacterial communities are immensely diverse and have been
shown to fluctuate with season (Fuhrman et al., 2006; Gilbert et al., 2010; Beman et
al., 2011; Steele et al., 2011; Gilbert et al., 2012; Chow et al., 2013; Jones et al., 2013;
Cram, 2014), nutrients (Fuhrman and Steele, 2008; Morris et al., 2010; Allen et al.,
2012), salinity (Hewson and Fuhrman, 2004; Fuhrman and Steele, 2008), light (Van
Mooy et al., 2004), viruses (Needham et al., 2013; Chow et al., 2014), and protists
(Steele et al., 2011; Gilbert et al., 2012; Jones et al., 2013; Chow et al., 2014). The
existence of strong seasonal reoccurrence in marine bacterial populations indicates
niche partitioning and a relatively low degree of functional redundancy within
microbial communities (Fuhrman et al., 2006, Fuhrman, 2009). Low functional
redundancy may allow for specific functional responses to be attributed to specific
Teel
17
bacterial communities. These findings strongly suggest that community dynamics
may be essential for the accurate modeling of bacterial dynamics, phytoplankton
dynamics, and carbon cycling. Fortunately, the discovery and confirmation of
seasonality in marine bacterial populations also indicates the potential for
predictive modeling of marine bacterial heterotrophic communities in global
ecosystem models.
1.4 Extrapolation of Time-series Data
The bulk of the research on bacterial community structure and its variability
with time and environmental drivers has occurred at time-series sites, such as the
San Pedro Ocean Time-series (SPOT) site, where Microbial Observatory (MO)
sampling has taken place monthly for over a decade (Fuhrman et al., 2006; Steele et
al., 2011; Chow et al., 2013; Cram et al., 2015). Previous analyses of the surface
bacterial communities at SPOT have identified strong statistical correlations
between a set of distinct bacterial operational taxonomic units (OTUs) and higher
BP, as well as strong negative correlations between oscillating communities of
bacterial OTUs (Chow et al., 2013; Cram, 2014); increases in PP have been shown to
result in dramatic shifts in bacterial community composition, which can in turn
change BP and BGE (Needham and Fuhrman, 2016); and mesocosm dilution
experiments with water collected from the San Pedro Channel have also identified
groups of bacterial OTUs that likely represent “grazer-limited” and “grazer
resistant” functional groups within the local heterotrophic marine bacterial
community (Cram, 2014; Cram et al., 2016). Differential growth and grazing
Teel
18
between common marine bacterial community members would result in complex
carbon transport through the microbial carbon pump and further support the
hypothesis that a dynamic bacterial component may be required to correctly model
oceanic carbon cycling.
Time-series findings such as these are an incredible utility for understanding
fundamentals of biological, chemical, and physical oceanography. A weakness,
though, is the fact that, by its nature, Eulerian sampling does not allow for a parcel of
water to be sampled over time, as Lagrangian sampling theoretically could. If
bacterial communities were homogenous, this would not be a problem, and all
observed community variation could be attributed to local ecological changes. In
regions with little variation in water mass source or type, bacterial communities
may change little over large spatial scales (Hewson et al., 2006). However, in the
Southern California Bight (SCB), where there are distinct onshore and offshore
signals due to variations in upwelling and the strength of the Southern California
Eddy, bacterial communities may be more easily predicted by water mass type than
by location (Figure 1.1; Huyer, 1983; Oey et al., 1999; Noble et al., 2009; Allen et al.,
2012). Therefore, if a significant change occurred in the microbial community
sampled at SPOT, it would not be immediately apparent whether this was the result
of a change to the microbial community within a single water mass or the
replacement of the entire microbial community due to advection of a different water
mass.
Additionally, many time-series measurements have low sampling frequency,
commonly sampling once per month. Sampling monthly allows for a sub-seasonal
Teel
19
frequency in the data collection, which may capture relevant changes in microbial
community structure and function, which have been found to be annually and
seasonally reoccurring (Fuhrman et al., 2006; Steele et al., 2011; Chow et al., 2013;
Chow et al., 2014; Giovannoni and Vergin, 2012; Cram et al., 2015); however, recent
publications have also shown that microbial community changes often occur on
time scales of days to weeks, especially following phytoplankton blooms (Teeling et
al., 2012; Needham et al., 2013; Needham and Fuhrman, 2016). Blooms in the San
Pedro Channel are strongly associated with wind driven mixing events that occur on
time scales of days or weeks as well (Huyer, 1983; Holt, 2011). The ability to catch
blooms with monthly sampling is theoretically poor, but the effects of blooms may
remain for months and be easily detected in the monthly dataset, as was seen in
recent work by Halewood et al., (2012) from the Santa Barbara Channel.
With these caveats in mind, the application of time-series data to developing
broad ecological principles that may be incorporated into models requires a firm
understanding of the spatial and temporal variability associated with that time-
series site.
1.5 Research Goals
Because of the complexity of marine microbial processing of carbon, many
different steps can be taken to improve the modeling of heterotrophy in global
ecosystem models. Due to the scale of global models, though, each addition of
complexity must be well justified. Ecological simplification has enabled these
models to function efficiently within coupled 3D global circulation models.
Teel
20
Computational constraints are continually decreasing, however, and the ability to
add to global models provides a great opportunity to increase the realism of these
models. Incorporating a realistic and dynamic bacterial component into global
ecosystem models should be given high priority due to the complexity and
importance of bacterial cycling of carbon, DIN, and iron (Azam et al., 1983, Azam,
1998, Tortell et al., 1999, Zehr and Kudela, 2011). Understanding how
oversimplification of microbial community structure is impacting modeled
ecosystem dynamics and how remineralization rates can be improved is vital for
increasing realism in long-term predictions. This multi-dataset analysis
investigated the variability of heterotrophic BCD and theoretical DOC cycling in
relation to fluctuations of bacterial community structure, PP, and environmental
variables.
The San Pedro Ocean Time-series (SPOT) site was an ideal location for these
proposed analyses because of its relevance to both coastal and oceanic sites and the
wealth of data that has been collected by researchers through the Microbial
Observatory (MO), Upwelling Regime In-Situ Ecosystem Efficiency cruises
UpRISEE), and the Ecology and Oceanography of Harmful Algal Blooms (ECOHAB)
program; however, in dynamic coastal upwelling zones, like the SCB, three major
questions must be addressed before time series results can be scaled to a larger
region: How does the time series location compare to the surrounding waters? What
is the spatial and temporal domain represented by discrete monthly sampling? How
can we use physical data to better inform the interpretation of time series data? In
Chapter 2, to answer these questions, temporal and spatial variation of
Teel
21
environmental and bio-optical variables were studied using a combination of in-situ
profile data and satellite data.
After studying the physical and temporal variability around the SPOT site,
results were directly incorporated into the analysis of bacterial community
functional data in Chapter 3 in order to disentangle the effects of environmental
drivers from those of community composition. The findings from this analysis were
then used to create parameters for bacterial remineralization of DOC, which were
synthesized into a test model in Chapter 4. The goal of my research in Chapter 4 was
two-fold, to determine: 1) how a dynamic heterotrophic bacterial component and
variable remineralization rates could be parameterized in numerical ecosystem
models, and 2) if the addition of a dynamic bacterial component with variable BGE
and community composition is justified, in terms of computational cost and
additional complexity, for global ecosystem models. Due to the interdisciplinary
nature of this research, the results presented here contribute to the science of time-
series analysis, optical oceanography, microbial ecology, and ecosystem modeling.

Teel
22
Figure 1. 1 The Southern California Bight
The Southern California Bight (SCB) from Point Conception to Mexico. Arrows
indicate mean regional surface currents. Top right insert shows SCB within
California. The San Pedro Ocean Time-series sit (SPOT) is represented by the star
located in the San Pedro Channel at 33˚ 33’ 00” N and -118˚ 24’ 00” W.) (Figure
adapted from Noble et al., 2009a.)
Teel
23
References
Allen, L. Z., E. E. Allen, J. H. Badger, J.P. McCrow, I. T. Paulsen, L. D.H. Elbourne, M.
Thiagarajan, D. B. Rusch, K. N. Nealson, S. J. Williamson, J. C. Venter, A. E. Allen.
(2012). Influence of nutrients and currents on the genomic composition of microbes
across an upwelling mosaic. The ISME Journal. 6:1403-1414.

Alonso-Saez L., J. Pinhassi, J. Pernthaler, and J.M. Gasol. (2010). Leucine-to-carbon
empirical conversion factor experiments: does bacterial community structure have
an influence? Environ Mircobiol. 12(11):2988-2997.

Aluwihare, L.I., D.J. Repeta, and R.F. Chen. (1997). A major biopolymeric component
to dissolved organic carbon in surface sea water. Nature 387:166–169.

Anderson T.R. and P.J. le B. Williams. (1999). A one-dimensional model of dissolved
organic carbon cycling in the water column incorporating combined biological-
photochemical decomposition. Global Biogeochemical Cycles. 13(2):337-349.

Anderson T.R. and H.W. Ducklow (2001). Microbial loop carbon cycling in ocean
environments studied using a simple steady-state model. Aquat. Microb. Ecol. 26:37-
49.

Azam, F., T. Fenchel, J.G. Field, J.S. Gray, L.A. Meyer-Reil, and F. Thingstad. (1983).
The ecological role of water-column microbes in the sea. Mar. Ecol. Progr. Ser.
10:257-263.

Azam F., D.C. Smith, G.F. Steward, and A. Hagström. (1993). Bacteria-organic matter
coupling and its significance for oceanic carbon cycling. Microb. Ecol. 28:167-179.

Azam F. (1998). Microbial control of ocean carbon flux: the plot thickens. Science.
280(5364):694-696.

Baines S.B. and M.L. Pace. (1991). The production of dissolved organic matter by
phytoplankton and its importance to bacteria: patterns across marine and
freshwater systems. Limnol. Oceanogr. 36(6):1078-1090.

Becker, J.W., P.M. Berube, C.L. Follett, J.B. Waterbury, S.W. Chisholm, E.F. DeLong,
and D.J. Repeta. (2014). Closely related phytoplankton species produce similar
suites of dissolved organic matter. Front. Microbiol. 5(111):1-14.
doi:10.3389/fmicb.2014.00111.

Beman, J. M., J. A. Steele, and J. A. Fuhrman (2011). Co-occurrence patterns for
abundant marine archaeal and bacterial lineages in the deep chlorophyl maximum
of coastal California. The ISME Journal. 5: 1077-1085.

Teel
24
Brown, M.V., M.S. Schwalbach, I. Hewson, J.A. Fuhrman.(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. Env.
Micro. 7(9):1466-1479.

Campbell B.J., L.-Y. Yu, J.F. Heidelberg, and D.L. Kirchman. (2011). Activity of
abundant and rare bacterial in a coastal ocean. Proc Natl Acad Sci USA.
108(31):12776-12781.

Campbell B.J. and D.L. Kirchman. (2013). Bacterial diversity, community structure,
and potential growth rates along an estuarine salinity gradient. ISME J. 7:210-220.

Carlson C.A., H.W. Ducklow, D.A. Hansell, and W.O. Smith Jr. (1998). Organic carbon
partitioning during spring phytoplankton blooms in the Ross Sea polynya and the
Sargasso Sea. Limol. Oceanogr. 43(3):375-386.

Carlson, C. A., S. J. Giovannoni, D. A. Hansell, S. J. Goldberg, R. Parsons, M. P. Otero, K.
Vergin, and B. R. Wheeler. (2002). Effect of nutrient amendments on
bacterioplankton production, community structure, and DOC utilization in the
northwestern Sargasso Sea. Aquatic Microbial Ecology. 30:19-36.

Carlson C.A., S.J. Giovannoni, D.A. Hansell, S.J. Goldberg, R. Parsons, and K. Vergin.
(2004). Interactions among dissolved organic carbon, microbial processes, and
community structure in the mesopelagic zone of the northwestern Sargasso Sea.
Limol. Oceanogr. 49(4):1073-1083.

Cetinic, I., G. Toro-Farmer, M. Ragan, C. Oberg, and B.H. Jones. (2009). Calibration
procedure for Slocum glider deployed optical instruments. Optics Express.
17(18):15420-15430.

Cherrier, J. and J.E. Bauer. (2004). Bacterial utilization of transient plankton-derived
dissolved organic carbon and nitrogen inputs in surface ocean waters. Aquat.
Microb. Ecol. 35:229-241.

Chow, C. E. T., R. Sachdeva, J. A. Cram, J. A. Steele, D. M. Needham, A. Patel, A. E.
Parada, and J. A. Fuhrman. (2013). Temporal variability and coherence over a
decade in the Southern California Bight. The ISME Journal. 7(12):2259-2273.

Chow, C. E. T., D. Y. Kim, R. Sachdeva, D. A. Caron, and J. A. Fuhrman. (2014). Top-
down controls on bacterial community structure: microbial network analysis of
bacteria, T4-like viruses and protists.

Countway P.D., P.D. Vigil, A. Schnetzer, S.D. Moorthi, and D.A. Caron. (2010).
Seasonal analysis of protistan community structure and diversity at the USC
Microbial Observatory (San Pedro Channel, North Pacific Ocean). Limnol. Oceanogr.
55(6):2381-2396.
Teel
25

Cram, J.A. (2014). Dynamics of marine bacterial communities from surface to
bottom and the factors controlling them. (PhD Dissertation). University of Southern
California.

Cram, J. A., A. Parada, J. A. Fuhrman. (2016). Dilution reveals how viral lysis and
grazing shape microbial communities. Limnology and Oceanography. 61: 889–905.
doi:10.1002/lno.10259

de Boyer Montégut, C., G. Madec, A.S. Fischer, A. Lazar, and D. Iudicone. (2004).
Mixed layer depth over the global ocean: an examination of profile data and profile-
based climatology. J. Geophys. Res. 109(C12003):1-20. doi:10.1029/2004JC002378.

del Giorgio, P.A. and J.J. Cole. (1998). Bacterial growth efficiency in natural aquatic
systems. Annu. Rev. Ecol. Syst. 29:503-541.

del Giorgio, P.A. and C.M. Duarte. (2002). Respiration in the open ocean. Nature.
420:379-384.

del Giorgio, P.A., M.L. Pace, and D. Fischer. (2006). Relationship of bacterial growth
efficiency to spatial variation in bacterial activity in the Hudson River. Aquat. Microb.
Ecol. 45:55-67.

del Giorgio, P. A., R. Condon, T. Bouvier, K. Longnecker, C. Bouvier, E. Sherr, and J. M.
Gasol. (2011). Coherent patterns of bacterial growth, growth efficiency, and leucine
metabolism along a northeastern Pacific inshore-offshore transect. Limnology and
Oceanography. 56(1):1-16.

DeLong E.F., D.G. Franks, and A.L. Alldredge. (1993). Phylogenetic diversity of
aggregate-attached vs. free-living marine bacterial assemblages. Limnol. Oceanogr.
38(5):924-934.

Doney, S.C. (1996). A synoptic atmospheric surface forcing data set and physical
upper ocean model for the U.S. JGOFS Bermuda Atlantic Time-Series Study site. J.
Geophys. Res. 101(C10):25615-25634.

Doney, S.C., I. Lima, J. K. Moore, K. Lindsay, M. J. Behrenfeld, T. K. Westberry, N.
Mahowald, D. M Glover, and T. Takahashi. (2009). Skill metrics for confronting
global upper ocean ecosystem-biogeochemistry models against field and remote
sensing data. Journal of Marine Systems. 76:95-112.
Dong, C.-M., E. Y. Idica, and J. C. McWilliams. (2009). Circulation and multiple-scale
variability in the Southern California Bight. Progress in Oceanography. 82:168-190.

Dupont, C.L., D.B. Rusch, S. Yooseph, M.J. Lombardo, R. A. Richter, R. Valas, M.
Novotny, J. Yee-Greenbaum, J.D. Selengut, D.H. Haft, A.L. Halpern, R.S. Lasken, K.
Teel
26
Nealson, R. Friedman, and J.C. Venter. (2012). Genomic insights to SAR86, an
abundant and uncultivated marine bacterial lineage. ISME J. 6:1186-1199.

Eppley, R.W., E.H. Renger, and W.G. Harrison. (1979). Nitrate and phytoplankton
production in southern California coastal waters. Limnol. Oceanogr. 24(3):483-494.

Fogg, G.E. (1983). The ecological significance of extracellular products of
phytoplankton photosynthesis. Botanica Marina. 26:3-14.

Forsythe, W.C., E.J. Rykiel Jr., R.S. Stahl, H. Wu, and R.M. Schoolfield. (1995). A model
comparison for daylength as a function of latitude and day of year. Ecological
Modelling. 80:87-95.

Fuhrman, J.A. and F. Azam. (1980). Bacterioplankton secondary production
estimates for coastal waters of British Columbia, Antarctica, and California. Applied
Environ. Microbiol. 39(6):1085-1095.

Fuhrman, J.A. and F. Azam (1982). Thymidine incorporation as a measure of
heterotrophic bacterioplankton production in marine surface waters: evaluation
and field results. Marine Biology. 66:109-120.

Fuhrman, J.A., I. Hewson, M.S. Schwalbach, J.A. Steele, M.V. Brown, S. Naeem (2006).
Annually reoccurring bacterial communities are predictable from ocean conditions.
Proc Natl Acad Sci USA 103: 13104–13109.

Fuhrman, J.A. and J. A. Steele. (2008). Community structure of marine
bacterioplankton: patterns, networks, and relationships to function. Aquatic
Microbial Ecology. 53:69-81.

Fuhrman J.A., J.A. Steele, I. Hewson, M.S. Schwalbach, M.V. Brown, J.L. Green, and J.H.
Brown (2008). A latitudinal diversity gradient in planktonic marine bacteria. Proc
Natl Acad Sci USA 105: 7774–7778.

Galand, P. E., I. Salter, and D. Kalenitchenko. Ecosystem productivity is associated
with bacterial phylogenetic distance in surface marine waters. Molecular Ecology.
24:5785-5795.

Gilbert, J. A., D. Field, P. Swift, S. Thomas, D. Cummings, B. Temperton, K. Weynberg,
S. Huse, M. Hughes, I. Joint, P. J. Somerfield, and M. Mühling. (2010). The taxonomic
and functional diversity of microbes at a temperate coastal site: a ‘multi-omics’
study of seasonal and diel temporal variation. PLoS One. 5:e15545.

Gilbert, J. A., J. A. Steele, J. G. Caporaso, L. Steinbrück, J. Reeder, B. Temperton, S.
Huse, A. C. McHardy, R. Knight, I. Joint, P. Somerfield, J. A. Fuhrman, and D. Field.
(2012). Defining seasonal marine microbial community dynamics. The ISME Journal.
6:298-308.
Teel
27

Giovannoni S.J. and K.L. Vergin. (2012). Seasonality in ocean microbial communities.
Science. 335:671-676.

Hagström, A., U. Larsson, P. Hörstedt, and S. Normark. (1979). Frequency of dividing
cells, a new approach to the determination of bacterial growth rates in aquatic
environments. Applied and Environ. Microbiol. 37(5):805-812.

Halewood, E.R., C.A. Carlson, M.A. Brzezinski, D.C. Reed, and J. Goodman. (2012).
Annual cycle of organic matter partitioning and its availability to bacteria across the
Santa Barbara Channel continental shelf. Aquat. Microb. Ecol. 67:189-209.

Hamersley M.R., K.A. Turk, A. Leinweber, N. Gruber, J.P. Zehr, T. Gunderson, and D.G.
Capone. (2011). Nitrogen fixation within the water column associated with two
hypoxic basins in Southern California Bight. Aquat. Microb. Ecol. 63:193-205.

Hansell, D.A., C.A. Carlson, and S. Suzuki. 2002. Dissolved organic carbon export with
North Pacific Intermediate Water formation. Global Biogeochemical Cycles 16:77–84.

Hansell, D.A., C. A. Carlson, D.J. Repeta, and R. Schlitzer. (2009). Dissolved organic
matter in the ocean. Oceanography. 22(4):202-211.

Hansell, D. A., Carlson, C. A., & Schlitzer, R. (2012). Net removal of major marine
dissolved organic carbon fractions in the subsurface ocean. Global Biogeochemical
Cycles, 26(1).

Hasumi, H. and T. Nagata. (2014). Modeling the global cycle of marine dissolved
organic matter and its influence on marine productivity. Ecological Modelling. 288:9-
24.

Hewson, I., and J. A. Fuhrman. (2004). Richness and diversity of bacterioplankton
species along an estuarine gradient in Moreton Bay, Australia. Applied and
Environmental Microbiology. 70(6):3425-3433.

Hewson, I., J.A. Steele, D.G. Capone, and J.A. Fuhrman. (2006). Temporal and spatial
scales of variation in bacterioplankton assemblages of oligotrophic surface waters.
Mar. Eco. Prog. Series 311:67-77.

Holt, B. (2011). “Coastal Upwelling and Harmful Algal Blooms in Southern California
(September-October 2011)” Physical Oceanography Distributed Active Archive
Center. Article published online at:
podaac.jpl.nasa.gov/oceanevents/hab_sc_septemter-october2011

Hoppe, H.-G., K. Gocke, R. Koppe, and C. Begler. (2002). Bacterial growth and
primary production along a north-south transect of the Atlantic Ocean. Nature.
416:168-171.
Teel
28

Hu H. and T. Liu. (2003). Oceanic thermal and biological responses to Santa Ana
winds. Geophysical Research Letters. 30(11)1596. Doi:10.1029/2003GL017208.

Hunt D.E., Y.J. Lin, M.J. Church, D.M. Karl, S.G. Tringe, L.K. Izzo, and Z.I. Johnson.
(2013). Relationship between abundance and specific activity of bacterioplankton in
open ocean surface waters. Appl. Environ. Microbiol. 79(1):177-184.

Huyer, A. (1983). Coastal upwelling in the California Current System. Prog. Oceanog.
12:259-284.

Iturriaga, R. and B.G. Mitchell. (1986). Chroococcoid cyanobacteria: a significant
component in the food web dynamics of the open ocean. Mar. Ecol. Prog. Ser. 28:291-
297.

Jacox, M. G., C. A. Edwards, M. Kahru, D. L. Rudnick, and R. M. Kudela. 2015. The
Potential for Improving Remote Primary Productivity Estimates through Subsurface
Chlorophyll and Irradiance Measurement. Deep-Sea Research II. 112:107-116.

Jiao, N.-Z., G.J. Herndl, D.A. Hansell, R. Benner, G. Kattner, S.W. Wilhelm, D.L.
Kirchman, M.G. Weinbauer, T.-W. Luo, F. Chen, and F. Azam. (2010). Microbial
production of recalcitrant dissolved organic matter: long-term carbon storage in the
global ocean. Nature Reviews: Microbiology. 8:593-599.

Jones, A. C., T. S. V. Liao, F. Z. Najar, B. A. Roe, K. D. Hambright, and D. A. Caron.
(2013). Seasonality and disturbance: annual pattern and response of the bacterial
and microbial eukaryotic assemblages in a freshwater ecosystem. Environmental
Microbiology. 15(9):2557-2572.

Khatiwala, S., T. Tanhua, S. Mikaloff Fletcher, M. Gerber, S.C. Doney, H.D. Graven, N.
Gruber, G.A. McKinley, A. Murata, A.F. Rios, and C.L. Sabine. Global ocean storage of
anthropogenic carbon. Biogeosciences. 10:2169-2191.

Kirchman, D.L., E K’nees, and R. Hodson. (1985). Leucine incorporation and its
potential as a measure of protein synthesis by bacteria in natural aquatic systems.
Appl. Environ. Microbiol. 49(3):599-607.

Kirchman, D.L., Y. Suzuki, C. Garside, and H.W. Ducklow. (1991). High turnover rates
of dissolved organic carbon during a spring phytoplankton bloom. Nature. 352:612-
614.

Kirchman, D. L., J. H. Rich, and R. T. Barber. (1995). Biomass and biomass production
of heterotrophic bacteria along 140˚W in the equatorial Pacific: effect of
temperature on the microbial loop. Deep-Sea Research II. 42(2-3):603-619.

Teel
29
Kolber, Z. and P.G. Falkowski. (1993). Use of Active Fluorescence to Estimate
Phytoplankton Photosynthesis in Situ. Limnol. Oceanogr.. 38: 1646–1665.

Krause, G. H. and E. Weis. (1991). Chlorophyll fluorescence and photosynthesis - the
basics. Annu. Rev. Plant Phys.. 42: 313–349.

Krempin, D.W. and C.W. Sullivan.(1981). The seasonal abundance, vertical
distribution, and relative microbial biomass of chroococcoid cyanobacteria at a
statin in southern California coastal waters. Canadian Journal of Microbiology.
27(12):1341-1344.

Lai, C.C., Y.W. Fu , H.B. Liu

, H.Y. Kuo

, K.W. Wang , C.H. Lin , J.H. Tai , G. T. F. Wong,
K.Y. Lee , T.Y. Chen , Y. Yamamoto , M.F. Chow , Y. Kobayashi , C.Y. Ko , and F.K. Shiah.
(2014). Distinct bacterial-production–DOC–primary-production relationships and
implications for biogenic C cycling in the South China Sea shelf. Biogeosciences.
11:147-156.

Larsen, P.E., S.M. Gibbons, and J.A. Gilbert. (2012). Modeling microbial community
structure and functional diversity across time and space. FEMS Microbiol. Lett.
332:91-98.

Lawrenz, E. and T.L. Richardson. (2010). How does the species used for calibration
affect chlorophyll a measurements by in situ fluorometry? Estuaries and Coasts.
doi:10.1007/s12237-010-9346-6.

Lavigne H., F. D’Ortenzio, H. Claustre, and A. Poteau. (2012). Towards a merged
satellite and in situ fluorescence ocean chlorophyll product. Biogeosciences. 9:2111-
2125.

Lennert-Cody C.E. and P.J.S. Franks. (2002). Fluorescence patches in high-frequency
internal waves. Mar. Ecol. Prog. Ser. 235:29-42.

Lignell, R. , H. Haario, M. Laine, and T.F. Thingstad. (2013). Getting the “right” values
for models of the pelagic microbial food web. Limnol. Oceanogr. 58(1):301-313.

Lindsay, K. (2012). The CESM ocean ecosystem scientific description.
http://www.cesm.ucar.edu/models/cesm1.1/pop2/doc/sci/ecosys_placeholder.ht
ml , National Center for Atmospheric Research.

López-Urrutia, A. and X. A. G. Morán. (2007). Resource limitation of bacterial
production distorts the temperature dependence of oceanic carbon cycling. Ecology.
88(4):817-822.

Lucas, A.J., C. L. Dupont, V. Tai, J.L. Largier, B. Palenik, and P.J.S. Franks. (2011). The
green ribbon: multiscale physical control of phytoplankton productivity and
Teel
30
community structure over a narrow continental shelf. Limnol. Oceanogr. 56(2):611-
626.

Marra, J. (2003). 14C-Uptake by phytoplankton, now and in the future. Limnol.
Oceanogr. Bull. 12(1):1-3.

Marra, J. (2009). Net and gross productivity: weighing in with 14C. Aquat. Microb.
Ecol. doi:10.3354/ame01306.

McCarren, J., J. W. Becker, D. J. Pepeta, Y. Shi, C. R. Young, R. R. Malmstrom, S. W.
Shisholm, and E. F. Delong. (2010). Microbial community transcriptomes reveal
microbes and metabolic pathways associated with dissolved organic matter
turnover in the sea. Proceedings of the National Academy of Sciences.
107(38):16420-16427.

Moore, J.K. , S.C. Doney, J.A. Kleypas, D.M. Glover, I.Y. Fung (2002). An Intermediate
Complexity Marine Ecosystem Model for the Global Domain. Deep-See Research II
49: 403 – 406.

Moore, J. K., Lindsay, K., Doney, S. C., Long, M. C., & Misumi, K. (2013). Marine
ecosystem dynamics and biogeochemical cycling in the Community Earth System
Model [CESM1 (BGC)]: Comparison of the 1990s with the 2090s under the RCP4. 5
and RCP8. 5 scenarios. Journal of Climate, 26(23), 9291-9312.

Moorthi, S.D., P.D. Countway, B.A. Stauffer, and D.A. Caron. (2009). Use of
quantitative real-time PCR to investigate the dynamics of the red tide dinoflagellate
Lingulodinium polyedrum. Microbial Ecology. 52(1):136-150.

Morán, X.A.G., M. Estrada, J.M. Gasol, and C. Pedrós-Alió. (2002). Dissolved primary
production and the strength of phytoplankton-bacterioplankton coupling in
contrasting regions. Microbial Ecology. 44(3):217-223.

Morris, R. M., B. L. Nunn, C. Fazar, D. R. Goodlett, Y. S. Ting, and G. Rocap. (2010).
Comparative metaproteomics reveals ocean-scale shifts in microbial nutrient
utilization and energy transduction. The ISME Journal. 4:673:685.

Mou, X., S. Sun, R. A. Edwards, R. E. Hudson, and M. A. Moran. (2008). Bacterial
carbon processing by generalist species in the coastal ocean. Letters to Nature.
451:708-711.

Müller, P., X.-P. Li, and K.K. Niyogi. (2001). Non-Photochemical Quenching. A
Response to Excess Light Energy. Plant Physiol.. 125: 1558–1566.
doi:10.1104/pp.125.4.1558.

Teel
31
Nagata, T. (2000) Production mechanisms of dissolved organic matter. In: DL
Kirchman(ed) Microbial Ecology of the Ocean. John Wiley and Sons. Chichester, UK.
121-152

Needham D.M., C.-E.T. Chow, J.A. Cram, R. Sachdeva, A. Parada, J.A. Fuhrman. (2013).
Short-term observations of marine bacterial and viral communities: patterns,
connections and resilience. ISME J 7: 1274–1285.

Needham, D.M. and J. A. Fuhrman. 2016. Pronounced Daily Succession of
Phytoplankton, Archaea, and Bacterial Following a Spring Bloom. Nature
Microbiology. 1:1605. doi:10.1038/nmicrobiol.2016.5.

Noble M.A., K.J. Rosenberger, P. Hamilton, and J.P. Xu. (2009). Coastal ocean
transport patterns in the central Southern California Bight, in Lee, H.J., and Normark,
W.R., eds., Earth Science in the Urban Ocean: The Southern California Continental
Borderland. Geological Society of America Special Paper 454: 193–226, doi:
10.1130/2009.2454(3.3)

Noble, M., B. Jones, P. Hamilton, J. Xu, G. Robertson, L. Rosenfeld, and J. Largier.
(2009). Cross-Shelf Transport Into Nearshore Waters Due to Shoaling Internal Tides
in San Pedro Bay, CA. Continental Shelf Research 29(15): 1768–1785.

Oey, L.Y. (1999). A forcing mechanism for the poleward flow off the Southern
California coast. J. Geophys. Res. Oceans. 104:13529-13539.

Ouverney, C.C. and J.A. Fuhrman. (1999). Combined microautoradiography-16S
rRNA probe technique for determination of radioisotope uptake by specific
microbial cell types in situ. Appl. Environ. Microbiol. 65(4):1746-1752.

Pomeroy, L.R. and D. Deibel. (1986). Temperature regulation of bacterial activity
during the spring bloom in Newfoundland coastal waters. Science. 233:359-361.

Riemann, L., G.F. Steward, and F. Azam. (2000). Dynamics of bacterial community
composition and activity during a mesocosm diatom bloom. Appl. Environ. Microbiol.
66(2):578-587.

Rivkin, R.B. and L. Legendre. (2001). Biogenic Carbon Cycling in the Upper Ocean:
Effects of Microbial Respiration. Science. 291:2398-2400.

Ryan J., D. Greenfield, R. Marin III, C. Preston, B. Roman, S. Jensen, D. Pargett, J. Birch,
C. Mikulski, G. Doucette, and C. Scholin. (2011). Harmful phytoplankton ecology
studies using an autonomous molecular analytical and ocean observing network.
Limnol. Oceanogr. 56(4):1255-1272.

Teel
32
Sarmiento, H. and J.M. Gasol. (2012). Use of phytoplankton-derived dissolved
organic carbon by different types of bacterioplankton. Environmental Microbiology.
14(9):2348-2360.

Schnetzer, A., P.E. Miller, R.A. Schaffner, B.A. Stauffer, B.H. Jones, S.B. Weisberg, P.M.
DiGiacomo, W.M. Berelson, and D.A. Caron. (2007). Blooms of Pseudo-nitzschia and
domoic acid in the San Pedro Channel and Los Angeles harbor areas of the Southern
California Bight, 2003-2004. Harmful Algae. 6:372-387.

Seegers, B.N., C. Scholin, E. Seubert, M. Howard, G. Robertson, D. A. Caron, and B. H.
Jones. The Role of Coastal Ocean Dynamics in the Initiation and Transport of
Harmful Algal Blooms in Southern California. (In Preparation).

Simon M. and F. Azam (1989). Protein content and protein synthesis rates of
planktonic marine bacteria. Mar. Ecol. Prog. Ser. 51:201-213.

Sinsabaugh, R.L., S. Manzoni, D.L. Moorhead, and A. Richter. (2013). Carbon use
efficiency of microbial communities: stoichiometry, methodology, and modeling.
Ecology Letters. 16:930-939.

Sintes, E., K. Stoderegger, V. Parada, and G.J. Herndl. (2010). Seasonal dynamics of
dissolved organic matter and microbial activity in the coastal North Sea. Aquat.
Microb. Ecol. 60:85-95.

Smith D.C. and F. Azam. (1992). A simple, economical method for measuring
bacterial protein synthesis rates in seawater using
3
H-leucine. Marine Microb. Food
Webs. 6(2):107-114.

Smith W.O., R.T. Barber, and S.A. Huntsman. (1977). Primary production off the
coast of northwest Africa: excretion of dissolved organic matter and its
heterotrophic uptake. Deep-Sea Res. 24:35-47.

Steele JA, PD Countway, L Xia, PD Vigil, JM Beman, DY Kim, C.-E.T. Chow, R.
Sachdeva, A.C. Jones, M.S. Schwalbach, J.M. Rose, I. Hewson, A. Patel, F. Sun, D.A.
Caron, and J.A. Fuhrman. (2011). Marine bacterial, archaeal and protistan
association networks reveal ecological linkages. ISME J 5: 1414–1425.

Stoderegger, K. and G.J. Herndl. (1998). Production and release of bacterial capsular
material and its subsequent utilization by marine bacterioplankton. Limnol.
Oceanogr. 43(5):877-884.

Suttle, C. A. (2007). Marine viruses—major players in the global ecosystem. Nature
Reviews Microbiology, 5(10), 801-812.

Teeling H., B.M. Fuchs, D. Becher, C. Klockow, A. Gardebrecht, C.M. Bennke, M.
Kassabgy, S. Huang, A.J. Mann, J. Waldmann, M. Weber, A. Klindworth, A. Otto, J.
Teel
33
Lange, J. Bernhardt, C. Reinsch, M. Hecker, J. Peplies, F.D. Bockelmann, U. Callies, G.
Gerdts, A. Wichels, K.H. Wiltshire, F.O. Glöckner, T. Schweder, and R. Amann. (2012).
Substrate-controlled succession of marine bacterioplankton populations induced by
a phytoplankton bloom. Science. 336: 608-611.

Teira, E., M.J. Pazó, P. Serret, and E. Fernández. (2001). Dissolved organic carbon
production by microbial populations in the Atlantic Ocean. Limnol. Oceanogr.
46(6):1370-1377.

Thingstad, T.F. and R. Lignell. (1997). Theoretical models for the control of bacterial
growth rate, abundance, diversity, and carbon demand. Aquat. Microb. Ecol. 13:19-
27.

Todd, R.E., D.L. Rudnick, and R.E. Davis. (2009). Monitoring the greater San Pedro
Bay region using autonomous underwater gliders during fall of 2006. J. Geophys. Res.
114. C06001. doi:10.1029/2008JC005086.

Tortell P.D., M.T. Maldonado, J. Granger, and N.M. Price. (1999). Marine bacteria and
biogeochemical cycling of iron in the oceans. FEMS Microbiol. Ecol. 29(1):1-11.

Van Mooy, B. A. S., A. H. Devol, and R. G. Keil. (2004). Relationship between bacterial
community structure, light, and carbon cycling in the eastern subarctic North
Pacific. Limnology and Oceanography. 49(4):1056-1062.

Van Wambeke, F., S. Bonnet, T. Moutin, P. Raimbault, G. Alarcón, and C. Guieu.
(2008). Factors limiting heterotrophic bacterial production in the southern Pacific
Ocean. Biogeosciences. 5:833-845.

Venrick, E.L. (2002). Floral patterns in the California Current System off southern
California: 1990-1996. J. Marine Research. 60:171-189.

Vichi, M., Pinardi, N., & Masina, S. (2007). A generalized model of pelagic
biogeochemistry for the global ocean ecosystem. Part I: Theory. Journal of Marine
Systems, 64(1), 89-109.

Vichi, M., Masina, S., & Navarra, A. (2007). A generalized model of pelagic
biogeochemistry for the global ocean ecosystem. Part II: Numerical
simulations. Journal of Marine Systems, 64(1), 110-134.

Vichi, M., & Masina, S. (2009). Skill assessment of the PELAGOS global ocean
biogeochemistry model over the period 1980–2000. Biogeosciences, 6(11), 2333-
2353.

Walsh, J.J., D.A. Dieterle, F.E. Muller-Karger, R. Bohrer, and W.P. Bissett. (1999).
Simulation of carbon-nitrogen cycling during spring upwelling in the Cariaco Basin.
Marine Science Faculty Publications. Paper 17.
Teel
34

Ward, C. S., C-M. Yung, K. M. Davis, S. K. Blinebry, T. C. Williams, Z. I. Johnson, and D.
E. Hunt. (2017). Annual community patterns are driven by seasonal switching
between closely related marine bacteria. The ISME Journal. 1-11.

Weitz, J.S., C.A. Stock, S.W. Wilhelm, L. Bourouiba, A. Buchan, M.L. Coleman, M.J.
Follows, J.A. Fuhrman, L.F. Jover, J.T. Lennon, M. Middelboe, D.L. Sonderegger, C.A.
Suttle, B.P. Taylor, T.F. Thingstad, W.H. Wilson, and K.E. Wommack. (2014). A
multitrophic model to quantify the effects of marine viruses on microbial food webs
and ecosystem processes. The ISME Journal . 1-13.

White, P. A., J. Kalff, J. B. Rasmussen, and J. M. Gasol. (1990). The Effect of
Temperature and Algal Biomass on Bacterial Production and Specific Growth Rate in
Freshwater and Marine Habitats. Microbial Ecology. 21:99-118.

Williams T.J., D. Wilkins, E. Long, F. Evans, M.Z. DeMeere, M.J. Raftery, and R.
Cavicchioli. (2013). The role of Flavobacteria in processing algal organic matter in
coastal East Antarctica revealed using metagenomics and metaproteomics. Environ.
Microbiol. 15(5):1302-1317.

Williams, P.J. leB. (1981). Microbial contribution to overall marine plankton
metabolism: direct measurements of respiration. Oceanol. Acta. 4(3):359-364.

Xiu P and F. Chai. (2014). Connections between physical, optical, and
biogeochemical processes in the Pacific Ocean. Progress in Oceanography. 122:30-
53.

Zehr J.P. and R.M. Kudela. (2011). Nitrogen cycle of the open ocean: from genes to
ecosystems. Annu. Rev. Mar. Sci. 3:197-225.

Ziveri, P., R.C. Thunell, and D. Rio. (1995). Export production of coccolithophores in
an upwelling region: results from San Pedro Basin, Southern California Borderlands.
Marine Micropaleontology. 24:335-358.

Teel
35
Chapter 2. Contextualizing time-series data: Using high-
resolution in-situ gliders to quantify short-term regional
oceanographic variability in the San Pedro Channel
Abstract
Oceanic time-series have provided insight into biological, physical, and chemical
processes and how these processes change over time. Though time-series sites are
increasing in number, the ability to incorporate these sites into larger scale analyses
is restricted by the fact that regional oceanographic features can often prevent the
extrapolation of local results. To increase their usability, time-series data must be
contextualized with high-frequency spatial and temporal studies of regional
oceanographic variability. To study the oceanographic context of the San Pedro
Ocean Time-series (SPOT) station in the San Pedro Channel (SPC), Slocum gliders
were deployed off the coast of Los Angeles, California from February to July of 2013
and 2014. The data were collapsed onto a standardized grid and primary and
secondary characteristics of glider profiles were analyzed by principal component
analysis to determine the processes impacting the SPC and the SPOT station. The
dataset was dominated by four water profile types: active upwelling, offshore
intrusion, subsurface bloom, and surface bloom. On average, waters across the SPC
were most similar to offshore water profiles, though this offshore influence
increased dramatically across the channel midway through the deployments. The
SPOT station was representative of the SPC 64% of the time, and SPOT was least
similar to SPC locations that were closest to the Palos Verdes Peninsula. Subsurface
blooms were common in both years with co-located chlorophyll and particle
Teel
36
maxima, and results suggested that these subsurface blooms contribute significantly
to the local primary production in the SPC. Satellite estimation of integrated
chlorophyll was poor, possibly due to the prevalence of subsurface blooms and
shallow optical depths during surface blooms. These results indicate that high-
resolution in-situ glider deployments can be used to determine the spatial domain
of time-series data, allowing for broader application of these datasets and greater
integration into modeling efforts.

2.1 Introduction
Time-series sites have been invaluable for providing insights into the
biological, chemical, physical, and ecological processes that occur in the world’s
oceans. These sites have also provided in-situ measurements that climate models
can use for development, validation, and ongoing improvement through hindcast
comparisons (Fasham et al., 1990; Doney et al., 1996; Spitz et al., 2001; Boyd and
Doney, 2002; Moore et al., 2002; Dugdale et al., 2002; Doney et al., 2009). The
predictive capabilities of these climate models are hindered, however, by a
mismatch in the spatial and temporal resolution of in situ data relative to the
modeled processes (Baklouti, et al., 2006; Alkire et al., 2012; Behrenfeld and Boss,
2013; Needham and Fuhrman, 2016). By characterizing the regional, seasonal, and
decadal variability of an individual site relative to the larger region, time-series data
can be extrapolated with greater confidence.
Teel
37
Incorporating data collected near coastal zones and continental margins into
ocean models is even further complicated by the highly dynamic and often unique
nature of these locations (Hofmann et al., 2011 and references therein). Coastal
variables such as freshwater discharge, sediment input, wind fields, upwelling flux,
and coastal hypoxia complicate both data-assimilative and predictive coastal
modeling. However, coastal ecosystems are likely to be some of the most at risk
from the effects of climate change, and, furthermore, it is now well recognized that
understanding coastal zone ecosystems is imperative for fully describing global
carbon budgets (Feely et al., 2008; Bauer et al., 2013 and references therein). For
these reasons, determining the best way to include data from these coastal sites has
become a priority for the time-series and modeling communities (Benway et al.,
2016).
Since most time-series are sampled at a single fixed location approximately
once per month, the overall dataset is assumed to represent the mean state of a
given geographical region as it varies with seasonal and annual cycles. To determine
if this assumption is correct and to allow for the extrapolation of coastal time-series
data to a larger region, high-resolution spatial and temporal monitoring of the
physical and biological variability around these time-series sites is required..
Because of the limitations with traditional in situ approaches, satellite imagery is
frequently used to characterize spatial and temporal variability, assuming a tight
coupling between surface and sub-surface variability (eg., DiGiacomo and Holt,
2001; Kahru et al., 2009; Nezlin et al., 2012). However, for many coastal regions
satellite observations are insufficient for assessing the biological and environmental
Teel
38
variability due to decoupling between surface and sub-surface dynamics, the
importance of fine spatial scale (<1 km) variability, and the presence of terrestrially
derived CDOM .
High-frequency in-situ sampling with gliders can provide uninterrupted
monitoring of the surface 100 to 1000 meters over tens of kilometers. These
datasets provide both an understanding of fine-scale spatial and temporal dynamics
as well as insight into the connectivity between surface and subsurface dynamics,
thereby aiding in the interpretation of satellite data. Here, we apply an eight month
Slocum electric glider dataset from the San Pedro Channel (SPC) to investigate the
representativeness of the coastal San Pedro Ocean Time-series (SPOT). This test-
case provides a framework for investigating how data from a single location relates
to nearby regional variation. We aim to show that high frequency sampling allows
for the intra-monthly dynamics to be resolved just as fine-scale spatial sampling can
provide a better understanding of the representative nature of a given time series
location
The SPOT station, which is located at 33˚ 33’ 00” N and 118˚ 24’ 00” W, sits in
the SPC between Catalina Island and the Palos Verdes Peninsula where the water
depth is approximately 900 meters. The SPC lies within the larger Southern
California Bight (SCB), which extends from Point Conception to Mexico (Noble et al.,
2009). Channel Islands and submarine canyons oceanographically separate the
eastern and western sides of the SCB. Within the SCB, the Southern California Eddy
is made of poleward-flowing surface currents that break off from the California
Current (Oey, 1999; Noble et al., 2009; Dong et al., 2009). The SCB is characterized
Teel
39
by strong seasonal variation, including a spring upwelling season and subsequent
phytoplankton blooms. Within the SPC, however, local upwelling and post-
upwelling bloom formation are less persistent or predictable than these features are
farther north. Post-upwelling blooms occur on the timescales of days to a couple of
weeks, and can be quickly followed by periods of very low surface chlorophyll. The
SPOT station has been sampled monthly for environmental and biological
parameters since 1998. Microbial communities at SPOT have been found to have
annually and seasonally recurring membership (Fuhrman et al., 2006; Steele et al.,
2011; Chow et al., 2013; Chow et al., 2014). Additional daily sampling has shown
that dominant microbial taxa vary at much shorter time-scales, indicating that
monthly sampling may only represent a persistent background community
(Needham et al., 2013; Needham et al., 2016).
2.2 Methods
2.2.1 Glider Deployments
To characterize the vertical physical and biological characteristics of the SPC
at high spatial and temporal resolution, a Teledyne-Webb G1 Slocum electric glider
was deployed from March through July of 2013 and 2014. This period of time, early
spring to early summer, is known to be highly dynamic for the SPC and the Southern
California Bight as a whole (Hayward and Venrick, 1998; DiLorenzo et al., 2003;
Mantyla et al., 2008; Schnetzer et al., 2013). Sampling during a period of high local
variability maximized the likelihood that local and offshore processes would be
captured in the dataset. In particular, local upwelling, post-upwelling surface
Teel
40
blooms, and the intrusion of offshore warm fronts were expected to occur within
this sampling period.
The glider was deployed on a 28 km path between Catalina Island and the
Palos Verdes Peninsula (Figure 2.1). This transect approximated a cross-channel
section of the SPC. The glider, with an average horizontal speed of 1 kilometer per
hour, completed a single cross-channel pass every 1.5-2 days and sampled near the
SPOT station on each of these passes. The glider profiles reached maximum depths
of 70 to 90 meters and surfaced to approximately 3 meters with the exception of
when the glider was crossing the major shipping lanes that are associated with this
region. When crossing the shipping lanes, the glider were prevented from rising
above 20 meters to avoid glider damage or loss. During these glider deployments,
chlorophyll a fluorescence was measured by a WetLabs EcoPuck FL3 fluorometer,
backscatter at wavelengths of 532, 660, and 880 nanometers was measured by a
WetLabs EcoPuck BB3 sensor, and a SeaBird flow-through CTD was used to
measure temperature, salinity, and pressure. Vertical resolutions for the WetLabs
pucks were approximately 0.3 m, while the vertical resolution for the SeaBird CTD
was approximately 0.6 m. The glider was recovered every 3-4 weeks for cleaning,
battery replacement, and recalibration using standard methods published in Cetinić
et al (2009).
2.2.2 Glider Analyses
Glider data was processed and analyzed using Mathworks Matlab release 2013a. All
chlorophyll a fluorescence data were calibrated as described in Cetinic et al (2009).
Teel
41
The glider data was projected onto a standardized grid along an idealized glider
transect to correct for current induced drift (Figure 2.1). Since the dominant
currents in the SPC generally flow northwest or southeast parallel to the shoreline
(eg., Huyer, 1983; Oey, 1999; Noble, 2009), the idealized transect was created
perpendicular to mean flow and to the coastline (Figure 2.1). Only glider data within
5 km of the idealized transect were used in this analysis (Figure 2.1). The transect
was divided into 62 500m horizontal bins, which approximated a single downcast
and upcast. The data was additionally binned vertically onto a 1 m grid (80 depths).
Temporal averaging was minimized by separating the overall glider dataset into 2-
day transect passes before completing vertical and horizontal binning. After binning,
each idealized glider transect contained 62 independent glider profiles with a
vertical resolution of 1 m, a horizontal resolution of 500 m, and a temporal
resolution of 48 hours. From these binned profiles, only profiles with data for >85%
of the vertical bins were used for further analyses. This criterion effectively
eliminated all profiles from under the shipping lanes, where the glider did not
sample in the top 20 meters. For included profiles, missing data were filled using 2D
interpolation from all neighboring bins. A total of 557 profiles from 2013 and 1049
profiles from 2014 were accepted for further analyses.
The temperature, salinity, and chlorophyll a data from each of the 1606
glider profiles were used to calculate secondary metrics that were used for
statistical analyses. Specifically, maximum chlorophyll fluorescence (MaxCHL),
depth of maximum chlorophyll fluorescence (zMaxChl), 70 meter integrated
chlorophyll (ChlInt70), depth of maximum backscatter (zMaxBB), maximum
Teel
42
backscatter (MaxBB), and ratio of integrated chlorophyll in the top 70 meters
relative to the top 20 meters (ChlInt70Per20) were calculated. In addition, the
maximum stratification index (MaxBVF) and depth of maximum stratification
(zMaxBVF) as measured by the Brunt Vaisala Frequency (BVF) were calculated as:
(eq 2.1) 𝐵𝑉𝐹=
!∗!"
!∗!"

where g is gravity in meters per squared second, ρ is the ambient density of in-situ
water in kilograms per cubic meter, dz is the change in depth in meters, and dρ is
the change in density in kilograms per cubic meter (Mann and Lazier, 2006). The
mixed layer depth (MLD) was defined as the depth where density was greater than
potential density calculated using 5 m salinity and 5 m temperature minus 0.4˚C
(modified from Sprintall and Tomczak, 1992). The mixed layer temperature
(MLTemp) was calculated as the mean temperature within the mixed layer. Finally,
the depth of the 12.5˚C isotherm (z12p5), which indicates nutrient-rich sub-
thermocline waters in the SPC (Hayward and Venrick, 1998), was determined.
Principal component analysis (PCA) was used to differentiate between the
major water profile types observed within the glider dataset. Specifically, 54 ideal
profiles were selected to define each of four dominant water profile types observed
in the SPC: early upwelling, coastal phytoplankton bloom, subsurface phytoplankton
bloom, and offshore intrusion (Table 2.1, Figure 22.). These profile types had been
observed qualitatively in glider curtain plots, and were selected from the data set
using MLTemp, ChlInt70Per20, ChlInt70, z12p5, maxCHL, zMaxChl, and MLD criteria
(Supplemental Table 2.1). A step-wise PCA was conducted to determine the relative
Teel
43
influence of each of the secondary characteristics on total observed variance within
the ideal profile dataset. Based on this analysis, four characteristics (zMaxBB,
MaxBB, zMaxBVF, and MaxBVF) were omitted from further PCA analyses as they did
not strongly affect overall dataset variance or the resulting PCA distribution. The
final PCA axes provided clear PC1 and PC2 components that were correlated with
the physical and biological characteristics of the water profiles, respectively
(Supplemental Figure 2.1). The remaining 1552 glider profiles were then projected
onto the PC1 and PC2 coordinates based on their secondary characteristics (Figure
2.4). Due to the comparable weights of PC1 and PC2 (49.8% and 32.7% respectively)
in this study, distance between points within the ideal coordinate system could be
used as an approximation of dissimilarity between profiles.
2.2.3 Ancillary Satellite Data
Level 3 mapped MODIS Aqua daily 9km photosynthetically active radiation
(PAR) measurements were acquired from the NASA Ocean Biology
(http://oceandata.sci.gsfc.nasa.gov/MODISA/L3BIN/). MODIS Aqua daily 1 km
chlorophyll (ChlSat) data and were acquired from NOAA CoastWatch West Coast
Regional Node (http://coastwatch.pfeg.noaa.gov/coastwatch/CWBrowser.jsp).
These data were then matched geographically and temporally with the in-situ glider
data. Of the 1606 final glider profiles from 2013 and 2014, 1170 matching PAR
measurements and 571 matching ChlSat measurements were available. PAR
measurements were then used in combination with the glider chlorophyll profiles to
estimate the light field at depths between 1 and 80 m for each glider profile. The
Teel
44
diffuse attenuation coefficient was calculated as a function of chlorophyll a
concentration at each depth (Jacox et al., 2015). The euphotic depth and the first
optical depth were then calculated for each glider profile from these light profiles
(Kirk, 1994).
2.3 Results
2.3.1 Cross-Channel Oceanographic Trends
Cross-channel comparisons of MLTemp, MLD, z12p5, and ChlInt70 were
used to identify persistent oceanographic changes along the length of the idealized
transect (Figure 2.3). Observed physical properties in both spring 2013 and spring
2014 displayed an onshore-offshore gradient (Figure 2.3 a., b., and c.). This gradient
could be seen most clearly in the z12p5 data, where there was a strong offshore tilt
in the mean depth of this isotherm. This tilt was consistent with equatorward flow
through the channel, but could also have been amplified closest to shore with
periods of active upwelling. Though the average depth of the cold, high-nutrient
waters was shallowest close to shore, the cross-channel data for ChlInt70 did not
display a strong cross-channel gradient (Figure 2.3d). Rather, ChlInt70 had a fairly
constant cross-channel concentration of about 100 mg/m
2
. It is important to note
that integrated chlorophyll alone cannot be used to assess the productivity of a
location, because it does not account for the vertical chlorophyll distribution or the
vertical light field. These cross-channel analyses also highlight high intra-bin
temporal variability over the course of the deployments, especially for ChlInt70 and
MLD (Figure 2.3b and 2.3d). The importance of this temporal and spatial pattern of
Teel
45
variability for SPOT is two fold: 1) a monthly time-series sampling scheme may
under-sample both biological and physical variability at SPOT, but 2.2) the similarity
in variance across the SPC suggests that, given sufficient sampling, SPOT data could
be representative of the average state of the SPC.
To look specifically at variability between glider profiles during these
deployments, it was necessary to define what types of profiles were common in the
SPC. Individual binned glider profiles from each 2-day transect were analysed to
identify these common profile types within the SPC. When these glider profiles
were analysed independent of location within the channel, four distinct water
profile types were apparent (Table 2.1, Figure 2.2). These profile types were
hallmarks of specific oceanographic events within the channel: early upwelling prior
to a significant biological response, surface phytoplankton blooms, subsurface
phytoplankton blooms, and offshore intrusion of relatively oligotrophic water.
These four profile types represented the extremes that occurred within the channel
during these glider deployments and each was uniquely defined using physical and
biological criteria (Supplemental Table 1).
Early upwelling profiles were characterized by MLTemps that were cooler
than 12.5˚C, a shallow chlorophyll maxima, and integrated chlorophyll that did not
exceed 85 mg chl m
-2
. The combination of these characteristics indicated that the
deep, cold, nutrient rich water had been recently mixed into the surface mixed layer.
During this study, the strongest surface bloom signals occurred after upwelling had
begun to relax and the 12.5 ˚C isotherm had returned to a depth of greater than 20
m. These surface blooms were also characterized by MLTemps between 13 and
Teel
46
17˚C, MLDs deeper than 10 m, and ChlInt70 values above 150 mg chl m
-2
. To further
differentiate these surface blooms from subsurface blooms within the dataset, the
integrated chlorophyll within the surface 20 m was directly compared to that within
the top 70 m. Here we define surface blooms as those with at least 50% of the total
integrated chlorophyll within the top 20 m and subsurface blooms as having less
than 5% of integrated chlorophyll in the top 20 m of the profile. Both subsurface
bloom profiles and offshore intrusion profiles shared high MLTemp, shallow MLD,
and deep z12p5, but the offshore intrusion profiles were relatively oligotrophic and
had values of ChlInt70 less than 80 mg chl m
-2
. On average, the MLD and zMaxChl
were deeper during subsurface blooms than during offshore intrusion.
The four water profile types described above capture the typical progression
of seasonal traits for the SPC (Figure 2.2). In this region, local upwelling in the
spring is followed by the appearance of surface phytoplankton blooms, which in
turn is followed by an offshore intrusion of relatively oligotrophic waters due to the
spin-up of the Southern California Eddy (Hickey 1979; Kim et al., 2009; Noble et al.,
2009). These four water profile types represent the extremes such that one would
anticipate that a monthly time-series would, for the most part, capture intermediate
states rather than these end members. Identifying these end-members, hereafter
referred to as ideal profiles, both helps characterize the overall biological and
physical variation seen in water column profiles within the SPC and provides a
means for quantifying the influence of coastal versus offshore waters in the channel.
PCA axes based on the 54 ideal profiles were used to investigate the
representativeness of SPOT to gauge the similarity between profiles collected at
Teel
47
different locations or times. The first principal component (PC1) and second
principal component (PC2) accounted for 49.8% and 32.7% of total variance,
respectively (Figure 2.4; Supplemental Figure 2.1). MLTemp and the zMaxChl
projected primarily onto PC1 while ChlInt70 and maxCHL projected onto PC2
suggesting that PC1 corresponded with an increased offshore signature while PC2
was inversely correlated with profile biomass (Figure 2.2 & 2.4; Supplemental
Figure 2.1). Specifically, PC1 was most positive when MLTemp was greater than 19
˚C and there was a strongly defined thermocline (Figure 2.2). PC2 was most negative
when the chlorophyll maximum was most clearly defined and ChlInt70 was greater
than 120 mg chl m
-3
(Figure 2.2). To determine the relative contribution of the 4
water profile types to the observed glider profiles, the 1606 final glider profiles
were projected onto these ideal axes (Figure 2.4a). As PC1 and PC2 explained
similar amounts of variance within this dataset, ‘distance’ between points in PC
coordinate space approximated dissimilarity between individual profiles.
Conversely, decreasing distance between profiles in PC coordinate space could be
attributed to increasing similarity in profile features.
2.3.2 Regional Similarity to SPOT Station
When all glider profiles were projected onto the PCA axes, 39.6% fell within
one of the ideal profile types, as determined by 95% confidence intervals (Table 1).
Similarly, 37% of the 54 SPOT samples collected associated with an ideal subgroup.
Of the four ideal profiles, the offshore-types dominated, with 24.8% of all profiles
falling within the subsurface bloom group and 12.1% aligning with offshore
Teel
48
intrusion profiles. SPOT samples followed a very similar trend to the SPC, being
characterized by subsurface bloom profiles in 25.9% of samples and offshore
intrusion profiles in 9.3% of the glider profiles. Coastal end members, early
upwelling and coastal blooms, were rare in the overall dataset as well as in the SPOT
samples themselves. Less than 4% of all samples associated these coastal end-
members. No profiles similar to early upwelling were detected at the SPOT station
itself, and only 1.8% of the SPOT station profiles could be characterized as
exhibiting surface bloom characteristics. These results indicate that for the majority
of locations and times sampled during the spring and early summer of 2013 and
2014, water profiles within the SPC most closely resembled offshore water profiles,
characterized by higher values of PC1; however, these results mostly likely
underrepresent the overall coastal influence within the SPC due to limited data
collection within the shipping lanes (Figure 2.1). These findings are consistent with
previous research conducted at SPOT, which indicated a general prevalence of open
ocean bacterial groups (Chow et al., 2013) and a notable scarcity of land-derived
organic matter (Collins et al., 2011).
To compare cross-channel differences in water profile properties, three bins
were selected: bin 10 (closest to Catalina), 28 (SPOT), and 58 (nearshore to
mainland). Physical and biological characteristics of the profiles collected at these
three locations maintained an onshore-offshore gradient on average, however,
when the variance at each location was taken into account, bins 10 and 28 were not
significantly different from one another (p-value < 0.01). These findings suggest
that, during this study, SPOT was at least partially detached from nearshore coastal
Teel
49
dynamics, such as early upwelling or upwelling driven surface blooms. These results
also indicate that samples taken at SPOT are in general a poor representation of
nearshore dynamics but a relatively good representation of the channel on the
whole.
Due to the seasonal variation in both local upwelling and the Southern
California Eddy in the SPC, temporal dynamics and succession play an important
role in driving regional productivity. To explore temporal connectivity between
SPOT and nearshore coastal bins, the evolution of profile characteristics over time
were analysed in PCA space (Figure 2.5). Early in each deployment, there was a
strong coastal influence, which coincided with early upwelling events and coastal
blooms. This influence was more apparent in the nearshore bins, as expected, but
also clearly affected the overall characteristics of vertical profiles taken at SPOT as
well. Cross-channel connectivity was more apparent when, during late April to early
May of each deployment, the nearshore and SPOT bins collectively experienced a
relative ‘cross-over’, when the water profiles shifted from coastally dominated
characteristics to offshore dominated. These results imply that while the profiles at
SPOT did not often resemble nearshore profiles, profile variability in both mid-
channel and nearshore locations was likely forced by the same coastal and offshore
oceanographic events .
To further investigate how changes in the SPOT profile characteristics are
related to changes in cross-channel profile types, we looked at the relationship
between all non-SPOT profiles and the corresponding SPOT profile. For this
analysis, the PCA coordinate plane was divided into four quadrants, which
Teel
50
represented low biomass and coastal dynamics (positive PC2, negative PC1), low
biomass and offshore dynamics (positive PC2, positive PC1), high biomass and
offshore dynamics (negative PC2, positive PC1), and high biomass with coastal
dynamics (negative PC2, negative PC1) (Figure 2.6). These four quadrants were then
used to investigate the cross channel distribution relative to the most recent SPOT
sample. The majority of samples, 64%, displayed similar profile characteristics to
the most-recent SPOT profile, and these profiles fell into the same PCA quadrant as
that respective SPOT-profile. Broken down by profile type, SPOT profiles were most
similar to cross-channel profiles when SPOT displayed sub-surface bloom profile
characteristics, as 78% of cross-channel samples were in the same quadrant as
these SPOT samples (Figure 2.6). SPOT profile characteristics coincided with 55% of
other profiles when exhibiting characteristics of offshore intrusion, 48% when
displaying surface bloom characteristics, and 33% of the channel in the one instance
when the SPOT profile displayed low biomass and more coastal physical properties
(Figure 2.6). The sample distributions in each of these four cases were statistically
different from one another (p<0.01), demonstrating that the profile characteristics
at the SPOT station were indicative of the state of the SPC as a whole. For example,
when SPOT profiles displayed highly offshore characteristics (Figure 2.6b and 2.6d),
no channel profiles displayed surface bloom profile characteristics. Conversely,
during periods when SPOT profiles were most similar to surface bloom profiles, no
channel samples were found to be representative of either the offshore intrusion or
subsurface bloom ideal profile types. These results suggest cross-channel
Teel
51
oceanographic connectivity and are consistent with the observed simultaneous
shifts in both SPOT and nearshore bins (Figure 2.5).
Due to strong alongshore currents within this region, it is not surprising that
the nearshore and offshore sides of the channel would differentially experience local
oceanographic changes. With these analyses, however, it was possible to verify that
the data collected at SPOT are generally representative of the SPC. The largest
variation was found when comparing the coastal bins to SPOT or other offshore
bins, with increased differences observed during periods of early upwelling in the
spring of 2013 and 2014. Without sampling the surface data within the shipping
lanes, it was impossible to determine the full profile behaviour of locations between
the coastal and SPOT bins via this method. However, subsurface glider temperature
data showed that patterns of weakened stratification during early upwelling events
can extend across the entire SPC. These observations imply that there would be
connectivity across the shipping lanes during upwelling events and are consistent
with the result that changes observed within profiles taken at SPOT can indicate
general regional oceanographic signals (Figure 2.6).
2.3.3 Surface to Sub-surface Connectivity
Our analyses indicated that subsurface bloom characteristics were a dominant
profile type within the SPC (occurring ~25% of the time), and that these subsurface
blooms may appear similar to offshore water intrusion from surface characteristics
alone. In many regions where subsurface chlorophyll maxima are not indicative of
particle maxima, phytoplankton growth at the chlorophyll maximum is thought to
Teel
52
account for minimal primary production. In this study, though, the subsurface
chlorophyll maximum was on average only 5 m deeper than the particle maximum,
as determined by the scattering maximum. The subsurface particle maxima
observed within this study are consistent with previous regional findings (Cullen
and Eppley, 1981). Close proximity between the particle and chlorophyll maxima
suggests that these subsurface phytoplankton community may contribute
significantly to local primary production (Cullen and Eppley, 1981). Since satellites
will theoretically miss all integrated chlorophyll below the first optical depth (OD1),
in-situ integrated glider chlorophyll was analysed above and below OD1 to assess
how the frequency of the subsurface bloom water profile type might impact satellite
estimates of chlorophyll and primary productivity.
The OD1 was calculated from satellite PAR and glider chlorophyll-based
diffuse attenuation coefficients for PAR. This method provides a conservative
estimate of OD1 by ignoring light attenuation caused by dissolved organic matter.
Average OD1 for these deployments was 12.3 m while the average euphotic depth,
as defined by the 1% light level, was 38.3 m. In-situ measurements of the euphotic
depth from temporally overlapping cruises showed average euphotic depth to be 40
m (Haskell et al., 2016). Our estimates for the first optical and euphotic depths are
also within the range of regional data collected during the CCE LTER process cruises
from 2006-2008 (Mitchell, 2008). Though MODIS Aqua chlorophyll a data were
acquired and compared with the in-situ glider data, only glider-to-glider chlorophyll
comparisons were used for further analyses due to low correlation between glider
and satellite integrated chlorophyll over OD1 (Supplemental Figure 2.2).
Teel
53
Overall 87% of integrated chlorophyll within the euphotic zone was located
beneath the OD1 during the 2013 and 2014 deployments. This percentage increased
to 92% for subsurface blooms and decreases to 82% for surface blooms. For these
deployments, the zChlMax was generally at or below the 10% light level, deepening
with more oligotrophic conditions. To determine which local oceanographic
scenarios were least likely to be accurately observed by satellites, total integrated
chlorophyll below OD1 for the 2013 and 2014 deployments was compared with the
profile characteristics using the ideal principlal component axes in order (Figure
2.7). By mass, the majority of the integrated chlorophyll that fell below OD1 was
associated with surface blooms (up to 130 mg m
-2
), rather than subsurface blooms.
In these cases, the first optical depth was shallower than the bloom thickness,
causing underestimation of the surface chlorophyll layer. During subsurface blooms
fewer PAR match-ups were available, hindering the comparison of above-and-below
OD1 integrated chlorophyll, but within the observed profiles, integrated chlorophyll
a values of up to 63 mg/m
2
existed between the first optical depth and the euphotic
depth (Figure 2.7).
With the observed frequency of subsurface blooms within the SPC during this
study and their colocation with backscattering maxima, it is probable that the
primary production associated with chlorophyll beneath the OD1 during subsurface
blooms is a non-negligible portion of the regional carbon budget. However, regional
carbon export estimates are still likely to be more heavily affected by
underestimation of integrated chlorophyll during surface blooms. This is due both
Teel
54
to the reduction of PAR at depth and because surface blooms have been found to be
associated with larger amounts of total carbon export (Buesseler, 1998).
2.4 Discussion
2.4.1 Regional Application of SPOT Data
This study identified four major profile types within the SPC during March
through July of 2013 and 2014, early upwelling, surface bloom, subsurface bloom,
and offshore intrusion. These four types represented the end-members of vertical
profiles observed within the SPC during consecutive glider deployments. The
statistical analysis of these profile types allowed for the identification of temporal
and spatial trends in local variability within the SPC and the contextualization of the
SPOT station within the SPC. Our results suggest that data collected at SPOT during
monthly time-series sampling should be representative for the SPC. As such,
ecological or physical trends that are found to be consistent for the entirety of the
SPOT time-series may be assumed to be applicable for the SPC and could give
relevant insight into offshore regions within the Southern California Eddy as well.
Given the magnitude of variability, monthly sampling appears to temporally
under-sample biological fluctuation within the SPC. Consequently, infrequent
events, such as coastal upwelling and surface blooms, are not captured by sampling
at SPOT, even during the coastally-dominated early Spring. From these results, data
collected monthly at the SPOT station would also hypothetically be insensitive to
other coastal drivers, such as anthropogenic stressors, sediment discharge, or
freshwater input. The offshore dominated nature of the SPOT station may be
Teel
55
unusual for coastal locations that are within such close proximity to land, however,
and similar studies would need to be conducted to determine the regional context
for other coastal time-series stations.
During this study, glider deployments were timed to sample only during the
most dynamic seasons for the SPOT station. However, our results suggest that a
year-long sampling scheme would likely find similar relationships given that the
sampling occurred during the season when coastal processes (local upwelling)
occurred at the highest frequency but still concluded that SPOT was most
representative of offshore water types. Building on these findings, and
understanding the rarity of coastal events in the SPC, higher-frequency time-series
sampling at SPOT during the spring season could be used to better monitor the
effects of nearshore processes on the physical and biological oceanography of the
SPC.
The similarity between the SPOT station and offshore profile locations
suggests that these locations experience similar oceanographic transport through
the SPC. The offshore state of SPOT during this study also suggests that time-series
data collected at SPOT would be relatively sensitive to climate-derived changes to
major current patterns and wind patterns, but may be insensitive to local terrestrial
changes such as decreased freshwater discharge or increased sedimentation. Within
the SPC, the spin-up of the Southern California Eddy is a major driver of
oceanographic change, and we have hypothesized this to be the cause of the profile
shift during late April and early May of 2013 and 2014 observed at both SPOT and
nearshore locations. Utilizing these observed profile changes, typified by a
Teel
56
transition from coastally-dominated to offshore-dominated water profile types, the
timing of the spin-up could be monitored historically or prospectively within cruise
profile data.
The onshore-offshore gradient in the physical profile characteristics
observed during these studies suggests that biological samples collected at SPOT
would be dominated by offshore taxa rather than nearshore taxa; however, profile
characteristics based only on temperature and chlorophyll supply only very limited
taxonomic information. Due to the complex upwelling dynamics within the SCB,
even during periods when the profile characteristics were more coastally
dominated at SPOT, it is likely that there would be taxonomic differences between
samples taken at SPOT and those taken closer to Palos Verdes. Profile data during
coastal events observed at SPOT would need to be paired with concurrent regional
mapping (eg., satellite imagery, high frequency radar surface current mapping, etc.)
to contextualize the SPOT samples. For a time-series location that was more
coastally dominated, the regional application of data from that time-series site
would likely require robust observations of the taxonomic variation over space and
time. These observations could still be completed with glider profiling with the
incorporation of backscatter, nitrate, and oxygen data collection.
2.4.2 Determining Regional Domains for Time-Series Sites
The methods used here to study the applicable domain for SPOT samples and
particular sensitivity of SPOT samples to regional oceanographic dynamics could
easily be applied to other time-series sites. The glider deployments in this study, as
Teel
57
previously noted, were designed to sample a dynamic period within the SPC with
high-spatial and high-temporal frequency. These high frequency samples enabled
identification and statistical analysis of regional connectivity and response to local
oceanographic events. In the case of the SPC, these events were upwelling, surface
blooms, subsurface blooms, and offshore intrusion. We expect these dominant
water profile types will vary with oceanographic region. For example, at the
Bermuda Atlantic Time-Series these features could include winter mixing, cyclonic
or anti-cyclonic eddies, and hurricanes. We have shown that high-frequency data
can be used to identify the major regional modes and profile types, which can
hypothetically be used to determine long-term trends within the time-series site
itself. These profile signatures therefore allow for a more quantitative description of
time-series observations and thereby an accurate method for detecting deviations
within the time-series. In the example of the SPOT station, increased nearshore
upwelling or delayed spin-up of the Southern California Eddy would each increase
the overall coastal signature of vertical profiles collected at SPOT. Historical vertical
profiles as well as future profiles could be analyzed with reference to coastal and
offshore signals to determine how the relative influences change over time.
Spatial contextualization of time-series sites during in-situ glider
deployments such as those used during this study can help to identify potentially
uncharacterized sources of biological complexity within the time-series location. In
the case of the SPOT station, the temporal switch from coastally-influenced profiles
to offshore-influenced profiles implies that samples collected during late April may
vary significantly from samples collected in early May simply due to changes in local
Teel
58
currents. In-situ platforms are particularly suited in the case of regional analysis
because they not only can detect potentially hidden subsurface dynamics, such as
the missed subsurface integrated chlorophyll at the SPOT station, but will also
collect data that is directly comparable to cruise profile data from time-series sites.
2.4.3 Incorporation of Time-Series Data into Future Modeling Efforts
Once the oceanographic modes and regional extent of a time-series site have
been determined, the monthly dataset can be interpreted within this new context,
allowing for greater regional extrapolation. This type of analysis can be applied to
any time-series site and is particularly important for integrating regional modeling
efforts into larger scale models. Without context for time-series data, physical or
temporal under-sampling at time-series stations may mask local drivers of
variability making it difficult to accurately scale-up local results to inform larger-
scale modeling efforts. Conversely, as discussed above for the SPOT station, monthly
sampling is likely to miss infrequent events that are important for local processes
making single (monthly sampled) time-series fairly ineffective for predictive
dynamic regions such as coastal systems. By providing insight into temporal and
spatial variability, analyses such as the one presented here can provide context for
time-series sites and increase the applicability of time-series data.
Understanding the main drivers for site variability would in turn highlight
the relative need for improved regional physical modeling, oceanographic
ecosystem modeling, or riverine modeling in accurately estimating the coastal
carbon dynamics. In the case of SPOT, the results of this study imply that accurate
Teel
59
modeling of the carbon dynamics will require precision in physical modeling of
major currents, physical modeling of local winds, and ecosystem modeling of both
surface and subsurface bloom communities. However, an improved understanding
of riverine flow and local sediment loads may only marginally improve model
results for the SCB. This insight into the local forcing of the biological and physical
dynamics at a site is a necessary prerequisite for predictive models that will be able
to accurately describe local features that may change dramatically with climate
change, such as oxygen minimum zones or eddy-stimulated primary production.
Determining how to simplify regional complexity has obvious benefit in the
case of coastal time-series sites, where each site is thought to represent a much
smaller range than open ocean sites and the questions of regional production are
still under heavy debate. By understanding the local variability, it is possible that the
data collected at these sites could be applied to larger ranges. However, the
application of similar techniques to open ocean time-series sites should not be
overlooked either, particularly because of short-lived dynamics that may vary with
climate change, such as the frequency of eddies, hurricanes, and submesoscale
processes. In this case as well, understanding the amount of local variability
attributed to specific regional features will allow for better predictive power within
climate models.
2.5 Conclusions
In-situ high-frequency regional data collection can contextualize time-series
sites and identify the major modes of regional variation. This not only allows for the
Teel
60
data to be more accurately extrapolated to provide larger scale estimates of critical
dynamics such as primary production or carbon export, but also identifies the most
crucial regional signals that would need to be correctly simulated to produce
accurate regional modeling. In the case of the San Pedro Ocean Time-series (SPOT)
station, this study identified four major regional water profile types and indicates
that time-series samples collected monthly would be most representative of
offshore water profiles. However, due to cross-channel connectivity within the San
Pedro Channel (SPC), higher frequency sampling at SPOT may also capture coastal
signals from nearshore events such as upwelling and surface blooms. Glider profile
data from the SPC also indicated that the integrated biomass of surface and
subsurface blooms would be underestimated by satellite chlorophyll measurements,
suggesting that accurate observation of regional dynamics within the SPC requires
in-situ sampling. This study suggests that time-series data collected at SPOT would
be relatively insensitive to coastal anthropogenic change but should be well located
for identifying the regional response to climate change. The methods described in
this study can be applied to other coastal and oceanic time-series sites in order to
identify the major modes of local variability, the region represented by the time-
series data, and the sensitivity of the site to anthropogenic change. Better
understanding of the spatial domain represented by global marine time-series sites
will aid in the extrapolation of local findings, in the improvement of regional
modeling, and in the coupling of regional and global modeling efforts.
Teel
61
Acknowledgments
We would like to acknowledge NOAA, NSF, and the Wrigley Institute for
Environmental Sciences for funding this research. Field work was partially
supported by grants 1260296 to Maria Prokopenko at Pomona College and 1260692
to Douglas Hammond at USC from the Chemical Oceanography program of the
National Science Foundation. We would also like to thank the glider team who
helped to collect this data. These collaborators made it possible to broaden the
regional sampling beyond a single season. Carl Oberg assisted with countless hours
of fieldwork - recovering, deploying, preparing, and repairing gliders. Nick Rollins of
the Hammond Laboratory was irreplaceable during glider deployments in both
2013 and 2014. The Caron laboratory, especially Alyssa Gellene, assisted with
culture preparation and glider calibration during the 2014 deployment. Ivona
Cetinic, Arvind Pereira, and the Sukhatme Laboratory were integral in developing
the Jones Laboratory data pipeline and glider network used during these glider
deployments. Dario Diehl and the Southern California Coastal Water Research
Project provided the use of their glider during this field campaign. SPOT cruise data,
collected by Troy Gunderson, the Fuhrman Laboratory, the Caron Laboratory, and
the crew of the Yellowfin, were made available for this research by Diane Kim and
the Wrigley Institute for Environmental Sciences.

Teel
62

Figure 2. 1 Glider Deployment Map and Idealized Glider Transect
Slocum electric gliders were deployed in 2013 and 2014 between the Palos Verdes
Peninsula and Catalina Island in the San Pedro Channel. The San Pedro Channel lies
within the Southern California Bight and contains the San Pedro Ocean Time-series
(SPOT) station, located at 33˚ 33’ 00” N and 118˚ 24’ 00” W (indicated by red dot).
Glider surfacings during deployments in 2013 and 2014 are indicated with grey
dots. An idealized transect was defined running perpendicular to mean flow (thick
black line). All glider profiles collected within 5 kilometers of the idealized transect
line (dashed black lines), were used to assess cross-channel variability of
oceanographic properties during these glider deployments. The bathymetry of the
study area ranged from ~ 20 - 900 m. The maximum glider dive depth was 90 m
during this deployment.
Teel
63

Figure 2. 2 Temperature and Chlorophyll Profiles for Ideal Profile Types
Averaged temperature and chlorophyll a profiles for the four end-member profile
types in the San Pedro Channel during spring and early summer in 2013 and 2014:
early upwelling (n=12), surface phytoplankton bloom (n=10), subsurface
phytoplankton bloom (n=15), and offshore water intrusion (n=17). Secondary
characteristics from these four water profile types were used to create principal
component axes for downstream analyses. The variability of glider profiles within
the channel was determined relative to these four profile types. The first principal
component axis was correlated with increasing offshore physical properties while
the second principal component axis was negatively correlated with increasing
biomass.
Teel
64

Figure 2. 3 Cross-channel Variation of Profile Characteristics
Teel
65
Cross-channel variation in mixed layer temperature, mixed layer depth, the depth of
the 12.5˚C isotherm, and vertically integrated chlorophyll (surface to 70 meters) for
all binned glider profiles from 2013 and 2014. Cross-channel whisker plots show
the median value for each bin (open circle), data between the 25
th
and 75
th

percentiles (blue box), data between the 9
th
and 91
st
percentiles (blue lines), and
outliers (blue dots). Low numbered bins correspond with the western side of the
San Pedro Channel (SPC), near Catalina. High numbered bins correspond with the
eastern side of the SPC, near the Palos Verdes Peninsula (PV). Bins 1-2 and 35-55
have been removed due to low sampling frequency. Bins 35-55 correspond with the
shipping lanes for the Port of Los Angeles. Cross-channel analysis shows that
isopycnal depth and mixed layer depth shoal from west to east, corresponding with
a mean equatorward flow through the channel. Corresponding mixed layer
temperatures decreased west to east across the channel on average during these
glider deployments as well. Integrated chlorophyll observations did not display a
linear cross-channel pattern, even though the average depth of nutrient rich, colder
waters was shallower near PV.

Teel
66

Teel
67
Figure 2. 4 Principal Component Analysis of Glider Profiles
The four end-member, or ideal, water profile types (Figure 3) were used to create
principal component axes. All glider profiles collected in 2013 and 2014 were then
projected onto these axes (grey dots). Physical variability was associated with PC1
(49.8% of total variance) and biological variability was associated with PC2 (32.7%
of total variance). Glider profiles from the SPOT location are shown in black
diamonds in 4a, and the mean and standard deviation of those samples are shown in
4b along with the mean and standard deviation for profiles collected at bin 10 near
Catalina Island and bin 58 near the Palos Verdes Peninsula.

Teel
68
All Samples Percentage SPOT Samples Percentage
Total Sample Size 1606 -- 54 --
Surface Bloom 19 1.2% 1 1.8%
Subsurface
Bloom
398 24.8% 14 25.9%
Offshore
Intrusion
194 12.1% 5 9.3%
Early Upwelling 25 1.6% 0 0.0%
Not Within 95%
CI
970 60.4% 34 63.0%
Table 2. 1 Distribution of Profile Types
After projecting all glider profiles from 2013 and 2014 onto the ideal PCA axes, the
number of glider profiles corresponding with each ideal group was determined. The
total dataset and SPOT specific profiles behaved similarly, however, only one coastal
bloom profile and no early upwelling profiles were observed at SPOT.
Teel
69

Figure 2. 5 Temporal Variation at SPOT and Coastal Locations
The glider profiles collected at individual locations, SPOT (bin 28) and nearshore
bins 55-62, were followed over time to analyze the temporal variability. Though
SPOT and the nearshore bins displayed a different temporal evolution during the
course of each deployment, there appeared to be a crossover point in mid-April for
coastal and mid-channel in both 2013 and 2014. This cross-over point can be
visualized as a movement from more coastal profile characteristics (negative PC1)
to more offshore characteristics (positive PC1).

Teel
70

Figure 2. 6 Investigation into Whole-Channel Oceanographic Events
To assess cross-channel profile behaviour relative to the profile characteristics at
the SPOT site, SPOT samples were divided into four groups according to their PC1
and PC2 coordinates. The PCA coordinate plane was divided into four quadrants to
represent increases in offshore profile characteristics (increasing PC1) and
increases in phytoplankton biomass (decreasing PC2). All non-SPOT glider profiles
were then grouped by the corresponding most recent SPOT profile.

Teel
71

Figure 2. 7 Integrated Chlorophyll below the First Optical Depth
When integrated chlorophyll below the first optical depth was compared with PCA
coordinates for the same glider profiles, the surface bloom profiles were those that
had the highest levels of integrated chlorophyll that would have been missed by
satellites. These instances were rare, but are likely to be associated with
underestimation of regional productivity and export. Subsurface bloom profiles
were associated with up to 200 milligrams per square meter of missing integrated
chlorophyll as well. However given the depth of these blooms the overall primary
production associated with them is theoretically much lower than during surface
blooms, and an equal mismatch during surface blooms would then be associated
with a much larger underestimation of regional production and export. These
results reiterate the need for in-situ vertical chlorophyll profiles for improving
regional satellite primary production estimation.

Teel
72
Water
Type
Criterion 1 Criterion
2
Criterion
3
Criterion 4 Criterion 5 Criterion 6
Upwelling
n = 12
MLTemp
<12.5
ChlInt70
<85
maxCHL
<20
zMaxChl
<45
N/A N/A
Surface
Bloom
n = 10
ChlInt70Per20
<2
ChlInt70
>150 &
< 450
z12p5
>20
MLTemp
>12.9 & <17
MLD
>10
maxCHL
>10
Subsurface
Bloom
n = 15
MLTemp
> 17
z12p5
> 35
ChlInt70
> 120
ChlInt70Per20
>20
maxCHL
> 4
N/A
Offshore
Intrusion
n = 17
ChlInt70
<80
MLTemp
>18
MLD
<12
z12p5
<40
ChlInt70Per20
>5 & <15
N/A
Supplemental Table 2. 1 Ideal Profile Characteristics
End-member profile types, early upwelling, surface bloom, subsurface bloom, and
offshore water intrusion, were selected numerically according to these criteria.
MLTemp refers to mixed layer temperature; ChlInt70 is the value of integrated
chlorophyll a over the top 70 meters of the water column; maxCHL is the maximum
value of calibrated chlorophyll a fluorescence in micrograms per liter along a single
profile; zMaxChl is the depth at which the maxCHL was observed; MLD refers to the
mixed layer depth; z12p5 is the shallowest depth at which the water temperature
was below a value of 12.5˚C and is associated with sub-thermocline, nutrient-rich
waters; and ChlInt70Per20 is a ratio of the amount of integrated chlorophyll in the
top 20 meters of a profile compared with the integrated chlorophyll in the top 70
meters of the profile.

Teel
73

Supplemental Figure 2. 1 Principal Component Analysis of Ideal Profiles
The secondary profile characteristics that defined the majority of the variance in the
end-member profiles were mixed layer depth (MLD), the maximum chlorophyll
value for a profile (maxChl), the amount of integrated chlorophyll in the top 70
meters of the profile (ChlInt70), the depth of the 12.5˚C isotherm (z12p5), the mixed
layer temperature (MLTemp), the depth of the chlorophyll maximum (zMaxChl),
and the ratio of integrated chlorophyll in the top 20 meters of the profile relative to
the integrated chlorophyll over the top 70 meters (chlInt70Per20). Using these
variables, the axes PC1 and PC2 explained a combined 82.5% of the total variance
observed between end-member profiles. Each end-member group, early upwelling,
offshore intrusion, surface bloom, and subsurface bloom, is presented here with a
95% confidence interval ellipse to define similar profile types. These PCA axes were
used for temporal and spatial analyses of all glider profiles collected within the SPC
during 2013 and 2014.

Teel
74

Supplemental Figure 2. 2 Optical Depth versus Glider-Satellite Mismatch
MODIS Aqua chlorophyll a data and calibrated glider chlorophyll a fluorescence were
compared using integrated chlorophyll values between the surface and the first
optical depth (OD1). Integrated chlorophyll over OD1 was generally 3-5x higher
when estimated from the glider data as compared to the satellite estimate. This
discrepancy was likely caused at least in part by deviation in fluorescence to
chlorophyll ratios between phytoplankton in-situ and phytoplankton used for
fluorometric calibrations. The glider chlorophyll fluorometer was calibrated with a
mixture of locally relevant phytoplankton cultures to minimize this error, but the
dominant phytoplankton species and related fluorescence to chlorophyll ratio was
almost certainly varying during the course of these deployments. More important,
there was no correlation between glider and satellite derived integrated
chlorophyll. This disagreement was particularly pronounced during periods with
extremely shallow optical depths such as those typically associated with surface
blooms. This mismatch was likely exacerbated by temporal and spatial discrepancy
between these datasets, since glider data was collected continuously and at 500km
resolution while satellite data was collected once daily at 1km resolution. Due to
these mismatches, further analyses were based solely on the glider-to-glider
chlorophyll comparisons, examining the chlorophyll profiles
Teel
75
References
Alkire, M.B., E. D’Asara, C. Lee, M.J. Perry, A. Gray, I. Cetinic, N. Briggs, E. Rehm, E.
Kallin, J. Kaiser, and A. Gonzalez-Posada. 2012. Estimates of Net Community
Production and Export Using High-Resolution Lagrangian Measurements of O2, NO3
-
,
and POC through the Evolution of a Spring Diatom Bloom in the North Atlantic.
Deep-Sea Research I. 64:157-174.

Baklouti, M., F. Diaz, C. Pinazo, V. Faure, and B. Queguiner. 2006. Investigation of
Mechanistic Formulations Depicting Phytoplankton Dynamics for Models of Marine
Pelagic Ecosystems and Description of a New Model. Progress in Oceanography.
71:1-33.

Behrenfeld, M. J. and E. S. Boss. 2014. Resurrecting the Ecological Underpinnings of
Ocean Phytoplankton Blooms. Annual Review of Marine Science. 6:167-194.

Benway, H., Alin, S., Boyer, E., Cai, W.-J., Coble, P., Cross, J., Friedrichs, M., Goñi, M.,
Griffith, P., Herrmann, M., Lohrenz, S., Mathis, J., McKinley, G., Najjar, R., Pilskaln, C.,
Siedlecki, S., Smith, R., 2016. A Science Plan for Carbon Cycle Research in North
American Coastal Waters. Report of the Coastal CARbon Synthesis (CCARS)
community workshop, August 19-21, 2014, Ocean Carbon and Biogeochemistry
Program and North American Carbon Program, 84 pp., DOI 10.1575/1912/7777.

Bauer, J.E., W.-J. Cai, P. A. Raymond, T.S. Bianchi, C. S. Hopkinson, and P. A. G. Regnier.
2013. The Changing Carbon Cycle of the Coastal Ocean. Nature. 504:61-70.

Boyd, P.W. and S. C. Doney. 2002. Modelling Regional Responses by Marine Pelagic
Ecosystems to Global Climate Change. Geophysical Research Letters. 29(16):1806.

Buesseler, K. O., J. E. Andrews, S. M. Pike, M. A. Charette, L. E. Goldson, M.A.
Brzezinksi, and V. P. Lance. 2005. Particle Export During the Southern Ocean Iron
Experiment (SOFeX). Limnology and Oceanography. 50(1):311-327.

Cetinić , I., G. Toro-Farmer, M. A. Ragan, C. Oberg, and B. H. Jones. 2009. Calibration
Procedure for Slocum Glider Deployed Optical Instruments. Optics Express.
17(18)15420-15430.

Chow, C.-E. T., R. Sachdeva, J.A. Cram, J.A. Steele, D. M. Needham, A. Patel, A.E. Parada,
and J. A. Fuhrman. 2013. Temporal Vairability and Coherence of Euphotic Zone
Bacterial Communities over a Decade in the Southern California Bight. The ISME
Journal. 7:2259-2273.

Chow C.-E. T., D. Y. Kim, R. Sachdeva, D. A. Caron, and J.A. Fuhrman. 2014. Top-down
Controls on Bacterial Community Structure: Microbial Network Analysis of Bacteria,
T4-like Viruses, and Protists. The ISME Journal. 8:816-829.
Teel
76

Collins, L. E., W. Berelson, D. E. Hammond, A. Knapp, R. Schwartz, and D. Capone.
2011. Particle Fluxes in San Pedro Basin, California: A Four-year Record of
Sedimentation and Physical Forcing. Deep Sea Research I. 58:898-914.

Cram, J.A. , C.-E. T. Chow, R. Sachdeva, D. M. Needham, A. E. Parada, J. A. Steele, and J.
A. Fuhrman. 2015. Seasonal and Interannual Variability of the Marine
Bacterioplankton Community Throughout the Water Column over Ten Years. The
ISME Journal. 9:563-580.

Cullen, J. J. and R. W. Eppley. 1981. Chlorophyll Maximum Layers of the Southern
California Bight and Possible Mechanisms of their Formation and Maintenance.
Oceanologica Acta. 4(1)23-32.

Cullen, J.J., E. Stewart, E. Renger, R. W. Eppley, and C. D. Winant. 1983. Vertical
Motion of the Thermocline, Nitracline, and Chlorohyll Maximum Layers in Relation
to Currents on the Southern Califronia Shelf. Journal of Marine Research. 41(2):239-
262.

Digiacomo and Holt, 2001 Satellite Observations of Small Coastal Ocean Eddies in
the Southern California Bight. Journal of Geophysical Research: Oceans.
106(C10):22521-22543.

DiLorenzo et al., 2003. Seasonal Dynamics of the Surface Circulation in the Southern
California Current System. Deep-Sea Research II.
Doney, S. C., D. M. Glover, and R. G. Najjar. 1996. A New Coupled, One-dimensional
Biological-physical Model for the Upper Ocean: Applications to the JGOFS Bermuda
Atlantic Time-Series (BATS) site. Deep Sea Research II. 43(2-3):591-624.
Doney, S. C., I. Lima, J.K. Moore, K. Lindsay, M. J. Behrenfeld, T. K. Westberry, N.
Mahowald, D.M. Glover, and T. Takahashi. 2009. Skill Metrics for Confronting Global
Upper Ocean Ecosystem-biogeochemistry Models against Field and Remote Sensing
Data. Journal of Marine Systems. 79:95-112.
Dong, C. M., E. Y Idica, and J. C. McWilliams. 2009. Circulation and multiple-scale
variability in the Southern California Bight. Progress in Oceanography. 82: 168-190.
Dugdale, R.C., R. T. Barber, F. Chai, T.-H. Peng, F.P. Wilkerson. 2002. One-dimensional
Ecosystem Model of the Equatorial Pacific Upwelling System. Part II: sensitivity
analysis and comparison with JGOFS EqPac data. Deep Sea Research II. 49:2747-
2768.
Fasham, M.J.R., H. W. Ducklow, and S.M. McKelvie. 1990. A Nitrogen-based Model of
Plankton Dynamics in the Oceanic Mixed Layer. Journal of Marine Research. 48:591-
639.
Teel
77
Feely, R.A., C.L. Sabine, J.M. Hernandez-Ayon, D. Ianson, and B. Hales. 2008. Evidence
for Upwelling of Corrosive “Acidified” Water onto the Continental Shelf. Science.
320:1490-1492.
Fuhrman, J. A., I. Hewson, M. S. Schwalbach, J. A. Steele, M. V. Brown, and S. Naeem.
2006. Annually Reoccurring Bacterial Communities are Predictable from Ocean
Conditions. Proceedings of the National Academy of Sciences. 103(35)13104-13109.
Hayward and Venrick, 1998. Nearsurface pattern in the California Current: coupling
between physical and biological structure. Deep-Sea Research II
Haskell, W. Z., M. G. Prokopenko, D. E. Hammond, R. H.R. Stanley, W. M. Berelson, J. J.
Baronas, J. C. Fleming, and L. Aluwihare. 2016. An Organic Carbon Budget for Coastal
Southern California Determined by Estimates of Vertical Nutrient Flux, Net
Community Production and Export. Deep-Sea Research I. 116:49-76.
Hickey, B. M. 1979. The California Current System – Hypotheses and Facts. Progress
in Oceanography. 8(4):191-279.

Hofmann, E. E., B. Cahill, K. Fennel, M. A.M. Freidrichs, K. Hyde, C. Lee, A. Mannino, R.
G. Najjar, J. E. O’Reilly, J. Wilkin, and J.-H. Xue. 2011. Modeling the Dynamics of
Continental Shelf Carbon. Annual Review of Marine Science. 3:93-122.

Huyer, A.. 1983. Coastal Upwelling in the California Current System. Progress in
Oceanography. 12(3):259-284.

Jacox, M. G., C. A. Edwards, M. Kahru, D. L. Rudnick, and R. M. Kudela. 2015. The
Potential for Improving Remote Primary Productivity Estimates through Subsurface
Chlorophyll and Irradiance Measurement. Deep-Sea Research II. 112:107-116.

Kahru, M. et al., 2009. Trends in Primary Production in the California Current
Detected with Satellite Data. Journal of Geophysical Research.

Kim et al., 2014 Monthly to interannual variability of microbial eukaryote
assemblages at four depths in the eastern North Pacific

Kirk, J.T. O., 1994, Light & Photosynthesis in Aquatic Ecosystems. 2
nd
Edition.
Cambridge Press. p. 22.

Lie, A. A., D. Y. Kim, and D. A. Caron. 2013. Small-scale temporal and spatial
variations in protistan community composition at the San Pedro Ocean Time-series
station off the coast of southern California. Aquatic Microbial Ecology. 70:93-110.
Lucas et al., 2011 Horizontal internal-tide fluxes support elevated phytoplankton
productivity over the inner continental shelf

Teel
78
Mann, K.H. and J.R.N. Lazier. 2006. Dynamics of Marine Ecosystems: Biological-
Physical Interactions in the Oceans. 3
rd
Edition. Blackwell Publishing. 74-75.

Mantyla, A. W., et al. 2008. Patterns and Controls of Chlorophyll-a and Primary
Productivity Cycles in the Southern California Bight. Journal of Marine Systems.

Mitchell, G. 2008. Photosynthetically active radiation (PAR) at depths from 0 to 100
meters, expressed as percentage of surface PAR, measured aboard CCE LTER
process cruises in the California current, 2006, 2007 and 2008.
http://oceaninformatics.ucsd.edu/datazoo/data/
ccelter/datasets?action=summary&id=61#metadata .

Moore, J.K., S. C. Doney, J. A. Kleypas, D.M. Glover, and I.Y.Fung. 2002. An
Intermediate Complexity Marine Ecosystem Model for the Global Domain. Deep-Sea
Research II. 49:403-462.

Needham, D. M. , C-E T. Chaw, J. A. Cram, R. Sachdeva, A. Parada, J. A. Fuhrman. 2013.
Short-term observations of marine bacterial and viral communities: patterns,
connections and resilience. The ISME Journal . 7:1274-1285.

Needham, D.M. and J. A. Fuhrman. 2016. Pronounced Daily Succession of
Phytoplankton, Archaea, and Bacterial Following a Spring Bloom. Nature
Microbiology. 1:1605. doi:10.1038/nmicrobiol.2016.5.

Nezlin et al., 2012. Phytoplankton Blooms Detected by SeaWIFS Along the Central
and Southern California Coast. Journal of Geophysical Research.

Noble, M.A., Rosenberger, K.J., Hamilton, P., and Xu, J.P., 2009, Coastal ocean
transport patterns in the central Southern California Bight, in Lee, H.J., and Normark,
W.R., eds., Earth Science in the Urban Ocean: The Southern California Continental
Borderland: Geological Society of America Special Paper 454, p. 193–226, doi:
10.1130/2009.2454(3.3).
Noble, M. A., K. J. Rosenberger, P. Hamilton, and J.P. Xu. 2009. Coastal ocean
transport patterns in the central Southern California Bight, in Lee, H.J., and Normark,
W.R., eds., Earth Science in the Urban Ocean: The Southern California Continental
Borderland: Geological Society of America Special Paper 454, p.193–226.

Oey, L.-Y. 1999. A Forcing Mechanism for the Poleward Flow off the Southern
California Coast. Journal of Geophysical Research. 104(C6):13529-13539.

Pawka, S.S. 1983. Island Shadows in Wave Directional Spectra. Journal of Geophysical
Research.
Schnetzer, A., et al., 2013. Coastal Upwelling Linked to Toxic Pseudo-nitzschia
autralis Blooms in Los Angeles Coastal Waters, 2005-2007. Journal of Plankton
Research.
Teel
79

Spitz, Y.H., J. R. Moisan, and M. R. Abbott. 2001. Configuring an Ecosystem Model
Using Data from the Bermuda Atlantic Time Series (BATS). Deep-Sea Research II.
48:1733-1768.

Sprintall, J. and M. Tomczak. 1992. Evidence of the Barrier Layer in the Surface
Layer of the Tropics. Journal of Geophysical Research. 97(C5):7305-7316.

Steele, J.A., P. D. Countway, L. Xia, P.D. Vigil, M. Beman, D. Y. Kim, C.-E. T. Chow, R.
Sachdeva, A. C. Jones, M. S. Schwalbach, J. M. Rose, I. Hewson, A. Patel, F.-Z. Sun, D. A.
Caron, and J. A. Fuhrman.

Thunell, R.C., C. H. Pilskaln, E. Tappa, and L. R. Sautter. 1994. Temporal variability in
sediment fluxes in the San-Pedro Basin, Southern California Bight. Continental Shelf
Research. 14(4):333-352.

Teel
80
Chapter 3. From Networks to Models: Investigation and
parameterization of bacterial carbon demand at the San Pedro
Ocean Time-series site (SPOT).
Abstract
Marine ecosystem models within global climate models have traditionally
excluded explicit bacterial dynamics; however, marine bacteria metabolism is
known to be one of the primary controls for the fate of marine carbon. The inclusion
of explicit bacterial dynamics within marine ecosystem modeling may be
computationally possible due to improvements in computational power and model
efficiency, but in order to justify this added complexity bacterial carbon dynamics
drivers of variability in bacterial carbon demand must be identified. The functional
variability within the free-living marine bacterial communities at the San Pedro
Ocean Time Series (SPOT) site sampled from the surface monthly between 2000 and
2011 were analyzed with regard to changes in both environmental and biological
parameters. On average, the bacterial community showed predictable seasonal
changes in community composition and peaked in abundance in the spring with a
one-month lag from peak chlorophyll concentrations. Bacterial growth efficiency
(BGE), estimated from bacterial production, was found to vary widely at the site
(1% to 40%). In a multivariate analysis, 47.6% of BGE variability was predicted
using primary production, bacterial community composition, and temperature.
These results indicate that community composition directly affects community
function and that bacterial carbon dynamics within models would not be accurately
depicted without a means of incorporating variable BGE.
Teel
81
3.1 Introduction
Oceanic drawdown of carbon dioxide from the atmosphere is accomplished
through a combination of both physical and biological processes. Predicting how the
oceanic carbon cycle will change as a result of climate change requires mechanistic
modeling of both the physical and biological processes, as they do not act
independently of one another. The biological drawdown of carbon from the
atmosphere is, however, poorly represented in current global scale oceanic models
(Laufkötter et al., 2015). Historically, computational limitations required that only a
small number of phytoplankton function groups and a single zooplankton group
could be included in order to complete model runs at the temporal and physical
scales required to make climate predictions (Moore et al., 2002; Doney et al., 2009;
Laufkötter et al., 2015). As computing powers increase the modeling community has
the ability to improve the oceanic ecosystem dynamics within global models.
Computational powers will never be infinite, of course, so each additional
complexity must be carefully selected for its benefit to both the accuracy of the
model output when compared with observational datasets and the mechanistic
power of predictive model runs.
One of the most significant missing mechanisms in the current global-scale
oceanic ecosystem models is a dynamic, mechanistic representation of the
remineralization of both dissolved organic carbon (DOC) and particulate organic
carbon (POC). Carbon remineralization in the ocean is dominated by heterotrophic
prokaryotes. The prokaryotic communities that complete this remineralization
fluctuate with season, temperature, nutrients, coincident protistan and viral
Teel
82
communities, and even the constitution of the DOC or POC itself (Carlson et al.,
2004; Fuhrman et al., 2006; Gilbert et al., 2010; Beman et al., 2011; Steele et al.,
2011; Gilbert et al., 2012; Halewood et al., 2012; Teeling et al., 2012; Chow et al.,
2013; Jones et al., 2013; Needham et al., 2013; Chow et al., 2014; Cram, 2014;
Klindworth et al., 2014; Galand et al., 2015; Kirchman, 2016).
General bacterial modeling assumes a black-box method, in which a single
bacterial biomass responds to environmental cues, such as temperature or DOC
concentration, without attention to community structure or the functional
variability that might be caused by that community structure (Hagström et al., 1988;
Thingstad and Lignell, 1997; Anderson and Williams, 1999; Walsh et al., 1999;
Anderson and Ducklow, 2001; Allen et al., 2004; Hasumi and Nagata, 2014; Weitz et
al., 2014). This technique is common, but may not be applicable for marine bacterial
communities, whose members vary vastly in maximum growth rates, growth
efficiencies, and nutrient limitation. Marine bacterial growth efficiency (BGE) alone
varies from values as low as 0.01 to values as high as 0.5 in-situ, but BGE is
incorporated into most ecosystem models as a fixed value, which in turn ranges
from 0.1 to 0.5 between models (del Giorgio and Cole, 1998; Anderson and Williams,
1998; Anderson and Williams, 1999; Xiu and Chai, 2014; Hasumi and Nagata, 2014;
Weitz et al, 2015). In global climate models, marine remineralization is further
abstracted, where bacterial biomass is only implicitly represented (Moore et al.,
2002; Doney et al., 2009; Laufkötter et al., 2015). For example, in the commonly
referenced ecosystem model embedded within the NCAR CESM, DOC
remineralization is modeled as either instantaneous, in the case of labile DOC, or a
Teel
83
temperature-dependent, in the case of semi-labile DOC, loss rate (Moore et al., 2002;
Moore et al., 2004; Doney et al., 2009);
With the influx of new data regarding marine bacterial community dynamics
and functional change, commonly used algorithms for bacterial growth and
remineralization need to be re-examined to determine how marine microbial DOC
cycling can be improved within oceanic ecosystem models. Before incorporating
heterotrophic bacterial into oceanic marine ecosystem models, it is crucial to
determine if the functional variability is solely a result of environmental drivers, as
assumed in current ecosystem models. Here, I utilize data from the Microbial
Observatory at the San-Pedro Ocean Time-Series site (SPOT) to re-evaluate
previously published bacterial algorithms, to investigate the functional variability of
the bacterial community over time, to isolate the key contributors to functional
variability, and to determine how functional response might best be parameterized
within this dataset.
3.2 Study Site
The SPOT site is located within the San Pedro Channel in the Southern
California Bight and experiences significant seasonal variability with deeper winter
mixed layer depths, increased surface primary production in spring, and
stratification that intensifies through summer and fall (Oey, 1999; Strub et al., 2000;
Noble et al., 2009; Dong et al., 2009). Though the SPOT site can experience coastal
blooms following regional upwelling events, the average oceanographic
characteristics of the SPOT site have been found to be more similar to locations
Teel
84
farther offshore when sampled on monthly timescales (Chapter 2; Teel et al., in
prep). Bacterial dynamics at the SPOT site have been studied through the Microbial
Observatory since 2001 and environmental data has been collected at this location
as part of the Wrigley Time Series since 1998. The location of the SPOT site also falls
within the California Cooperative Oceanic Fisheries Investigations (CalCOFI)
transect region, allowing for the application of CalCOFI ancillary datasets where
available (calcofi.org/data.html). Direct investigations of composition of the free-
living bacterial community have been completed at SPOT monthly since 2001
(Fuhrman et al., 2006; Chow et al., 2013; Cram, 2014; Cram et al., 2015). These
investigations found that bacterial dynamics at SPOT have strong seasonality with
reoccurring communities, which is representative of global trends (Fuhrman et al.,
2006; Gilbert et al., 2010; Gilbert et al., 2012; Chow et al., 2013; Cram et al., 2015;
Fuhrman et al., 2015; Ward et al., 2017). Recent research at SPOT has also found
strong oscillation within the surface bacterial communities and correlation of these
oscillations to both physical and biological parameters (Figure 3.1, Table 3.1, Table
3.2, modified from Cram, 2014). Finally, the bacterial communities at SPOT also
appear to be on average more similar to oceanic communities than to coastal
communities, indicating that the functional dynamics observed at SPOT may be
applicable beyond the Southern California Bight (Chow et al., 2013).
Teel
85
3.3 Data and Methods
3.3.1 Microbial Observatory Data
Data collected at the SPOT site during Microbial Observatory (MO)
investigations and Wrigley Time Series collection from 2001 through 2011 were
used in this study through collaboration with the Fuhrman Laboratory and the
Wrigley Institute of Environmental Studies. The MO data used in this study included
5m: bacterial abundance (BA), bacterial production (BP), viral abundance (VA),
bottle-measured chlorophyll a concentration (CHL), nitrate concentration (NO3),
phosphate concentration, temperature, the relative abundance of free-living marine
bacterial OTUs as determined by automated ribosomal intergenic spacer analysis
(ARISA) fragment lengths, and ARISA-derived calculations for diversity (Inverse
Simpson index and Shannon index), evenness (Peilou’s evenness index), and OTU
richness. For a subset of ARISA OTUs, putative genealogical identifiers were
provided based off of clone library comparisons with ITS fragment lengths. A total of
114 surface samples were available between 2001 and 2011. Sample collection and
processing has been previously described in detail (Chow et al., 2013).
Cram (2014) identified two primary, negatively correlated bacterial
communities or modules in the surface waters at SPOT (Mod1 and Mod2) (Figure
3.1). Membership for Mod2 was also further divided into Mod2A and Mod2B to
depict the subsets of highly correlated bacterial OTUs within Mod2 (Figure 3.1).
Heterotrophic membership and cyanobacterial membership within groups were
Teel
86
analyzed independently. Cyanobacterial membership was established through
putative OTU identification provided with the ARISA OTU data.
Relative abundance for each group described above (Mod1, Mod2, Mod2A,
Mod2B, Mod2A-het, Mod2B-het, and Cyano) was calculated as the sum of the
relative abundances of each of the component OTUs within a single 5 m surface
sample. Multiplication of relative abundance by the total bacterial abundance from a
sample was used to estimate the total abundance of these groups.
3.3.2 Satellite Chlorophyll a and Photosynthetically Active Radiation data
MODIS Aqua level 3 mapped 9-km resolution satellite photosynthetically
active radiation data (satPAR) from 2003 through 2014 were downloaded from the
NASA Ocean Color website (http://oceandata.sci.gsfc.nasa.gov/ ). The satPAR data
are shown in Figures 3.2a and 3.2b in Einstein per square meter per day and fit both
the expected total daily sum (Figure 3.2a) and annual trend (Figure 3.2b). PAR data
were consistent with estimates of PAR from CalCOFI Process Cruises (Mitchell,
2008).
Daily and monthly averaged MODIS Aqua mapped high-resolution (0.05
degree) surface CHL data (satCHL) were acquired from NOAA CoastWatch
(http://coastwatch.pfeg.noaa.gov/coastwatch/CWBrowser.jsp).
3.3.3 Wrigley Time Series Data
Temperature, chlorophyll a fluorescence, and salinity data from Wrigley
Time Series depth profiles from 2001-2011 were used. Mixed layer depth (MLD)
was calculated after Sprintall and Tomczak (1992). Specifically, the mixed layer
Teel
87
depth (MLD) was defined as the depth where density was greater than potential
density calculated using 5 m salinity and 5 m temperature minus 0.4˚C. These values
were determined through rigorous comparisons of variations of the equation and
the physical location of the thermocline over the time-series.
Wrigley Time Series chlorophyll a concentrations (botCHL) were used to
calibrate fluorescence profiles. Bottle chlorophyll samples were taken
approximately every 10 m in the surface 50 m. The value of chlorophyll fluorescence
at 500 m was used to set the dark calibration. Any non-zero values at this depth
were assumed to equate to the background noise within the fluorometer itself and
were used to adjust water column profiles to account for this noise. The dark-
corrected in situ chlorophyll fluorescence data were then compared against the
bottle chlorophyll data and calibration factors for the fluorescence data were
calculated through linear regression. Post-calibration chlorophyll data is compared
against pre-calibration data and the bottle data in Figure 3.3. For this time-series,
the chlorophyll fluorometer read approximately 3x higher than bottle data for years
1998 through 2003 and read approximately 1.2x higher than bottle data between
2003 and 2014. While bottle chlorophyll data was used whenever possible,
calibrated chlorophyll fluorescence data from 5m was used when bottle data was
missing (N=51 of 114).
3.3.4 Primary Production Estimates
Primary production is a key parameter for understanding ecosystem
dynamics and carbon cycling but was unfortunately missing from the MO and
Teel
88
Wrigley datasets. While surface satellite primary production can help fill this gap,
these satellite based estimates are unlikely to capture the biological variability
within the mixed layer or the euphotic zone due to subsurface chlorophyll maxima
and particle maxima that are commonly observed at SPOT (Cullen, 1982; Teel et al.,
in prep).
Here depth profiles of primary production (PPz) were estimated after Jacox
et al. (2015). The Jacox et al. (2015) algorithm was developed for the Southern
California Bight, utilizing CalCOFI deck-board C-14 carbon fixation incubations
study region, in-situ chlorophyll a fluorescence profiles, and satellite PAR data
(Jacox et al., 2015). For this study, depth resolved PP(z), was calculated as:
eq (3.1) 𝑃𝑃 𝑧 =𝑝𝐵 𝑧 ∗𝑐ℎ𝑙 𝑧 ∗𝑑𝑖𝑟𝑟
where pB(z) is the regionally tuned carbon fixation rate in mgC per mgChl per hour,
chl(z) is the chlorophyll concentration (mgChl/m3), and dirr is day length (hours
per day). dirr, was calculated as:
eq (3.2) 𝑡ℎ𝑒𝑡𝑎= 0.2163108+2∗arctan 0.9671396∗tan 0.00860∗ 𝐽−186
eq (3.3) 𝑝ℎ𝑖= arcsin 0.39795∗cos 𝑡ℎ𝑒𝑡𝑎
eq (3.4) 𝑑𝑖𝑟𝑟= 24−
!"
!
arccos
!"#
!"
!"#
!!"#
!!
!"#
∗!"# !"#
!"#
!!
!"#
∗!"# !"#

where J is Julian day of sample date, L is the sample latitude, and P is set to 0.2667 to
account for standard definitions of sunset and sunrise.
The regionally tuned carbon fixation rate, pB(z), was estimated as:
eq (3.5) 𝑝𝐵(𝑧)= 2.9∗ (
!"# !
!"# ! !!.!
)
where photosynthetically active radiation (PAR) at a depth z was calculated as:
Teel
89
eq (36) 𝑘 𝑃𝐴𝑅 𝑧 = 0.04+ 0.0088∗𝑐ℎ𝑙 𝑧 + 0.054∗ 𝑐ℎ𝑙 𝑧
!.!"

eq (3.7) 𝑃𝐴𝑅 𝑧 =𝑃𝐴𝑅 𝑠𝑎𝑡 ∗𝑒
!! !"# ! ∗!

using observed chlorophyll concentrations and satellite surface PAR. The resulting
primary production estimates for 2003 - 2014 are shown in Figure 3.4. Integrated
primary production was calculated for the mixed layer and for the upper 90 m,
which was considered to be the maximum euphotic depth for the SPOT site based on
regional and local data. Euphotic depths calculated with this same equation for
k(PAR(z)) during automated glider deployments in 2013 and 2014 were consistent
with simultaneous ship-based measurements of the euphotic zone (Chapter 2; Teel
et al., in prep), supporting the application of a chlorophyll-based attenuation
coefficient at SPOT. The primary production estimates for SPOT were within the
bounds of regional estimates (Jacox et al., 2015). The seasonal averages clearly show
the significant contribution of the sub-surface chlorophyll maximum to total water
column primary production at SPOT is found below 20m, up to 84% of total
production (22% to 62%) during the SPOT MO cruises from 2000 to 2011.
3.3.5 Bacterial Growth and Growth Efficiency
Bacterial production was quantified at SPOT as part of the Microbial
Observatory using both leucine and thymidine incorporation rates. Uptake rates
were converted into bacterial production (BP) in cells per milliliter per day as
described in Chow et al. (2014) and used to predict bacterial growth rate (µ) as:
eq (3.8) µ=
!"
!"

Teel
90
where BA is bacterial abundance in cells per milliliter. These estimates of µ were
compared against the environmentally-driven algorithms for bacterial growth from
White et al., (1991) where:
eq (3.9) log µ =−1.30+𝑇 ∗ (0.052+0.016∗ log 𝐶𝐻𝐿 )
where µ is the specific growth rate, T is temperature, CHL is the chlorophyll a
concentration, and log values are base 10.
Bacterial growth efficiency (BGE), bacterial carbon demand (BCD) are
defined as:
eq (3.10) 𝐵𝐺𝐸=
!"
!"!!"

eq (3.11) 𝐵𝐶𝐷=𝐵𝑃+𝐵𝑅
where BR is bacterial respiration and BP is bacterial production. Bacterial
respiration measurements are exceedingly difficult to make and so are not routinely
made. However, BGE, bacterial growth rates (µ), and the ratio of leucine
incorporation to thymidine incorporation (Leu:TdR) have been shown to be
strongly correlated (del Giorgio et al. 2011) suggesting that BGE can be estimated
from BP and/or Leu:TdR.
While del Giorgio et al. (2011) observed a significant negative relationship
between Leu:TdR and both BGE and µ, this relationship was not observed in the
SPOT dataset (Figure 3.5). Rates of Leu:TdR in surface samples at SPOT were higher
than those reported by del Giorgio et al (2011), averaging 19.1 moles Leucine to
moles Thymidine incorporated. At SPOT, Leu:TdR was not significantly correlated
with BA, BP predicted by leucine (BP-Leu), µ calculated from (BP-Leu), virus
concentration, chlorophyll concentration, vertically integrated primary production,
Teel
91
temperature, salinity or nitrate (P-values between 0.06 and 0.99). The Leu:TdR ratio
was very weakly correlated with salinity, BP predicted by TdR, and µ calculated
from BP (TdR) (R
2
0.07, 0.10 and 0.09, respectively; P-values < 0.01). The
relationship between Leu:TdR and BP may be obscured by variation in the microbial
community composition, since alternation between high nucleic acid and low
nucleic acid cells would alter the ratio of protein synthesis to DNA synthesis. Due to
the lack of strong correlations, the Leu:TdR method for estimating BGE was not used
in this study.
BGE is commonly calculated from BP or temperature (T) following either
delGiorgio and Cole (1998) or Rivkin and Legendre (2001) as:
eq (3.12) 𝐵𝐺𝐸
!"#$%&'(%& !"# !"#$,!""#
=
!.!"#!!.!"∗!"
!.!!!!

eq (3.13) 𝐵𝐺𝐸
!"#$"% !"# !"#"$%&",!""#
= 0.374−0.0104∗𝑇
BGE at SPOT was estimated using both algorithms and compared against regional
measurements of BGE (delGiorgio et al., 2011; Halewood et al., 2012; James et al.,
2017). Both the mean (12%) and range (2% to 35%) of BGE values estimated from
the delGiorgio and Cole (1998) algorithm were consistent with previous measured
BGE values and so all downstream analyses were completed using the BGE
predicted from the delGiorgio and Cole, 1998 meta-analysis.
Teel
92
3.4. Results
3.4.1 “Black Box” Algorithms
Marine microbial ecosystem models that represent bacterial dynamics
assume that environmental drivers are the first order processes determining
bacterial community functionality and so fluctuations in the bacterial community
composition are not represented (Hagström et al., 1988; Fasham et al., 1990; White
et al., 1991; Thingstad and Lignell, 1997; Anderson and Williams, 1999; Walsh et al.,
1999; Anderson and Ducklow, 2001; Allen et al., 2004; Hasumi and Nagata, 2014;
Weitz et al., 2014). Aggregating all heterotrophs into a single ‘black-box’ can be
beneficial as it limits both the number of free parameters and the number of
calculations required per time-step. To determine whether a black-box technique
would be sufficient for the prediction of bacterial functional dynamics at SPOT, a
suite of correlative analyses were performed to test ‘black-box’ assumptions.
Bacterial growth rates are widely calculated from CHL and temperature after
White et al. (1991). However, when this algorithm was applied to the MO dataset, no
correlation was found between predicted growth rate and measured growth rate
(Figure 3.6). The lack of relationship was observed for bacterial production
calculated both from 3H-Thymidine incorporation rates and from 3H-Leucine
incorporation rates (Leucine: R
2
0.0042, P-value 0.5; Thymidine: R
2
0.0189, P-value
0.23). The White et al. algorithm was derived from a meta-analysis of estuarine,
coastal and oceanic ecosystems, including locations where total production
numbers and growth rates were both much higher and much lower than what is
Teel
93
commonly found at the SPOT site. Minimum and maximum values within the
saltwater environments used in the meta-analysis spanned from 4e-6 mmolC m
-3

day
-1
to as high as 2.3e4 mmolC m
-3
day
-1
(White et al., 1991); whereas the range
observed at SPOT was from 1.3e-2 - 5.9 mmolC m
-3
day
-1
.
Based on the assumption that additional environmental and biological
parameters at SPOT may improve a regression-based prediction of bacterial
function, a step-wise linear regression was used to attempt to predict bacterial
production and bacterial growth rates at SPOT (Figure 3.7). Initial stepwise multiple
linear regression analyses identified BA, day length (dirr), CHL, and, to a lesser
extent, Leu:TdR and NO3 as potential predictors for BP and µ. The regressions
varied slightly depending upon the use of either 3H-Thymidine or 3H-Leucine
uptake based bacterial production estimates, where 3H-Thymidine based
production was more strongly correlated with predicted bacterial production
(Figure 3.7). For Thymidine-based bacterial production, rates were best predicted
by BA, CHL, dirr, and Leu:TdR with the regression accounting for 42.7% of the
variance seen within the production measurements (P-value <0.01). For Leucine-
based production measurements, rates were best predicted by BA, CHL, and dirr
with the regression accounting for 35.0% of the observed variance (P-value <0.01).
These linear regression algorithms both over-predicted bacterial production and
were dominated by the correlation between bacterial production and bacterial
abundance, which explained 32% of the variance within the Thymidine based
bacterial production measurements (P-value <0.01) and 17% of the variance within
the Leucine based production measurements (P-value <0.01). Linear regressions for
Teel
94
bacterial µ at SPOT, which were largely insensitive to the correlation between
production and abundance, with daylight, nitrate, and Leu:TdR explaining only 18%
(P-value <0.01) and 5% (P-value 0.04) of the observed variance within Thymidine
and Leucine based growth rates, respectively (Figure 3.7). Temperature, which has
been found to correlate with both bacterial growth rate and BP in previous studies
(Pomeroy and Deibel, 1986; White et al., 1991; Azam et al., 1993; Rivkin and
Legendre, 2001; Jiao et al., 2010), showed no correlation with either µ or BP in the
MO dataset (Figure 3.8).
Seasonality in physical dynamics in the Southern California Bight was
hypothesized to be the primary driver of seasonality in bacterial dynamics.
Specifically, SPOT is characterized by relatively deep mixed layers during the winter,
the shallowing of the mixed layer throughout the spring accompanied by periodic
upwelling events, and increased stratification during the summer and fall coinciding
with the strengthening of the Southern California Eddy (Oey, 1999; Strub et al.,
2000; Noble et al., 2009; Dong et al., 2009). Because of both the increase in local
production following spring blooms and the ephemeral nature of those blooms,
bacterial production was expected to have a non-linear correlation with dirr, where
peak values were expected to coincide with spring samples. Maximum BP did occur
most often during the spring season; however, minimum values showed no
seasonality and elevated BP lasted well into summer months, generally increased
with dirr (Figure 3.9). Hypothetically, BP during the summer could be elevated due
to photodegradation of semi-labile carbon compounds into labile carbon, advection
of labile carbon into the region, or due to a community shift that enables bacterial
Teel
95
uptake and remineralization of semi-labile carbon compounds. A positive linear
correlation between BP and the standard deviation of dirr was also observed
indicating that the maximum BP at 5 m may have a threshold limitation that is
relieved by increased dirr. This may imply energy limitation during winter and
nutrient or carbon limitation during spring and summer.
Estimation of bacterial production and growth rates provide a proxy for the
amount of carbon available for higher trophic level grazing. However, in order to
calculate total bacterial carbon demand (BCD) one must account bacterial growth
efficiency (BGE), which accounts for organic matter remineralization through
bacterial respiration (BR). While only a weak correlation was observed between BP
and CHL (R
2
= 0.10, p << 0.01), a much stronger positive correlation was observed
between estimated BGE and integrated primary production over 90 m (R
2
= 0.43, p
<< 0.01) (Figure 3.10). We suggest that this is because integrated primary
production is a better proxy for water profile state and associated phytoplankton
community characteristics than chlorophyll alone. This result, in combination with
the poor performance of the White 1991 algorithm at the SPOT site, supports the
conclusion that surface chlorophyll concentrations alone are poor predictors of
bacterial functional dynamics. Moreover, this suggests that phototrophic and
heterotrophic dynamics at SPOT are tightly linked indicating that labile DOC
production is a primary driver of bacterial dynamics as previously suggested
(Fuhrman et al., 1980; Hagstrom et al., 1988; Yooseph et al., 2010; Halewood et al.,
2012; Wear et al., 2015; Kirchman, 2016; Ward et al., 2017). The correlation
between primary production and BGE explained approximately 43% of observed
Teel
96
variance in BGE (Figure 3.10). Due to the algorithm used to calculate primary
production, this correlation also applies to integrated chlorophyll, dirr, and surface
PAR. This also suggests that there might a secondary correlation between BGE and
inorganic nitrate, which drive phytoplankton growth rates.
3.4.2 Community Composition and Function
With the general understanding of seasonality and bacterial dynamics at
SPOT, we further investigated these relationships with regard to fluctuating
bacterial community dynamics. Specifically, utilizing the Microbial Observatory
dataset of bacterial OTU abundance over time, changes in bacterial production and
bacterial growth efficiency were compared with changes in bacterial community
composition.
Diversity
As an initial examination of the effect of microbial community shifts on BGE,
four metrics of community diversity were compared against BGE data: OTU
richness, evenness, and two metrics of alpha diversity, the Shannon Index and the
Inverse Simpson index (Figure 3.11). Due to rapid succession of bacterioplankton
observed following phytoplankton bloom events, which theoretically corresponds to
increased niche space, a positive relationship might be expected between
bacterioplankton diversity, primary production, BGE and BP (Teeling et al., 2012;
Klindworth et al., 2014; Needham and Fuhrman, 2016); however, no correlation was
observed between any of these diversity metrics and BGE at the SPOT site (Figure
3.11). Mesocosm experiments have previously shown both positive and negative
Teel
97
relationships between diversity and production, and recent research has suggested
that the diversity of the most active fraction of the community can be positively
correlated with overall community production (Galand et al., 2015). The active
community is often not numerically abundant, however, indicating that whole-
community diversity metrics may not be sensitive to changes in active-fraction
diversity (Winter et al., 2010; Campbell and Kirchman, 2013; Galande et al., 2015).
Bacterial Surface Modules
A disconnect between the diversity of the whole community and the diversity
of the active population might indicate a lack of correlation between community
composition and community function; however, the persistent reoccurrence and
oscillation of bacterial communities over time at various time-series sites indicates
low functional redundancy and a direct link between community composition and
function (Fuhrman et al., 2006; Cram et al., 2015; Fuhrman et al., 2015; Ward et al.,
2017). Two strongly negatively correlated bacterial communities, or modules,
determined through network analyses at the SPOT site within the surface samples
collected between 2000 and 2011 are one such case (Figure 3.1, Reproduced from
Cram, 2014). Network analyses are invaluable for identifying the members of large
bacterial communities whose relative abundance may be good proxies for larger
ecological dynamics; however, the impact of those dynamics can be hard to assess
from networks alone. For example, the negative correlation between modules 1 and
2 does not indicate how often the OTUs were observed or how much of the bacterial
biomass is represented within each module.
Teel
98
To determine whether these observed modules could be utilized as
indicators of overall bacterial community dynamics and carbon cycling, the previous
network analyses were contextualized with module-specific measurements of
biological and environmental change. These two bacterial communities (Mod1 and
Mod2) account for on average 17.5% (0.7 to 64.4%) and 24.5% (0.7 to 63.7%) of the
bacterial biomass, respectively (Figure 3.12). The negative correlation between Mod
I and Mod II was primarily driven by their seasonal alternation, rather than by low
frequency of occurrence of one module. Total Mod1 abundance correlated positively
with total bacterial abundance (R
2
0.41, P-value <0.01) and chlorophyll a (R
2
0.11, P-
value <0.01) (Figure 3.1 reproduced from Cram, 2014; Figure 3.12). Within an
average annual cycle, peak Mod 1 abundance lagged peak chlorophyll
concentrations (i.e. spring bloom) by approximately one-month (Figure 3.12a).
Theoretically this ecological lag is driven by increased DOC availability during
bloom senescence, which has been correlated to both increased sloppy feeding by
phytoplankton grazers and increased direct extracellular phytoplankton release as a
result of nutrient stress (Hagstrom et al., 1988; Wear et al., 2015a; Wear et al.,
2015b).
These results indicate that, on average, increases in Mod1 community
abundance may be an indicator for overall rapid bacterial growth in response to
local primary production; while periods when the Mod2 community were abundant,
conversely, may indicate low primary production and a generally more oligotrophic
oceanographic state. Consistent with these hypotheses, when Mod1 and Mod2
community abundance were directly compared with bacterial growth, high Mod1
Teel
99
abundance was associated with the highest growth rates and greatest growth rate
variability. However, neither Mod1 nor Mod2 community abundance was linearly
correlated with community growth rate (Figure 3.13). The maximum observed
growth rates were three to four times lower for samples in which Mod2 made up
over 40% of the community compared to those when Mod2 made up less than 10%
of the community. The observed community dynamics for Mod2 were most strongly
associated with the subgroup, Mod2 Group B (Figure 3.1); specifically the
abundance of Mod2 was strongly correlated with the abundance of Mod2B (R
2
0.73,
P-value <0.01; vs Mod2A R
2
0.10, P-value <0.01) (Figure 3.14). In addition, when
only heterotrophic members of the Mod2 OTUs were compared with the total
percent of cyanobacterial OTUS from each sample, Mod2 groupB showed a weak
positive correlation with cyanobacterial relative abundance (R
2
0.06, P-value 0.01),
while no relationship was observed for Mod2 groupA (R
2
5.9e-4, P-value 0.8)
(Figure 3.14). Though abundance and activity are not equivalent, these results may
indicate that Mod2A abundance is correlated with a separate low-chlorophyll, low-
production ecosystem state than the cyanobacterial-dominated state that Mod2B is
predicted to describe.
Variation in BGE
To further investigate the potential changes in bacterial heterotrophic
function associated with Mod1 and Mod2 communities, percent abundance of each
of these modules was compared directly against estimated BGE for each sample.
Mod1 percent abundance was found to be weakly positively correlated (R
2
0.02; P-
Teel
100
value 0.15) with BGE while Mod2 percent abundance was found to be significantly
negatively correlated (R
2
0.18; P-value << 0.01) with BGE (Figure 3.15). Though
these correlations were extremely weak, a two-sample T-test showed that high
Mod2 and low Mod2 community BGE estimates were significantly different from
one another (P-value 0.01); this result was not biased by the choice of community,
since high Mod1 and low Mod1 communities also displayed significantly different
BGE distributions (P-value <0.01). These relationships were also consistent with
lower community BGE being associated with more oligotrophic sites and higher
community BGE being associated with more productive sites (delGiorgio and Cole,
1998).
The effect of community composition on BGE was more apparent when the
relationship between temperature and BGE was investigated relative to community
composition (Figure 3.16). At the SPOT site, when bulk BGE estimates were
compared with sample temperature, no correlation was found, contradicting
previous work that showed strong negative correlation between BGE and
temperature (R
2
0.02, P-value 0.18) (Figure 3.16; Rivkin and Legendre, 2001). When
samples were separated into low (<24.7%) and high (>24.7%) percent abundance
of Mod2, a weakly positive correlation was observed between BGE and Temperature
for low Mod2 samples (R
2
0.13, P-value 0.03) and a weak negative correlation was
observed for samples with high Mod2 percent abundance (R
2
0.05, P-value 0.26)
(Figure 3.16b; Figure 3.16c). Furthermore, the distributions of BGE were
significantly different for samples with low percent abundance of Mod2 relative to
those with high Mod 2 abundance (two sample t-test: P-value <0.01) (Figure 3.16b;
Teel
101
Figure 3.16c). The different relationships between BGE and temperature for the two
communities suggest different, community-dependent, dynamics. This is consistent
with previous work which suggested that BGE should vary with community as the
result of bacterial substrate preference, whereby different bacterial OTUs have the
ability to remineralize different suites of carbon substrates produced by different
phytoplankton (Carlson et al., 2004; Halewood et al., 2012; Teeling et al., 2012;
Klindworth et al., 2014; Needham and Fuhrman, 2016; Ward et al., 2017).
Classically, higher temperatures are associated with higher respiration rates and
lower efficiencies (Somero and Hochachka, 1971; Rivkin and Legendre, 2001).
However, for bacteria, temperature has been shown in meta-analyses to positively
correlate with bacterial growth rates, which are in turn positively correlated with
bacterial production and growth efficiency (delGiorgio and Cole, 1998; Kirchman et
al., 2009). To further complicate the story, however, it has also been proposed that
some fast-growing marine bacterial OTUs may be able to increase their growth rate
by sacrificing efficiency while in high-nutrient and high-DOC conditions (Halewood
et al., 2012).
Within the Southern California Bight, temperature and nitrate are strongly
negatively correlated (Lucas et al., 2011; Chapter 4). The high correlation between
lower temperatures and higher nutrients indicates that this last hypothesis may
apply for the Mod1 bacterial OTUs during and after phytoplankton blooms.
Alternatively, positive correlation between temperature and BGE for Mod1 OTUs
may also be a result of higher labile DOC production during bloom senescence,
which would follow restratification and temperature increases (Halewood et al.,
Teel
102
2012; Wear et al., 2015). These phenotypic strategies may coexist as well, resulting
in intra-module oscillation in relative abundance.
3.4.3 BGE and viral dynamics
To further examine biological drivers of community differentiation, viral
abundance over the time-series was compared with bacterial growth rates, BGE, and
BGE of high-Mod2 and low-Mod2 communities (Figure 3.17). The relationship
between viral abundance and growth rate or efficiency for marine bacterial
communities is currently not well understood. Theoretically, viral infection and cell
lysis should be associated with reductions in BGE through the viral utilization of
bacterial cellular machinery, which would increase the upkeep costs of the cell,
increasing respiration relative to growth (Thingstad et al., 2014). Laboratory
experiments have shown both increased growth rate and decreased growth
efficiency in response to increased viral abundance (Midelboe et al., 1996);
however, mesocosm and in-situ viral abundance has also been positively correlated
with bacterial abundance and higher chlorophyll values, which would indicate a
potential positive correlation between viral abundance and BGE (Riemann et al.,
2000; Cram et al., 2015). At SPOT, BGE was positively correlated with viral
abundance (R
2
0.13, P-value <0.01) for the whole surface time-series, indicating that
the latter explanation may apply (Figure 3.17a). The positive correlation between
BGE and viral abundance was much stronger in high-Mod2 communities (R
2
0.21, P-
value <<0.01) indicating that the membership of the community and the stability of
the environmental drivers may affect this correlation as well (Figure 3.17d).
Teel
103
3.4.4 Linear Regression of BGE at SPOT
A step-wise multiple linear regression analysis was done to determine the
primary factors contributing to BGE variability in the SPOT MO dataset (Figure
3.18). Using integrated primary production temperature and percent Mod2
abundance alone, 50% of the variability in BGE could be explained (P-value <0.01).
where integrated primary production was the strongest single predictor for BGE at
the SPOT site, explaining 43% of observed variance. The combination of
temperature and percent abundance of Mod2 explained an additional 7% of the
variance (P-value <0.01). The equation for predicted BGE was as follows:
𝐵𝐺𝐸= 0.00018∗𝑃𝑃𝑖𝑛𝑡+0.0042∗𝑇−0.0018∗𝑀𝑜𝑑2%−0.0053
where PPint is defined as the integrated primary production over the maximum
euphotic depth (90m), T is temperature in degrees Celsius, and Mod2% is the
percent abundance for Mod2 OTUs. Linear regression using percent abundance of
Mod1 community abundance explained a similar of variance (45%) within the
dataset due to the strong negative correlations between Mod1 and Mod2
abundance. These results demonstrate that, even though Mod2 abundance is
negatively correlated with primary production, variance of bacterial community
function (e.g. BGE) cannot be fully represented by primary production alone.
3.5 Discussion
This study bridges between theory and correlative analyses to explore
hypothesized mechanistic bacterial function within time-series data. Individual
results from this study specifically reinforce the theories that marine bacterial
Teel
104
community function can be tied to community abundance, and that “black box”
algorithms cannot represent this variability. Tying these results to bacterial carbon
demand through BGE showed that even at a single location bacterial carbon
dynamics can vary greatly with environmental and biological drivers. This study
suggests that at least two bacterial communities, a fast-growing opportunistic
bacterial community, similar to Mod1, and a slow-growing oligotrophic community,
similar to Mod2, would be required to capture basic bacterial carbon uptake
dynamics.
Here we leveraged correlation network analyses (Cram, 2014) in order to
quickly identify two largely independent groups of bacterial OTUs that displayed
distinct seasonal dynamics at SPOT. Mod1 abundance was negatively correlated
with Mod2 abundance and temperature, and positively correlated with chlorophyll
a concentration, and integrated primary production. In general, samples with high
percent abundance of Mod1 had higher BGE and displayed a weak positive
correlation between BGE and temperature. This relationship between BGE and
temperature was supported by recent field studies in the Santa Barbara Channel
(Halewood et al., 2012; Wear et al., 2015), but contradicted both the common Q10
theory of increased respiration with increased temperature as well as a commonly
cited global analysis of bacterial respiration versus temperature (Rivkin and
Legendre, 2001).
The membership within the Mod1 community, as well as its environmental
and functional correlations, indicated that this module could be hypothetically
described as a community of bloom-associated OTUs that flourish in relatively
Teel
105
mesotrophic conditions. Though Mod1 OTUs could be expected to increase in
abundance during and after spring blooms, the stochastic and ephemeral nature of
phytoplankton blooms would cause a large degree of variance in how specific OTU
abundance would correlate to overall bacterial abundance. Higher overall
community BGE could be a result of both specific phenotypic variation within the
community or the increased lability of DOC compounds during and after blooms
(Carlson et al., 2004; Halewood et al., 2012; Teeling et al., 2012; Wear et al., 2015;
Ward et al., 2017). Given the tight coupling between bacterial and protistan OTUs
observed in previous studies, the mechanism is likely to be a combination of these
factors (Sher et al., 2011; Chow et al., 2013; Chow et al., 2014; Needham et al.,
2016); however, additional measurements of both community composition on finer
time-scales and DOC production over the course of a bloom are necessary in order
to tease the relative impact of these two mechanisms apart.
The second module, Mod2, represented a more traditionally oligotrophic set
of OTUs, including multiple members of the SAR11 and SAR86 clades (Figure 3.1).
As would be expected for an oligotrophic community, the abundance of these
organisms was relatively constant year round, though with a slight increase during
the fall when temperatures were highest and the mixed layer depths were most
shallow, representing a more heavily stratified oceanographic state. Mod2 percent
abundance was also negatively correlated with BGE, BP, and primary production.
These results describe a community of slow-growing bacterial OTUs with high
respiration costs relative to production. Low BGE in this community could be a
result of reduced influx of labile-DOC compounds, higher respiration costs due to
Teel
106
higher average temperatures, or due to high maintenance costs, such as defense
from viral and grazing mortality (delGiorgio and Cole, 1998; Rivkin and Legendre,
2001; Yooseph et al., 2010; Winter et al., 2010; Halewood et al., 2012; Kirchman,
2016). The connection between viral abundance and BGE in communities with
higher Mod2 abundance may be related to this maintenance cost as well, since anti-
viral defenses would increase maintenance costs and reduce efficiency (Yooseph et
al., 2010; Kirchman, 2016).
These two groups of bacterial OTUs, behaving in general terms as an
oligotrophic set and an opportunistic set of bacterial OTUs, have been commonly
described in studies of marine ecosystems (Yooseph et al., 2010; Winter et al., 2010;
Kirchman, 2016; Ward et al., 2017). Importantly, however, in this study these
communities co-exist, are commonly found throughout the year, and are not
inherently prescribed by their environmental drivers alone. The Mod2 community is
not restricted to oligotrophic conditions or warm temperatures, and Mod1 OTUs
appear within the free-living bacterial community even during periods of low
primary production. The coexistence of these alternating communities combined
with the variation in primary production throughout the year produced highly
variable community BGE, ranging from 2% to 35% -- nearly the entire range of
global marine BGE measurements. The coexistence over time also suggests that the
observed seasonal dynamics were not simply driven by alternating physical
dynamics due to the weakening and strengthening of the Southern California Eddy,
which dominates the oceanographic circulation patterns within the San Pedro
Channel.
Teel
107
For the purpose of modeling bacterial remineralization of DOC produced
during phytoplankton blooms, one of the most interesting results from this study
was the one-month lag in bacterial abundance, specifically Mod1 abundance,
following the spring bloom (Figure 3.12). This result is supported by observations of
increased DOC production towards the end of spring blooms in the Santa Barbara
Channel (Halewood et al., 2012; Wear et al., 2015) and implies that the biological
interactions between phytoplankton and bacterial OTUs may be much more
complex than a monthly time-series can fully resolve. The rapid transition between
phytoplankton and bacterial OTUs has been shown in multiple field studies (Teeling
et al., 2012; Klindworth et al., 2014; Needham et al., 2016); however to date the
suite of DOC compounds produced during phytoplankton blooms has not been
synchronously sampled with community observations to explore how changes in
both the extracellular release and lability of phytoplankton exuded DOC may be
driving the bacterial community dynamics.
3.6 Conclusions
As the modeling community moves toward added realism within marine
ecosystem models, the results from this study indicate that one of the most crucial
but continuously ignored factors in bacterial carbon modeling is the wide variation
in BGE that can be observed both globally and at a single location. Restricting BGE to
a fixed value that approximates respiration costs for an average bacterial cell
creates bias within the modeled remineralization dynamics. As a first-order
approximation, BGE could be modeled as a function of primary production. The
Teel
108
correlations between community composition, bacterial production, growth rates,
and calculated BGE observed in this study indicate that a more mechanistic
understanding of both biological and environmental drivers of community
composition is necessary in order to accurately model bacterial remineralization.
Here we used correlative network analyses coupled with statistical analyses of
potential mechanistic relationships to understand the drivers of bacterial dynamics
at SPOT. As the global microbial community composition dataset for oceanic time-
series sites grows, similar network analyses could be created and utilized to study
how function and community co-vary globally.
Acknowledgments
We would like to acknowledge NSF and the University of Southern California for
funding this research. This study was completed in close collaboration with the
Fuhrman Laboratory, and the SPOT bacterial community dataset would not exist
without countless hours of work put in by its members. SPOT cruise data, collected
by Troy Gunderson, the Fuhrman Laboratory, the Caron Laboratory, and the crew of
the Yellowfin, were made available for this research by Diane Kim and the Wrigley
Institute for Environmental Sciences.
Teel
109

Figure 3. 1 Surface Modules, Modified from Cram, 2014
Two statistically significant negatively correlated modules, Mod1 and Mod2 were
identified through extended local similarity analysis (eLSA). Circles represent
bacterial OTUs, as defined by automated ribosomal intergenic-spacer analysis
(ARISA), sampled at 5m depth at the San Pedro Ocean Time-series Station (SPOT)
from 2000-2011 as part of the Microbial Observatory project. Squares represent
ancillary variables, both environmental and biological. Dashed lines indicate
negative correlations, and solid lines indicate positive correlations. Arrows indicate
a time-lag associated with the positive or negative correlation. The size of each
bacterial OTU, circle, indicates the average abundance of that member throughout
the time-series. Abbreviations are as follows in Table 1 and Table 2. The Mod2
module was further subdivided into Mod2A and Mod2B for correlative analyses.

Teel
110
Abbreviation
AEG169
Altero
Bacter
Plastid
Cronob
Cyanob
E01-9C
Flavob
Fluvii
Formos
Hyd24-
Marino
Microb
Nitros
NS2b
NS5
NS9
OCS116
OCS155
OM43
OTU
Owenwe
PAUC34
Piscir
Pro
Pro_HL(I)
Pro_LL(IV)
Rhodob
Roseob
SAR11
SAR116
SAR324
MGA
SAR86
Shewan
Sva099
Thioba
Thioth
ZD0405
ZD0417
Parameter
AEGEAN-169
Alteromonas
Bacteriovoraceae
Chloroplast
Cronobacter
Cyanobacteria
E01-9C-26
Flavobacteriaceae
Fluviicola
Formosa
Hyd24-01
Marinoscillum
Microbacteriaceae
Nitrospina
NS2b
NS5
NS9
OCS116
OCS155
OM43
Unidentified OTU
Owenweeksia
PAUC34f or Marine Group A
Piscirickettsiaceae
Prochlorococcus
Prochlorococcus High Light Strain (I)
Prochlorococcus Low Light Strain (IV)
Rhodobacteraceae
Roseobacter
SAR11
SAR116
SAR324
Marine Group A/ SAR 406
SAR86
Shewanella
Sva0996
Thiobacillus
Thiothrix
ZD0405
ZD0417
Taxonomy Note
SAR11 Clustered
Gammaprotobacteria
Deltaproteobacteria

Gammaproteobacteria

Gammaprotobacteria

Flavobacteria
Flavobacteria
Gammaprotobacteria
Sphinobacteria
Actinobacteria
Deltaproteobacteria
Flavobacteria
Flavobacteria
Flavobacteria
Alphaproteobacteria
Actinobacteria
Betaproteobacteria

Flavobacteria

Gammaproteobacteria
Cyanobacteria

Alphaproteobacteria
Alphaproteobacteria
Alphaproteobacteria
Alphaproteobacteria
Gammaprotobacteria

Gammaproteobacteria
Gammaproteobacteria
Actinobacteria
Betaproteobacteria
Gammaproteobacteria
Gammaprotobacteria
Gammaprotobacteria
Table 3. 1 Operational Taxonomic Unit Abbreviations, from Cram, 2014
ARISA-defined free-living bacterial OTUs used within the eLSA study of samples
taken at 5m depth between 2000 and 2011 at the San Pedro Ocean Time-series
Station. Abbreviations correspond to names used in Figure 1.

Teel
111

Abbreviation
Bact
bc.shift
CDOM
Chl_A_Sat
CHL-A
Cmax_Depth
DL
DDL
Depth
Elapsed_Days
InvSimpson
Leu
MEI
MLD
NO2
NO3
Oxygen
PAR
PielouJ
PO4
Prim_Prod
P*
qAmoA
qArch
qGroup1
RDate
Richness1000
Salinity
st.shift
ShannonH
SiO3
SSHD
Temperature
Thy
TurnoverLeu
TurnoverThy
Upwelling
VBR
Vir
WaveHeight
σθ
Parameter
Bacterial Abundance
Bray Curtis Rate of Community Change
Colored Dissolved Organic Matter
Monthly Mean Satellite Chlorophyll a
Chlorophyll A Bottle Concentration
Depth of Chlorophyll Maximum
Day Length
Day Length Rate of Change
Depth
Elapsed Days since Study Began
Inverse Simpson Biodiversity Index
Bacterial Production (Leucine)
Multivariate El-Niño Souther Oscillation Index
Mixed Layer Depth
Nitrite Concentration
Nitrate Concentration
Oxygen Concentration
Photosynthetically Active Radiation
Peilou’s Evenness Index
Phosphate Concentration
Satellite Estimated Primary Production
P-Star (N:P Ratio Below Redfield)
AmoA Gene Abundance
Archaeal Abundance
Archaea Group I Abundance
Date
Richness (OTUs with >0.1% Abundance)
Salinity
Change in Salinity or Temperature (Euclidean Distance)
Shannon Biodiversity Index
Silicate Concentration
Sea Surface Height Differential
Temperature
Bacterial Production (Thymidine)
Bacterial Turnover Time (Leucine)
Bacterial Turnover Time (Thymidine)
Bakun Upwelling Index For 33˚N -119˚W
Virus to Bateria Ratio
Viral Abundance
Mean Wave Height on Sampling Date
Sigma-Theta, Seawater Density
Collection Note

MODIS Aqua

MODIS Aqua

MODIS Aqua

Table 3. 2 Ancillary Abbreviations, from Cram, 2014
Ancillary variables (squares) used within the eLSA study of samples taken at 5m
depth between 2000 and 2011 at the San Pedro Ocean Time-series Station.
Abbreviations correspond to names used in Figure 1.

Teel
112

Figure 3. 2 Satellite PAR data, MODIS Aqua
Daily MODIS Aqua 9km photosynthetically active radiation (PAR) data for the SPOT
site from 2002-2013 are shown here in both histogram (top) and scatter versus
Julian day (bottom). PAR values were directly utilized for the estimation of depth-
resolved primary production at the SPOT site.

Teel
113

Figure 3. 3 Chlorophyll Calibration
Comparison of chlorophyll a fluorescence values with bench-top fluorometer
measurements from concurrent bottle samples. Calibrated fluorescence data is
shown in red, raw data is shown in blue, and the 1:1 ratio line is displayed in solid
black. Values of chlorophyll fluorescence below 500 m were assumed to be
instrumental noise and subtracted from the overall fluorescence profile for dark
calibration. Afterwards, linear regression between bottle and in-situ fluorescence
data was used for calibration. Calibration factors showed that the fluorescence data
was approximately 3x higher than bottle data from 1998 through early 2003 and
approximately 1.2x higher than bottle data from late 2003 through 2014.

Teel
114

Figure 3. 4 Depth Resolved Seasonal Primary Prediction
Vertically-resolved seasonally-averaged values of daily primary production between
the surface and 90 m from 2002 through 2013 at the San Pedro Ocean Time-series
Station (SPOT) are shown here. Primary production per meter was calculated as a
function of surface photosynthetically-active radiation, day length, the calibrated
vertical chlorophyll a profile, and a regionally tuned carbon fixation rate (Jacox et al.,
2015). Total integrated primary production was highest on average in the spring
season, consistent with previous results. Subsurface primary production contributes
significantly to integrated primary production during summer months. Regionally,
this subsurface primary production contributes to mismatches between satellite
and in-situ estimation of primary production (Jacox et al., 2015; Teel et al., in prep).
On average the depth of the production maximum follows the same seasonal pattern
as the depth of the nitricline, which is deepest during fall months and most shallow
during winter months.

Teel
115

Figure 3. 5 Leucine to Thymidine Incorporation Ratios
Recent findings off the coast of Oregon, reproduced above from delGiorgio et al.
(2011) (top panels) were contrasted with those from the surface and subsurface
chlorophyll maximum (SCM) at the SPOT site (lower panels). Sampling by
delGiorgio et al. (2011) showed significant correlations between the uptake ratio of
3H-Leucine and 3H-Thymidine when compared with both bacterial growth rate and
bacterial growth efficiency; however, when the uptake ratio was compared with
growth rate at the San Pedro Ocean Time-Series station, no strong correlation
existed. After calculating BGE at SPOT using the delGiorgio and Cole 1998 algorithm,
no correlation was found between uptake ratio and BGE either. The lack of
consistent correlation between growth rate and uptake ratio across studies may be
explained by the range of observed growth rates in each study, reported here in per
hour change. At SPOT growth rates represent data collected over a ten-year time-
series and the range of growth rates has a much lower average than those observed
off of the coast of Oregon. The samples reported by delGiorgio et al., 2011 were also
collected during a short period of time during one season, while the SPOT dataset
spans all seasons. Theoretically, the uptake ratio should represent the ratio of new
protein to new DNA, and given that DNA replication is thought to increase at a
higher rate during cell division relative to protein building, a negative correlation
between uptake ratio and growth would be logical (delGiorgio et al., 2011);
however, in the SPOT dataset the overall variance is more likely driven by
community composition changes, allowing for variable uptake ratios that may not
correspond with growth or BGE.

Teel
116

Figure 3. 6 White et al. (1991) Predicted Growth Rates at the SPOT Site
The commonly referenced algorithm published by White et al. in 1991 relating
bacterial growth to chlorophyll a concentration and temperature was applied to the
data collected at the San Pedro Ocean Time-series Site (SPOT). The predicted
growth rates did not fit the SPOT dataset regardless of whether the growth rates at
SPOT had been calculated using 3H-Thymidine incorporation (TdR, right panel) or
3H-Leucine incorporation (Leu, left panel). Overall, the predicted values did not
correlate with the measured values, and the range of the predicted values was
approximately four times too low. The lack of relationship was observed for
bacterial production calculated both from 3H-Thymidine incorporation rates and
from 3H-Leucine incorporation rates (Leucine: R2 0.0042, P-value 0.5; Thymidine:
R2 0.0189, P-value 0.23).

Teel
117

Figure 3. 7 Linear Regressions of Bacterial Production and Growth
Stepwise linear regression was used to estimate bacterial production and bacterial
growth rate as functions of biological and environmental parameters. The
correlation between bacterial abundance (BA) and bacterial production (BP) (BA
and BP calculated from Leucine incorporation: R2 0.40, P-value 0.0001; Thymidine
based: R2 0.56, P-value 3.2e-9) allowed for a function of BA, day length (dirr),
chlorophyll a concentration (CHL), and the 3H-Leucine to 3H-Thymidine ratio
(Leu:TdR) to predict BP within surface samples at SPOT. Predictive multiple linear
regressions explained 35% of explained variance in the BP calculated from 3H-
Leucine incorporation (P-value 7.4e-9) and 42% of the variance in the BP calculated
from 3H-Thymidine incorporation (P-value 6.3e-10). Predictions of growth rate (µ)
from dirr, Leu:TdR, and nitrate (NO3) were not strongly correlated with observed
values (µ from Leucine incorporation: R2 0.05, P-value 0.04; µ from Thymidine
incorporation: R2 0.18, P-value 0.0001).
Teel
118

Figure 3. 8 Temperature versus Bacterial Production and Growth
Temperature is commonly assumed to regulate bacterial community production and
growth (Kirchman et al., 2009 and references therein); however at the SPOT site the
relationships between temperature and bacterial production (BP) and between
temperature and growth rate (µ ) were investigated and no strong correlation was
found. In culture studies, increased temperature has been correlated with increased
growth and production up to the maximum tolerated temperature threshold
(Ratkowsky et al., 1983). In-situ elevated production and growth appears to be
correlated with values between 14˚C and 15˚C, which indicates that available
nitrate, which is negatively correlated with temperature regionally, may be a
stronger predictor of growth at the SPOT site.

Teel
119

Figure 3. 9 Seasonality of Bacterial Production
Bacterial production was compared with Julian day and day length at SPOT to assess
the seasonality of bacterial dynamics at SPOT. The highest bacterial production
values were observed during the Spring, corresponding with the regional upwelling
period and associated increased regional primary production. On average, bacterial
production remained elevated throughout the summer. The variance observed
within bacterial production samples increased with day length, indicating that
longer days may be associated with the relaxation of an energetic limitation, such as
temperature or light limitation.
Teel
120

Figure 3. 10 Correlation between Bacterial and Phytoplankton Measurements
The correlations between bacterial production (BP) and chlorophyll a concentration
(CHL) and bacterial growth efficiency (BGE) and integrated primary production
(PPint) were compared for all SPOT surface samples from 2000-2011. In this
dataset, the correlation between BP and CHL was significantly weaker than the
correlation between BGE and PPint. The rate parameters, BGE and PPint, are
logically more interconnected due to their direct relationship with DOC cycling
(delGiorgio and Cole, 1998). The relationship between BP and CHL has been
referenced in many models, though, and is often assumed to be a strong, positive,
linear correlation. These results indicate that the in-situ measurement of CHL is a
poor predictor of bacterial growth or function at SPOT.
Teel
121

Figure 3. 11 BGE versus Diversity, Evenness, and OTU Richness
Four diversity metrics were assessed in relation to the calculated bacterial growth
efficiency (BGE) for each surface sample from 2000-2011 at SPOT. The correlation
between diversity and activity is debated for marine bacterial communities
(Campbell and Kirchman, 2013; Galand et al., 2015). At the SPOT site no correlation
was found between BGE and any metric of diversity.

Teel
122

Figure 3. 12 Average Monthly Bacterial Abundance and Satellite Chlorophyll
Monthly average abundance of bacterial cells, surface module 1 (Mod1) cells, and
surface module 2 (Mod2) cells at SPOT from 2000-2011 were compared with the
monthly-averaged daily chlorophyll a concentration, estimated from the MODIS
Aqua satellite from 2002-2011. Bacterial abundances were determined through
microscopic cell counts (Cram, 2014; Cram et al., 2015). Mod1 and Mod2
abundances were calculated from individual OTU percent abundances, which were
calculated from automated ribosomal intergenic spacer analysis (ARISA) (Cram,
2014; Cram et al., 2015). Overall, Mod1 and Mod2 compose an average of 42.1% of
the bacterial community (min 10.0%; max 73.4%) and the abundance of Mod1 is
negatively correlated with the abundance of Mod2. Mod1 abundance is positively
correlated with bacterial abundance. Bacterial abundance also experiences a one-
month lag relative to the spring-bloom peak in satellite chlorophyll a that occurs
between March and April.
Teel
123

Figure 3. 13 Growth Rate and Abundance of Mod1 and Mod2
Growth rate, as calculated from bacterial production divided by bacterial
abundance, was compared with the percent abundances of bacterial OTUs belonging
to either surface module 1 (Mod1) or module 2 (Mod2) from Cram, 2014 (Figure 1).
Overall Mod2 abundance was weakly negatively correlated with growth rate. There
was also a much smaller range of growth rates observed when Mod2 abundance
was high or Mod1 abundance was low. These results indicated that the bacterial
community function is more predictable during periods of high Mod2 abundance.

Teel
124

Figure 3. 14 Mod2 Subgroups and Cyanobacterial Abundance
Surface module 2 (Mod2) was originally designated as a group of bacterial OTUs
that negatively correlated with surface module 1 (Mod1) bacterial OTUs (Cram,
2014), and not all of the Mod2 OTUs were strongly positively correlated with each
other. Two major subsets of Mod2, Mod2A and Mod2B, were analyzed relative to
Mod2 percent abundance and the percent abundance of cyanobacterial OTUs to
investigate the statistical relevance of each sub-module. Only heterotrophic
members of Mod2A and Mod2B were compared with cyanobacterial abundance to
avoid any potential self-correlation. Mod2 was strongly correlated with the
abundance of Mod2B (R2 0.73, P-value 4.5e-30; vs. Mod2A R2 0.10, P-value 0.001).
Mod2 groupB was positively correlated with cyanobacterial relative abundance (R2
0.06, P-value 0.01), while no relationship was observed for Mod2 groupA (R2 5.9e-4,
P-value 0.8). Mod2B percent abundance was also generally at least twice as high as
Mod2A percent abundance, indicating that it statistically dominated Mod2
correlations.
Teel
125

Figure 3. 15 BGE for High and Low Mod2 Communities
Percent abundance of surface modules 1 and 2, Mod1 and Mod2, were compared
with bacterial growth efficiency (BGE) estimates, calculated from measured
bacterial production within surface samples from the SPOT site between 2000 and
2011. Mod2 percent abundance was found to be significantly negatively correlated
with BGE (R
2
0.18; P-value << 0.01), and Mod1 percent abundance was found to be
weakly positively correlated with BGE (R
2
0.02; P-value 0.15). Though these
correlations were generally weak, a two-sample T-test showed that high Mod2 and
low Mod2 community BGE distributions were significantly different from one
another (p-value < 0.01); this result was not biased by the choice of community,
since high Mod1 and low Mod1 communities also displayed significantly different
BGE distributions (p-value 0.04).
Teel
126

Figure 3. 16 Community Dependence of BGE versus Temperature
No significant correlation was found between calculated bacterial growth efficiency
(BGE) and temperature for surface samples collected at the SPOT site from 2000-
2011. (Top panel; R2 0.02, P-value 0.18). High and low percent abundance were
determined using the average abundance for each module. When the samples were
divided into high Mod2 percent abundance (>24.7%) and low Mod2 percent
abundance (<24.7%), temperature was weakly positively correlated with BGE for
low Mod2 samples (bottom left panel: R2 0.13, P-value 0.03) and was weakly
negatively correlated with BGE for high Mod2 samples (bottom right panel: R2 0.05,
P-value 0.26). Furthermore, the average BGE was significantly greater for samples
with low percent abundance of Mod2 (0.13) relative to those with high Mod 2
abundance (0.09) (two sample t-test: P-value <0.01).
Teel
127

Figure 3. 17 Viral Abundance and Bacterial Growth Efficiency
Viral abundance was compared with (a) community growth rate (R2 0.02, P-val
0.17), (b) calculated bacterial growth efficience (BGE) for the whole community (R2
0.13 p-val 5.0e-4), (c) BGE for samples with low Mod2 percent abundance (<24.7%)
(R2 0.19 p-val 2.0e-3), and (d) BGE for samples with high Mod2 percent abundanct
(>24.7%) (R2 0.21 p-val 6.6e-3). Though viral abundance showed no strong
correlation with growth rate, it was positively correlated with BGE. The correlation
was further improved when either low or high Mod2 samples were examined
individually.
Teel
128

Figure 3. 18 Predicted BGE as fn(PPint, T, %Mod2)
Through step-wise multiple linear regression analysis, bacterial growth efficiency
(BGE) was found to be most accurately predicted as a function of integrated primary
production, temperature, and percent abundance of Module 2 (Mod2) operational
taxonomic units.
Teel
129
References
Anderson T.R. and P.J. le B. Williams. (1999). A one-dimensional model of dissolved
organic carbon cycling in the water column incorporating combined biological-
photochemical decomposition. Global Biogeochemical Cycles. 13(2):337-349.

Anderson T.R. and H.W. Ducklow (2001). Microbial loop carbon cycling in ocean
environments studied using a simple steady-state model. Aquat. Microb. Ecol. 26:37-
49.

Allen, J. I., J. R. Siddorn, J. C. Blackford, and F. J. Gilbert. (2004). Turbulence as a
control on the microbial loop in a temperate seasonally stratified marine systems
model. Journal of Sea Research. 52:1-20.

Azam F., D.C. Smith, G.F. Steward, and A. Hagström. (1993). Bacteria-organic matter
coupling and its significance for oceanic carbon cycling. Microb. Ecol. 28:167-179.

Beman, J. M., J. A. Steele, and J. A. Fuhrman (2011). Co-occurrence patterns for
abundant marine archaeal and bacterial lineages in the deep chlorophyl maximum
of coastal California. The ISME Journal. 5: 1077-1085.

Campbell B.J. and D.L. Kirchman. (2013). Bacterial diversity, community structure,
and potential growth rates along an estuarine salinity gradient. ISME J. 7:210-220.

Carlson C.A., S.J. Giovannoni, D.A. Hansell, S.J. Goldberg, R. Parsons, and K. Vergin.
(2004). Interactions among dissolved organic carbon, microbial processes, and
community structure in the mesopelagic zone of the northwestern Sargasso Sea.
Limol. Oceanogr. 49(4):1073-1083.

Chow, C.-E. T., R. Sachdeva, J.A. Cram, J.A. Steele, D. M. Needham, A. Patel, A.E. Parada,
and J. A. Fuhrman. 2013. Temporal Vairability and Coherence of Euphotic Zone
Bacterial Communities over a Decade in the Southern California Bight. The ISME
Journal. 7:2259-2273.

Chow C.-E. T., D. Y. Kim, R. Sachdeva, D. A. Caron, and J.A. Fuhrman. 2014. Top-down
Controls on Bacterial Community Structure: Microbial Network Analysis of Bacteria,
T4-like Viruses, and Protists. The ISME Journal. 8:816-829.

Cram, J.A. (2014). Dynamics of marine bacterial communities from surface to
bottom and the factors controlling them. (PhD Dissertation). University of Southern
California.

Cram, J.A. , C.-E. T. Chow, R. Sachdeva, D. M. Needham, A. E. Parada, J. A. Steele, and J.
A. Fuhrman. 2015. Seasonal and Interannual Variability of the Marine
Teel
130
Bacterioplankton Community Throughout the Water Column over Ten Years. The
ISME Journal. 9:563-580.

del Giorgio, P.A. and J.J. Cole. (1998). Bacterial growth efficiency in natural aquatic
systems. Annu. Rev. Ecol. Syst. 29:503-541.

del Giorgio, P. A., R. Condon, T. Bouvier, K. Longnecker, C. Bouvier, E. Sherr, and J. M.
Gasol. (2011). Coherent patterns of bacterial growth, growth efficiency, and leucine
metabolism along a northeastern Pacific inshore-offshore transect. Limnology and
Oceanography. 56(1):1-16.

Doney, S.C., I. Lima, J. K. Moore, K. Lindsay, M. J. Behrenfeld, T. K. Westberry, N.
Mahowald, D. M Glover, and T. Takahashi. (2009). Skill metrics for confronting
global upper ocean ecosystem-biogeochemistry models against field and remote
sensing data. Journal of Marine Systems. 76:95-112.
Dong, C. M., E. Y Idica, and J. C. McWilliams. 2009. Circulation and multiple-scale
variability in the Southern California Bight. Progress in Oceanography. 82: 168-190.
Fasham, M.J.R., H. W. Ducklow, and S.M. McKelvie. 1990. A Nitrogen-based Model of
Plankton Dynamics in the Oceanic Mixed Layer. Journal of Marine Research. 48:591-
639.
Fuhrman, J. A., Ammerman, J. W., & Azam, F. (1980). Bacterioplankton in the coastal
euphotic zone: distribution, activity and possible relationships with
phytoplankton. Marine Biology, 60(2), 201-207.

Fuhrman, J. A., I. Hewson, M. S. Schwalbach, J. A. Steele, M. V. Brown, and S. Naeem.
2006. Annually Reoccurring Bacterial Communities are Predictable from Ocean
Conditions. Proceedings of the National Academy of Sciences. 103(35)13104-13109.
Fuhrman, J. A., J. A. Cram, and D. M. Needham. (2015). Marine microbial community
dynamics and their ecological interpretation. Nature Reviews Microbiology. 13:139-
146.

Galand, P. E., I. Salter, and D. Kalenitchenko. Ecosystem productivity is associated
with bacterial phylogenetic distance in surface marine waters. Molecular Ecology.
24:5785-5795.

Gilbert, J. A., D. Field, P. Swift, S. Thomas, D. Cummings, B. Temperton, K. Weynberg,
S. Huse, M. Hughes, I. Joint, P. J. Somerfield, and M. Mühling. (2010). The taxonomic
and functional diversity of microbes at a temperate coastal site: a ‘multi-omics’
study of seasonal and diel temporal variation. PLoS One. 5:e15545.

Teel
131
Gilbert, J. A., J. A. Steele, J. G. Caporaso, L. Steinbrück, J. Reeder, B. Temperton, S.
Huse, A. C. McHardy, R. Knight, I. Joint, P. Somerfield, J. A. Fuhrman, and D. Field.
(2012). Defining seasonal marine microbial community dynamics. The ISME Journal.
6:298-308.

Hagström, A., F. Azam, A. Andersson, J. Wikner, and F. Rassoulzadegan. (1988).
Microbial loop in an oligrotrophic pelagic marine ecosystem: possible roles of
cyanobacteria and nanoflagellates in the organic fluxes. Marine Ecology – Progress
Series. 49:171-178.

Halewood, E.R., C.A. Carlson, M.A. Brzezinski, D.C. Reed, and J. Goodman. (2012).
Annual cycle of organic matter partitioning and its availability to bacteria across the
Santa Barbara Channel continental shelf. Aquat. Microb. Ecol. 67:189-209.

Hasumi, H. and T. Nagata. (2014). Modeling the global cycle of marine dissolved
organic matter and its influence on marine productivity. Ecological Modelling. 288:9-
24.

Jacox, M. G., C. A. Edwards, M. Kahru, D. L. Rudnick, and R. M. Kudela. 2015. The
Potential for Improving Remote Primary Productivity Estimates through Subsurface
Chlorophyll and Irradiance Measurement. Deep-Sea Research II. 112:107-116.

James, A. K., U. Passow, M. A. Brzezinski, R. J. Parsons, J. N. Trapani, and C. A. Carlson.
(2017). Elevated pCO2 enhances bacterioplankton removal of organic carbon. PLoS
ONE 12(3): e0173145. doi:10.1371/journal.pone.0173145joo
Jiao, N.-Z., G.J. Herndl, D.A. Hansell, R. Benner, G. Kattner, S.W. Wilhelm, D.L.
Kirchman, M.G. Weinbauer, T.-W. Luo, F. Chen, and F. Azam. (2010). Microbial
production of recalcitrant dissolved organic matter: long-term carbon storage in the
global ocean. Nature Reviews: Microbiology. 8:593-599.

Jones, A. C., T. S. V. Liao, F. Z. Najar, B. A. Roe, K. D. Hambright, and D. A. Caron.
(2013). Seasonality and disturbance: annual pattern and response of the bacterial
and microbial eukaryotic assemblages in a freshwater ecosystem. Environmental
Microbiology. 15(9):2557-2572.

Kirchman, D. L., X. A. G. Moran, and H. Ducklow. (2009). Microbial growth in the
polar oceans – role of temperature and potential impact of climate change. Nature
Reviews Microbiology. 7:451-459.

Kirchman, D. L. (2016). Growth Rates of Microbes in the Oceans. Annual Review of
Marine Science. 8:285-309.

Klindworth, A., A. J. Mann, S. Huang, A. Wichels, C. Quast, J. Waldmann, H. Teeling,
and F. O. Glöckner. (2014). Diversity and activity of marine bacterioplankton during
Teel
132
a diatom bloom in the North Sea assessed by total RNA and pyrotag sequencing.
Marine Genomics. 18:185-192.

Laufkötter, C., M. Vogt, N. Gruber, O. Aumont, L. Bopp, S. C. Doney, J. P. Dunne, J.
Hauck, J. G. John, I. D. Lima, R. Seferian, and C. Völker. (2015). Projected decreases in
future marine export production: the role of the carbon flux through the upper
ocean ecosystem. Biogeosciences Discussions . 12:19941-19998.

Lucas, A. J., C. L. Dupont, V. Tai, J. L. Largier, B. Palenik, and P. J. S. Franks. (2011). The
green ribbon: Multiscale physical control of phytoplankton productivity and
community structure over a narrow continental shelf. Limnology and Oceanography.
56(2):611-626.

Moore, J. K., S. C. Doney, J. A. Kleypas, D. M. Glover, and I. Y. Fung. (2002). An
intermediate complexity marine ecosystem model from the global domain. Deep-Sea
Research II.49:403-462.

Moore, J. K., S. C. Doney, K. Lindsay. (2004). Upper ocean ecosystem dynamics and
iron cycling in a global three-dimensional model. Global Biogeochemical Cycles. 18,
GB4028, doi:10.1029/2004GB002220.

Needham, D. M. , C-E T. Chaw, J. A. Cram, R. Sachdeva, A. Parada, J. A. Fuhrman. 2013.
Short-term observations of marine bacterial and viral communities: patterns,
connections and resilience. The ISME Journal . 7:1274-1285.

Needham, D.M. and J. A. Fuhrman. 2016. Pronounced Daily Succession of
Phytoplankton, Archaea, and Bacterial Following a Spring Bloom. Nature
Microbiology. 1:1605. doi:10.1038/nmicrobiol.2016.5.

Noble, M.A., Rosenberger, K.J., Hamilton, P., and Xu, J.P., 2009, Coastal ocean
transport patterns in the central Southern California Bight, in Lee, H.J., and Normark,
W.R., eds., Earth Science in the Urban Ocean: The Southern California Continental
Borderland: Geological Society of America Special Paper 454, p. 193–226, doi:
10.1130/2009.2454(3.3).
Oey, L.-Y. 1999. A Forcing Mechanism for the Poleward Flow off the Southern
California Coast. Journal of Geophysical Research. 104(C6):13529-13539.

Pomeroy, L.R. and D. Deibel. (1986). Temperature regulation of bacterial activity
during the spring bloom in Newfoundland coastal waters. Science. 233:359-361.

Ratkowsky, D. A., R. K. Lowry, T. A. McMeekin, A. N. Stokes, and R. E. Chandler.
(1983). Model for Bacterial Culture Growth Rate Throughout the Entire Biokinetic
Temperature Range. Journal of Bacteriology. 154(3):1222-1226.

Teel
133
Riemann, L., G. F. Steward, and F. Azam (2000). Dynamics of Bacterial Community
Composition and Activity during a Mesocosm Diatom Bloom. Applied and
Environmental Microbiology. 66(2):578-587.

Rivkin, R.B. and L. Legendre. (2001). Biogenic Carbon Cycling in the Upper Ocean:
Effects of Microbial Respiration. Science. 291:2398-2400.

Sprintall, J. and M. Tomczak. 1992. Evidence of the Barrier Layer in the Surface
Layer of the Tropics. Journal of Geophysical Research. 97(C5):7305-7316.

Steele, J.A., P. D. Countway, L. Xia, P.D. Vigil, M. Beman, D. Y. Kim, C.-E. T. Chow, R.
Sachdeva, A. C. Jones, M. S. Schwalbach, J. M. Rose, I. Hewson, A. Patel, F.-Z. Sun, D. A.
Caron, and J. A. Fuhrman.

Strub, P.T., James, C., 2000]. Altimeter-derived variability of surface velocities in the
California Current System: 2. Seasonal circulation and eddy statistics. Deep Sea
Research Part II: Topical Studies in Oceanography 47, 831–870. 10.1016/S0967-
0645(99)00129-0.
Teel, E. N., X. Liu, B. N. Seegers, M. A. Ragan, B. H. Jones, and N. M. Levine.
Contextualizing time-series data: Using high-resolution in-situ gliders to quantify
short-term regional oceanographic variability in the San Pedro Channel. In
Preperation.

Teeling H., B.M. Fuchs, D. Becher, C. Klockow, A. Gardebrecht, C.M. Bennke, M.
Kassabgy, S. Huang, A.J. Mann, J. Waldmann, M. Weber, A. Klindworth, A. Otto, J.
Lange, J. Bernhardt, C. Reinsch, M. Hecker, J. Peplies, F.D. Bockelmann, U. Callies, G.
Gerdts, A. Wichels, K.H. Wiltshire, F.O. Glöckner, T. Schweder, and R. Amann. (2012).
Substrate-controlled succession of marine bacterioplankton populations induced by
a phytoplankton bloom. Science. 336: 608-611.

Thingstad, T.F. and R. Lignell. (1997). Theoretical models for the control of bacterial
growth rate, abundance, diversity, and carbon demand. Aquat. Microb. Ecol. 13:19-
27.

Walsh, J.J., D.A. Dieterle, F.E. Muller-Karger, R. Bohrer, and W.P. Bissett. (1999).
Simulation of carbon-nitrogen cycling during spring upwelling in the Cariaco Basin.
Marine Science Faculty Publications. Paper 17.

Ward, C. S., C-M. Yung, K. M. Davis, S. K. Blinebry, T. C. Williams, Z. I. Johnson, and D.
E. Hunt. (2017). Annual community patterns are driven by seasonal switching
between closely related marine bacteria. The ISME Journal. 1-11.

Wear, E. K., C. A. Carlson, L. A. Windecker, and M. A. Brzezinski (2015). Roles of
diatom nutrient stress and species identity in determining the short- and long-term
Teel
134
bioavailability of diatom exudates to bacterioplankton. Marine Chemistry. 117:335-
348.

Wear, E. K., C. A. Carlson, A. K. James, M. A. Brzezinski, L. A. Windecker, and C. E.
Nelson. (2015). Synchronous shifts in dissolved organic carbon bioavailability and
bacterial community responses over the course of an upwelling-driven
phytoplankton bloom. Limnology and Oceanography. 60:657-677.

Weitz, J.S., C.A. Stock, S.W. Wilhelm, L. Bourouiba, A. Buchan, M.L. Coleman, M.J.
Follows, J.A. Fuhrman, L.F. Jover, J.T. Lennon, M. Middelboe, D.L. Sonderegger, C.A.
Suttle, B.P. Taylor, T.F. Thingstad, W.H. Wilson, and K.E. Wommack. (2014). A
multitrophic model to quantify the effects of marine viruses on microbial food webs
and ecosystem processes. The ISME Journal . 1-13.

White, P. A., J. Kalff, J. B. Rasmussen, and J. M. Gasol. (1991). The Effect of
Temperature and Algal Biomass on Bacterial Production and Specific Growth Rate in
Freshwater and Marine Habitats. Microbial Ecology. 21:99-118.

Winter, C., T. Bouvier, M. G. Weinbauer, and T. F. Thingstad. (2010). Trade-offs
between competition and defense specialists among unicellular planktonic
organisms: the “kill the winner” hypothesis revisited. Microbial and Molecular
Biology Reviews. 74(1):42-57.

Xiu P and F. Chai. (2014). Connections between physical, optical, and
biogeochemical processes in the Pacific Ocean. Progress in Oceanography. 122:30-
53.

Yooseph, S. K. H. Nealson, D. B. Rusch, J. P. McCrow, C. L. Dupont, M. Kim, J. Johnson,
R. Montgomery, S. Ferriera, K. Beeson, S. J. Williamson, A. Tovchigrechko, A. E. Allen,
L. A. Zeigler, G. Sutton, E. Eisenstadt, Y-H. Rogers, R. Friedman, M. Frazeier, and J. C.
Venter. (2010). Genomic and fuctional adaptation in surface ocean planktonic
prokaryotes. Nature. 468:60-67.

Teel
135
Chapter 4. Modeling the effects of free-living marine bacterial
community composition on heterotrophic remineralization rates
and biogeochemical carbon cycling
Abstract
The dissolved organic carbon (DOC) reservoir in the ocean is approximately
equal to that of atmosphere CO2. Cycling of DOC within the ocean is largely
controlled by heterotrophic marine prokaryotes, however, these organisms are
absent from nearly all global ocean ecosystem models. While our understanding of
marine microbial ecology has improved rapidly with the development of enhanced
molecular techniques and rate measurements, the majority of these ecological
advancements have not been incorporated into ocean ecosystem models. Bacterial
community dynamics were incorporated into a classical zero-dimensional mixed
layer nutrient-phytoplankton-zooplankton-detritus (NPZD) model to test the
biogeochemical importance of explicitly representing mechanistic bacterial
dynamics in an ocean ecosystem model. The model was run for a test site (San
Pedro Ocean Time-Series site) forced with in situ observations and remote sensing
products.
Both the base NPZD and NPZD+bact models captured reasonable seasonal
cycles of phytoplankton biomass, growth, and grazing observed at the site. Bacterial
dynamics enhanced regenerated production and allowed for the partial decoupling
of bacterial production and primary production. Picophytoplankton growth and
grazing were most strongly affected by the inclusion of explicit bacterial dynamics,
indicating that picophytoplankton dynamics in the base model have been tuned to
Teel
136
compensate for the absence of dynamic bacterial remineralization and bacterial
grazing. The observed seasonal fluctuations between oligotrophic and opportunistic
bacterial biomass could only be recreated with the inclusion of improved DOC
dynamics, including a proxy for DOC lability. The base and bacterial models
exhibited significantly different ecosystem and biogeochemical responses to
reductions in seasonal upwelling flux, predicting opposite changes in
picophytoplankton fall biomass under increased stratification. We show that the
inclusion of bacterial dynamics substantially, and non-linearly, impacts regenerated
production such that the base model, which could be tuned to capture these missing
dynamics under present day conditions, resulted in significantly different dynamics
under changing conditions. Given the sensitivity of modeled picophytoplankton to
the inclusion of explicit bacterial dynamics, it is likely that estimates of carbon
cycling in oligotrophic regions would be most impacted by the inclusion of explicit
bacterial dynamics in global ocean ecosystem models. The decoupling of primary
production and bacterial production also impacts the time-scales of
remineralization in regions of high productivity, potentially extending the duration
of blooms and altering carbon export out of the surface ocean.
4.1 Introduction
Heterotrophic remineralization in the surface ocean is dominated by the
activity of prokaryotic microbes; however, current global ocean ecosystem models
lack explicit and mechanistic bacterial dynamics. Instead, bacterial heterotrophic
activity is represented as simple loss equations applied to standing stocks of
dissolved and particulate organic matter (Moore et al., 2002; Moore et al., 2004;
Teel
137
Doney et al., 2009). In smaller-scale or steady-state ocean ecosystem models,
explicit bacterial dynamics have been incorporated in order to improve specific
output products. Models that predict DOC reactivity or partitioning often include a
simple mechanistic bacterial biomass pool that actively converts DOC between
labile, semi-labile, and refractory pools (Anderson and Williams, 1999; Walsh et al.,
1999; Hasumi and Nagata, 2014). Oceanic viral dynamics and microbial loop
models rely upon explicit bacterial carbon and growth to support viral and
zooplankton biomass (Hagström et al., 1988; Thingstad and Lignell, 1997; Anderson
and Ducklow, 2001; Allen et al., 2004; Weitz et al., 2014). Other models have
incorporated explicit bacterial biomass in order to better estimate ocean-optics (Xiu
and Chai, 2014), C:N ratios (Walsh et al., 1999), and competition between bacteria
and phytoplankton over inorganic nitrogen sources (Bratbak and Thingstad, 1985;
Fasham et al., 1990; Walsh et al., 1999; Weitz et al., 2014). In each of these models,
bacterial biomass has been treated as a single group (‘black box’) with average
parameters for Michaelis-Menton growth on organic matter and simplistic
mortality. The majority of the models also used a fixed growth efficiency with only
three models implementing variable bacterial growth efficiency as a function of
phosphorous or nitrogen availability (Bratbak and Thingstad, 1985; Fasham et al.,
1990; Vichi et al., 2007a; Vichi et al., 2007b; Vichi et al., 2009). To our knowledge, no
ocean ecosystem model has modeled variable bacterial community dynamics in
attempts to simulate variable bacterial heterotrophic function in the surface ocean.
The advancement of molecular techniques have allowed for direct
observations of fluctuation in bacterial community composition and genetic
Teel
138
functional potential with time, environment, and biological parameters. This new
understanding of marine microbial ecosystem dynamics challenges the assumption
that bacterial heterotrophic dynamics can be accurately represented by a simple
loss equation dependent only upon the available organic matter and temperature.
Here we will argue that bacterial dynamics including variable growth, grazing,
substrate affinity and growth efficiency, are necessary for accurately capturing the
microbial loop and regenerated production. Specifically, observed oceanic bacterial
growth efficiencies (BGE) vary between 1% and 58% as a function of temperature,
community composition, DOC lability, and DOC source (delGiorgio and Cole, 1998;
Rivkin and Legendre, 2001; Halewood et al., 2012; Sinsabaugh, et al., 2013; James et
al., 2017; Chapter 3). With variable BGE, remineralization and export could be better
depicted, improving the timescales and mapping of primary production and DOC
dynamics within global models.
The diversity of microbes that make up phototrophic and heterotrophic
communities in the surface ocean is immense and has been shown to change rapidly
in response to the physical, chemical, and biological dynamics (Teeling et al., 2012;
Klindworth et al., 2014; Galand et al., 2015; Behrenfeld et al., 2016; Needham and
Fuhrman, 2016). While characterizing this diversity can elucidate intricate
ecological dependencies, some degree of reduction of ecosystem dynamics is
necessary in any global model as the vast diversity of marine prokaryotic
metabolisms, cell sizes, substrate affinities, and growth parameters is too complex
to incorporate into ocean ecosystem models. In this study, we propose that a
tractable solution to the incorporation of variable and mechanistic bacterial
Teel
139
heterotrophic dynamics into ocean ecosystem models lies in the ability to define
bacterial community types, which can be done through network and correlation
analyses. We focus on a case study from the San Pedro Ocean Time-series (SPOT)
site where the free-living heterotrophic community can be divided into two
functional groups, a fast-growing opportunistic community and a slow-growing
oligotrophic community (Chapter 3).
Just as phototrophic plankton species have long been classified into two
broad categories, fast-growing opportunitrophs and slow-growing oligotrophs, a
similar classification can be applied to free-living heterotrophic marine bacterial
communities (Chapter 3; Yooseph et al, 2010; Kirchman, 2016; Ward et al, 2017).
The distinction that we make here is that we define a community, a composite of
diverse species, as either opportunistic or oligotrophic. This allows us to focus on
the aggregated biogeochemical function of the diverse community and how it varies
from other heterotrophic communities without having to explicitly represent each
individual bacterial species. The division of the heterotophic bacteria into distinct
subgroups is supported by in situ and mesocosm experiments from oceanic and
coastal environments. Though members of communities appear to be present year-
round, the relative abundances of the community aggregates are negatively
correlated with each other and the fluctuation in community dynamics has been
associated with changes in community function as well as season and
phytoplankton dynamics (Ward et al., 2017; Chapter 3). In general, slow-growing
oligotrophic bacteria dominate the community under low-biomass, low-nutrient
conditions, where bacterial growth efficiencies are estimated to be minimal.
Teel
140
Conversely, fast-growing opportunistic bacteria are thought to be grazed to small
numbers in marine ecosystems except during periods of high organic matter
availability, when they can outcompete the oligotrophs (del Giorgio and Cole, 1998;
Winter et al., 2010; Yooseph et al., 2010).
Heterotrophic bacterial community dynamics are known to be driven by
both ‘bottom-up’ (DOC cycling) and ‘top-down’ (grazing) processes. Increased DOC
production, both by increased zooplankton sloppy feeding and through direct
extracellular release by phytoplankton, has been linked to enhanced bacterial
abundance (Fuhrman et al., 1980; Hagström et al., 1988; Wear et al., 2015). As the
dissolved compounds exuded by phytoplankton can vary greatly (Becker et al.,
2014), it has been hypothesized that changes in PER and DOC composition may be
important drivers of variation in bacterial community composition (Halewood et al.,
2012; Teeling et al., 2012; Chow et al., 2013; Kim et al., 2014; Klindworth et al.,
2014; Wear et al., 2015; Ward et al., 2017). In addition, research has shown that
bacterial grazing (‘top-down’ control) will selectively reduce the abundance of large
fast-growing bacterial cells opening up a niche for smaller slow-growing bacterial
cells and altering both community composition and total bacterial biomass (Hahn
and Hofle, 1990; Kirchman 2000; Kirchman 2009; Winter et al., 2010; Yooseph et al.,
2010; Cram et al., 2016).
In this study, we leverage the network and statistical analyses of the SPOT
microbial observatory data to add two dynamic bacterial heterotrophic functional
groups and improved DOC cycling into a nutrient-phytoplankton-zooplankton-
detrital model (NPZD). By comparing the model with explicit bacterial community
Teel
141
dynamics to a base model that uses fixed remineralization rates, we can specifically
address the following research questions: (1) What role do dynamic and fluctuating
heterotrophic bacterial communities play in biogeochemical cycling? (2) Does
explicitly including heterotrophic bacterial dynamics affect the predictive power of
ocean ecosystem models? (3) How have current ocean ecosystem models
compensated for the absence of explicit bacterial dynamics? Lastly, (4) which
oceanographic regions are most likely to be impacted by the incorporation of
explicit mechanistic bacterial dynamics within ocean ecosystem models?
4.2 Methods
4.2.1 Base Ecological Model
The bacterial model developed in this study builds off of the NCAR CESM
nutrient-phytoplankton-zooplankton-detritus (NPZD) model with some minor
modification, here after referred to as the ‘base model’ (Figure 4.1; Moore et al.,
2002; Doney et al., 2009; Levine, submitted). The model tracked two phytoplankton
functional groups, a large ‘diatom-like’ phytoplankton and a ‘cyanobacteria like’
pico-plankton, one zooplankton group, nitrate, ammounium, dissolved organic
carbon (DOC), and particular organic carbon (POC). Phytoplankton growth rates
were calculated as:
(eq 4.1)
!"!!"#
!"
= 𝑖𝑔𝑟𝑜𝑤𝑡ℎ𝐶− 𝑖𝑙𝑜𝑠𝑠
where iloss is a combined mortality term describing losses from respiration,
grazing, and cell lysis, and igrowth C is defined as:
(eq 4.2) 𝑖𝑔𝑟𝑜𝑤𝑡ℎ𝐶= 𝜇
!"#
𝛾
!
𝛾
!
𝛾
!
𝑖𝑝ℎ𝑦𝑡𝑜𝐶
Teel
142
such that the maximum phytoplankton growth rate (µmax) is limited by nitrate and
ammonium (𝛾
!
), temperature (𝛾
!
), and light (𝛾
!
) (Table 4.1; Moore et al., 2002;
Doney et al., 2009; Levine, submitted). Nutrient limitation was modeled with
Michaelis-Menton kinetics. Temperature limitation was calculated following
Dutkiewicz et al., (2012) as:
(eq 4.3) 𝑇𝑓𝑢𝑛𝑐=max 0.33∗ 1.04
!
∗𝑒
!!"∗ !!!"#$
−0.3 ,0.01 ;
where T is the mixed layer temperature, Topt represents the optimum temperature
for phytoplankton growth, and Tb corresponds to the total range of temperatures at
which the phytoplankton can survive. Tb and Topt were adjusted to represent the
ranges and optimum observed for large and small phytoplankton in the Southern
California Bight; where the range and optimum for large phytoplankton was 7 ˚ -
25˚C and 15°C, respectively, and the range and optimum for picophytoplankton was
15 - 40˚C and 22°C, respectively. Available photosynthetically active radiation
(PAR) was calculated as:
(eq 4.4) 𝑃𝐴𝑅= 𝑖𝑃𝐴𝑅 ∗
!!!
!!
!
∗ !
!
!
∗!

where iPAR is surface incident light, updated hourly, and Kz is the attenuation
coefficient at depth z calculated using the Southern California Bight validated
relationship (Parsons et al., 1984; Jacox et al. 2015) of:
(eq 4.5) 𝐾
!
= 0.04+0.0088∗𝑐ℎ𝑙
!
+0.054∗𝑐ℎ𝑙
!
!.!"

where chlz is the concentration of chlorophyll in mmol per cubic meter.
Remineralization of labile DOC was assumed to occur instantaneously in the
model (i.e. POC was converted directly to DIC) while the remineralization of semi-
Teel
143
labile DOC and POC was modeled at a fixed rate of 0.1 per day. Zooplankton grazing
(grazeC) was calculated as:
(eq 4.6) 𝑔𝑟𝑎𝑧𝑒𝐶= 𝑖𝑧𝑜𝑜𝐶 ∗𝑔𝑟𝑧𝑚𝑎𝑥 ∗ (
!"!!"#$
!"!!"#$!!"#!
)
Where izooC is the zooplankton biomass, grzmax is the maximum grazing rate,
iphytoC is the phytoplankton biomass, and zgrz is a half-saturation constant for
grazing; the values of both grzmax and zgrz vary by phytoplankton functional type
(Table 4.1; Moore et al., 2002; Doney et al. 2009; Levine, submitted).
4.2.2 Bacterial Model
The bacterial model modified three aspects of the base model: 1) improved
DOC cycling 2) explicit representation of two bacterial functional groups 3) slight
modification of phytoplankton growth and grazing parameters (Table 4.1).
1.1 4.2.2.1 DOC cycling
In the ocean, DOC exists in a continuum between labile DOC and refractory
DOC. In situ observations suggest that DOC is re-set during winter mixing events to
low concentrations of semi-labile DOC that is mixed into the surface ocean from
depth (Hansell and Carlson, 2009). The DOC pool then builds up during the spring
and summer as a result of increased primary production. One approach is to sub-
divide DOC into labile, semi-labile, and refractory pools (eg., Anderson and Williams,
1999). Here we propose an alternative approach. We use the winter-time seasonal
DOC minimum value as an implicit way of representing differences in DOC lability.
Specifically, when the total DOC is at or below this threshold, it is assumed to be
composed entirely of semi-labile DOC. When DOC is above this threshold it is
Teel
144
assumed to be composed of a combination of labile and semi-labile compounds. We
set the seasonal threshold was to 54 mmol m
-3
based off of regional data for the
Southern California Bight (Table 4.1; Halewood et al., 2012).
To represent the gradual change in lability of DOC over the season that has
been observed in the Southern California Bight (Halewood et al., 2012; Wear et al.,
2015), we add a scalar to the labile DOC pool, which is a proxy for the age of the
DOC. Specifically, labile DOC is defined as:
(eq 4.7) 𝐷𝑂𝐶
!"#$!%
=𝛼 𝐷𝑂𝐶−𝐷𝑂𝐶
!"#!$%#&

where
(eq 4.8) 𝑎=min 1,max 𝑎
!"#
,
!"#
!"#$%

where αmin is the minimum threshold for α, iER is the DOC produced through
extracellular release by large phytoplankton within the model model (see below),
and ERmax represents the maximum of iER and was set to 0.2. As oligotrophic
bacteria are thought to have higher affinity for semi-labile DOC, αmin was set to 1 for
oligotrophic bacteria (such that α always equaled 1) and 0.1 for opportunistic
bacteria.
A more significant change to the model was the explicit inclusion of DOC exudation
as an additional component of primary production. Phytoplankton growth in the
base model was represented as the change in cellular carbon content with time
(d(iphytoC)/dt). These biomass growth rates have been largely based on and
validated with comparison to either cellular growth rates in the lab or
14
C–POC-
based primary production measurements in the field (Moore et al., 2002; Doney et
al., 2009). These measurements do not capture the extracellular release of labile-
Teel
145
DOC from phytoplankton that is thought to be quickly remineralized by bacteria. As
a result, this carbon was not tracked in the base model as it would transfer directly
from DIC back to DIC; however, this labile DOC plays a critical role in the microbial
loop and has been hypothesized to be an important mechanism in driving the timing
of bacterial dynamics (Halewood et al., 2012; Teeling et al., 2012; Wear et al., 2015).
Here we explicitly represent the direct release of extracellular DOC as an additional
pool of labile DOC. In order to maintain the same rate of phytoplankton biomass
growth (igrowthC ) as the base model, release of extracellular DOC was added to
gross primary production such that:
(eq 4.9) 𝑖𝑃𝑃𝑡𝑜𝑡𝐶= 𝑖𝑔𝑟𝑜𝑤𝑡ℎ𝐶+ 𝑎𝑢𝑡𝑜𝑃𝐸𝑅𝐷𝑂𝐶
where autoPERDOC is the extracellular DOC which was modeled as a function of PER
(percent extracellular release) of gross primary production.
Studies of PER have shown that PER dynamics vary between phytoplankton
functional types and within a single phytoplankton OTU in response to nutrient
stress (Baines and Pace, 1991; Nagata, 2000; Halewood et al., 2012; Wear et al.,
2015a; Wear et al., 2015b). In addition, PER has been shown to be inversely related
to primary production with the highest PER values found in oligotrophic
environments or produced by oligotrophic cyanobacteria (Berman and Holm-
Hansen, 1974; Baines and Pace, 1991; Nagata, 2000; Becker et al., 2014). Here PER
release was modeled as a function of per-cell nitrogen stress. Specifically, the PFT
specific percent extracellular release of DOC was modeled as a sinusoidal function of
this stress as:
Teel
146
(eq 4.10) 𝑖𝑃𝐸𝑅=
!"#$!
!"#
!
∗ sin
!"#$%%!!"#
!"#$!
!"#$!!
!"
−
!"
!
+𝑃𝐸𝑅
!"#
+
!"#$!
!"#
!

(eq 4.11) 𝑆𝑡𝑟𝑒𝑠𝑠=
!"!!"#$
(!"#!!!"#!)

(eq 4.12) 𝑅𝑎𝑛𝑔𝑒
!"#
=𝑃𝐸𝑅𝑚𝑎𝑥−𝑃𝐸𝑅min
(eq 4.13) 𝑅𝑎𝑛𝑔𝑒
!"#$!!
= 𝑚𝑎𝑥𝑆𝑡𝑟𝑒𝑠𝑠−𝑅𝑒𝑝
where Stress is the per-cell Nitrogen stress experienced by each PFT, PERmax and
PERmin represent the maximum and minimum PER values associated with that PFT,
maxStress is the value of Stress above which the phytoplankton will exhibit PERmax,
and Rep is the value of Stress at which conditions are theoretically replete and PER
values would be minimum. PERmin was set at 5%, which is similar to rates observed
in replete laboratory conditions. Picophytoplankton maximum PER was set to 30%,
similar to what has been observed in oligotrophic regions (Berman and Holm-
Hansen, 1974; Nagata, 2000; Becker et al., 2014). For large phytoplankton in the
model, the maximum value of PER was set to 20%, similar to values observed in the
Santa Barbara Channel (Table 1; Wear et al., 2015).
Finally, in the base model, 85% of all DOC produced from phytoplankton and
zooplankton loss was routed directly into the DIC pool as immediately respired
labile DOC (Moore et al., 2002; Doney et al., 2009; Levine, submitted). In reality, this
loss term is the combined effect of phytoplankton and zooplankton respiration and
the release and remineralization of labile DOC. To explicitly account for these
dynamics, the direct DIC loss from phytoplankton and zooplankton was minimized
to represent only the respiration ratios for phytoplankton (10%) and zooplankton
Teel
147
(10%) and the remaining DOC was released so that it was available for uptake by
the heterotrophic bacteria.
1.2 4.2.2.2 Bacterial Community Dynamics
Two explicit heterotrophic bacterial groups were added to the model, a fast
growing opportunotrophic community and a slower growing oligotrophic
community, both of which were designed to represent active free-living marine
bacterial heterotrophs in the surface ocean (Figure 4.2). The oligotrophic and
opportunistic communities resemble reoccurring alternating communities that have
been linked to variable community function as well as composition (Chapter 3;
Fuhrman et al., 2006; Yooseph et al., 2010; Cram et al., 2015; Ward et al., 2017).
Their parameterization included variable grazing, growth, and bacterial growth
efficiency (Table 4.1). Bacterial growth was modeled using Michaelis-Menten
kinetics as:
(eq 4.14) 𝐵𝑃
!
= µ
!"# (!)
!"#
!"#$!%
!"#
!"#$!%
!!
!"#(!)
𝐵𝑎𝑐𝑡
!

where BPi is the bacterial growth for group i, Bacti is bacterial biomass for group i,
µmax is the maximum growth rate for group i, DOClabile is the concentration of labile
DOC, and KDOC(i) is the half-saturation constant for DOC uptake by bacterial group i.
Here, µmax(i) was selected to represent maximal bacterial growth rates on labile DOC.
Change in bacterial biomass over time were defined as total bacterial
production minus losses from both grazing and non-grazing mortality.
(eq 4.15)
!"#$%
!"
=𝐵𝑃 − 𝑙𝑜𝑠𝑠
!"#$
−𝑔𝑟𝑎𝑧𝑒
!"#$

Teel
148
The non-grazing mortality does not include respiration, which is directly related to
BP through BGE as
(eq 4.16) 𝐵𝐶𝐷=
!"
!"#

and
(eq 4.17) 𝑖𝐷𝑂𝐶
!"#$%
= 𝐵𝑅=𝐵𝐶𝐷−𝐵𝑃
where BCD represents the total bacterial carbon demand, or uptake, iDOCremin is the
fraction of the DOC pool remineralized to DIC during this time-step by bacterial
uptake, BR is bacterial respiration, BP is production as described above, and BGE is
calculated at each time-step as a function of temperature and community.
Oligotrophic bacteria have a higher affinity for semi-labile DOC than
opportunotrophic bacteria but also have a lower maximum growth rate and growth
efficiency (Table 4.1; Chapter 3; Ducklow 2000, Lankiewicz et al., 2015; Kirchman
2016). Conversely, opportunistic bacteria have higher growth rates, higher growth
efficiency, and their abundance is thought to be controlled through high grazing
rates and the accessibility of labile DOC (Jooseph et al., 2010; Kirchman, 2016).
We parameterize these dynamics using the seasonal DOC threshold
(described above). Specifically, below the seasonal DOC threshold, oligotrophic
bacteria growth was modeled as:
(eq 4.18) 𝐵𝑃
!
= µ
!"#
∗𝐵𝑎𝑐𝑡
!

where BPi is the bacterial growth for group i, Bacti is bacterial biomass for group i,
and µmin is the minimum growth rate for group i. Above the seasonal DOC threshold
bacterial growth followed the Michaelis-Menten equations previously described.
Teel
149
Bacterial growth efficiency (BGE) has been shown to vary in situ from 1% in
some oligotrophic regions to as high as 50% in high-nutrient conditions (delGiorgio
and Cole, 1998; Rivkin and Legendre, 2001). BGE at SPOT was best predicted as a
function of primary production, community composition, and temperature (Chapter
3). In the bacterial model, the relationships observed at SPOT were used to
dynamically vary BGE as a function of temperature for each of the two bacterial
groups. The observed positive correlation between BGE and primary production
was an emergent property of the model, since labile DOC increased with increased
primary production, allowing for additional growth by the opportunistic bacterial
group, which has higher overall BGE. BGE as a function of temperature for each
group was described as follows:
(eq 4.19)
𝐵𝐺𝐸
!""!
=
!"#$%&'!
!""!
!
∗ sin
!"#!!"#!
!""!
!"#$%!
!""!
!"
−
!"
!
+𝐵𝐺𝐸𝑚𝑖𝑛
!""!
+
!"#$%&'!
!""!
!

(eq 4.20)
𝐵𝐺𝐸
!"#$!
=
!"#$%&'!
!"#$!
!
∗cos
!"#!!"#!
!"#$!
!"#$%!
!"#$!
!"
+𝐵𝐺𝐸𝑚𝑖𝑛
!"#$!
+
!"#$%&'!
!"#$!
!

where BGErange is the difference between the maximum and minimum possible
BGE values for each bacterial group, BGEmin is the minimum BGE value for each
group, MLT is the mixed layer temperature, Tmin is the temperature below which
BGE for each group is no longer variable, and Trange is the temperature range over
which BGE is variable for each bacterial group.
Teel
150
Bacterial loss included respiration, non-grazing mortality, and grazing.
Mortality due to viral lysis is implicitly included in the model as part of the fixed
non-grazing mortality rate (0.1 per day), based off of minimal estimates for viral
mortality from bacterioplankton (Fuhrman et al., 1999). This is likely an
underestimation of viral losses and is certainly an oversimplification of viral
dynamics (Fuhrman et al., 1999; Suttle, 2007); however, incorporation of further
viral dynamics was beyond the scope of this project. Variable grazing rates for
opportunistic and oligotrophic bacterial groups were designed to mimic the Cryptic
Escape and Kill The Winner hypotheses, which describe the tendency for
bacterivores to selectively graze larger faster-growing bacterial cells (Hahn and
Hofle, 1990; Winter et al., 2010; Yooseph et al., 2010; Cram et al., 2016). While
zooplankton grazing rates were higher on opportunistic bacteria than on
oligotrophic bacteria (Table 4.1), the routing of bacterial carbon through the
zooplankton carbon pool was equal for both bacterial groups and was based on
pico-phytoplankton grazing within the BEC model (Moore et al., 2002; Doney et al.,
2009). Bacterial grazing was modeled after phytoplankton grazing as:
(eq 4.21) 𝑔𝑟𝑎𝑧𝑒
!"#$
=𝑍µ
!"#
!"#$
!"#$!!"#
𝑍𝑜𝑜
(eq 4.22) 𝑙𝑜𝑠𝑠
!"#$
=𝑚𝑜𝑟𝑡
!"#$
𝐵𝑎𝑐𝑡
where grazebact is the amount of carbon lost from the bacterial biomass through
grazing, Zoo is the zooplankton biomass, Zµmax is the maximum grazing rate for each
bacterial group, Bact is the bacterial biomass, lossbact is the all non-grazing losses
from the bacterial biomass, mortbact is the fixed non-grazing mortality.
Teel
151
4.2.3 Physical Framework and Environmental Drivers
The ecological frameworks (described above) were embedded into a mixed
layer model with seasonally varying mixed layer depth, driven by observed mixed
layers from the SPOT site (Figure 4.3; Chapter 3). Temperature evolved
prognostically through heating and cooling and was calculated as:
(eq 4.23)
!"
!"
= ∆𝑇
!!
∗𝑖𝑓𝑥 + ∆𝑇
!!
∗𝑖𝑑𝑓𝑠 + 𝑤𝑎𝑟𝑚− 𝑐𝑜𝑜𝑙
where ∆Tbb is the difference in temperature between the mixed layer box and the
bottom boundary condition, ifx is proportional to upwelling flux, idfs is proportional
to the diffusive flux, warm is heating from absorbed solar radiation, and cool is
radiative cooling.
The model was forced with repeating climatology derived annual averages of
surface wind stress (National Centers for Environmental Prediction (NCEP)), air
temperature (NCEP), and PAR (MODIS Aqua) based off of all available data between
2000 and 2011 (Figure 4.3). Daily values of wind stress and air temperature were
used for radiative cooling. Satellite surface PAR from MODIS Aqua was converted
from daily average values in µEinsteins per square meter per day to watts per
square meter using daylength calculated according to methods described in
Forsythe et al. (1995) and Jacox et al. (2015). Daily average values of watts per
square meter were further converted to hourly values according to calculations
described in Stull (1988). Shortwave radiation was estimated from hourly PAR
values, under the assumption that PAR was 45% of total shortwave radiation
(Moore et al., 2002). Advective exchange with a bottom boundary condition was
Teel
152
driven by a seasonally varying upwelling flux calculated from the Bakun Upwelling
Index (33˚N and 119˚W) as:
(eq 4.24) 𝐹𝑙𝑢𝑥 = 𝑓+ (𝐵𝑎𝑘𝑢𝑛∗
!"#
!"
!
!
)
where Flux is in meters per day, Bakun values were originally in units of cubic
meters per second per 100 meters of coastline, Flux is in meters per day, spd is the
conversion factor for seconds per day, and f was a mean correction factor (f=0.3)
that accounted for uncertainty in the bottom boundary condition and horizontal
transport. With this flux, the model reproduced observed mixed layer temperatures
with an R
2
value of 0.89 (Figure 3).
A fixed bottom boundary value for dissolved organic carbon concentration
was derived from regional estimates (Halewood et al., 2012). Seasonally varying
bottom boundary temperatures were calculated from SPOT data as the average
value over the 10 meters beneath the mixed layer (Figure 4.3). Nitrate and
temperature are highly negatively correlated in the deep waters with temperatures
less than 14.5 ˚C in the Southern California Bight (Lucas et al., 2010; Seegers et al.,
2015; Figure 4.4). We use the SPOT data to derive a relationship between
temperature and nitrate for the bottom boundary condition where temperature was
less than 14°C where:
(eq 4.25) NO3 = -4.1*TEMP + 61
This is comparable to other reported regression between NO3 and Temperature for
the region (Lucas et al., 2010). When bottom temperatures were greater than 14˚C,
bottom nitrate was assumed to be zero.
Teel
153
4.2.4 Validation Dataset
Base and bacterial model output was compared against observations of
surface nitrate, mixed layer temperatures, bacterial growth rates, bacterial
production, bacterial abundance, phytoplankton growth rates, phytoplankton
grazing rates, bacterial grazing rates, and zooplankton biomass from the SPOT site
(Chapter 3; Cram et al., 2015; Caron et al., 2017; Connell et al., 2017). Because
monthly in situ sampling of chlorophyll at SPOT are likely to temporally under-
sample regional phytoplankton blooms (Chapter 2; Teel et al., in prep), seasonal
variation in phytoplankton carbon was estimated for SPOT using satellite
chlorophyll data (MODIS Aqua). Monthly averaged satellite chlorophyll values were
scaled by linear regression against in situ chlorophyll values and further converted
to phytoplankton biomass with a site-specific in situ carbon to chlorophyll
conversion factor (Haskell et al., 2017).
Cyanobacterial abundance was estimated from bacterial abundance
observed at the SPOT site multiplied by the percent abundance of cyanobacteria for
each sample, as determined through automated ribosomal intergenic spacer
analysis (ARISA) (Chapter 3; Cram et al., 2015). Cyanobacterial abundances were
then converted to biomass with published site-specific estimates for carbon per cell
of Prochlorococcus (90 fg per cell), carbon per cell of Synechococcus (200 fg per cell),
and the ratio (1 to 2) of Prochlorococcus to Synechococcus at the SPOT site (Caron et
al., 2017).
Key model parameters for phytoplankton and bacterial growth and grazing
were optimized using the SPOT datasets. Final values for optimized parameters are
Teel
154
given in Table 4.1, when not specified, default parameters from the NCAR CESM BEC
model were used.
4.3 Results
4.3.1 Phytoplankton Dynamics:
The base model was able to reproduce reasonable seasonal phytoplankton
dynamics (Figure 4.5) with no significant model drift. The phytoplankton biomass
predicted by the base model (annual average 1.7 mmolC m
-3
; range 0.2 to 2.9 mmol
m
-3
) is consistent with estimates from the SPOT site based on scaled satellite
chlorophyll data (annual average 1.9 mmol m
-3
) and flow cytometric data (annual
average 1.8 mmol m
-3
)(Figure 4.6; Haskell et al., 2016; Caron et al., 2017). The
modeled annual cycle for picophytoplankton biomass (annual average 0.4 mmol m
-3
,
range 0.1 to 1.0 mmol m
-3
) corresponds well with monthly mean values predicted
from cyanobacterial abundance at the SPOT site (annual average 0.8 mmol m
-3
;
range 0.1 to 1.5 mmol m
-3
) (Figure 4.6). The magnitude of mixed layer nitrate values
within the model (annual average 0.8 mmol m
-3
; range 1e-3 to 2.5 mmol m
-3
) is in
good agreement with average annual surface nitrate measurements from the SPOT
site (annual average 0.4 mmol m
-3
; range 0 to 1.9 mmol m
-3
), with peak values
occurring in the spring (Figure 4.5).
Rates of phytoplankton growth (µavg 0.44 day
-1
; range 0.2 to 0.8 day
-1
) and
grazing (mavg 0.15 day
-1
; range 0.02 to 0.34 day
-1
) within the base model were
within the ranges measured at the SPOT site (µavg 0.36 day
-1
; range -1.36 to 1.53
day
-1
) (Figure 4.8; Connell et al., 2017); however modeled zooplankton biomass is
Teel
155
lower (annual average 0.28 mmol m
-3
; range 0.04 to 0.72 mmol m
-3
) in the base
model than published zooplankton measurements for the SPOT site (annual average
0.91 mmol m
-3
; range 0.24 to 2.21 mmol m
-3
) (Figure 4.5; Caron et al., 2017). The
base model also showed slightly stronger coupling (R
2
0.79; P-value <0.01) with
phytoplankton biomass than in-situ (R
2
0.70; P-value <0.01) (Figure 4.5; Caron et al.,
2017). Low zooplankton biomass in the base model could be the result of a number
of factors. The absence of bacterial dynamics in this base model limits the prey-
availability for zooplankton grazing within the base model. Additionally, the SPOT
site has a persistent subsurface chlorophyll maximum, indicating that a
phytoplankton population below the shallow summer mixed layer may support
surface summer zooplankton populations (Chow et al., 2013; Cram et al., 2014;
Caron et al., 2017). With the mixed layer model used in this study, we do not resolve
subsurface populations nor do we account for vertical migration of zooplankton
grazers, such that modeled zooplankton biomass only represents the carbon added
to zooplankton within the mixed layer.
When dynamic heterotrophic bacteria were added to the base model without
re-optimizing the phytoplankton growth and grazing parameters, the
picophytoplankton biomass crashed due to overgrazing. Specifically, bacterial
grazing within the model supported a higher zooplankton biomass, which in turn
suppressed the picophytoplankton biomass and enabled large phytoplankton
dominance within the model. Because of the enhanced zooplankton biomass within
the bacterial model, for stable seasonal phytoplankton dynamics to be restored, the
maximum grazing rate parameters for picophytoplankton needed to be reduced by
Teel
156
50% (Table 4.1). In addition, the nitrate half-saturation uptake constant for large
phytoplankton was decreased slightly, yielding a higher affinity for nitrate, to offset
the increased grazing during spring. The ammonium half-saturation uptake constant
was increased (lower affinity for ammonium) in order to suppress diatom growth
on regenerated nitrogen from remineralized DOC during the summer and fall
stratified periods. After re-optimization, seasonal phytoplankton alternations, total
phytoplankton biomass (annual average 2.2 mmolC m
-3
; range 0.2 to 3.2 mmolC
m
-3
), cyanobacterial biomass (annual average 0.64 mmolC m
-3
; range 0.09 to 1.6
mmolC m
-3
), surface nitrate concentration (annual average 0.63 mmol NO3 m
-3
;
range 1e-2 to 1.86 mmol NO3 m
-3
), and phytoplankton growth (annual average 0.57
day
-1
; range 0.13 to 0.95 day
-1
), and grazing rates (annual average 0.23 day
-1
; range
0.04 to 0.46 day
-1
) within the bacterial model were all comparable with the base
model and observations (Figure 4.9; Figure 4.10).
4.3.2 Bacterial Dynamics
The bacterial model was able to capture seasonal alternation of oligotrophic
and opportunistic bacterial communities that has been observed at both coastal and
oceanic time-series sites (Figure 4.11; Figure 4.12; Yooseph et al., 2010; Cram, 2014;
Cram et al., 2014; Halewood et al., 2012; Wear et al., 2015; Ward et al., 2017). The
community dynamics resulted in variable bacterial growth efficiency, which was
modeled as a function of both community and temperature (Figure 4.12). Bacterial
biomass (annual average 0.8 mmolC m
-3
; range 0.2 to 1.3 mmol m
-3
), growth (annual
average 0.35 day
-1
; range 0.09 to 0.7 day
-1
), production (annual average 4.9e5 cells
Teel
157
ml
-1
day
-1
; range 2.4e5 to 7.3e5 cells ml
-1
day
-1
), growth efficiencies (annual average
22%; range 18% to 31%), and grazing rates (annual average 0.18 day
-1
; range 0.01
to 0.49 day
-1
) within the bacterial model were within the range of observed values
at the SPOT site (Figure 4.13; Figure 4.2; Cram et al., 2015; Connell et al., 2017).
However, on average, bacterial biomass within the model was low while bacterial
production, growth and efficiency were higher than the in-situ measurements
(Figure 4.13). This indicates that the bacterial pool was turning over too quickly
and that bacterial mortality due to grazing or viral lysis may have been
overestimated in the model.
The alternation of bacterial communities within the bacterial model was, to a
large extent, driven by variable grazing rates on the two communities. Opportunistic
growth (annual average 1.2 day
-1
; range 0.1 to 3.1 day
-1
) and grazing (annual
average 0.3 day
-1
; range 0.05 to 0.6 day
-1
) were higher overall than oligotrophic
growth (annual average 0.9 day
-1
; range 0.5 to 1.5 day
-1
) and grazing (annual
average 0.1 day
-1
; range 0.02 to 0.2 day
-1
). This resulted in opportunistic bacteria
responding more rapidly to increased nutrients but being grazed more heavily
during periods of higher stratification and lower nutrients. These grazing dynamics
are supported by mesocosm experiments performed near the SPOT site and by the
cryptic escape hypothesis, which theorizes that the low abundance and low growth
rates of oligotrophic bacteria are inherently a form of grazer defense because total
biomass is too low to be selectively grazed (Yooseph et al., 2010; Cram et al., 2016)
While variable growth and grazing sets up general seasonal oscillations in
the bacterial communities, more specific temporal ecological dynamics observed at
Teel
158
the SPOT site, and also in the Santa Barbara Channel, could only be captured by the
model with an improved representation of DOC dynamics (Figure 4.11; Chapter 3;
Halewood et al., 2012). Specifically, it has been shown that the C:N ratio of DOC
increases throughout the summer as the more labile lower C:N DOC is preferentially
remineralized (Halewood et al., 2012). In this model, the term α is a proxy for age of
the DOC pool (equations 4.7 and 4.8). α was used to decrease the opportunistic
bacterial growth rates such that when the DOC pool was older and more
recalcitrant, opportunistic bacterial growth rates were lower. This model dynamic is
supported by in-situ and laboratory studies, which indicate that the composition of
DOC varies seasonally and that DOC composition can directly affect bacterial
community abundance and metabolism (Halewood et al., 2012; Wear et al., 2015;
Ward et al., 2017).
The incorporation of improved DOC dynamics into the model provided a
more accurate representation of the observed ‘bloom’ in opportunistic bacteria
immediately following the bloom of large phytoplankton (Figure 4.11). In addition,
the bacterial model was able to capture the temporal decoupling of bacterial
production from primary production that has been observed at oligotrophic and
coastal sites, including the SPOT site (Figure 4.11; Wear et al., 2015; Viviani and
Church, 2017).
4.3.3 Sensitivity tests
A set of model experiments was conducted to determine the sensitivity of the
model dynamics to variable bacterial growth rates, percent extracellular release,
Teel
159
and the use of multiple bacterial groups. The model was most sensitive to changes in
percent extracellular release and the DOC-aging related reduction in opportunistic
bacterial growth rates (Figure 4.14). In addition, without this proxy for DOC lability,
the opportunistic bacterial community dominated throughout the year, supporting
higher zooplankton biomass throughout the summer as well. Without dynamic DOC
cycling, the timing of peak opportunistic bacterial abundance was shifted later in the
season towards higher temperatures, as the bacterial growth efficiency for
opportunistic bacteria increased with temperature.
Changes in bacterial growth efficiency and number of modeled bacterial
community types had limited effects on modeled large phytoplankton biomass,
small phytoplankton biomass, and zooplankton biomass (Figure 4.15; Figure 4.16).
In model runs in which variable growth efficiencies were replaced with fixed values
for the opportunistic and oligotrophic bacterial groups (akin to previous modeling
approaches), the seasonality of the bacterial oscillation was maintained by the PER-
related constraints; however, the bacterial biomass associated with each group
varied dramatically as a function of the efficiencies selected (Figure 4.15). These
results suggest that use of fixed bacterial growth efficiencies may be sufficient for
capturing observed bacterial dynamics. However, we demonstrate that the model
was very sensitive to the selected growth efficiency suggesting that fixing bacterial
growth efficiencies at the values that best predict present day biomass values at a
given location would reduce the predictive capabilities of the model across the wide
range of temperatures and nutrient availability that are seen globally. It also
questions the confidence of model predictions of future ecosystem dynamics. This is
Teel
160
particularly true since observed growth efficiencies vary by two orders of
magnitude.
To test the importance of multiple bacterial functional groups, the model was
run with only one functional type, opportunistic or oligotrophic . Simulations with
only oligotrophic bacterial dynamics were able to capture most of the seasonal
features produced by the model with both bacterial groups (Figure 4.16). The model
with only opportunistic bacterial dynamics tended to enhance zooplankton biomass
and repress summer phytoplankton growth via a reduction in regenerated nutrients
(Figure 4.16). These results indicate that there may be a way to represent bacterial
community dynamics within a single bacterial group that behaves primarily as an
oligotrophic functional group; perhaps with the ability to increase growth rates
when PER was high. However, as with the results from fixed growth efficiency, the
disparity between oligotrophic and opportunistic function would increase globally
as the ranges of environmental and biological drivers widen. By not explicitly
representing opportunistic bacteria from the model, it is likely that regions of high
upwelling production would experience over-stimulated nutrient remineralization
and under-represented zooplankton biomass.
4.4 Discussion
4.4.1 Impact of future changes
Heterotrophic bacteria in the surface ocean control the fate of a large portion
of phytoplankton-derived and land-derived organic carbon (Jiao et al., 2010; Bauer
et al., 2013); however, there is a trade-off between attempting to represent the
Teel
161
immense diversity of heterotrophic bacteria, being able to functionally describe this
diversity, and working within the computational limitations of global ecosystem
models. Therefore, the inclusion of bacterial dynamics into these models requires a
framework that captures the first order biogeochemical function of dynamic
bacterial communities and ensures maximum predictive power, while minimizing
the number of free parameters and the computational cost. Here we compare two
model formulations, one with and one without bacterial dynamics. Both models
were able to capture the seasonal dynamics and biomass values observed at the test
site (SPOT) after the optimization of several key ecosystem parameters. The base
model has also been shown in previous studies to globally represent present-day
mean global oceanic ecosystem dynamics (Moore et al., 2002; Moore et al., 2004;
Doney et al., 2009). However, the goals of this research were not only to determine
how bacterial heterotrophic community dynamics could be incorporated into
ecosystem models, but also to identify which features of the current NPZD models
are compensating for the lack of bacterial dynamics and how the inclusion of
bacterial dynamics might improve the predictive capability of these models.
As expected, bacterial dynamics significantly changed the rate of regenerated
production and so had the largest impact on picoplankton dynamics, which,
theoretically, support a larger fraction of their productivity through the uptake of
regenerated ammonium. We have shown that the model formulation that does not
include dynamic heterotrophic bacteria accounts for the impact of these missing
heterotrophic dynamics on picophytoplankton growth and grazing with by
modifying these parameter values.
Teel
162
To examine the potential impact of explicitly representing dynamic
heterotrophs on ecosystem model prediction of future changes in biogeochemical
cycling, the base and bacterial models were run with reducing seasonal flux (Figure
4.17; Figure 4.18). As flux was dampened, a proxy for increased stratification, the
large phytoplankton biomass was reduced non-linearly in both models. While the
sign of change was consistent between the base and bacterial models, the bacterial
model always predicted a larger decrease in large phytoplankton biomass as a
response to reduced flux (Figure 4.17). Critically, the shape of the response curves
in the summer and fall differ significantly between the two models, with the base
model showing a linear decrease in biomass in the fall with decreased flux, while the
bacterial model displays a non-linear response. For picophytoplankton, the
variation between model results is even more dramatic. The base model predicts
that picophytoplankton will gain an increasing advantage (increase abundance)
with increased stratification across all seasons in the base model. . This is consistent
with global climate model predictions that picophytoplankton will be favored under
future, warmer and more stratified, conditions. In the bacterial model, this group
gains an earlier advantage but then, under higher levels of stratification, the sign of
response in the summer and fall switches and the model predicts decreases in
picophytoplankton biomass (Figure 4.18). Mechanistically, the bacterial model is
capturing the shift from a fall picophytoplankton peak to spring picophytoplankton
peak, which is not captured by the base model (Figure 4.18). Furthermore, these
results highlight the importance of accurately representing dynamics in models and
Teel
163
cautions against predictions by models that get the ‘right answer for the wrong
reason’ particularly when underlying dynamics are non-linear.
The above sensitivity tests focused on changes in nutrient flux into the
surface ocean. Changes in upwelling inherently impact mixed layer temperatures,
with a 50% reduction in flux corresponding to a 2˚C increase in mixed layer
temperatures. Future change might result in enhanced warming, in addition to
reduced nutrient flux, which has the potential to further impact remineralization
rates through increases in bacterial growth efficiency in oligotrophic communities
(Chapter 3; Rivkin and Legendre, 2001).
Future changes in efficiency may be compounded by increases in pCO2, which
were recently shown to reduce bacterial growth efficiency (James et al., 2017). The
sensitivity of both the model results to BGE, and of BGE to environmental
parameters that are predicted to be impacted by future climate changes, indicates
that the modeling approach of using a fixed remineralization rate will become even
less applicable to the future ocean. Finally, recent work has suggested that photo-
heterotrophy might represent a significant source of energy acquisition in
oligotrophic gyres, and that this process might become increasingly important in the
future as seasonal photo-autotrophic primary production is significantly reduced in
highly-stratified regions (Gomez-Consarnau et al., submitted). The explicit inclusion
of bacterial dynamics is a necessary precursor for the representation of photo-
heterotrophy in ecosystem modeling. Our model results predict that regions with
extremely low primary production are likely to be strongly affected by the
incorporation of explicit bacterial dynamics in ocean ecosystem models; however,
Teel
164
because of the ecological lag between primary production and bacterial production
that has been observed both in situ and in our bacterial model, the temporal and
spatial scales for DOC-remineralization are likely to also be impacted by bacterial
dynamics in regions of high primary production.
4.4.2 Next Steps
Currently the DOC pool in the bacterial model represents both labile and
semi-labile DOC; however, only a seasonal threshold and the α parameter have been
utilized to characterize the changes in bacterial growth in response to varying
lability of the DOC (see Methods). To improve modeled DOC cycling both within the
bacterial model and when incorporated globally, DOC lability could be represented
as relative age, with the assumption that newer DOC will be more labile and have a
lower C:N ratio (Jiao et al., 2010; Halewood et al., 2012). Utilizing DOC age as a proxy
for lability would also allow for variable C:N ratios within DOC pools.
Another advantage of including a variable C:N within the DOC pools would be
to allow for nitrogen competition between phytoplankton and heterotrophic
bacteria, as has been modeled previously (Fasham et al., 1990). Currently, within
the bacterial model, heterotrophic bacteria acquire all nitrogen from the dissolved
organic matter, according Redfield ratios and proportional to the amount of
dissolved organic carbon uptake that is required for growth (see Methods); however,
it is known that marine bacteria do not, on average, have the same C:N ratio as
phytoplankton, and that heterotrophic bacteria can actively take up inorganic
nitrogen sources when under nitrogen-limited conditions (Kirchman, 2009). Due to
Teel
165
the assumption of the Redfield ratio across all carbon reservoirs in the model,
bacterial remineralization of nitrogen during the summer and fall is theoretically
higher in the bacterial model than it would be in situ. Incorporation of variable C:N
dynamics would allow for the bacterial biomass to utilize semi-labile DOC sources
with high C:N ratios and would restrict the amount of regenerated nitrogen that was
available after remineralization (Touratier et al., 1999).
This study has focused exclusively on free-living heterotrophic bacteria.
However, particle-attached bacterial play a critical role in bacterial community
dynamics and remineralization rates. Particle-attached bacteria are thought to be
generally genetically and phenotypically distinct from free-living bacteria (DeLong
et al., 1993), and their high remineralization rates of sinking particulate matter have
been associated with the creation of oxygen minimum zones (Wright et al., 2012,
and references therein). Current particle remineralization estimates within the
NCAR CESM model rely upon fixed percent loss from POC over time (Moore et al.,
2002; Moore et al., 2004; Doney et al., 2009). Recent studies have focused on
accurately predicting transfer efficiency using correlations to environmental
variables, such as nutrient distributions (eg., Weber et al., 2016); however, the
incorporation of bacterial dynamics into these models would allow for the
mechanistic representation of variable rates of POC remineralization as a function of
environmental drivers, particle type, bacterial metabolism, and bacterial biomass.
Teel
166
4.5 Conclusions
The incorporation of mechanistic remineralization of DOC by two
heterotrophic bacterial groups into an NPZD model impacted both bottom-up and
top-down dynamics. The evaluation of picophytoplankton growth and grazing
dynamics within the model indicated that both parameters were enhanced in the
base model to compensate for the lack of explicit bacterial dynamics. In addition, the
DOC cycling within the base model was insufficient to support realistic bacterial
biomass, and a proxy for lability was required in order to simulate observed
seasonal alternation between oligotrophic and opportunistic bacterial community
types. When base and bacterial models were run under progressively weaker
seasonal upwelling flux, the impact of explicitly including bacterial dynamics was
non-linear, and, in the case of fall picophytoplankton biomass, predicted opposite
responses to reduced upwelling flux relative to the base model. This study
demonstrates that the incorporation of dynamic and mechanistic bacterial
heterotrophic remineralization of DOC should improve the predictive capabilities of
ocean ecosystem models in a changing climate, especially as bacterial growth
efficiencies are likely to decline with increasing temperature, increasing pCO2, and
increasing stratification.
Acknowledgments
We would like to acknowledge NSF and the University of Southern California for
funding this research. This study was completed in close collaboration with the
Fuhrman Laboratory, and the SPOT bacterial community dataset would not exist
Teel
167
without countless hours of work put in by its members. SPOT cruise data, collected
by Troy Gunderson, the Fuhrman Laboratory, the Caron Laboratory, and the crew of
the Yellowfin, were made available for this research by Diane Kim and the Wrigley
Institute for Environmental Sciences.

Teel
168

Figure 4. 1 Base Ecosystem Model Schematic
This study utilized a base ecosystem model designed after Moore et al. (2002),
Moore et al. (2004), and Doney et al. (2009). This base model tracks carbon and
nitrogen dynamics and contains a small number of phytoplankton functions types
(n=3), a single zooplankton type, dissolved and particulate detrital pools, and a fixed
remineralization parameter that is represented as a function of time and organic
matter concentration (Levine, submitted).

Teel
169

Figure 4. 2 Bacterial Ecosystem Model Schematic
The bacterial model was based on the base ecosystem model, with three major
changes: (1) two explicit bacterial heterotrophic groups were added to the model,
with variable growth, grazing, and growth efficiency (BGE), in order to
mechanistically remineralize the dissolved organic matter, (2) all implicit bacterial
remineralization of dissolved organic matter was removed from the model, (3)
phytoplankton extracellular exudation was explicitly parameterized as an additional
component of primary production, percent extracellular release (PER), which varies
as a function of per-cell nitrogen stress. Mechanistic particulate organic carbon
remineralization was beyond the scope of this study and was kept in its original
form, as a fixed loss rate (Levine, submitted). SF represents DOC production through
sloppy feeding. BCD is the total bacterial carbon demand required to support
bacterial growth at variable BGE. Growth and grazing rates shown were annual
averages for the entire phytoplankton or bacterial biomass.
Teel
170

Figure 4. 3 Seasonal Cycles of Environmental Drivers
Average annual cycles of environmental boundary conditions were created from all
available data between 2000 and 2011. Annual cycles were smoothed with a low-
pass filter to minimize intra-seasonal variability. Temperature evolved
prognostically through heating and cooling, which were calculated as functions of
wind speed, air temperature, bottom boundary temperature, upwelling flux,
diffusive flux, and solar radiative heating. Resultant modeled sea surface
temperatures (SST) were well correlated with the observed seasonal cycle at SPOT
(R2 value of 0.89).
Teel
171

Figure 4. 4 Correlation between NO3 and Temperature at SPOT
Within the Southern California Bight, nitrate (NO3) and temperature (TEMP) have
been found to negatively correlate at values of temperature less than approximately
14˚ C (Lucas et al., 2010; Seegers et al., 2015). This relationship held true for the
SPOT site as well, where at TEMP less than 14˚C NO3 values could be reasonably
predicted by linear regression (TEMP> 14˚C: NO3 = -4.1*TEMP + 61; TEMP< 14˚C:
NO3 = 0). Within both the base ecosystem and bacterial models, bottom boundary
NO3 in mmol per cubic meter was calculated with this function.

Teel
172
Parameter varBACT==0 (547) varBACT==4 (556)
maxPCdia 5.4 5.5
KNO3dia 1.25 0.8
KNH4dia 0.13 1.0
grzMaxdia 4.9 4.7
zgrzDia 3.1 3.1
maxPCpico 1.25 1.25
KNO3pico 0.23 0.23
KNH4pico 0.003 0.003
grzMaxpico 2.5 1.25
zgrzPico 3.1 3.1
muMaxOligo - 0.8
KDOColigo - 2.3
grzMaxOligo - 1.0
zgrzOligo - 2.1
muMinOligo - 0.1
maxOligoBGE - 0.35
muMaxOppo - 7.5
KDOCoppo - 7
grzMaxOppo - 2.8
zgrzOppo - 2.1
maxOppoBGE - 0.4
maxPERpico - 0.3
minPERpico - 0.1
PERstressPico - 30
PERrepletePico - 0.2
maxPERdia - 0.2
minPERdia - 0.05
PERstressDia - 15
PERrepleteDia - 0.2
maxPERdiazo - 0.1
minPERdiazo - 0.05
PERstressDiazo - 150
PERrepleteDiazo - 100
AutoResp - 0.12
AlphaMin_oppo - 0.1
AlphaMin_oligo - 1
SeasonalDOC - 54
DeepDOC - 6
Table 4. 1 Model Parameterization for Base and Bacterial Models
These values represent the final phytoplankton and bacterial parameters that varied
with either phytoplankton functional type or bacterial community type.

Teel
173

Figure 4. 5 Base Ecosystem Seasonal Biomass Results
Shown here, from top left to bottom right, are three years of model output from the
base ecosystem model for phytoplankton biomass, zooplankton biomass, dissolved
organic carbon (DOC), particulate organic carbon (POC), nitrate (NO3), and
ammonium (NH4). All values shown are in mmol per cubic meter. 1After initial spin-
up, model years two and three were stable and responded to seasonal oscillations in
environmental drivers. Alternation between large and small phytoplankton
biomass resembles observations reported for the SPOT site (Cram et al., 2015;
Caron et al., 2017; Connell et al., 2017). Zooplankton biomass was tightly coupled
with large phytoplankton biomass. DOC increased in the mixed layer box model as
semi-labile DOC was created through phytoplankton lysis, sloppy feeding, and
grazer mortality.
Teel
174

Figure 4. 6 Base Model Phytoplankton versus SPOT Estimates
Estimated annual cycles of phytoplankton biomass from the SPOT site were
compared with the predicted annual cycle of phytoplankton biomass within the
base ecosystem model output. Modeled biomass is shown in blue and whisker plots
represent estimated biomass values from the SPOT site. MODIS Aqua monthly
averaged satellite chlorophyll values between 2002 and 2011 were scaled with a
local satellite-to-in-situ chlorophyll for the SPOT site and further converted to
biomass using the recently published Haskell et al. (2017) chlorophyll to carbon
conversion for the SPOT site.

Teel
175

Figure 4. 7 Base Model Picophytoplankton versus SPOT Estimates
Picophytoplankton growth and mortality within the base ecosystem model are
derived from cyanobacterial equations. The cyanobacterial biomass at SPOT shown
here was estimated utilizing the ARISA dataset from 2000-2011 (Chow et al., 2013;
Cram , 2014; Crame et al., 2015) and has been converted from percent abundance,
to cellular abundance, and then to biomass assuming an average biomass per cell of
163 fg/cell. This value for biomass per cell is derived from the published
generalization that the SPOT cyanobacterial populations are roughly one third
Prochloroccus (90 fg per cell) and two thirds Synechococcus (200 fg per cell) (Caron
et al., 2017). The estimated cyanobacterial biomass values are represented by the
background whisker plot and the modeled annual picophytoplankton biomass is
represented in blue for the base ecosystem model.

Teel
176

Figure 4. 8 Base Model Phytoplankton Growth Rates
Base model values for growth rates and grazing rates of large (Diatom-like) and
small (Picophytoplankton-like) phytoplankton groups are shown. The ranges of in-
silico growth and grazing rates are within the bounds of observed growth rates at
the SPOT site (Connell et al., 2017).

Teel
177

Figure 4. 9 Bacterial Model Seasonal Biomass Results
The annual cycle (model years 2 and 3) of modeled large phytoplankton biomass
(PHYTO L), small phytoplankton biomass (PHYTO S), zooplankton biomass (ZOO),
bacterial biomass (BACT), and mixed-layer nitrate (NO3) are shown in
concentration values of mmol per cubic meter. All biological biomass values are
shown in carbon units.

Teel
178

Figure 4. 10 Bacterial Model Phytoplankton Growth Rates
Bacterial model values for growth and grazing of large phytoplankton (Diatom-like),
small phytoplankton (Picophytoplankton-like), and bacteria are shown. The ranges
of in-silico growth rates are within the bounds of observed growth rates at the SPOT
site (Cram et al., 2015; Connell et al., 2017). Diatom growth rates are slightly higher
within the bacterial model as compared with the base ecosystem model.

Teel
179

Figure 4. 11 Ecological Lag within Bacterial Model and at SPOT
One of the major model goals for the bacterial model was to mechanistically
recreate the ecological lag observed in-situ between phytoplankton abundance and
bacterial abundance. Secondly, the bacterial model was designed to mimic the
alternation of two coexisting and competing bacterial community types, the
opportunistic community, similar to Module 1 observed at the SPOT site, and the
oligotrophic community, similar to Module 2 observed at the SPOT site. Here we
show that the model represents strong oscillation between oligotrophic and
opportunistic types as well as recreating an ecological lag between phytoplankton
biomass and bacterial biomass.
Teel
180

Figure 4. 12 Bacterial Oscillation within Bacterial Model
These results show the alternating bacterial (BACT) populations (Oppo and Oligo)
within the model and the corresponding oscillations in bacterial growth efficiency,
grazing, non-grazing mortality, respiration, and production. In the bottom panels,
the primary production (GROWTH C, right) and DOC released as a function of
increased nutrient stress (PER DOC, left) are shown for all phytoplankton functional
types.
Teel
181

Figure 4. 13 Bacterial Model Bacteria versus SPOT Measurements
Here model predicted bacterial abundance (BA), production (BP), and growth rate
(µ) have been plotted in green with the direct measurements from surface samples
at the SPOT site between 2000 and 2011. SPOT values for BP and µ correspond to
those calculated from 3H-Leucine incorporation. Median bacterial abundance at the
SPOT site from 2000-2011 was 1.5 million cells per mL (Cram et al., 2015; Caron et
al., 2017). Within the bacterial model, median bacterial abundance was estimated to
be 0.81 million cells per mL. For these model estimates, oligotrophic bacteria were
assumed to contain 12 fg per cell while opportunistic bacteria were assumed to
contain 20 fg per cell. Modeled bacterial production and growth rates were higher
than observations, but were within the bounds of observations and showed similar
seasonal fluctuation.
Teel
182

Figure 4. 14 Bacterial Model without Lability Restriction
Two bacterial model runs are shown here to depict the effect of variable percent
extracellular release (PER) and the lability scalar for DOC uptake by bacterial groups
within the model (alpha). In thin lines, the bacterial model was run with variable
PER, as a function of phytoplankton functional type and nitrogen stress, and
variable minimum alpha that was fixed at 0.1 for opportunotrophs and 1 for
oligotrophs. Overlaid in thick lines is the model result when variable PER is set to
zero and the alpha parameter is set to 1; these changes correspond with a reduction
in overall DOC production by phytoplankton and an increase in available DOC for
opportunotrophic uptake. Bacterial dynamics were more sensitive to the alpha
parameter than to PER, leading to opportunistic overgrowth on semi-labile DOC
during the model year.
Teel
183

Figure 4. 15 Bacterial Model without Variable BGE
Model runs depicted here show the sensitivity of bacterial oscillation to prescribed
fixed bacterial growth efficiency. Opportunistic bacterial biomass (BACT C, oppo) is
shown in blue lines, while oligotrophic bacterial biomass (BACT C, oligo) is shown in
red lines. Seasonal alternation between bacterial groups remains relatively constant
and the bacterial biomass itself remained within a factor of 4 among these
experiments, however, changes in fixed bacterial growth efficiencies strongly affects
the dominance of opportunistic or oligotrophic bacteria within the model, and
thereby the resultant BGE values for the bacterial community as a whole. With well
known seasonal dynamics for a single location, fixed bacterial growth efficiencies
may be able to predict overall community BGE, however, the application of fixed
values to wider scale (eg., global) modeling would underestimate bacterial
functional dynamics.

Teel
184

Figure 4. 16 Bacterial Model with Single Bacterial Group
Three bacterial model runs are depicted here indicating the model run with both
oligotrophic and opportunistic bacterial groups (black lines), only the opportunistic
bacterial group (blue lines), and only the oligotrophic bacterial group (red lines). On
average, the oligotrophic group better predicted the seasonality and biomass values
of the bacterial model with two groups; however, without opportunistic bacterial
grazing, the connection of bacterial carbon to higher trophic levels is weakened.
These results indicate that there may be an effective way to create a single bacterial
biomass module that can switch between oligotrophic and opportunistic
functionality.

Teel
185

Figure 4. 17 Flux Sensitivity Analysis: Large Phytoplankton Biomass
Base and bacterial models were run with dampened seasonal flux to investigate
model responses to increased stratification. Seasonal flux was reduced by 10% for
each consecutive model run and the results of change in large phytoplankton
(Diatom) biomass are shown for base and bacterial models in blue and green lines,
respectively. For diatom-dominated seasons within the model (spring and winter),
biomass losses were similar between base and bacterial models. For summer and
fall, however, diatom biomass with the bacterial model was reduced more quickly
and non-linearly in response to dampened upwelling flux.
Teel
186

Figure 4. 18 Flux Sensitivity Analysis: Picophytoplankton Biomass
Base and bacterial models were run with dampened seasonal flux to investigate
model responses to increased stratification. Seasonal flux was reduced by 10% for
each consecutive model run and the results of change in small phytoplankton
(cyanobacteria) biomass are shown for base and bacterial models in blue and green
lines, respectively. Base and bacterial models predicted vastly different fall patterns
for cyanobacterial abundance under low flux, indicating that the predictive ability of
current base model ecosystem dynamics may break down under intensified
stratification.

Teel
187
References
Allen, J. I., J. R. Siddorn, J. C. Blackford, and F. J. Gilbert. (2004). Turbulence as a
control on the microbial loop in a temperate seasonally stratified marine systems
model. Journal of Sea Research. 52:1-20.

Anderson T.R. and H.W. Ducklow (2001). Microbial loop carbon cycling in ocean
environments studied using a simple steady-state model. Aquat. Microb. Ecol. 26:37-
49.

Anderson T.R. and P.J. le B. Williams. (1999). A one-dimensional model of dissolved
organic carbon cycling in the water column incorporating combined biological-
photochemical decomposition. Global Biogeochemical Cycles. 13(2):337-349.

Baines S.B. and M.L. Pace. (1991). The production of dissolved organic matter by
phytoplankton and its importance to bacteria: patterns across marine and
freshwater systems. Limnol. Oceanogr. 36(6):1078-1090.

Bauer, J. E., Cai, W. J., Raymond, P. A., Bianchi, T. S., Hopkinson, C. S., & Regnier, P. A.
(2013). The changing carbon cycle of the coastal ocean. Nature, 504(7478), 61-70.

Becker, J.W., P.M. Berube, C.L. Follett, J.B. Waterbury, S.W. Chisholm, E.F. DeLong,
and D.J. Repeta. (2014). Closely related phytoplankton species produce similar
suites of dissolved organic matter. Front. Microbiol. 5(111):1-14.
doi:10.3389/fmicb.2014.00111.

Behrenfeld, M. J., Y.X. Hu, R. T. O’Malley, E. S. Boss, C. A. Hostetler, D. A. Siegel, J. L.
Sarmiento, J. Schulien, J. W. Hair, X.M. Lu, S. Rodier, and A. J. Scarino. (2016). Annual
boom-bust cycles of polar phytoplankton biomass revealed by space-based lidar.
Nature Geosciences. DOI: 10.1038/NGEO2861

Berman, T. and O. Holm-Hansen. (1974). Release of Photoassimilated Carbon as
Dissolved Organic Matter by Marine Phytoplankton. Marine Biology. 38:305-310.

Bratbak, G. and T. F. Thingstad. (1985). Phytoplankton-bacteria interactions: an
apparent paradox? Analysis of a model system with both competition and
commensalism. Marine Ecology Progress Series. 25:23-30.

Caron, D. A., P. E. Connell, R. A. Schaffner, A. Schnetzer, J. A. Fuhrman, P. D. Countway,
and D. Y. Kim. (2017). Planktonic food web structure at a coastal time-series site: I.
Partitioning of microbial abundance and carbon biomass. Deep-Sea Research I.
121:14-29.

Chow, C.-E. T., R. Sachdeva, J.A. Cram, J.A. Steele, D. M. Needham, A. Patel, A.E. Parada,
and J. A. Fuhrman. 2013. Temporal Vairability and Coherence of Euphotic Zone
Teel
188
Bacterial Communities over a Decade in the Southern California Bight. The ISME
Journal. 7:2259-2273.

Connell, P. E., V. Campbell, A. G. Gellene, S. K. Hu, and D. A. Caron. (2017). Planktonic
food web structure at a coastal time-series site: II. Spatiotemporal variability of
microbial trophic activites. Deep-Sea Research I. 121:210-223.

Cram, J.A. (2014). Dynamics of marine bacterial communities from surface to
bottom and the factors controlling them. (PhD Dissertation). University of Southern
California.

Cram, J.A. , C.-E. T. Chow, R. Sachdeva, D. M. Needham, A. E. Parada, J. A. Steele, and J.
A. Fuhrman. 2015. Seasonal and Interannual Variability of the Marine
Bacterioplankton Community Throughout the Water Column over Ten Years. The
ISME Journal. 9:563-580.

Cram, J. A., A. Parada, J. A. Fuhrman. (2016). Dilution reveals how viral lysis and
grazing shape microbial communities. Limnology and Oceanography. 61: 889–905.
doi:10.1002/lno.10259

del Giorgio, P.A. and J.J. Cole. (1998). Bacterial growth efficiency in natural aquatic
systems. Annu. Rev. Ecol. Syst. 29:503-541.

Doney, S.C., I. Lima, J. K. Moore, K. Lindsay, M. J. Behrenfeld, T. K. Westberry, N.
Mahowald, D. M Glover, and T. Takahashi. (2009). Skill metrics for confronting
global upper ocean ecosystem-biogeochemistry models against field and remote
sensing data. Journal of Marine Systems. 76:95-112.

Dutkiewicz, S., B. A. Ward, F. Monteiro, and M. J. Follows. (2012). Interconnection of
nitrogen fixers and iron in the Pacific Ocean: Theory and numerical simulations.
Global Biogeochemical Cycles. 26, GB1012, doi:10.1029/2011GB004039.

Ducklow, H. W. (2000) “Bacterial Production and Biomass in the Oceans.” pp. 85-
120. In: D. L. Kirchman, ED. Microbial Ecology of the Oceans. New York: Wiley.

Fasham, M.J.R., H. W. Ducklow, and S.M. McKelvie. 1990. A Nitrogen-based Model of
Plankton Dynamics in the Oceanic Mixed Layer. Journal of Marine Research. 48:591-
639.
Fogg, G.E. (1983). The ecological significance of extracellular products of
phytoplankton photosynthesis. Botanica Marina. 26:3-14.

Forsythe, W.C., E.J. Rykiel Jr., R.S. Stahl, H. Wu, and R.M. Schoolfield. (1995). A model
comparison for daylength as a function of latitude and day of year. Ecological
Modelling. 80:87-95.
Teel
189

Fuhrman, J. A., Ammerman, J. W., & Azam, F. (1980). Bacterioplankton in the coastal
euphotic zone: distribution, activity and possible relationships with
phytoplankton. Marine Biology, 60(2), 201-207.

Fuhrman, J. A. (1999). Marine viruses and their biogeochemical and ecological
effects. Nature. 399:541-548.
Gomez-Consarnau, L, NM Levine, LS Cutter, D Wang, B Seegers, J Aristegui, JA
Fuhrman, JM Gasol, SA Sanudo-Wilhelmy. (Submitted) Marine proteorhodopsins
rival chlorophyll in solar energy capture.

Hagström, A., F. Azam, A. Andersson, J. Wikner, and F. Rassoulzadegan. (1988).
Microbial loop in an oligrotrophic pelagic marine ecosystem: possible roles of
cyanobacteria and nanoflagellates in the organic fluxes. Marine Ecology – Progress
Series. 49:171-178.

Hahn, M. W., and M. G. Höfle. (1999). Flagellate Predation on a Bacterial Model
Community: Interplay of Size-Selective Grazing, Specific Bacterial Cell Size, and
Bacterial Community Compositoin. Applied and Environmental Microbiology.
65(11):4863-4872.

Halewood, E.R., C.A. Carlson, M.A. Brzezinski, D.C. Reed, and J. Goodman. (2012).
Annual cycle of organic matter partitioning and its availability to bacteria across the
Santa Barbara Channel continental shelf. Aquat. Microb. Ecol. 67:189-209.

Hansell, D.A., C. A. Carlson, D.J. Repeta, and R. Schlitzer. (2009). Dissolved organic
matter in the ocean. Oceanography. 22(4):202-211.

Haskell, W. Z., Prokopenko, M. G., Hammond, D. E., Stanley, R. H., Berelson, W. M.,
Baronas, J. J., Fleming, J. C., and Aluwihare, L. (2016). An organic carbon budget for
coastal Southern California determined by estimates of vertical nutrient flux, net
community production and export. Deep Sea Research Part I: Oceanographic
Research Papers, 116, 49-76.

Haskell, W. Z., Prokopenko, M. G., Hammond, D. E., Stanley, R. H., & Sandwith, Z. O.
(2017). Annual cyclicity in export efficiency in the inner Southern California
Bight. Global Biogeochemical Cycles, 31(2), 357-376.

Hasumi, H. and T. Nagata. (2014). Modeling the global cycle of marine dissolv
ed organic matter and its influence on marine productivity. Ecological Modelling.
288:9-24.

James AK, Passow U, Brzezinski MA, Parsons RJ, Trapani JN, Carlson CA (2017)
Elevated pCO
2
enhances bacterioplankton removal of organic carbon. PLoS ONE
12(3): e0173145. doi:10.1371/journal.pone.0173145
Teel
190
Jiao, N.-Z., G.J. Herndl, D.A. Hansell, R. Benner, G. Kattner, S.W. Wilhelm, D.L.
Kirchman, M.G. Weinbauer, T.-W. Luo, F. Chen, and F. Azam. (2010). Microbial
production of recalcitrant dissolved organic matter: long-term carbon storage in the
global ocean. Nature Reviews: Microbiology. 8:593-599.

Kim, D. Y., Countway, P. D., Jones, A. C., Schnetzer, A., Yamashita, W., Tung, C., &
Caron, D. A. (2014). Monthly to interannual variability of microbial eukaryote
assemblages at four depths in the eastern North Pacific. The ISME journal, 8(3), 515-
530.

Kirchman, D. L., X. A. G. Moran, and H. Ducklow. (2009). Microbial growth in the
polar oceans – role of temperature and potential impact of climate change. Nature
Reviews Microbiology. 7:451-459.

Kirchman, D. L. (2016). Growth Rates of Microbes in the Oceans. Annual Review of
Marine Science. 8:285-309.

Klindworth, A., A. J. Mann, S. Huang, A. Wichels, C. Quast, J. Waldmann, H. Teeling,
and F. O. Glöckner. (2014). Diversity and activity of marine bacterioplankton during
a diatom bloom in the North Sea assessed by total RNA and pyrotag sequencing.
Marine Genomics. 18:185-192.

Lankiewicz, T. S., Cottrell, M. T., & Kirchman, D. L. (2016). Growth rates and rRNA
content of four marine bacteria in pure cultures and in the Delaware estuary. The
ISME journal, 10(4), 823-832.

Levine, NM (submitted) Life in a patchy ocean: the impact of fine-scale
spatiotemporal heterogeneity on ecosystem dynamics.

Moore, J. K., S. C. Doney, J. A. Kleypas, D. M. Glover, and I. Y. Fung. (2002). An
intermediate complexity marine ecosystem model from the global domain. Deep-Sea
Research II.49:403-462.

Moore, J. K., S. C. Doney, K. Lindsay. (2004). Upper ocean ecosystem dynamics and
iron cycling in a global three-dimensional model. Global Biogeochemical Cycles. 18,
GB4028, doi:10.1029/2004GB002220.

Nagata, T. (2000) Production mechanisms of dissolved organic matter. In: DL
Kirchman(ed) Microbial Ecology of the Ocean. John Wiley and Sons. Chichester, UK.
121-152

Needham, D.M. and J. A. Fuhrman. 2016. Pronounced Daily Succession of
Phytoplankton, Archaea, and Bacterial Following a Spring Bloom. Nature
Microbiology. 1:1605. doi:10.1038/nmicrobiol.2016.5.

Teel
191
Rivkin, R.B. and L. Legendre. (2001). Biogenic Carbon Cycling in the Upper Ocean:
Effects of Microbial Respiration. Science. 291:2398-2400.

Sinsabaugh, R.L., S. Manzoni, D.L. Moorhead, and A. Richter. (2013). Carbon use
efficiency of microbial communities: stoichiometry, methodology, and modeling.
Ecology Letters. 16:930-939.

Suttle, C. A. (2007). Marine viruses—major players in the global ecosystem. Nature
Reviews Microbiology, 5(10), 801-812.

Stull, R. B. (1988). “Boundary Conditions and Surface Forcing.” pp. 251-294. In: An
Introduction to Boundary Layer Meteorology. Ed. R. B. Stull. Springer. ISBN: 978-94-
009-3027-8

Teel, E. N., X. Liu, B. N. Seegers, M. A. Ragan, B. H. Jones, and N. M. Levine.
Contextualizing time-series data: Using high-resolution in-situ gliders to quantify
short-term regional oceanographic variability in the San Pedro Channel. In
Preperation.

Teeling H., B.M. Fuchs, D. Becher, C. Klockow, A. Gardebrecht, C.M. Bennke, M.
Kassabgy, S. Huang, A.J. Mann, J. Waldmann, M. Weber, A. Klindworth, A. Otto, J.
Lange, J. Bernhardt, C. Reinsch, M. Hecker, J. Peplies, F.D. Bockelmann, U. Callies, G.
Gerdts, A. Wichels, K.H. Wiltshire, F.O. Glöckner, T. Schweder, and R. Amann. (2012).
Substrate-controlled succession of marine bacterioplankton populations induced by
a phytoplankton bloom. Science. 336: 608-611.

Thingstad, T.F. and R. Lignell. (1997). Theoretical models for the control of bacterial
growth rate, abundance, diversity, and carbon demand. Aquat. Microb. Ecol. 13:19-
27.

Touratier, F., Legendre, L., & Vézina, A. (1999). Model of bacterial growth influenced
by substrate C: N ratio and concentration. Aquatic Microbial Ecology, 19(2), 105-118.

Vichi, M., Pinardi, N., & Masina, S. (2007). A generalized model of pelagic
biogeochemistry for the global ocean ecosystem. Part I: Theory. Journal of Marine
Systems, 64(1), 89-109.

Vichi, M., Masina, S., & Navarra, A. (2007). A generalized model of pelagic
biogeochemistry for the global ocean ecosystem. Part II: Numerical
simulations. Journal of Marine Systems, 64(1), 110-134.

Vichi, M., & Masina, S. (2009). Skill assessment of the PELAGOS global ocean
biogeochemistry model over the period 1980–2000. Biogeosciences, 6(11), 2333-
2353.

Teel
192
Walsh, J.J., D.A. Dieterle, F.E. Muller-Karger, R. Bohrer, and W.P. Bissett. (1999).
Simulation of carbon-nitrogen cycling during spring upwelling in the Cariaco Basin.
Marine Science Faculty Publications. Paper 17.

Ward, C. S., C-M. Yung, K. M. Davis, S. K. Blinebry, T. C. Williams, Z. I. Johnson, and D.
E. Hunt. (2017). Annual community patterns are driven by seasonal switching
between closely related marine bacteria. The ISME Journal. 1-11.

Wear, E. K., C. A. Carlson, A. K. James, M. A. Brzezinski, L. A. Windecker, and C. E.
Nelson. (2015). Synchronous shifts in dissolved organic carbon bioavailability and
bacterial community responses over the course of an upwelling-driven
phytoplankton bloom. Limnology and Oceanography. 60:657-677.

Wear, E. K., C. A. Carlson, L. A. Windecker, and M. A. Brzezinski (2015). Roles of
diatom nutrient stress and species identity in determining the short- and long-term
bioavailability of diatom exudates to bacterioplankton. Marine Chemistry. 117:335-
348.

Weitz, J.S., C.A. Stock, S.W. Wilhelm, L. Bourouiba, A. Buchan, M.L. Coleman, M.J.
Follows, J.A. Fuhrman, L.F. Jover, J.T. Lennon, M. Middelboe, D.L. Sonderegger, C.A.
Suttle, B.P. Taylor, T.F. Thingstad, W.H. Wilson, and K.E. Wommack. (2014). A
multitrophic model to quantify the effects of marine viruses on microbial food webs
and ecosystem processes. The ISME Journal . 1-13.

Winter, C., T. Bouvier, M. G. Weinbauer, and T. F. Thingstad. (2010). Trade-offs
between competition and defense specialists among unicellular planktonic
organisms: the “kill the winner” hypothesis revisited. Microbial and Molecular
Biology Reviews. 74(1):42-57.

Wright, J. J., Konwar, K. M., & Hallam, S. J. (2012). Microbial ecology of expanding
oxygen minimum zones. Nature Reviews Microbiology, 10(6), 381-394.

Xiu P and F. Chai. (2014). Connections between physical, optical, and
biogeochemical processes in the Pacific Ocean. Progress in Oceanography. 122:30-
53.

Yooseph, S. K. H. Nealson, D. B. Rusch, J. P. McCrow, C. L. Dupont, M. Kim, J. Johnson,
R. Montgomery, S. Ferriera, K. Beeson, S. J. Williamson, A. Tovchigrechko, A. E. Allen,
L. A. Zeigler, G. Sutton, E. Eisenstadt, Y-H. Rogers, R. Friedman, M. Frazeier, and J. C.
Venter. (2010). Genomic and fuctional adaptation in surface ocean planktonic
prokaryotes. Nature. 468:60-67.

Teel
193
Bibliography

Allen, J. I., J. R. Siddorn, J. C. Blackford, and F. J. Gilbert. (2004). Turbulence as a
control on the microbial loop in a temperate seasonally stratified marine systems
model. Journal of Sea Research. 52:1-20.

Allen, L. Z., E. E. Allen, J. H. Badger, J.P. McCrow, I. T. Paulsen, L. D.H. Elbourne, M.
Thiagarajan, D. B. Rusch, K. N. Nealson, S. J. Williamson, J. C. Venter, A. E. Allen.
(2012). Influence of nutrients and currents on the genomic composition of microbes
across an upwelling mosaic. The ISME Journal. 6:1403-1414.

Alonso-Saez L., J. Pinhassi, J. Pernthaler, and J.M. Gasol. (2010). Leucine-to-carbon
empirical conversion factor experiments: does bacterial community structure have
an influence? Environ Mircobiol. 12(11):2988-2997.

Aluwihare, L.I., D.J. Repeta, and R.F. Chen. (1997). A major biopolymeric component
to dissolved organic carbon in surface sea water. Nature 387:166–169.

Anderson T.R. and H.W. Ducklow (2001). Microbial loop carbon cycling in ocean
environments studied using a simple steady-state model. Aquat. Microb. Ecol. 26:37-
49.

Anderson T.R. and P.J. le B. Williams. (1999). A one-dimensional model of dissolved
organic carbon cycling in the water column incorporating combined biological-
photochemical decomposition. Global Biogeochemical Cycles. 13(2):337-349.

Azam, F., T. Fenchel, J.G. Field, J.S. Gray, L.A. Meyer-Reil, and F. Thingstad. (1983).
The ecological role of water-column microbes in the sea. Mar. Ecol. Progr. Ser.
10:257-263.

Azam F., D.C. Smith, G.F. Steward, and A. Hagström. (1993). Bacteria-organic matter
coupling and its significance for oceanic carbon cycling. Microb. Ecol. 28:167-179.

Azam F. (1998). Microbial control of ocean carbon flux: the plot thickens. Science.
280(5364):694-696.

Baines S.B. and M.L. Pace. (1991). The production of dissolved organic matter by
phytoplankton and its importance to bacteria: patterns across marine and
freshwater systems. Limnol. Oceanogr. 36(6):1078-1090.

Bauer, J. E., Cai, W. J., Raymond, P. A., Bianchi, T. S., Hopkinson, C. S., & Regnier, P. A.
(2013). The changing carbon cycle of the coastal ocean. Nature, 504(7478), 61-70.

Becker, J.W., P.M. Berube, C.L. Follett, J.B. Waterbury, S.W. Chisholm, E.F. DeLong,
and D.J. Repeta. (2014). Closely related phytoplankton species produce similar
Teel
194
suites of dissolved organic matter. Front. Microbiol. 5(111):1-14.
doi:10.3389/fmicb.2014.00111.

Behrenfeld, M. J., Y.X. Hu, R. T. O’Malley, E. S. Boss, C. A. Hostetler, D. A. Siegel, J. L.
Sarmiento, J. Schulien, J. W. Hair, X.M. Lu, S. Rodier, and A. J. Scarino. (2016). Annual
boom-bust cycles of polar phytoplankton biomass revealed by space-based lidar.
Nature Geosciences. DOI: 10.1038/NGEO2861

Beman, J. M., J. A. Steele, and J. A. Fuhrman (2011). Co-occurrence patterns for
abundant marine archaeal and bacterial lineages in the deep chlorophyl maximum
of coastal California. The ISME Journal. 5: 1077-1085.

Berman, T. and O. Holm-Hansen. (1974). Release of Photoassimilated Carbon as
Dissolved Organic Matter by Marine Phytoplankton. Marine Biology. 38:305-310.

Bratbak, G. and T. F. Thingstad. (1985). Phytoplankton-bacteria interactions: an
apparent paradox? Analysis of a model system with both competition and
commensalism. Marine Ecology Progress Series. 25:23-30.

Brown, M.V., M.S. Schwalbach, I. Hewson, J.A. Fuhrman.(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. Env.
Micro. 7(9):1466-1479.

Campbell B.J., L.-Y. Yu, J.F. Heidelberg, and D.L. Kirchman. (2011). Activity of
abundant and rare bacterial in a coastal ocean. Proc Natl Acad Sci USA.
108(31):12776-12781.

Campbell B.J. and D.L. Kirchman. (2013). Bacterial diversity, community structure,
and potential growth rates along an estuarine salinity gradient. ISME J. 7:210-220.

Carlson C.A., H.W. Ducklow, D.A. Hansell, and W.O. Smith Jr. (1998). Organic carbon
partitioning during spring phytoplankton blooms in the Ross Sea polynya and the
Sargasso Sea. Limol. Oceanogr. 43(3):375-386.

Carlson, C. A., S. J. Giovannoni, D. A. Hansell, S. J. Goldberg, R. Parsons, M. P. Otero, K.
Vergin, and B. R. Wheeler. (2002). Effect of nutrient amendments on
bacterioplankton production, community structure, and DOC utilization in the
northwestern Sargasso Sea. Aquatic Microbial Ecology. 30:19-36.

Carlson C.A., S.J. Giovannoni, D.A. Hansell, S.J. Goldberg, R. Parsons, and K. Vergin.
(2004). Interactions among dissolved organic carbon, microbial processes, and
community structure in the mesopelagic zone of the northwestern Sargasso Sea.
Limol. Oceanogr. 49(4):1073-1083.

Teel
195
Caron, D. A., P. E. Connell, R. A. Schaffner, A. Schnetzer, J. A. Fuhrman, P. D. Countway,
and D. Y. Kim. (2017). Planktonic food web structure at a coastal time-series site: I.
Partitioning of microbial abundance and carbon biomass. Deep-Sea Research I.
121:14-29.

Cetinic, I., G. Toro-Farmer, M. Ragan, C. Oberg, and B.H. Jones. (2009). Calibration
procedure for Slocum glider deployed optical instruments. Optics Express.
17(18):15420-15430.

Cherrier, J. and J.E. Bauer. (2004). Bacterial utilization of transient plankton-derived
dissolved organic carbon and nitrogen inputs in surface ocean waters. Aquat.
Microb. Ecol. 35:229-241.

Chow, C.-E. T., R. Sachdeva, J.A. Cram, J.A. Steele, D. M. Needham, A. Patel, A.E. Parada,
and J. A. Fuhrman. 2013. Temporal Vairability and Coherence of Euphotic Zone
Bacterial Communities over a Decade in the Southern California Bight. The ISME
Journal. 7:2259-2273.

Chow C.-E. T., D. Y. Kim, R. Sachdeva, D. A. Caron, and J.A. Fuhrman. 2014. Top-down
Controls on Bacterial Community Structure: Microbial Network Analysis of Bacteria,
T4-like Viruses, and Protists. The ISME Journal. 8:816-829.

Connell, P. E., V. Campbell, A. G. Gellene, S. K. Hu, and D. A. Caron. (2017). Planktonic
food web structure at a coastal time-series site: II. Spatiotemporal variability of
microbial trophic activites. Deep-Sea Research I. 121:210-223.

Countway P.D., P.D. Vigil, A. Schnetzer, S.D. Moorthi, and D.A. Caron. (2010).
Seasonal analysis of protistan community structure and diversity at the USC
Microbial Observatory (San Pedro Channel, North Pacific Ocean). Limnol. Oceanogr.
55(6):2381-2396.

Cram, J.A. (2014). Dynamics of marine bacterial communities from surface to
bottom and the factors controlling them. (PhD Dissertation). University of Southern
California.

Cram, J. A., Chow, C. E. T., Sachdeva, R., Needham, D. M., Parada, A. E., Steele, J. A., &
Fuhrman, J. A. (2015). Seasonal and interannual variability of the marine
bacterioplankton community throughout the water column over ten years. The ISME
journal, 9(3), 563-580.

Cram, J. A., A. Parada, J. A. Fuhrman. (2016). Dilution reveals how viral lysis and
grazing shape microbial communities. Limnology and Oceanography. 61: 889–905.
doi:10.1002/lno.10259

Teel
196
de Boyer Montégut, C., G. Madec, A.S. Fischer, A. Lazar, and D. Iudicone. (2004).
Mixed layer depth over the global ocean: an examination of profile data and profile-
based climatology. J. Geophys. Res. 109(C12003):1-20. doi:10.1029/2004JC002378.

del Giorgio, P.A. and J.J. Cole. (1998). Bacterial growth efficiency in natural aquatic
systems. Annu. Rev. Ecol. Syst. 29:503-541.

del Giorgio, P.A. and C.M. Duarte. (2002). Respiration in the open ocean. Nature.
420:379-384.

del Giorgio, P.A., M.L. Pace, and D. Fischer. (2006). Relationship of bacterial growth
efficiency to spatial variation in bacterial activity in the Hudson River. Aquat. Microb.
Ecol. 45:55-67.

del Giorgio, P. A., R. Condon, T. Bouvier, K. Longnecker, C. Bouvier, E. Sherr, and J. M.
Gasol. (2011). Coherent patterns of bacterial growth, growth efficiency, and leucine
metabolism along a northeastern Pacific inshore-offshore transect. Limnology and
Oceanography. 56(1):1-16.

DeLong E.F., D.G. Franks, and A.L. Alldredge. (1993). Phylogenetic diversity of
aggregate-attached vs. free-living marine bacterial assemblages. Limnol. Oceanogr.
38(5):924-934.

Doney, S.C. (1996). A synoptic atmospheric surface forcing data set and physical
upper ocean model for the U.S. JGOFS Bermuda Atlantic Time-Series Study site. J.
Geophys. Res. 101(C10):25615-25634.

Doney, S.C., I. Lima, J. K. Moore, K. Lindsay, M. J. Behrenfeld, T. K. Westberry, N.
Mahowald, D. M Glover, and T. Takahashi. (2009). Skill metrics for confronting
global upper ocean ecosystem-biogeochemistry models against field and remote
sensing data. Journal of Marine Systems. 76:95-112.
Dong, C. M., E. Y Idica, and J. C. McWilliams. 2009. Circulation and multiple-scale
variability in the Southern California Bight. Progress in Oceanography. 82: 168-190.

Ducklow, H. W. (2000) “Bacterial Production and Biomass in the Oceans.” pp. 85-
120. In: D. L. Kirchman, ED. Microbial Ecology of the Oceans. New York: Wiley.

Dupont, C.L., D.B. Rusch, S. Yooseph, M.J. Lombardo, R. A. Richter, R. Valas, M.
Novotny, J. Yee-Greenbaum, J.D. Selengut, D.H. Haft, A.L. Halpern, R.S. Lasken, K.
Nealson, R. Friedman, and J.C. Venter. (2012). Genomic insights to SAR86, an
abundant and uncultivated marine bacterial lineage. ISME J. 6:1186-1199.

Dutkiewicz, S., B. A. Ward, F. Monteiro, and M. J. Follows. (2012). Interconnection of
nitrogen fixers and iron in the Pacific Ocean: Theory and numerical simulations.
Global Biogeochemical Cycles. 26, GB1012, doi:10.1029/2011GB004039.
Teel
197

Eppley, R.W., E.H. Renger, and W.G. Harrison. (1979). Nitrate and phytoplankton
production in southern California coastal waters. Limnol. Oceanogr. 24(3):483-494.
Fasham, M.J.R., H. W. Ducklow, and S.M. McKelvie. 1990. A Nitrogen-based Model of
Plankton Dynamics in the Oceanic Mixed Layer. Journal of Marine Research. 48:591-
639.
Fogg, G.E. (1983). The ecological significance of extracellular products of
phytoplankton photosynthesis. Botanica Marina. 26:3-14.

Forsythe, W.C., E.J. Rykiel Jr., R.S. Stahl, H. Wu, and R.M. Schoolfield. (1995). A model
comparison for daylength as a function of latitude and day of year. Ecological
Modelling. 80:87-95.

Fuhrman, J. A., Ammerman, J. W., & Azam, F. (1980). Bacterioplankton in the coastal
euphotic zone: distribution, activity and possible relationships with
phytoplankton. Marine Biology, 60(2), 201-207.

Fuhrman, J.A. and F. Azam. (1980). Bacterioplankton secondary production
estimates for coastal waters of British Columbia, Antarctica, and California. Applied
Environ. Microbiol. 39(6):1085-1095.

Fuhrman, J.A. and F. Azam (1982). Thymidine incorporation as a measure of
heterotrophic bacterioplankton production in marine surface waters: evaluation
and field results. Marine Biology. 66:109-120.

Fuhrman, J. A. (1999). Marine viruses and their biogeochemical and ecological
effects. Nature. 399:541-548.

Fuhrman, J.A., I. Hewson, M.S. Schwalbach, J.A. Steele, M.V. Brown, S. Naeem (2006).
Annually reoccurring bacterial communities are predictable from ocean conditions.
Proc Natl Acad Sci USA 103: 13104–13109.

Fuhrman, J.A. and J. A. Steele. (2008). Community structure of marine
bacterioplankton: patterns, networks, and relationships to function. Aquatic
Microbial Ecology. 53:69-81.

Fuhrman J.A., J.A. Steele, I. Hewson, M.S. Schwalbach, M.V. Brown, J.L. Green, and J.H.
Brown (2008). A latitudinal diversity gradient in planktonic marine bacteria. Proc
Natl Acad Sci USA 105: 7774–7778.

Fuhrman, J. A., J. A. Cram, and D. M. Needham. (2015). Marine microbial community
dynamics and their ecological interpretation. Nature Reviews Microbiology. 13:139-
146.

Teel
198
Galand, P. E., I. Salter, and D. Kalenitchenko. Ecosystem productivity is associated
with bacterial phylogenetic distance in surface marine waters. Molecular Ecology.
24:5785-5795.

Gilbert, J. A., D. Field, P. Swift, S. Thomas, D. Cummings, B. Temperton, K. Weynberg,
S. Huse, M. Hughes, I. Joint, P. J. Somerfield, and M. Mühling. (2010). The taxonomic
and functional diversity of microbes at a temperate coastal site: a ‘multi-omics’
study of seasonal and diel temporal variation. PLoS One. 5:e15545.

Gilbert, J. A., J. A. Steele, J. G. Caporaso, L. Steinbrück, J. Reeder, B. Temperton, S.
Huse, A. C. McHardy, R. Knight, I. Joint, P. Somerfield, J. A. Fuhrman, and D. Field.
(2012). Defining seasonal marine microbial community dynamics. The ISME Journal.
6:298-308.

Giovannoni S.J. and K.L. Vergin. (2012). Seasonality in ocean microbial communities.
Science. 335:671-676.

Gomez-Consarnau, L, NM Levine, LS Cutter, D Wang, B Seegers, J Aristegui, JA
Fuhrman, JM Gasol, SA Sanudo-Wilhelmy. (submitted) Marine proteorhodopsins
rival chlorophyll in solar energy capture.

Hagström, A., U. Larsson, P. Hörstedt, and S. Normark. (1979). Frequency of dividing
cells, a new approach to the determination of bacterial growth rates in aquatic
environments. Applied and Environ. Microbiol. 37(5):805-812.

Hagström, A., F. Azam, A. Andersson, J. Wikner, and F. Rassoulzadegan. (1988).
Microbial loop in an oligrotrophic pelagic marine ecosystem: possible roles of
cyanobacteria and nanoflagellates in the organic fluxes. Marine Ecology – Progress
Series. 49:171-178.

Hahn, M. W., and M. G. Höfle. (1999). Flagellate Predation on a Bacterial Model
Community: Interplay of Size-Selective Grazing, Specific Bacterial Cell Size, and
Bacterial Community Compositoin. Applied and Environmental Microbiology.
65(11):4863-4872.

Halewood, E.R., C.A. Carlson, M.A. Brzezinski, D.C. Reed, and J. Goodman. (2012).
Annual cycle of organic matter partitioning and its availability to bacteria across the
Santa Barbara Channel continental shelf. Aquat. Microb. Ecol. 67:189-209.

Hamersley M.R., K.A. Turk, A. Leinweber, N. Gruber, J.P. Zehr, T. Gunderson, and D.G.
Capone. (2011). Nitrogen fixation within the water column associated with two
hypoxic basins in Southern California Bight. Aquat. Microb. Ecol. 63:193-205.

Hansell, D.A., C.A. Carlson, and S. Suzuki. 2002. Dissolved organic carbon export with
North Pacific Intermediate Water formation. Global Biogeochemical Cycles 16:77–84.

Teel
199
Hansell, D.A., C. A. Carlson, D.J. Repeta, and R. Schlitzer. (2009). Dissolved organic
matter in the ocean. Oceanography. 22(4):202-211.

Hansell, D. A., Carlson, C. A., & Schlitzer, R. (2012). Net removal of major marine
dissolved organic carbon fractions in the subsurface ocean. Global Biogeochemical
Cycles, 26(1).

Haskell, W. Z., Prokopenko, M. G., Hammond, D. E., Stanley, R. H., Berelson, W. M.,
Baronas, J. J., Fleming, J. C., and Aluwihare, L. (2016). An organic carbon budget for
coastal Southern California determined by estimates of vertical nutrient flux, net
community production and export. Deep Sea Research Part I: Oceanographic
Research Papers, 116, 49-76.

Haskell, W. Z., Prokopenko, M. G., Hammond, D. E., Stanley, R. H., & Sandwith, Z. O.
(2017). Annual cyclicity in export efficiency in the inner Southern California
Bight. Global Biogeochemical Cycles, 31(2), 357-376.

Hasumi, H. and T. Nagata. (2014). Modeling the global cycle of marine dissolv
ed organic matter and its influence on marine productivity. Ecological Modelling.
288:9-24.

Hewson, I., and J. A. Fuhrman. (2004). Richness and diversity of bacterioplankton
species along an estuarine gradient in Moreton Bay, Australia. Applied and
Environmental Microbiology. 70(6):3425-3433.

Hewson, I., J.A. Steele, D.G. Capone, and J.A. Fuhrman. (2006). Temporal and spatial
scales of variation in bacterioplankton assemblages of oligotrophic surface waters.
Mar. Eco. Prog. Series 311:67-77.

Holt, B. (2011). “Coastal Upwelling and Harmful Algal Blooms in Southern California
(September-October 2011)” Physical Oceanography Distributed Active Archive
Center. Article published online at:
podaac.jpl.nasa.gov/oceanevents/hab_sc_septemter-october2011

Hoppe, H.-G., K. Gocke, R. Koppe, and C. Begler. (2002). Bacterial growth and
primary production along a north-south transect of the Atlantic Ocean. Nature.
416:168-171.

Hu H. and T. Liu. (2003). Oceanic thermal and biological responses to Santa Ana
winds. Geophysical Research Letters. 30(11)1596. Doi:10.1029/2003GL017208.

Hunt D.E., Y.J. Lin, M.J. Church, D.M. Karl, S.G. Tringe, L.K. Izzo, and Z.I. Johnson.
(2013). Relationship between abundance and specific activity of bacterioplankton in
open ocean surface waters. Appl. Environ. Microbiol. 79(1):177-184.

Teel
200
Huyer, A. (1983). Coastal upwelling in the California Current System. Prog. Oceanog.
12:259-284.

Iturriaga, R. and B.G. Mitchell. (1986). Chroococcoid cyanobacteria: a significant
component in the food web dynamics of the open ocean. Mar. Ecol. Prog. Ser. 28:291-
297.

James AK, Passow U, Brzezinski MA, Parsons RJ, Trapani JN, Carlson CA (2017)
Elevated pCO
2
enhances bacterioplankton removal of organic carbon. PLoS ONE
12(3): e0173145. doi:10.1371/journal.pone.0173145

Jacox, M. G., C. A. Edwards, M. Kahru, D. L. Rudnick, and R. M. Kudela. 2015. The
Potential for Improving Remote Primary Productivity Estimates through Subsurface
Chlorophyll and Irradiance Measurement. Deep-Sea Research II. 112:107-116.

James, A. K., U. Passow, M. A. Brzezinski, R. J. Parsons, J. N. Trapani, and C. A. Carlson.
(2017). Elevated pCO2 enhances bacterioplankton removal of organic carbon. PLoS
ONE 12(3): e0173145. doi:10.1371/journal.pone.0173145joo
Jiao, N.-Z., G.J. Herndl, D.A. Hansell, R. Benner, G. Kattner, S.W. Wilhelm, D.L.
Kirchman, M.G. Weinbauer, T.-W. Luo, F. Chen, and F. Azam. (2010). Microbial
production of recalcitrant dissolved organic matter: long-term carbon storage in the
global ocean. Nature Reviews: Microbiology. 8:593-599.

Jones, A. C., T. S. V. Liao, F. Z. Najar, B. A. Roe, K. D. Hambright, and D. A. Caron.
(2013). Seasonality and disturbance: annual pattern and response of the bacterial
and microbial eukaryotic assemblages in a freshwater ecosystem. Environmental
Microbiology. 15(9):2557-2572.

Khatiwala, S., T. Tanhua, S. Mikaloff Fletcher, M. Gerber, S.C. Doney, H.D. Graven, N.
Gruber, G.A. McKinley, A. Murata, A.F. Rios, and C.L. Sabine. Global ocean storage of
anthropogenic carbon. Biogeosciences. 10:2169-2191.

Kim, D. Y., Countway, P. D., Jones, A. C., Schnetzer, A., Yamashita, W., Tung, C., &
Caron, D. A. (2014). Monthly to interannual variability of microbial eukaryote
assemblages at four depths in the eastern North Pacific. The ISME journal, 8(3), 515-
530.

Kirchman, D.L., E K’nees, and R. Hodson. (1985). Leucine incorporation and its
potential as a measure of protein synthesis by bacteria in natural aquatic systems.
Appl. Environ. Microbiol. 49(3):599-607.

Kirchman, D.L., Y. Suzuki, C. Garside, and H.W. Ducklow. (1991). High turnover rates
of dissolved organic carbon during a spring phytoplankton bloom. Nature. 352:612-
614.
Teel
201

Kirchman, D. L., J. H. Rich, and R. T. Barber. (1995). Biomass and biomass production
of heterotrophic bacteria along 140˚W in the equatorial Pacific: effect of
temperature on the microbial loop. Deep-Sea Research II. 42(2-3):603-619.

Kirchman, D. L., X. A. G. Moran, and H. Ducklow. (2009). Microbial growth in the
polar oceans – role of temperature and potential impact of climate change. Nature
Reviews Microbiology. 7:451-459.

Kirchman, D. L. (2016). Growth Rates of Microbes in the Oceans. Annual Review of
Marine Science. 8:285-309.

Klindworth, A., A. J. Mann, S. Huang, A. Wichels, C. Quast, J. Waldmann, H. Teeling,
and F. O. Glöckner. (2014). Diversity and activity of marine bacterioplankton during
a diatom bloom in the North Sea assessed by total RNA and pyrotag sequencing.
Marine Genomics. 18:185-192.

Kolber, Z. and P.G. Falkowski. (1993). Use of Active Fluorescence to Estimate
Phytoplankton Photosynthesis in Situ. Limnol. Oceanogr.. 38: 1646–1665.

Krause, G. H. and E. Weis. (1991). Chlorophyll fluorescence and photosynthesis - the
basics. Annu. Rev. Plant Phys.. 42: 313–349.

Krempin, D.W. and C.W. Sullivan.(1981). The seasonal abundance, vertical
distribution, and relative microbial biomass of chroococcoid cyanobacteria at a
statin in southern California coastal waters. Canadian Journal of Microbiology.
27(12):1341-1344.

Lai, C.C., Y.W. Fu , H.B. Liu

, H.Y. Kuo

, K.W. Wang , C.H. Lin , J.H. Tai , G. T. F. Wong,
K.Y. Lee , T.Y. Chen , Y. Yamamoto , M.F. Chow , Y. Kobayashi , C.Y. Ko , and F.K. Shiah.
(2014). Distinct bacterial-production–DOC–primary-production relationships and
implications for biogenic C cycling in the South China Sea shelf. Biogeosciences.
11:147-156.

Lankiewicz, T. S., Cottrell, M. T., & Kirchman, D. L. (2016). Growth rates and rRNA
content of four marine bacteria in pure cultures and in the Delaware estuary. The
ISME journal, 10(4), 823-832.

Larsen, P.E., S.M. Gibbons, and J.A. Gilbert. (2012). Modeling microbial community
structure and functional diversity across time and space. FEMS Microbiol. Lett.
332:91-98.

Laufkötter, C., M. Vogt, N. Gruber, O. Aumont, L. Bopp, S. C. Doney, J. P. Dunne, J.
Hauck, J. G. John, I. D. Lima, R. Seferian, and C. Völker. (2015). Projected decreases in
future marine export production: the role of the carbon flux through the upper
ocean ecosystem. Biogeosciences Discussions . 12:19941-19998.
Teel
202

Lawrenz, E. and T.L. Richardson. (2010). How does the species used for calibration
affect chlorophyll a measurements by in situ fluorometry? Estuaries and Coasts.
doi:10.1007/s12237-010-9346-6.

Lavigne H., F. D’Ortenzio, H. Claustre, and A. Poteau. (2012). Towards a merged
satellite and in situ fluorescence ocean chlorophyll product. Biogeosciences. 9:2111-
2125.

Lennert-Cody C.E. and P.J.S. Franks. (2002). Fluorescence patches in high-frequency
internal waves. Mar. Ecol. Prog. Ser. 235:29-42.

Levine, NM (submitted) Life in a patchy ocean: the impact of fine-scale
spatiotemporal heterogeneity on ecosystem dynamics.

Lignell, R. , H. Haario, M. Laine, and T.F. Thingstad. (2013). Getting the “right” values
for models of the pelagic microbial food web. Limnol. Oceanogr. 58(1):301-313.

Lindsay, K. (2012). The CESM ocean ecosystem scientific description.
http://www.cesm.ucar.edu/models/cesm1.1/pop2/doc/sci/ecosys_placeholder.ht
ml , National Center for Atmospheric Research.

López-Urrutia, A. and X. A. G. Morán. (2007). Resource limitation of bacterial
production distorts the temperature dependence of oceanic carbon cycling. Ecology.
88(4):817-822.

Lucas, A.J., C. L. Dupont, V. Tai, J.L. Largier, B. Palenik, and P.J.S. Franks. (2011). The
green ribbon: multiscale physical control of phytoplankton productivity and
community structure over a narrow continental shelf. Limnol. Oceanogr. 56(2):611-
626.

Marra, J. (2003). 14C-Uptake by phytoplankton, now and in the future. Limnol.
Oceanogr. Bull. 12(1):1-3.

Marra, J. (2009). Net and gross productivity: weighing in with 14C. Aquat. Microb.
Ecol. doi:10.3354/ame01306.

McCarren, J., J. W. Becker, D. J. Pepeta, Y. Shi, C. R. Young, R. R. Malmstrom, S. W.
Shisholm, and E. F. Delong. (2010). Microbial community transcriptomes reveal
microbes and metabolic pathways associated with dissolved organic matter
turnover in the sea. Proceedings of the National Academy of Sciences.
107(38):16420-16427.

Moore, J. K., S. C. Doney, J. A. Kleypas, D. M. Glover, and I. Y. Fung. (2002). An
intermediate complexity marine ecosystem model from the global domain. Deep-Sea
Research II.49:403-462.
Teel
203

Moore, J. K., S. C. Doney, K. Lindsay. (2004). Upper ocean ecosystem dynamics and
iron cycling in a global three-dimensional model. Global Biogeochemical Cycles. 18,
GB4028, doi:10.1029/2004GB002220.

Moore, J. K., Lindsay, K., Doney, S. C., Long, M. C., & Misumi, K. (2013). Marine
ecosystem dynamics and biogeochemical cycling in the Community Earth System
Model [CESM1 (BGC)]: Comparison of the 1990s with the 2090s under the RCP4. 5
and RCP8. 5 scenarios. Journal of Climate, 26(23), 9291-9312.

Moorthi, S.D., P.D. Countway, B.A. Stauffer, and D.A. Caron. (2009). Use of
quantitative real-time PCR to investigate the dynamics of the red tide dinoflagellate
Lingulodinium polyedrum. Microbial Ecology. 52(1):136-150.

Morán, X.A.G., M. Estrada, J.M. Gasol, and C. Pedrós-Alió. (2002). Dissolved primary
production and the strength of phytoplankton-bacterioplankton coupling in
contrasting regions. Microbial Ecology. 44(3):217-223.

Morris, R. M., B. L. Nunn, C. Fazar, D. R. Goodlett, Y. S. Ting, and G. Rocap. (2010).
Comparative metaproteomics reveals ocean-scale shifts in microbial nutrient
utilization and energy transduction. The ISME Journal. 4:673:685.

Mou, X., S. Sun, R. A. Edwards, R. E. Hudson, and M. A. Moran. (2008). Bacterial
carbon processing by generalist species in the coastal ocean. Letters to Nature.
451:708-711.

Müller, P., X.-P. Li, and K.K. Niyogi. (2001). Non-Photochemical Quenching. A
Response to Excess Light Energy. Plant Physiol.. 125: 1558–1566.
doi:10.1104/pp.125.4.1558.

Nagata, T. (2000) Production mechanisms of dissolved organic matter. In: DL
Kirchman(ed) Microbial Ecology of the Ocean. John Wiley and Sons. Chichester, UK.
121-152

Needham, D. M. , C-E T. Chaw, J. A. Cram, R. Sachdeva, A. Parada, J. A. Fuhrman. 2013.
Short-term observations of marine bacterial and viral communities: patterns,
connections and resilience. The ISME Journal . 7:1274-1285.

Needham, D.M. and J. A. Fuhrman. 2016. Pronounced Daily Succession of
Phytoplankton, Archaea, and Bacterial Following a Spring Bloom. Nature
Microbiology. 1:1605. doi:10.1038/nmicrobiol.2016.5.

Noble, M.A., Rosenberger, K.J., Hamilton, P., and Xu, J.P., 2009, Coastal ocean
transport patterns in the central Southern California Bight, in Lee, H.J., and Normark,
W.R., eds., Earth Science in the Urban Ocean: The Southern California Continental
Borderland: Geological Society of America Special Paper 454, p. 193–226, doi:
Teel
204
10.1130/2009.2454(3.3).
Noble, M., B. Jones, P. Hamilton, J. Xu, G. Robertson, L. Rosenfeld, and J. Largier.
(2009). Cross-Shelf Transport Into Nearshore Waters Due to Shoaling Internal Tides
in San Pedro Bay, CA. Continental Shelf Research 29(15): 1768–1785.

Oey, L.-Y. 1999. A Forcing Mechanism for the Poleward Flow off the Southern
California Coast. Journal of Geophysical Research. 104(C6):13529-13539.

Ouverney, C.C. and J.A. Fuhrman. (1999). Combined microautoradiography-16S
rRNA probe technique for determination of radioisotope uptake by specific
microbial cell types in situ. Appl. Environ. Microbiol. 65(4):1746-1752.

Pomeroy, L.R. and D. Deibel. (1986). Temperature regulation of bacterial activity
during the spring bloom in Newfoundland coastal waters. Science. 233:359-361.

Ratkowsky, D. A., R. K. Lowry, T. A. McMeekin, A. N. Stokes, and R. E. Chandler.
(1983). Model for Bacterial Culture Growth Rate Throughout the Entire Biokinetic
Temperature Range. Journal of Bacteriology. 154(3):1222-1226.

Riemann, L., G. F. Steward, and F. Azam (2000). Dynamics of Bacterial Community
Composition and Activity during a Mesocosm Diatom Bloom. Applied and
Environmental Microbiology. 66(2):578-587.

Rivkin, R.B. and L. Legendre. (2001). Biogenic Carbon Cycling in the Upper Ocean:
Effects of Microbial Respiration. Science. 291:2398-2400.

Ryan J., D. Greenfield, R. Marin III, C. Preston, B. Roman, S. Jensen, D. Pargett, J. Birch,
C. Mikulski, G. Doucette, and C. Scholin. (2011). Harmful phytoplankton ecology
studies using an autonomous molecular analytical and ocean observing network.
Limnol. Oceanogr. 56(4):1255-1272.

Sarmiento, H. and J.M. Gasol. (2012). Use of phytoplankton-derived dissolved
organic carbon by different types of bacterioplankton. Environmental Microbiology.
14(9):2348-2360.

Schnetzer, A., P.E. Miller, R.A. Schaffner, B.A. Stauffer, B.H. Jones, S.B. Weisberg, P.M.
DiGiacomo, W.M. Berelson, and D.A. Caron. (2007). Blooms of Pseudo-nitzschia and
domoic acid in the San Pedro Channel and Los Angeles harbor areas of the Southern
California Bight, 2003-2004. Harmful Algae. 6:372-387.

Seegers, B.N., C. Scholin, E. Seubert, M. Howard, G. Robertson, D. A. Caron, and B. H.
Jones. The Role of Coastal Ocean Dynamics in the Initiation and Transport of
Harmful Algal Blooms in Southern California. (In Preparation).

Teel
205
Simon M. and F. Azam (1989). Protein content and protein synthesis rates of
planktonic marine bacteria. Mar. Ecol. Prog. Ser. 51:201-213.

Sinsabaugh, R.L., S. Manzoni, D.L. Moorhead, and A. Richter. (2013). Carbon use
efficiency of microbial communities: stoichiometry, methodology, and modeling.
Ecology Letters. 16:930-939.

Sintes, E., K. Stoderegger, V. Parada, and G.J. Herndl. (2010). Seasonal dynamics of
dissolved organic matter and microbial activity in the coastal North Sea. Aquat.
Microb. Ecol. 60:85-95.

Smith D.C. and F. Azam. (1992). A simple, economical method for measuring
bacterial protein synthesis rates in seawater using
3
H-leucine. Marine Microb. Food
Webs. 6(2):107-114.

Smith W.O., R.T. Barber, and S.A. Huntsman. (1977). Primary production off the
coast of northwest Africa: excretion of dissolved organic matter and its
heterotrophic uptake. Deep-Sea Res. 24:35-47.

Sprintall, J. and M. Tomczak. 1992. Evidence of the Barrier Layer in the Surface
Layer of the Tropics. Journal of Geophysical Research. 97(C5):7305-7316.

Steele, J.A., P. D. Countway, L. Xia, P.D. Vigil, M. Beman, D. Y. Kim, C.-E. T. Chow, R.
Sachdeva, A. C. Jones, M. S. Schwalbach, J. M. Rose, I. Hewson, A. Patel, F.-Z. Sun, D. A.
Caron, and J. A. Fuhrman.

Stoderegger, K. and G.J. Herndl. (1998). Production and release of bacterial capsular
material and its subsequent utilization by marine bacterioplankton. Limnol.
Oceanogr. 43(5):877-884.

Strub, P.T., James, C., 2000]. Altimeter-derived variability of surface velocities in the
California Current System: 2. Seasonal circulation and eddy statistics. Deep Sea
Research Part II: Topical Studies in Oceanography 47, 831–870. 10.1016/S0967-
0645(99)00129-0.
Stull, R. B. (1988). “Boundary Conditions and Surface Forcing.” pp. 251-294. In: An
Introduction to Boundary Layer Meteorology. Ed. R. B. Stull. Springer. ISBN: 978-94-
009-3027-8

Suttle, C. A. (2007). Marine viruses—major players in the global ecosystem. Nature
Reviews Microbiology, 5(10), 801-812.

Teel, E. N., X. Liu, B. N. Seegers, M. A. Ragan, B. H. Jones, and N. M. Levine.
Contextualizing time-series data: Using high-resolution in-situ gliders to quantify
short-term regional oceanographic variability in the San Pedro Channel. In
Preperation.
Teel
206

Teeling H., B.M. Fuchs, D. Becher, C. Klockow, A. Gardebrecht, C.M. Bennke, M.
Kassabgy, S. Huang, A.J. Mann, J. Waldmann, M. Weber, A. Klindworth, A. Otto, J.
Lange, J. Bernhardt, C. Reinsch, M. Hecker, J. Peplies, F.D. Bockelmann, U. Callies, G.
Gerdts, A. Wichels, K.H. Wiltshire, F.O. Glöckner, T. Schweder, and R. Amann. (2012).
Substrate-controlled succession of marine bacterioplankton populations induced by
a phytoplankton bloom. Science. 336: 608-611.

Teira, E., M.J. Pazó, P. Serret, and E. Fernández. (2001). Dissolved organic carbon
production by microbial populations in the Atlantic Ocean. Limnol. Oceanogr.
46(6):1370-1377.

Thingstad, T.F. and R. Lignell. (1997). Theoretical models for the control of bacterial
growth rate, abundance, diversity, and carbon demand. Aquat. Microb. Ecol. 13:19-
27.

Todd, R.E., D.L. Rudnick, and R.E. Davis. (2009). Monitoring the greater San Pedro
Bay region using autonomous underwater gliders during fall of 2006. J. Geophys. Res.
114. C06001. doi:10.1029/2008JC005086.

Tortell P.D., M.T. Maldonado, J. Granger, and N.M. Price. (1999). Marine bacteria and
biogeochemical cycling of iron in the oceans. FEMS Microbiol. Ecol. 29(1):1-11.

Touratier, F., Legendre, L., & Vézina, A. (1999). Model of bacterial growth influenced
by substrate C: N ratio and concentration. Aquatic Microbial Ecology, 19(2), 105-118.

Van Mooy, B. A. S., A. H. Devol, and R. G. Keil. (2004). Relationship between bacterial
community structure, light, and carbon cycling in the eastern subarctic North
Pacific. Limnology and Oceanography. 49(4):1056-1062.

Van Wambeke, F., S. Bonnet, T. Moutin, P. Raimbault, G. Alarcón, and C. Guieu.
(2008). Factors limiting heterotrophic bacterial production in the southern Pacific
Ocean. Biogeosciences. 5:833-845.

Venrick, E.L. (2002). Floral patterns in the California Current System off southern
California: 1990-1996. J. Marine Research. 60:171-189.

Vichi, M., Pinardi, N., & Masina, S. (2007). A generalized model of pelagic
biogeochemistry for the global ocean ecosystem. Part I: Theory. Journal of Marine
Systems, 64(1), 89-109.

Vichi, M., Masina, S., & Navarra, A. (2007). A generalized model of pelagic
biogeochemistry for the global ocean ecosystem. Part II: Numerical
simulations. Journal of Marine Systems, 64(1), 110-134.

Teel
207
Vichi, M., & Masina, S. (2009). Skill assessment of the PELAGOS global ocean
biogeochemistry model over the period 1980–2000. Biogeosciences, 6(11), 2333-
2353.

Walsh, J.J., D.A. Dieterle, F.E. Muller-Karger, R. Bohrer, and W.P. Bissett. (1999).
Simulation of carbon-nitrogen cycling during spring upwelling in the Cariaco Basin.
Marine Science Faculty Publications. Paper 17.

Ward, C. S., C-M. Yung, K. M. Davis, S. K. Blinebry, T. C. Williams, Z. I. Johnson, and D.
E. Hunt. (2017). Annual community patterns are driven by seasonal switching
between closely related marine bacteria. The ISME Journal. 1-11.

Wear, E. K., C. A. Carlson, L. A. Windecker, and M. A. Brzezinski (2015). Roles of
diatom nutrient stress and species identity in determining the short- and long-term
bioavailability of diatom exudates to bacterioplankton. Marine Chemistry. 117:335-
348.

Wear, E. K., C. A. Carlson, A. K. James, M. A. Brzezinski, L. A. Windecker, and C. E.
Nelson. (2015). Synchronous shifts in dissolved organic carbon bioavailability and
bacterial community responses over the course of an upwelling-driven
phytoplankton bloom. Limnology and Oceanography. 60:657-677.

Weitz, J.S., C.A. Stock, S.W. Wilhelm, L. Bourouiba, A. Buchan, M.L. Coleman, M.J.
Follows, J.A. Fuhrman, L.F. Jover, J.T. Lennon, M. Middelboe, D.L. Sonderegger, C.A.
Suttle, B.P. Taylor, T.F. Thingstad, W.H. Wilson, and K.E. Wommack. (2014). A
multitrophic model to quantify the effects of marine viruses on microbial food webs
and ecosystem processes. The ISME Journal . 1-13.

White, P. A., J. Kalff, J. B. Rasmussen, and J. M. Gasol. (1991). The Effect of
Temperature and Algal Biomass on Bacterial Production and Specific Growth Rate in
Freshwater and Marine Habitats. Microbial Ecology. 21:99-118.

Williams T.J., D. Wilkins, E. Long, F. Evans, M.Z. DeMeere, M.J. Raftery, and R.
Cavicchioli. (2013). The role of Flavobacteria in processing algal organic matter in
coastal East Antarctica revealed using metagenomics and metaproteomics. Environ.
Microbiol. 15(5):1302-1317.

Williams, P.J. leB. (1981). Microbial contribution to overall marine plankton
metabolism: direct measurements of respiration. Oceanol. Acta. 4(3):359-364.

Winter, C., T. Bouvier, M. G. Weinbauer, and T. F. Thingstad. (2010). Trade-offs
between competition and defense specialists among unicellular planktonic
organisms: the “kill the winner” hypothesis revisited. Microbial and Molecular
Biology Reviews. 74(1):42-57.

Teel
208
Wright, J. J., Konwar, K. M., & Hallam, S. J. (2012). Microbial ecology of expanding
oxygen minimum zones. Nature Reviews Microbiology, 10(6), 381-394.

Xiu P and F. Chai. (2014). Connections between physical, optical, and
biogeochemical processes in the Pacific Ocean. Progress in Oceanography. 122:30-
53.

Yooseph, S. K. H. Nealson, D. B. Rusch, J. P. McCrow, C. L. Dupont, M. Kim, J. Johnson,
R. Montgomery, S. Ferriera, K. Beeson, S. J. Williamson, A. Tovchigrechko, A. E. Allen,
L. A. Zeigler, G. Sutton, E. Eisenstadt, Y-H. Rogers, R. Friedman, M. Frazeier, and J. C.
Venter. (2010). Genomic and fuctional adaptation in surface ocean planktonic
prokaryotes. Nature. 468:60-67.

Zehr J.P. and R.M. Kudela. (2011). Nitrogen cycle of the open ocean: from genes to
ecosystems. Annu. Rev. Mar. Sci. 3:197-225.

Ziveri, P., R.C. Thunell, and D. Rio. (1995). Export production of coccolithophores in
an upwelling region: results from San Pedro Basin, Southern California Borderlands.
Marine Micropaleontology. 24:335-358.

Abstract (if available)

Abstract The oceans have taken up approximately half of anthropogenic carbon emissions thereby buffering the impact of climate change

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Marine biogenic halocarbons: potential for heterotrophic bacterial production and seasonality at San Pedro Ocean Time-series

PDF

The impact of mesoscale and submesoscale physical processes on phytoplankton biomass, community composition, and carbon dynamics in the oligotrophic ocean

PDF

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

PDF

Temporal variability of marine archaea across the water column at SPOT

PDF

Identifying functional metabolic guilds: a computational approach to classifying heterotrophic diversity in the marine system

PDF

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

PDF

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

PDF

The development of novel measures of landscape diversity in assessing the biotic integrity of lotic communities

PDF

Ecological patterns of free-living and particle-associated prokaryotes, protists, and viruses at the San Pedro Ocean Time-series between 2005 and 2018

PDF

Enhancing recovery of understudied and uncultured lineages from metagenomes

PDF

Investigations of animal-sediment-microbe interactions at two different environments: coastal lagoons and methane seeps

PDF

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

PDF

Patterns of molecular microbial activity across time and biomes

PDF

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

PDF

Microbial ecology in the deep terrestrial biosphere: a geochemical, metagenomic and culture-based approach

Asset Metadata

Creator Teel, Elizabeth Nelson (author)

Core Title Multi-dataset analysis of bacterial heterotrophic variability at the San Pedro Ocean Time-series (SPOT): an investigation into the necessity and feasibility of incorporating a dynamic bacterial c...

Contributor Electronically uploaded by the author (provenance)

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Marine Biology and Biological Oceanography

Publication Date 07/22/2017

Defense Date 04/03/2017

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

Tag bacteria,BGE,Carbon,ecosystem,efficiency,gliders,heterotroph,marine,microbial,Model,OAI-PMH Harvest,oceanography,Production,SPOT,time series

Language English

Advisor Levine, Naomi (committee chair), Fuhrman, Jed (committee member), Hammond, Doug (committee member), Jones, Burton (committee member)

Creator Email elizabeth.teel@gmail.com,eteel@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-410018

Unique identifier UC11265399

Identifier etd-TeelElizab-5592.pdf (filename),usctheses-c40-410018 (legacy record id)

Legacy Identifier etd-TeelElizab-5592.pdf

Dmrecord 410018

Document Type Dissertation

Rights Teel, Elizabeth Nelson

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