AN ABSTRACT OF THE THESIS OF

advertisement
AN ABSTRACT OF THE THESIS OF
Samuel R. Laney for the degree of Master of Science in Oceanography presented on
August 25, 2000. Title: Laboratory Investigations of the Natural Fluorescence of
Marine Phytoplankton. Redacted for Privacy
Abstract approved:
Mark R. Abbott
The influence of nitrate availability and irradiance on phytoplankton natural
fluorescence was investigated in laboratory cultures of the marine diatom
Thalassiosira weissflogii (Bacillariophyceae). Two cultures of phytoplankton,
differing only in the nitrate-limited growth rate, were compared to learn how increases
in irradiance influences natural fluorescence. Instrumentation was developed for these
experiments to maintain and manipulate physical and chemical conditions in the
culture vessel while unobtrusively and continuously monitoring the natural
fluorescence response. Variable fluorescence in each culture was simultaneously
measured to provide a concurrent data set of photosynthetic physiology. Analysis of
results from these experiments suggest that gross features in the relationship between
fluorescence and irradiance contain information about the way in which phytoplankton
perceive the relative abundance of environmental nitrate. This relationship becomes
less informative at higher irradiance levels due to the increasing influence of
photoinhibition and photoprotection.
© Copyright by Samuel R. Laney
August 25, 2000
All Rights Reserved
Laboratory Investigations of the Natural Fluorescence of Marine Phytoplankton
by
Samuel R. Laney
A THESIS
submitted to
Oregon State University
in partial fulfillment of
the requirements for the
degree of
Master of Science
Presented August 25, 2000
Commencement June 2001
Master of Science thesis of Samuel R. Laney presented on August 25, 2000
APPROVED:
Redacted for Privacy
Maj or Professor, representing Oceanography
Redacted for Privacy
Dean of College of Oceanograpflic and Atmospheric Sciences
Redacted for Privacy
Dean of Grte School
I understand that my thesis will become part of the permanent collection of Oregon
State University libraries. My signature below authorizes release of my thesis to any
reader upon request.
Redacted for Privacy
Samuel R. Laney, Author
A CKNOWLEDGMENT
This thesis would have been severely restricted in scope and my education at
COAS would have been substantially poorer had not Ricardo Letelier been such an
encouraging and active mentor. The substantial difference between an academic and a
scholar is clearly seen in Ricardo, who has been an excellent role model for the latter.
I am indebted to Mark Abbott, my advisor, for supporting my research at all
stages and at many different levels. Mark always provided a greater framework into
which this research would fit and the resources to complete the tasks at hand
efficiently. Although it was easy for me to lose myself in the details of physiology,
Mark always had the big-picture, goal-oriented questions that returned this thesis to
oceanography when physiology became less useful than interesting.
Many of my fellow students at COAS know how important they have been to
me both in and out of the classroom. Over my three years of thesis research, repeated
discussions with fellow students have led to, I feel, a greater degree of clarity in my
discussion and a greater degree of focus in my presentation. My thanks to all who
showed me my shortcomings and helped me improve.
No work ever truly stands alone. Many people were consulted, badgered,
bribed, intenogated, plied, wheedled or otherwise involved in this thesis, all of whom
added a piece to the puzzle. Although too numerous to enumerate here, everyone who
helped me at any level, no matter how small, is very much appreciated and has been
very much the mortar holding these bricks together.
Support for this work has been provided by the National Aeronautics and
Space Administration (NAS5-3 1360).
CONTRIBUTION OF AUTHORS
Dr. Ricardo Letelier and Dr. Mark Abbott were both involved in the design, research,
analysis and writing of each manuscript. Dr. Russ Desiderio was extensively involved
in the analysis and writing of the first manuscript. Dr. Dale Kiefer and C. R. Booth
provided comments and helpful suggestions during preparation of the first manuscript
TABLE OF CONTENTS
1. Introduction
.1
1.1
Background
1.2
Research Objectives and Overview ................................................................. 3
1.3
Fundamental Findings and Relevance ............................................................. 6
.1
2. Measuring the Natural Fluorescence of Phytoplankton Cultures ............................. 8
2.1
Abstract ........................................................................................................... 8
2.2
Introduction ..................................................................................................... 9
2.2.1
2.2.2
2.3
System Design ............................................................................................... 12
2.3.1
2.3.2
2.3.3
2.4
Motivation ...........................................................................................9
Requirements ..................................................................................... 11
Chemostat System ............................................................................. 12
Fluorescence Detection System ......................................................... 15
Control and Acquisition System ........................................................ 16
Testing and Evaluation .................................................................................. 17
2.4.1
2.4.2
2.4.3
2.4.4
The Ambient Light Field ................................................................... 17
Artifacts in Fnat Measurements .......................................................... 17
Artifacts in Fstjm Measurements ......................................................... 21
Artifacts in Fluorescence Action Spectra .......................................... 23
2.5
Live Culture Data .......................................................................................... 25
2.6
Summary ....................................................................................................... 29
2.7
References ..................................................................................................... 30
3. The Influence of Light and Nitrate on the Natural Fluorescence of the Marine
Diatom Thalassiosira weissflogii (Bacillariophyceae) ................................................ 32
TABLE OF CONTENTS (continued)
3.1
Abstract
3.2
Introduction ................................................................................................... 33
3.3
Methods ......................................................................................................... 36
3.4
Results ........................................................................................................... 41
3.4.1
3.4.2
3.4.3
3.5
.32
Culture Response to an Increase in Irradiance .................................. 41
Nitrate-Driven Responses under LL and HIlL .................................... 47
Diel Variability during Intensive Sampling ...................................... 54
Discussion ..................................................................................................... 54
3.5.1
3.5.2
3.5.3
Correlation between 1' and [chl a] .................................................... 54
Nitrate-driven Variability in 1 ........................................................... 58
Physiological Sources of Variability in Fnat ...................................... 61
3.6
Conclusions ................................................................................................... 65
3.7
References ..................................................................................................... 66
4. Assessment of Reaction Center Connectivity in Phytoplankton using Fast
Repetition Rate Fluorometry ....................................................................................... 68
4.1
Abstract ......................................................................................................... 68
4.2
Introduction ................................................................................................... 69
4.2.1
4.2.2
4.3
Methods ......................................................................................................... 74
4.3.1
4.3.2
4.3.3
4.4
Background ....................................................................................... 69
Motivation ......................................................................................... 71
PSII Electron Flow Simulations ........................................................ 74
Estimates of p in Continuous Cultures .............................................. 75
Numerical Simulations using Actual Culture Data ........................... 76
Results ........................................................................................................... 77
4.4.1
4.4.2
in Different Models of PSI! Structure .......................................... 77
Comparing Software Performance with Simulated Data .................. 79
Ct,
TABLE OF CONTENTS (continued)
4.4.3
4.4.4
4.5
Variable Fluorescence Time Series from Cultures ............................ 80
Assimilation of Time Series Results ................................................. 83
Discussion ..................................................................................................... 85
4.5.1
4.5.2
.........
Connectivity and the Analysis of Variable Fluorescence
85
Does p Truly Reflect PSII Connectivity7 .......................................... 87
4.6
Conclusions ................................................................................................... 88
4.7
Appendices .................................................................................................... 89
4.7.1
4.7.2
4.8
Linear Approximation of Equation 4.2 ............................................. 89
Corrections to Kolber et al. (1998) .................................................... 90
References ....................................................................................................91
5. SUMMARY ............................................................................................................93
5.1
Practical Importance of this Research ........................................................... 93
5.2
Implications of the Observed Relationships in
5.3
Natural Fluorescence and Photosynthetic Rate ............................................. 96
5.4
Implications of Connectivity ......................................................................... 99
5.5
Future Work ..................................................................................................99
Fnat
95
.......................................
BIBLIOGRAPHY ...................................................................................................... 102
LIST OF FIGURES
Figure
2.1 Functional diagram of NFC .................................................................................. 14
2.2 Spectral distribution of growth irradiance in the chemostat for four
different intensity levels ....................................................................................... 18
2.3 Correlation between the apparent natural fluorescence of pure media
versus irradiance level .......................................................................................... 20
2.4 Apparent weak stimulated fluorescence of pure water as a function of
excitationwavelength ........................................................................................... 24
2.5 A) Natural fluorescence signal versus Julian day. Each day is roughly
sinusoidal in shape. B) Natural fluorescence normalized to incident
irradiance .............................................................................................................. 26
2.6 Scatter plots of irradiance normalized natural fluorescence versus
concentration of chlorophyll a .............................................................................. 28
3.1 Cell abundance, [chl a] and cell specific [chi a] in the low growth rate
(LGR) culture ....................................................................................................... 42
3.2 Cell abundance, [chl a] and cell specific [chi a] during the HGR
experiment ............................................................................................................ 42
3.3 Irradiance and natural fluorescence parameters during the LGR culture
experiment ............................................................................................................ 44
3.4 Irradiance and natural fluorescence during the HGR culture experiment ............ 45
3.5 Photosynthetic physiological parameters measured by variable
fluorescence during the LGR culture experiment ................................................. 48
3.6 Photosynthetic physiological parameters measured by variable
fluorescence during the IIGR culture experiment ................................................49
3.7 Photosynthetic physiological parameters during a period when nitrate
availability per cell decreased under LL ............................................................... 51
LIST OF FIGURES (continued)
Figure
3.8 Photosynthetic physiological parameters during a period when nitrate is
depleted by and then provided to HGR culture experiment under LL ................. 52
3.9 Photosynthetic physiological parameters during a period when nitrate
delivery ceased to the HGR culture experiment under I-IL................................... 53
3.10 Sub-diel variability during the LGR and HGR experiments .............................. 55
3.11 Diurnal maximum (*) and lOAM (A) cJ regressed against [chl a] for two
periods of steady state growth during LGR under LL (a) and I-IL (b), and
three periods of positive population growth: LGR shift to HL (c), HGR-LL
(d) and HGR-HIL (e)............................................................................................. 57
3.12 Long term changes in the ratio of PM to AM c1i ................................................. 59
3.13 Additional empirical metrics to examine ct under HL conditions ...................... 62
3.14 The relationship between the irradiance of maximal Fch1 and the
threshold at which photosynthesis switches between light limited and light
saturated (EK) in the forenoon (*) and the afternoon (x) ...................................... 64
4.1 Typical FRRF variable fluorescence data .............................................................. 71
4.2 Fit space for the current physiological model ....................................................... 73
4.3 ctø as a function of irradiance and p for five different numerical
simulations of a population of PSII ...................................................................... 78
4.4 Best fit estimates of F0, Fm, tYp511 and p using simulated, noise-corrupted
data ........................................................................................................................ 81
4.5 Results from processing variable fluorescence time series data using the
commercial FRSv1.6 (dotted lines) and the custom v3 software (solid
lines) ..................................................................................................................... 82
4.6 EK and EPAR from LL conditions (left) and HL conditions (right) for the T.
weissflogiiculture data ......................................................................................... 84
47
pe
vs. EPAR curves from PSII electron flow models .............................................. 86
LIST OF TABLES
Table
2.1 Signal Contamination Analysis .......................................................... 19
4.1 1p for five different simulations when p = 0 and when p = 0.4 (in
parenthesis) at four selected irradiance intensities .................................... 79
Laboratory Investigations of the Natural Fluorescence of Marine
Phytoplankton
1. Introduction
"Photosynthesis is an exquisite amalgam of physics, chemistry and biology, and chlorophyll
fluorescence has become one of the most widely used, non-invasive techniques to probe the fate of light
energy during the photosynthetic process."
-Preface to the Special Volume on Chlorophyll Fluorescence, Australian Journal of Plant Physiology
22, 1995, following the Robertson Symposium on Chlorophyll Fluorescence, held 27-29 May 1994,
Canberra, ACT.
1.1 Background
To say that Earth is a blue-green planet is something of an oversimplification.
Careful measurement of the upwelling radiance leaving "blue" water will demonstrate
the existence of a small fraction of red light. A similar but stronger red signal can be
observed in the "green" color given to both the ocean and land by photosynthetic
organisms. This emission of red light is the fluorescence of chlorophyll a molecules
stimulated either directly or indirectly by solar excitation. Plant physiologists, marine
ecologists, remote sensing specialists and others refer to this emission as chlorophyll
fluorescence.
G.G. Stokes is credited for the 1852 discovery of the fluorescent properties of
chlorophyll a and phycobilin in red algae. He originally termed this phenomenon
dispersive reflection but expressed dissatisfaction with this phrase and later coined the
term "fluorescence", analogous to opalescence1. In 1874, N.J.0 Muller observed that
the fluorescence of chlorophyll a in live plants was much weaker than the fluorescence
'Referring to optical properties exhibited by the minerals fluor-spar and opal, respectively.
emitted from a similar amount of chlorophyll a in solution. From this observation an
inverse relationship between photosynthesis and fluorescence was predicted. These
two related discoveries, the former describing a biophysical phenomenon and the latter
inferring its physiological ramifications, were the beginnings of a new field of inquiry
using fluorescence to explore photosynthesis.
The fluorescence of chlorophyll a in vivo has been described as a "rich...
(and) ambiguous signal" because of the depth of information obtained from
fluorescence analysis methodologies (Govindjee 1995). This richness and ambiguity
is due partly to the configuration and functionality of photosystems, biophysical
structures in photosynthetic organisms responsible for harvesting and processing light
energy. Since photosystems are not and cannot be perfectly efficient, some of the
light energy absorbed by these structures is lost to the environment as a fluorescence
emission. Variability in this fluorescence signal on a per cell basis reflects
physiological variability in the structure and function of these photosystems, and so
physiological photosynthetic adaptation to changes in the physical and chemical
environment may be monitored using natural fluorescence variability, if photosystems
are directly or indirectly affected.
Oceanographic field measurements of fluorescence can be divided into two
categories based on the excitation source used to stimulate fluorescence. The first
category encompasses all techniques in which artificial light sources are used to
stimulate fluorescence, generally with excitation fluxes that greatly exceed natural
irradiance intensities. The fluorescence emission that results from purposeful
stimulation is often referred to as active fluorescence. Similarly, the naturally
occurring ambient light field also stimulates chlorophyll fluorescence, albeit at a much
weaker intensity, often referred to as naturalfluorescence2. Natural fluorescence can
be readily measured with passive radiometers both in situ and remotely; the recent
launch of the MODIS spectroradiometer is presently providing basin scale images of
2
Also sometimes called
passive or solar fluorescence.
3
phytoplankton natural fluorescence. The capability to measure this property on such
large spatial and temporal scales makes natural fluorescence the most likely candidate
for global monitoring of phytoplankton physiology and ecology, and more sensors are
scheduled for launch within the next decade.
For the foreseeable future, investigation of how ocean physics and chemistry
influences photosynthesis and primary production on the basin scale will be largely an
examination of natural fluorescence. Whereas remotely sensed ocean color provides
an index of phytoplankton biomass, natural fluorescence can provide an added
physiological dimension to the present generation of productivity models. This signal
can potentially be used to infer how environmental factors such as temperature,
irradiance quality and intensity, nutrient availability and other species-specific factors
affect oceanic photosynthesis both globally and synoptically. However, the
relationship between the observed natural fluorescence variability and the underlying
changes in physiology are not simple or well understood. For oceanographic
purposes, natural fluorescence research needs to focus specifically on uncovering
these fundamental relationships with respect to the dominant sources of physical and
chemical variability in the ocean.
1.2 Research Objectives and Overview
Whereas active fluorescence methods have been thoroughly examined both in
the field and in the laboratory, natural fluorescence methods have not enjoyed such
extensive examination. To date, all published investigations of natural fluorescence
have been performed in field experiments on natural populations, where
environmental variability and lack of experimental control complicate interpretation
and preclude the determination of causative relationships between environmental
factors and the observed natural fluorescence signal. Controlled laboratory
experiments can resolve both of these issues and thus are a critical tool to exploring
the physiological relationships between changes in the environmental and responses in
phytoplankton natural fluorescence.
The central goal of this research has been to examine how changes in nitrate
and light availability are reflected in the natural fluorescence signal of a model
phytoplankton species. Light and nitrate are the two dominant controls on
phytoplankton abundance in most of the ocean, limiting growth in both the Blackman
and von Leibig sense. An extensive literature backed by a large body of experimental
evidence documents the influence of these two factors, yet how these two fundamental
factors influence natural fluorescence remains largely unknown. Investigating the
relationship between light, nitrate and natural fluorescence in this study has been a
three step process, including
the design, development and characterization of instrumentation and
protocols to measure phytoplankton natural fluorescence in controlled
laboratory cultures,
the design, execution and analysis of experiments to show how the
natural fluorescence of a model species behaves both during steady
state and during changes in the irradiance and nitrate environment, and
the concunent measurement and proper interpretation of photosynthetic
physiology to identify the underlying physiological variability which
controls the observed natural fluorescence signal.
This thesis is divided into three sections following the outline given above.
The first section details the design and evaluation of instrumentation and protocols for
measuring natural fluorescence in tightly controlled laboratory cultures of
phytoplankton. Past attempts to measure natural fluorescence in the laboratory have
produced inconclusive results, but in our hands the natural fluorescence signal could
5
be measured within an ambient light field and separated from instrument artifacts.
The instrumentation used in these experiments provides for the manipulation of
culture environmental properties such as pH, temperature, irradiance and dilution rate
over ranges encompassing most of the natural conditions experienced in the ocean.
By independently manipulating specific environmental properties, the natural
fluorescence response to individual factors can be directly linked to experimental
treatments, providing a causal relationship between environmental forcing and natural
fluorescence response. Since the light source used in these experiments was
programmed to mimic a typical diurnal irradiance profile, the physiological and
fluorescence responses observed in any experiment will more accurately represent the
responses expected in field populations.
The second section describes a pair of experiments performed to determine the
specific influence of irradiance and nitrate availability on phytoplankton natural
fluorescence. Two cultures of the marine diatom Thalassiosira weissflogii
(Bacillariophyceae) were grown under identical irradiance protocols in a physical and
chemical environment that differed only with respect to nitrate availability. Irradiance
was initially set low in each culture to force light limited photosynthesis, but after
several volume turnovers irradiance was increased fivefold to light saturating levels.
From these experiments were obtained concurrent time series of environmental
parameters, natural fluorescence and photosynthetic physiology, representing the
forcing factors, the observed response and the underlying causes of this response,
respectively. Analysis of results from these experiments indicate that nitrate and
irradiance have different effects on the natural fluorescence character of the model
phytoplankton. Simple metrics were formed to track variability in natural
fluorescence over the scale of days, and certain metrics appear to indicate the onset
and relief from nitrate starvation. Although simple metrics are adequate under low
irradiance conditions, physiological adaptations during high irradiance conditions
substantially degrade their performance. As a result, although single or twice-daily
remote sensing acquisitions may be adequate to compute these metrics under low
irradiance growth where photosynthesis is light limited, more complex formulations
may be required for regions of the ocean characterized by light saturated
photosynthesis.
Analysis of the variable fluorescence data collected during these experiments
led to certain discoveries regarding the numerical and physiological interpretation of
measurements collected with a Fast Repetition Rate fluorometer (FRRF). The third
section describes the influence of a specific physiological parameter, the connectivity
of reaction centers, which can be recovered from FRRF measurements. Because this
parameter appeared to be poorly retrieved using the commercial software supplied
with FRRF, the theory behind FRRF-based determination of connectivity and the
numerics of this software were reviewed. Mathematical errors and an interdependence
between two key parameters were found in the published literature, and inspection of
the numerical methods used to estimate these parameters uncovered weaknesses in the
fitting algorithm. A custom analysis package was written to correct these problems
and using this package a better time series of concurrent physiological variability was
obtained.
1.3 Fundamental Findings and Relevance
There are four major conclusions resulting from the research presented in this
thesis. First, the physiologically driven component of natural fluorescence variability
in laboratory cultures can be readily separated from the influence of inadiance and
biomass. Second, interpretation of this physiologically driven variability shows that
changes in the natural fluorescence signal can be ascribed to specific environmental
stimuli. Third, even under light limited conditions characteristic of the "linear"
portion of the photosynthesis-irradiance relationship, fundamental photosynthetic
properties of phytoplankton vary substantially over the time scale of hours. Fourth,
variability in one of these photosynthetic parameters, p (an index of reaction center
7
connectivity), appears to have a more substantial impact on photosynthetic quantum
efficiency than previously assumed.
This research shows that natural fluorescence has the potential to recover
information about the physiology of phytoplankton from remote sensing platforms,
but we are only beginning to discover how to examine this natural fluorescence to
obtain useful physiological properties. More detailed analyses of the physiological
controls on natural fluorescence are required before variability in this optical signal
can be applied to basin scale estimates of primary productivity from space. Natural
fluorescence variability needs to be further examined as a function of other taxonomic
classes, using a wider range of limiting nutrients and with more complex models of
diurnal irradiance. The experiments discussed in this thesis are valuable because they
represent a baseline case where environmental variability is minimal, but application
of natural fluorescence techniques to field conditions will necessarily require an
expansion of these investigations along several dimensions.
2. Measuring the Natural Fluorescence of Phytoplankton Cultures3
"When you can measure what you are speaking about and express it in numbers you know something
about it, but when you cannot measure it, when you cannot express it in numbers, your knowledge is of
a meagre and unsatisfactory kind."
Lord Kelvin
2.1 Abstract
A laboratory instrument, the Natural Fluorescence Chemostat (NFC), was
developed to measure the natural fluorescence of phytoplankton cultures. With this
instrument, the physical and chemical environment of a culture can be manipulated
with respect to temperature, pH, nutrient delivery rate and light intensity, during which
time the natural fluorescence signal and a weak stimulated fluorescence signal are
continuously recorded. The geometry and spectral distribution of the artificial light
field minimize the contribution of scattering to the natural fluorescence signal.
Preliminary investigations with the marine diatom T. weissflogii (Bacillariophyceae)
and the marine chlorophyte Dunaliella tertiolecta (Chlorophyceae) indicate that the
instrument can detect natural fluorescence signals in broadband artificial light fields as
bright as 1250 j.xmol photons m2 s1. Since the physiological controls on natural
fluorescence are not well understood, laboratory experiments are essential for
investigating how ocean physics and chemistry influence this signal. This instrument
provides a quantitative means to examine how the magnitude and kinetics of
phytoplankton natural fluorescence vary in response to changes in the physical and
chemical environment.
This manuscript has been submitted for publication as Laney et al. 2000. Measuring the Natural
Fluorescence of Phytoplankton Cultures. J. Atinos. Oceanic Tech. In Review.
2.2 Introduction
2.2.1 Motivation
Photosynthetic organisms have evolved complex structures of pigments and
proteins to absorb light energy within the visible range of wavelengths, the so called
"photosynthetically active radiation" (PAR) between approximately 400 to 700 nm. A
fraction of this absorbed energy is not used for photosynthesis and instead is lost to the
environment in the form of heat and light. At in vivo temperatures, the energy
dissipated as light is emitted almost completely from chlorophyll a molecules
organized in the photosynthetic structure called Photosystem II. This phenomenon is
known as chlorophyll fluorescence and represents an optical signal unique to
photosynthetic organisms.
For phytoplankton in a natural light field, this emission is often referred to as
"sun-stimulated", "solar-induced", "natural" or "passive" fluorescence. For
simplicity, we will reserve the term "sun-stimulated" specifically for situations in
which fluorescence is excited by a solar ambient light field, and we will use "natural"
for the more general phenomenon. Consequently, natural fluorescence will be evident
in any light field conducive to phytoplankton growth, such as that provided by
broadband lamps in an incubator or in a culture vessel. In contrast, so-called "active"
fluorescence results from purposeful, artificial stimulation by light sources such as a
laser in a LIDAR system or an LED in a fluorometer. These excitation sources
typically "overload" a photosynthetic system with high irradiance so that the measured
fluorescence signal is maximal, whereas a natural fluorometer simply records the
fluorescence signal from a photosynthetic system "under load" in an ambient light
field at natural irradiance levels. Variability in the spectral quality and delivery rate of
excitation light, as well as issues of measurement technique, have led to significant
debate in the oceanographic community regarding the merits and drawbacks of each of
10
these similar, but subtly different, fluorescence signals with respect to their physical
and physiological interpretation (Kiefer and Reynolds 1992).
Natural fluorescence is a commonly measured bio-optical property in
oceanographic research. Operationally, natural fluorometers are attractive because
they have lower power requirements than their active counterparts, making them
suitable for long term drifter and mooring deployments (Abbott and Letelier 1997a,b;
Abbott and Letelier 1998). Since phytoplankton fluorescence can be detected in the
upwelling radiance distribution, natural fluorescence can be remotely sensed from
both aircraft and spacecraft (Neville and Gower 1977; Letelier and Abbott 1996).
Satellite sensors like NASA's Moderate Resolution Imaging Spectroradiometer
(MODIS), the NASDA Global Imager (GLI) and the ESA Medium Resolution
Imaging Spectrometer (MERIS) will obtain natural fluorescence measurements from
orbit and thus greatly extend our ability to monitor this property on a global scale.
Natural fluorescence has been used to estimate chlorophyll concentrations
(Kiefer Ct al. 1989; Abbott and Letelier 1997a), instantaneous rates of photosynthesis
(Kiefer et al. 1989; Chamberlin et al. 1990; Chamberlin and Marra 1992; Stegmann et
al. 1992; Doerffer 1993), and ecological responses to environmental transients (Abbott
and Letelier 1998). For dynamic regions of the ocean, natural fluorescence appears to
be a promising method for observing phytoplankton dynamics on time scales shorter
than those that characterize chlorophyll synthesis (Letelier Ct al. 1996).
In contrast to the relatively large amount of field work with sun-stimulated
fluorescence, to the best of our knowledge there are no published laboratory
measurements of natural fluorescence. Controlled experiments with phytoplankton
cultures can provide insight into how the physical and chemical environment
influences the natural fluorescence signal, and such information would be valuable for
understanding the sources of variability observed in field studies of sun-stimulated
fluorescence. Sources of physical and chemical variability in the ocean include
11
irradiance intensity, spectral quality, nutrient availability and temperature, and culture
experiments can be used to investigate each of these sources independently. To date,
laboratory investigations of natural fluorescence have been hindered by a lack of
instrumentation suitable for measuring this signal within limitations dictated by
phytoplankton culturing techniques.
In order to overcome such limitations, we obtained and redesigned a system for
monitoring fluorescence in cultures originally designed and built by Biospherical
Instruments Inc. We refer to this system as the Natural Fluorescence Chemostat
(NFC) and have used it to obtain detailed laboratory time series of natural
fluorescence under manipulated environmental conditions.
2.2.2 Requirements
Given our specific research interests, we identified three priorities regarding
the redesign of the prototype instrumentation. First, we wished to measure the natural
fluorescence signal in phytoplankton cultures with optical densities similar to those
found in the oligotrophic ocean (i.e. a chlorophyll concentration
0.05 tg L').
Although we can easily observe this signal in phytoplankton cultures at higher
chlorophyll concentrations, the optical properties and nutrient dynamics in such dense
cultures do not necessarily approximate natural conditions. Second, we wished to
measure this signal over the range of irradiances which occur naturally in the euphotic
zone. We are especially interested in dynamics at the very near surface (e.g. the upper
6 to 10 m), where physiological responses to such high light levels have a substantial
but poorly understood effect on natural fluorescence (Cullen and Lewis 1995). Since
at these depths commercially available field instruments cannot discriminate natural
chlorophyll fluorescence from Raman scattering of water or elastic scattering from
water and/or particles (Kiefer et al., 1989; Hu and Voss, 1998), natural fluorescence
remains significantly undersampled at such light levels. Third, since variability in
natural fluorescence can result from wavelength dependent changes in phytoplankton
12
absorption, we wished to monitor natural fluorescence as a function of excitation
wavelength, i.e. the action spectrum of natural fluorescence. To obtain this action
spectrum, a very weak active fluorescence must be stimulated within the culture
vessel, but this stimulation may not in any way disturb the physiological and/or optical
state of the culture.
2.3 System Design
We obtained prototype equipment from prior work performed by Dr. Dale
Kiefer (University of Southern California) and C. R. Booth (Biospherical Instruments,
San Diego, CA). The equipment and design were evaluated and reengineered to
address our specific measurement needs. The entire system is shown in Figure 2.1
2.3.1 Chemostat System
Phytoplankton are grown inside the NFC in an upright Pyrex cylinder, 250 mm
high by 110 mm in diameter, with a culture volume of approximately 2.5 L. The
nutrient delivery rate is controlled by a variable speed peristaltic manifold pump
(Insmatec Reglo), and so although we refer to this instrument as a "chemostat", which
implies a constant rate of nutrient delivery (Kubitschek 1970), we can grow the
sample culture with nutrient delivery rates that range from zero (i.e. batch mode) to
any aperiodic and/or semicontinuous rate that can be programmed into the control
system. An integral water jacket is connected to a recirculating water bath (Neslab
RTE-21 1) which maintains the culture temperature at a preset level. The culture is
continuously bubbled with filtered air, but since daytime photosynthesis can
significantly reduce the concentration of dissolved CO2 in the culture, a 2% CO2
mixture is automatically metered into the culture when the pH exceeds a programmed
threshold. The culture is stirred continuously and kept as axenic as possible by initial
sterilization, positive pressure and sub-micron particulate filters on all input and
13
output lines. The culture vessel is shielded from room light by a black aluminum
shell, and dark curtains and light traps are used to minimize the amount of light
leaking through apertures in the shell.
The ambient light field in the chemostat can be manipulated with respect to
both spectral distribution and intensity. A halogen light source in the base of the NFC
(housing: Source 4 PAR-MCM; lamp: Ushio HPL575/1 15X) is dimmed or brightened
by computer following a user selected protocol. A lens integral to the lamp housing
has a dichroic coating which removes a significant fraction of the infrared (IR) lamp
output. Fresnel lenses external to the lamp housing focus light into a series of optical
filters to select for the wavelengths which best mimic our desired visible spectrum, i.e.
a blue-green output with extremely little red light at the chlorophyll fluorescence
wavelengths. Firstly, the lamp output is reflected off a cold minor at 900 to remove
more IR. Next it is passed through an JR absorbing filter (3mm Schott KG-3), a bluegreen bandpass filter (6mm Schott BG-39), and finally through a 590 nm shortpass
interference filter (manufacturer unknown). This filter set can be modified to provide
different spectral distributions if desired. Immersed in the culture is a scalar PAR
sensor (Biospherical Instruments QSL-100), which monitors inadiance and provides
real time feedback for the inadiance protocol. Two fans ventilate the inside of the
NFC to remove heat that builds up from the growth lamp.
A recirculating system provides small volumes of phytoplankton culture to
external instruments. Presently we place in line a Fast Repetition Rate fluorometer
(FRRF, Chelsea Instruments Fastracka) modified with a custom flow through cuvette
to continuously measure the variable fluorescence character of the culture.
Periodically and during calibration we incorporate an absorption-attenuation meter
(WetLabs AC9) to measure the inherent optical properties (lOPs) of the culture.
Culture samples can be removed at any time for discrete analyses including pigment
distribution, spectral absorption, nutrient concentration and bacterial abundance.
14
/
N
-
I
Figure 2.1 Functional diagram of NFC. A growth lamp and housing, B Fresnel
lenses, C growth lamp output, D cold mirror, E heat absorbing glass, F bluegreen bandpass filter, G cutoff interference filter, H IR beam dump, I
monochromator source lamp, J monochromator, K optical chopper, L reference
photodiode (behind 45° beamsplitter), M bandpass and cutoff filters, N lens, 0
excitation beam output, P chemostat culture vessel, Q pH and temperature sensor, R
- scalar PAR sensor, S fluorescence emission filters, T slit apertures, U
photomultiplier, V single board computer, W video monitor, X keyboard, Y
ventilation light trap, Z ventilation fans, AA media carboy, BB - manifold pump,
CC media in/out and stirring rod, DD lock in amplifier system. Not shown:
recirculating water bath, metered CO2 supply, sample recirculating system. Thin
arrows indicate direction of information flow between computer, sensors and controls.
Open arrows indicate light paths. Dashed arrows indicate media flow.
15
2.3.2 Fluorescence Detection System
Two separate fluorescence signals are measured during the course of an
experiment: a natural fluorescence (Fnat) resulting from the broadband ambient light
field and a stimulated fluorescence (Fstjm) resulting from a weak monochromatic
source. A photomultiplier tube (PMT, Hamamatsu R374HA) measures both
fluorescences at 90° to the growth irradiance through a fused silica window formed
into the side of the culture vessel. The PMT is preceded by a 685 nm bandpass, 30 nm
full-width half-maximum interference filter (Omega Optical 685 WB3O) and a
longpass filter (3mm Schott RG645) to select for the wavelengths of chlorophyll
fluorescence Afluor. Two slit apertures limit the detector field of view to a solid angle
of approximately 1 .22e-2 sr. The gain of the detector is adjusted by varying the
photomultiplier dynode supply voltage.
The manner in which fluorescence varies as a function of incident wavelength
is known as the fluorescence action spectrum. We measure this action spectrum in the
culture by stimulating fluorescence at discrete wavelength bands (typically 5 nm
bandpass) between 430 and 550 nm. Light from a halogen light source (housing:
Instruments S.A. AH1O; lamp: Osram 64460) is passed through a monochromator
(Instruments S.A. H10) and is modulated by an optical chopper (Photon Technology
Inc. 0C4000) at 200 Hz. An interference filter (SWPO15, manufacturer unknown) and
a blue-green bandpass filter (3mm of Schott BG-39) are used to further attenuate any
red light in the excitation beam so as to minimize crosstalk between fluorescence
excitation and emission. A reference photodiode measures a fraction of this filtered
monochromator output to monitor the long term performance of the halogen source.
The remaining output ('mono) is focused by a lens into the culture vessel where it
stimulates a fluorescence Fstim. Since 'mono illuminates only
0.5% of the culture
volume and is very weak (1.8 .tW at 485 nm, Newport Instruments 840-SL), which in
quantum units is equivalent to 0.05 imol photons m2 s, we assume that 'mono is too
weak to contribute to significant physiological artifacts in Fnat.
16
Because 'mono is so weak, the stimulated fluorescence Fstjm is actually
comparable in magnitude to the noise in the natural fluorescence signal Fnat. To
recover Fstim from the total photomultiplier output, we use a lock in amplifier system
(LIA, EG&G PARC Models 181 & 7265) referenced to the optical chopper frequency.
A chopping frequency of 200 Hz was chosen after examining the distribution of total
system noise in the photomultiplier output in frequency space.
To produce a large overlap between the stimulated and observed sample
volumes, the propagation direction of the excitation beam is oriented in the general
direction of the photomultiplier. The angle between the two is 166°, which is small
enough to minimize the effect of forward scattering by particles but large enough to
produce a noticeable enhancement in the amount of Fstim reaching the photomultiplier.
2.3.3 Control and Acquisition System
A single board computer (SBC, Ampro LittleBoard P5i ) and a data acquisition
card (Computer Boards Inc. PC1O4IDAS16Jr) are used to perform sampling and
control. This computer is integral to the NFC and is directly connected to a local area
network (LAN) to provide easy access to data and programs. A Windows program
written in Visual C (Microsoft Corp.) contains protocols for sampling the entire suite
of sensors and for manipulating key environmental parameters (e.g. temperature, pH,
irradiance, and nutrient availability). Although these parameters can be adjusted by
computer with a temporal resolution of 1 s, hysteresis ultimately limits the rate at
which each individual parameter can be adjusted. All analog voltage signals,
including the natural fluorescence Fnat, are low pass filtered with a cutoff frequency of
0.53 Hz to minimize aliasing at our sampling rate of 1 s. The voltage resolution of the
analog channels is at best 0.15 mV. The stimulated fluorescence Fstim is digitally
filtered by the lock in amplifier with a time constant of 10 s and then communicated
serially to the computer with a nominal voltage resolution of 0.1 tV.
17
2.4 Testing and Evaluation
2.4.1 The Ambient Light Field
The spectral character of lamps typically varies as a function of intensity. To
characterize this variability we measured the growth light output as a function of
irradiance and wavelength using a spectrometer (Ocean Optics Inc., S2000) and the
integral scalar PAR sensor. These data were corrected as a function of wavelength for
grating efficiency and detector sensitivity, and the resulting corrected quantum
spectral distribution is shown in Figure 2.2 for four different irradiance levels (850,
108, 7, and 0.65 tmol photons m2 s'). The peak wavelength Xpeak clearly shifts
toward the red as the light is dimmed and this effect is most noticeable at very low
irradiance levels. Although we believe that this shift in Apeak may lead to physiological
artifacts at very low intensities, we assume that this effect is minimal in our present
experiments because cultures are exposed to such low light levels for a comparatively
short amount of time. However, we are mindful of this potential physiological artifact
when interpreting variability in Fnat at very low irradiance levels.
2.4.2 Artifacts in Fnat Measurements
Electrical noise in the room leads to a measurable voltage on the Fnat channel
in the absence of any source of light. We measured this voltage during calibration and
found it to have a mean magnitude of 0.04 1 mV and its distribution to be strongly
skewed toward zero. Consequently, we estimate the inherent noise level in Fnat to be
on the order of 0.1 mV. This noise magnitude is less than the voltage resolution of the
Fnat analog channel, and so we neglect instrument noise in the analysis of Fnat.
II
Ambient Light Field Spectral Distribution
11
1
800 umol photons m-2 s-I
200
A7
E
0.8
U065
A.
05
C
04
$lè'
T+
0
400
420
440
460
480
500
520
540
560
580
600
Wavelength (nm)
Figure 2.2 Spectral distribution of growth irradiance in the chemostat for four
different intensity levels.
The filters in the optical path of the growth lamp output remove virtually all
light at the chlorophyll fluorescence wavelengths (i.e., the emission filter 30 nm
fluorescence bandwidth
XfluOr
centered at 685 nm). We estimate that the proportion of
quanta in the ambient light field within
Xfluor
is only about 4e-10 (Table 2.1). II one
tenth of all molecular scattering interactions in water are Raman (Mobley, 1994), the
majority of signal detected by the photomultiplier in the absence of particles will result
from Raman scattering, given the spectral distribution of the growth irradiance and the
spectral transmittance of the fluorescence filters. Leakage through the emission filters
at wavelengths outside of Afluor would contribute only a negligible amount to the total
measured Fnat.
19
Table 2.1 Signal Contamination Analysis. A signal contamination analysis for
scattered light detected by the photomultiplier. The column marked (a) includes a
scattering dependence of A-4. The columns marked (b) are relative values at the PMT
detection wavelengths. The final column shows the relative proportion of these
scattering sources in the measured PMT output.
Wavelength
of lamp
output (nm):
Relative
proportion
of photons
in the
lamp
output:
Relative
scattering
efficiency
to get to
PMT
detector':
Relative
transmission
of emission
filtersb:
Relative
spectral
response
Product:
relative
magnitude
670- 700
=
=0.2
1.0
1.0
0.02
670- 700
4e-10
1.0
1.0
1.0
4e-10
1.0
<10
<i
<10.0
<106
PMT
detection
bandwidth
(nm):
ofPMTb:
545-565
nm:
Raman
excitation
670-700 nm:
at
chlorophyll
fluorescence
wavelength___________
400-700 nm:
Growth light
which may
leak through
emission
filters
400- 700
To obtain a correction for this Raman scattering, we measured the apparent
natural fluorescence
Fnat
in pure media (italics are used here to denote measurements
which include or are solely composed of known artifacts). Assuming that pure media
has optical properties similar to those of pure seawater, Fnat will be proportional to the
Raman scattering as a function of irradiance within the chemostat vessel. We
measured Fnar over the irradiance range which demonstrated the most significant shift
in Xpeak (EPAR
= 0.25 to 75 imol photons m2 s1). These data show that
Fnat
is well fit
over this range by a linear function of EPAR (Figure 2.3; r2 > 0.999, n = 13001):
Fnat
= 4.02e-4
EPAR
[p.mol photons m2 s'} + 0.0036 V
2.1
20
Apparent Natural Fluorescence of Pure Media vs. Irradiance
0.04
.! 0.035
y = 0.0004x + 0.0036
0
C)
R = 0.9999
0.03
C.)
C
C)
0.025
C)
0
0.02
C)
4-
0.015
C)
z
4-
C
C)
I-
C)
0.005
0
0
10
20
30
40
50
60
Irradiance (umol photons m2
70
80
s1)
Figure 2.3 Correlation between the apparent natural fluorescence of pure media
versus irradiance level.
We attribute the nonzero intercept to room light leaking into the instrument,
because improved sealing resulted in undetectable Fizat at
EPAR
very near to 0 pmol
photons m2 s1. Residuals of this fit were less than 0.2 mV, which is comparable to
the voltage resolution of the analog measurement channels. The slope of this
relationship is then the proportionality constant of Raman fluorescence per unit
irradiance for our instrument. Given that no single measurement of Fnat will be
resolved to better than 0.15 mV, we use a value of 0.4 mV (jtmol photons)' m2 s to
remove the contribution of Raman scattering from Fnat to get Fnat, the actual natural
fluorescence signal of interest.
21
In addition to such molecular processes, scattering by particles may also
contribute to the apparent natural fluorescence signal. We estimate a worst case
scenario of particle scattering using the model of Gordon and Morel (1983). With a
reference wavelength Xref of 510 nm to represent the chemostat growth irradiance at
Xpeak,
this model predicts a particulate scattering coefficient
bpart
of 5.622 m1 at a
chlorophyll concentration of 100 jtg L'. We then use a "typical" particulate phase
function (Mobley, 1994) to calculate the scattering at 90°, which is the observation
angle of the fluorescence detector. At this unusually high chlorophyll concentration,
scattering by particulates at 90° is 0.024 m1 sr1. The Raman scattering of quanta at
this angle and wavelength is
sr1 (Bartlett
1.le-5et mal. 1998), approximately three
orders of magnitude smaller than the particulate scattering. However, the fluorescence
emission filters will block the particulate scattering with a much greater efficiency
than the Raman scattering, approximately by a factor of 1e8. Accounting for this
effect, Raman scattering contributes orders of magnitude more photons than elastic
scattering even at these high chlorophyll concentrations, and so correction for total
scattering using only Equation 2.1 is justified.
2.4.3 Artifacts in Fstjm Measurements
Similar to Fnat, electrical noise in the room and instrument can be observed in
the Fstim signal in the absence of any source of light. We measured this noise signal
during calibration and found it to have a mean magnitude of 11.5 pV and relatively
little variability ( = 2.13 jtV, n = 6960). The observed noise is normally distributed,
and so we estimate an inherent noise level in Fstim as three standard deviations above
mean noise, i.e. =18 tV. This noise is significant compared to the nominal voltage
resolution of this measurement, and so we correct for this instrument noise in our
analysis of Fstim.
22
Like Fnat, the weak stimulated fluorescence signal Fstim can also be increased
by Raman, molecular or particulate scattering. Since the detector sensitivity and the
optical geometry for this signal are functions of excitation wavelength, the relative
magnitudes of these artifacts in Fstim may be different than those in Fnat.
Measurements of Fstim are used at present only for longer term comparisons on the
order of days, and so we can limit our discussion of the scattering error in Fstim to
measurements performed during simulated nighttime only, when irradiance dependent
scattering processes can be ignored.
To correct for molecular scattering and obtain the true stimulated fluorescence
Fstim, we measured the apparent stimulated fluorescence Fstjm of pure media in the
dark. At an excitation wavelength Amoflo of 485 nm, Fstjm had a mean of 10.6 p.V and
substantial variability (c = 2.4 tV, n = 3965). Using a series of short pass and long
pass interference filters at different locations in the optical train, we determined that
this signal was due primarily to scattered excitation light which generated fluorescence
in the emission filters. This signal is small compared to the total Fcjjm measured with
live cultures, but we incorporate it in our correction analysis for completeness.
Particulate scattering may have a larger influence on
than Fnat because of
the nearly collinear geometry of the optical train and the enhanced scattering of
particles at forward angles. We again use Gordon and Morel's model to calculate a
worst case scattering coefficient of 5.91 m1 for a culture of 100 tg U' of chlorophyll
at X = 485 nm. The photomultiplier observes a field 2.8° wide in the horizontal,
centered at 14° from the direction of beam travel (i.e. 166° from the horizontal axis of
the beam). Using again the "typical" phase function value at 14° of 4.893e-1 sr1, we
find the volume scattering function of the excitation beam at 14°, fpart(14°), to be 2.89
m1 sf1. Since the observed scattering due to particles is generated along the path of
the beam as it travels through the culture vessel, the volume scattering function can be
interpreted as a relationship between incident irradiance and the resulting radiance
scattered per unit path length into an angle 14° from the direction of propagation. The
23
solid angle of the detector and the pathlength of travel through the culture vessel are
known, and so multiplying I3part(14°) by these values provides a rough estimate of the
fraction of excitation photons scattered by particles into the photomultiplier field of
view. We calculate this ratio to be 3.4e-3. Since the irradiance of the monochromator
output is known (1.8 .tW at A = 485 nm), the optical power scattered into the
photomultiplier field of view by particles is then approximately 7.2 nW. The
fluorescence emission filters will attenuate scattering at this wavelength by at least
1e8, which reduces this signal to =7e-17 W. The anode radiant sensitivity of this
photomultiplier (3.4e4 A W1) and the subsequent typical current to voltage gain of the
lock in amplifier system (106 A V') are insufficient to bring this signal to levels
above the measurable instrument noise in Fstjm. Consequently, particulate scattering is
considered to have a negligible contribution to the measured stimulated fluorescence
at any wavelength in the visible.
2.4.4 Artifacts in Fluorescence Action Spectra
To generate a fluorescence action spectrum, the center wavelength of the
Fstjm
excitation source is periodically scanned between 430 and 550 nm. During these
scans the intensity of the excitation beam 'mono varies because throughput efficiency
and the output of monochromator source lamp are not constant as a function of
wavelength. We measured the apparent stimulated fluorescence
Fsjjm
in pure water as
a function of wavelength and found three distinct peaks between A = 400 and 750 nm
(Figure 2.4). Analysis of these peaks upon insertion of longpass and shortpass
interference filters at key points in the optical path yielded the following conclusions.
The peak centered around 550 nm is due to the excitation of Raman scattering of water
into the fluorescence signal wavelengths Afluor. The broader peak around 470 nm is a
result of fluorescence stimulated in the emission filters by this wavelength. The small
peak around 680 nm is the fraction of
excitation filters. Since
'mono
'mono
which remains after passing through the
is measured constantly by the reference photodiode of
known spectral response, we correct
Fstjm
for both Raman scattering and filter
fluorescence by normalizing the apparent fluorescence by the excitation quanta at each
wavelength of interest.
Apparent Stimulated Fluorescence vs. Excitation Wavelength
3.00E-05
Raman
Scattering
Filter fluorescence
at excitation
wavelengths P
2.50E-05
0
>
2.00E-05
C
Leakage at
emission
wavelengths
C)
U
1.50E-05
/
)
U-
1
/
/
.00E-05
/
+
//
/
E
5.00E-06
I
/1
O.00E+OO
400
450
500
550
600
650
700
Wavelength (nm)
Figure 2.4 Apparent weak stimulated fluorescence of pure water as a function of
excitation wavelength. The peaks result from filter fluorescence, Raman emission and
red leakage through the excitation filters from the monochromator at 480, 550 and 685
nm respectively.
25
2.5 Live Culture Data
Using the described instrumentation, we have performed culture experiments
on the marine diatom Thalassiosira weissflogii (Bacillariophyceae) and the marine
chiorophyte Dunaliella tertiolecta (Chlorophyceae). A 45 day series of
Fnat
from a
highly concentrated culture of the marine diatom T. weissflogii is presented in Figure
2.5a as an example.
The irradiance model in this experiment was sinusoidal, delivered on a 12:12
light-dark cycle with a daily maximum irradiance of 75 tmol photons m2 s1. The
maximum chlorophyll concentration over this period was 130 p.g U', and although
such high chlorophyll concentrations are rarely found in nature, these preliminary
results allow us to define typical patterns in
Fnat
for comparison with dilute cultures.
To examine the variability in natural fluorescence during this experiment, we
normalize
Fnat
to PAR irradiance to form the coefficient of fluorescence I(X,
XE),
following the terminology of Gordon (1979). We use this ratio as a proxy for
fluorescence quantum efficiency 1(XF,
Strictly speaking, TI(XF,
XE)
XE)
for these data (Figure 2.5b).
is the ratio of emitted fluorescence to absorbed (not
incident) energy, but since we observed no significant variability in the fluorescence
action spectrum during this period, we make the simplifying assumption that
phytoplankton spectral absorption varies little and thus
Since neither
Fnat
nor c1(XF,
XE)
1(XF, XE)
estimates ri(X, XE).
is normalized to chlorophyll, long term variability in
both of these signals reflects changes in chlorophyll biomass over time. Most of the
long term change in amplitude seen in Figure 2.5 is due to variability in chlorophyll
biomass in the culture vessel.
26
Natural Flu orescene vs. Julian Day
a
0
C
LI
UI
Julian Day of Year
PAR Normaled FIuoresencevs. Julian Day
-
Ui
=
C,)
0
210
215
220
225
230
235
Julian Day of Year
240
245
250
255
Figure 2.5 A) Natural fluorescence signal versus Julian day. Each day is roughly
sinusoidal in shape. B) Natural fluorescence normalized to incident irradiance.
Dilute erythromycin was added after days 223 and 239. A glitch in the program is
apparent on day 224.
27
Although I(X, XE) mimics Fnat in some respects, there are periods during the
experiment where important differences occur. The chemical environment in this
culture was twice manipulated during this experiment by adding small amounts of a
dilute antibiotic solution (erythromycin in distilled water) in order to evaluate the
potential use of this antibiotic to control bacterial populations within the culture
vessel. These additions occurred before the simulated daytime of days 224 and 240.
When measured in a fluorometer (Turner Designs AU-b), the antibiotic solution
exhibited no measurable fluorescence in the chlorophyll wavelengths Xfluor, and so we
discount the possibility that this solution could directly influence measurement of
fluorescence. Although no immediate apparent response in the
observed after these additions,
4(XF, XE)
Fnat
signal can be
clearly increases in the days following each
treatment.
The
c1(XF, XE)
signal is plotted against predawn chlorophyll concentration [chl
a] in Figure 2.6, which allows the influence of [chl a] to be assessed separately, and no
significant correlation exists between chlorophyll concentration and c1(XF,
scale and with this culture
(r2
XE)
at this
= 0.57). However, patterns exist in the grouping of
these data which are indicated by different symbols. During the first 12 days of the
experiment chlorophyll biomass in the culture increased exponentially (growth rate
= 0.16 d1,
r2
> 0.98, n = 4), and 4(XF,
XE)
t
measurements from the days during this
time when we have chlorophyll data (open circles) define a line which is explicitly
drawn. During this period, it is apparent that the relationship between chlorophyll
biomass and
growth,
(XF, XE)
(XF, XE)
is approximately linear. After this initial period of exponential
becomes decoupled from chlorophyll biomass and c1(XF,
XE)
is
distributed near, above or below this line. The first and second erythromycin additions
are represented by triangles and squares respectively. Closed symbols represent the
day before the addition of erythromycin and open symbols represent the four
following days. The transient increase in
c1(XF, XE)
for several days after each
treatment suggests that the addition of erythromycin either directly or indirectly led to
enhanced fluorescence properties on a per chlorophyll and per irradiance basis.
Fnai/EpvS. Chlorophyll Concentration
35.
3.
c
>
1.5
;
I
I
0
20
______
40
60
80
100
120
140
[chi a] ug L-i
._______
Figure 2.6 Scatter plots of irradiance normalized natural fluorescence versus
concentration of chlorophyll a. The drawn line defines the period of exponential
growth. Open circles represent conditions of exponential growth. The first and
second erythromycin additions are represented by triangles and squares respectively,
where a closed symbol represents the day before the treatment and open symbols
represent the four days following the treatment.
2.6 Summary
The time series data presented here represent the first detailed look into the
natural fluorescence response of phytoplankton cultures to controlled environmental
disturbances. Using the NFC we can manipulate independently many of the factors
known to influence the natural fluorescence character of phytoplankton. The ability to
measure natural fluorescence at high irradiance levels extends our ability to investigate
how this signal evolves under such conditions. Simultaneous measurement of the
fluorescence action spectrum allows us to correlate variability in natural fluorescence
with specific changes in how the pigment ensemble fluoresces as a function of
wavelength. Such information is crucial to improving the physiological models of
natural fluorescence that incorporate data obtained either by in situ or remote
instruments.
During development and evaluation we noted several features which we will
attempt to improve, if possible. Although the theoretical and observational scattering
arguments are convincing enough for a preliminary evaluation, we desire a more direct
method of assessing the scattering contribution to
Presently we are designing a
Fnat.
calibration protocol that uses additional emission filters to measure scattering at
wavelengths near to
Secondly, we would like to extend the irradiance range
within the culture vessel to 2000 .tmol photons m2
s
and improve the precision with
which this irradiance is controlled. Thirdly, we believe that eliminating emission filter
fluorescence will improve not only our measurements of
Fstim
but those of
Fnat
as well.
Other emission filter sets are presently being evaluated.
Acknowledgments. We are grateful to Rocky Booth for the loan of the
prototype NFC, and we appreciate the input of Dale Kiefer and John Morrow
regarding the original natural fluorescence chemostat project. We thank Lisa Eisner
for her advice and comments during the preparation of this manuscript. Curt Vandetta
provided technical support and Claudia Mengelt assisted in sampling during several of
the experiments. This research was supported by the National Aeronautics and Space
Administration (NAS5-3 1360) as was the original construction of the prototype
(NAS7-969).
2.7 References
Abbott, M.R., and R.M. Letelier, 1997a: Going with the Flow The Use of Optical
Drifters to Study Phytoplankton Dynamics. Monitoring Algal Blooms: New
Techniques for Detecting Large-Scale Environmental Changes. M. Kahru and
C.W. Brown, Eds., Landes Bioscience, pp. 143-168.
Abbott, M.R., and R.M. Letelier, 1997b: Bio-optical drifters Scales of variability of
chlorophyll and fluorescence. Society of Photo-Optical Instrumentation
Engineers 2963, 2 16-221.
Abbott, M.R., and R.M. Letelier, 1998: Decorrelation scales of chlorophyll as
observed from bio-optical drifters in the California Current. Deep-Sea
Research 1145, 1639-1667.
Bartlett, J.S., K.J. Voss, S. Sathyendranath, and A. Vodacek, 1998: Raman scattering
by pure water and seawater. Applied Optics 37, 3324-3332.
Chamberlin, W.S., C.R. Booth, D.A. Kiefer, J.H. Morrow, and R.C. Murphy, 1990:
Evidence for a simple relationship between natural fluorescence,
photosynthesis and chlorophyll in the sea. Deep-Sea Research 37, 95 1-973.
Chamberlin, S., and J. Marra, 1992: Estimation of photosynthetic rates from
measurements of natural fluorescence: analysis of the effects of light and
temperature. Deep-Sea Research 39, 1695-1706.
Cullen, J.J., and M.R. Lewis, 1995: Biological processes and optical measurements
near the sea surface: Some issues relevant to remote sensing. J. Geophys. Res.
100, 13,255-13,266.
Doerffer, R., 1993: Estimation of primary production by observation of solarstimulated fluorescence. ICES Mar. Sci. Synzp. 197, 104-113.
Gordon, H.R., 1979: Diffuse reflectance of the ocean: the theory of its augmentation
by chlorophyll a fluorescence at 685 nm. Appl. Optics 18, 1161-1166.
31
Gordon, H.R., and A. Morel, 1983: Remote Assessment of Ocean Color for
Interpretation of Satellite Visible Imagery, a Review, Lecture Notes on
Coastal and Estuarine Studies, Volume 4, Springer Verlag.
Hu, C., and K.J. Voss, 1998: Measurement of solar-stimulated fluorescence in natural
waters. Limnology and Oceanography 43, 1198-1206.
Kiefer, D.A., and R.A. Reynolds, 1992: Advances in understanding phytoplankton
fluorescence and photosynthesis. Primary Productivity and Bio geochemical
Cycles in the Sea. P.G. Falkowski and A.D. Woodhead, Eds., Plenum Press,
pp. 155-174.
Kiefer, D.A., W.S. Chamberlin, and C.R. Booth, 1989: Natural fluorescence of
chlorophyll a: relationship to photosynthesis and chlorophyll concentration in
the western South Pacific gyre. Limnology and Oceanography 34, 868-881.
Kubitschek, H. E., 1970: Introduction to Research with Continuous Cultures.
Prentice-Hall Inc., Englewood Cliffs, N.J.
Letelier, R.M., and M.A. Abbott, 1996: An Analysis of Chlorophyll Fluorescence
Algorithms for the Moderate Resolution Imaging Spectrometer (MODIS).
Remote Sens. Environ. 58, 2 15-223.
Letelier, R.M., M.R. Abbott, and D.M. Karl, 1996: Chlorophyll natural fluorescence
response to upwelling events in the Southern Ocean. Geophys. Res. Lets., 24,
409-412.
Mobley, C.D., 1994: Light and Water, Radiative Transfer in Natural Waters,
Academic Press, 594 pp.
Neville, R.A., and J.F.R. Gower, 1977: Passive Remote Sensing of Phytoplankton via
Chlorophyll a Fluorescence. Journal of Geophysical Research 82, 3487-3493.
Stegmann, P.M., M.R. Lewis, C.O. Davis, and J.J. Cullen, 1992: Primary production
estimates from recordings of solar-stimulated fluorescence in the Equatorial
Pacific at 150 degrees West. Journal of Geophysical Research 97C, 627 638.
32
3. The Influence of Light and Nitrate on the Natural Fluorescence of
the Marine Diatom Thalassiosira weissflogii (Bacillariophyceae)
"Why do we always have to use pestles and mortars?"
-An unidentified student of photosynthesis.
3.1 Abstract
The influence of irradiance and nitrate availability on phytoplankton natural
fluorescence was examined in two nitrate limited continuous cultures of the marine
diatom Thalassiosira weissflogii (Bacillariophyceae). Each culture was maintained
for a period of 6-8 weeks under identical chemical and physical conditions except for
the media dilution rate. The dilution rate was constant and low in the first culture,
leading to nutrient-limited growth, but underwent a sudden increase from zero in the
second culture leading to temporarily nutrient-replete conditions. Both cultures
experienced identical irradiance histories consisting of an initial period of low diurnal
maximum irradiance
(Emax
80 .tmol photons m2
1),
which is increased fivefold
after approximately five volume turnovers. In both cultures, this shift up in irradiance
was accompanied by increases in cell abundance, decreases in cell-specific
chlorophyll a concentration, increases in the ratio of particulate C:N, and significant
decreases in the ratio of fluorescence to irradiance (cP) around solar noon. Simple
metrics were developed to quantify long term changes in c1; these metrics were not
always more strongly correlated with chlorophyll a ([chi a]) than with cell abundance.
Other metrics derived from C1 were found to reflect physiological adaptation to
changes in nitrate availability consistent with ecological population-resource models.
The point at which photosynthesis switches from light limited to light saturated was
observed to coincide with the maximum forenoon ct. These preliminary results
suggest that the physiologically driven variability in natural fluorescence can be
33
determined using techniques applicable to satellite remote sensing, and that simple,
easily derived metrics can be used to infer ecological and physiological responses of
marine phytoplankton from their natural fluorescence signature.
3.2 Introduction
Oceanographers have long realized the applicability of chlorophyll a
fluorescence to studies of phytoplankton distribution and production (e.g. Kiefer 1973;
Gordon 1979). In oceanographic studies, chlorophyll fluorescence is typically
measured with active instruments that use intense artificial light sources to stimulate
fluorescence emission. However, any light that is ultimately absorbed by chlorophyll
will stimulate fluorescence, which is why a chlorophyll fluorescence signal is evident
in the upper ocean due to ambient solar radiation. This naturally occurring
fluorescence is appropriately called natural fluorescence4 and has been correlated,
with varying degrees of success, with chlorophyll biomass (Gower and Borstad 1981;
Kiefer et al. 1989; Abbott and Letelier 1997a; Abbott and Letelier 1997b) and the rate
of carbon fixation (Kiefer et al. 1989; Chamberlin and Marra 1992; Chamberlin et al.
1990; Stegmann et al. 1992).
Using radiometers mounted on profilers, moorings and drifters, this natural
fluorescence signal (Fnat) can be measured throughout most of the euphotic zone.
Further, Fnat can be detected in the oceanic upwelling radiance distribution, which
allows aircraft and satellite deployed sensors to measure this property remotely over
large spatial scales. The recent deployment of the MODIS sensor on NASA' s Terra
satellite extends this measurement to global and yearly scales, providing a means to
acquire Fnat synoptically on the basin scale in most oceanic regions. For ecological
studies, natural fluorescence is additionally attractive because its measurement
requires no manipulation of the phytoplankton under study. Solar irradiance is used
Sometimes also called passive, sun-stimulated or solar fluorescence.
34
for excitation and so the measured natural fluorescence signal by definition reflects
phytoplankton fluorescence properties under ambient and natural conditions. The
practical significance of this bias is an ongoing source of debate, but the strong
spectral dependence of many photosynthetically relevant physiological properties
hints that active fluorometric methods may contain substantial biases under certain
conditions.
The ease with which Fnat is measured is offset by the difficulty encountered in
its interpretation. Fnat is predominantly a strong function of ambient irradiance and
chlorophyll biomass, and variability in these factors dominate the total measured
natural fluorescence variability throughout the euphotic zone (except perhaps in
certain ecological circumstances Letelier et al. 1997). Equation 3.1 shows how
natural fluorescence can be expressed as
Fnat
=Eo.[chl].a*.TY
3.1
where ambient solar irradiance E0, chlorophyll biomass [chl a], the chlorophyll
specific absorption coefficient a* and the absolute quantum efficiency of fluorescence
F
all affect the absolute magnitude of Fnat. Since both irradiance and chlorophyll are
known to vary over several orders of magnitude, these two factors are presumed to be
the dominant controls on variability in Fnat.
F
and a* vary as well, but not to the
same degree, and so the physiologically-driven variability in Fnat due to these two
minor influences can be examined by normalizing Fnat with independent assessments
of [chl
a]
and E0.
The physiologically-determined variability in Fnat is a complex function of
many environmentally and genetically determined factors. Fnat is influenced by
nutrient availability, irradiance quality, intensity, ambient temperature and species
composition. In the ocean these factors often vary in concert over a wide range of
temporal and spatial scales, but the resulting variability in Fnat is not random.
35
Phytoplankton respond to changes in the physical and chemical environment in an
ordered fashion, and the result is a similarly ordered change in photosynthetic
properties and Fnat. If relationships can be identified relating physiologically-driven
variability in Fnat to changes in photosynthetic structure or function, natural
fluorescence may become a useful tool with which to identify specific physiological
and ecological responses evidenced by phytoplankton when physical or chemical
conditions in the ocean vary.
To apply natural fluorescence to questions of marine phytoplankton physiology
or photosynthesis, however, the specific influence of these individual environmental
factors on Fat must be identified and quantified (Chamberlin and Marra 1992; Letelier
et al. 1997; Cullen and Lewis 1995). Laboratory experiments are crucial to better
understanding these influences, but there have been no published experiments
detailing such investigations to date. This lack of publications is due primarily to
technical difficulties encountered in measuring natural fluorescence in phytoplankton
cultures, and so we have developed and characterized a system for measuring Fnat and
other physiological properties under controlled and manipulated laboratory conditions
(Natural Fluorescence Chemostat, or NFC). Preliminary results from experiments
using the NFC indicate that changes in the physical and chemical environment lead to
specific and consistent changes in Fnat. Since the NFC can be programmed to maintain
irradiance levels with realistic time history, experiments performed with this
equipment are more appropriate for determining how phytoplankton react to realistic
changes in the light environment. This is important because the photosynthetic
apparatus has a constant dynamic response to changes in the light and nutrient
environment.
In this paper we examine how variability in two important environmental
factors, nitrate and irradiance, influences the natural fluorescence signal of a model
species of marine diatom. Availability of both nitrate and irradiance is a well
documented constraint on the distribution of phytoplankton biomass and primary
productivity in much of the ocean, and the manner in which each controls
photosynthesis and growth is well established. The proper application of Fnat to basin
scale studies of phytoplankton distribution and photosynthetic potential relies strongly
on accurate laboratory characterization of the relationship between measured
variability in Fnat and the environmental factors which might to these observed signals.
We designed an experiment to compare the natural fluorescence response of
phytoplankton to changes in the irradiance and nitrate environment. Our primary goal
is to determine if there is a significant difference in Fnat which can be attributed
directly to these experimental treatments. Secondly, if differences are discovered, we
are interested in exploring the character and kinetics of these changes, to see if
discernable changes in the natural fluorescence properties of phytoplankton have any
practical application. Since the physical location of fluorescence emission in
phytoplankton, Photosystem II, responds both kinetically and structurally to short and
long term variability in nitrate availability and irradiance, natural fluorescence may
potentially be used to monitor changes in these environmental factors, simply by
observing how these factors influence the natural fluorescence properties of
phytoplankton.
3.3 Methods
Phytoplankton species and media An inoculum of the marine diatom
Thalassiosira weissflogii (Bacillariophyceae) was obtained from the ProvasoliGuillard Culture Collection (strain CCMP 1051, Bigelow Laboratory for Ocean
Sciences, Boothbay Harbor, ME). Cultures were grown in a modified IMR medium
(Eppley at al. 1967), to which phosphate and silicate were added to a final
concentration of IMR/2. Nitrate was added to a final concentration of IMRI2O.
Natural Fluorescence Chemostat
Cultures were maintained and manipulated
in the Natural Fluorescence Chemostat apparatus described in Section 2 of this thesis.
Twin experiments were performed to examine the natural fluorescence properties of
37
phytoplankton under low growth rate (LGR) and high growth rate (HGR) conditions,
where each culture experiences first a low light (LL) and later a high light (HL)
environment. Irradiance in each experiment was modeled as a sinusoid, 12 hours in
duration with a maximum daily irradiance intensity Emax of
(LL) and
80 tmol photons m2
400 tmol photons m2 s (filL). Environmental conditions in each culture
vessel were identical in all other respects. The spectral composition of the growth
irradiance remained constant, culture temperature was 20±0.1 °C, pH was maintained
below 8.2 and cultures were continuously stirred and bubbled with filtered air.
The LGR experiment began on day 21 i. The culture vessel was sterilized,
filled with media, inoculated with T. weissflogii and continuously diluted at a rate of
0.11 d. The culture was allowed to stabilize for 47 subsequent days ( 5 volume
turnovers) where the first 15 days show acclimation to low nitrate availability under
low light conditions. Before sunrise on day 258 the maximum daily irradiance was
programmed to the HL level, which continued until the termination of the experiment
on day 270.
The HGR experiment began on day 320. Initial dilution rate was zero and the
culture grew in batch mode, decreasing the level of relative available nitrate and
slowing population growth. On day 328, dilution rate was increased to 0.74 d' for the
following 13 days (9.6 volume turnovers). On day 341 irradiance was increased to HL
levels, which continued until day 350 when dilution rate was again set to zero for the
following ten days. The initial period of batch mode growth and the final ten days
represent periods of nitrate starvation under LL and HL, respectively.
Variable fluorescence measurements A small volume of the culture ( 8 mL)
was continuously circulated through the dark chamber of a Fast Repetition Rate
fluorometer (FRRF, Fastracka, Chelsea Instruments, UK) using a custom flow
assembly. Recirculation rate was
5.8 mL min1 and so cells were remained in the
In this paper date is represented sequentially by day of year and time is represented decimally.
flow through system for about 90 s. Given the small volume and the rapid flow we
assume that recirculation had a negligible impact on the physiological state of the
whole culture. Since cells were measured within one minute of leaving the culture
vessel, we also assume that FRRF measurements reflect approximately the same level
of nonphotochemical quenching as that which occurred in the culture vessel. Variable
fluorescence was measured during both experiments with 1 minute resolution.
The instrumental response of the FRRF was characterized following methods
recommended by the manufacturer. Photosynthetic parameters of Photosystem II
(PSII) were estimated by fitting a physiological model to the calibrated variable
fluorescence data. A custom analysis tool was developed for this experiment to
perform this fitting and is described in Chapter 4 of this thesis. Bad fits were
identified and rejected using Pearson's chi-squared statistic (X2). This model retrieved
five photosynthetic physiological parameters: the initial fluorescence (F0, relative
units), the maximal fluorescence (Fm, relative units),the functional absorption cross
section of PSU
A2 photon'), a connectivity factor (p, dimensionless), and the
overall time constant of QA reoxidation
(t,
jts), calculated using a single exponential
decay model. We form the photochemical conversion efficiency Fv/Fm as
(Fm
F0)
Fm' and the photosynthetic saturation irradiance EK as (sii 'r)1, scaled to obtain the
proper units.
Cell abundance and size distribution Cell abundance and size distribution
were measured daily on discrete samples taken from the culture vessel within the hour
before programmed sunrise. Between four to eleven 0.5 mL sample replicates were
processed each day using a Coulter Counter (Model ZBI, Coulter Electronics Inc.,
Hialeah FL), a pulse sampler and a multichannel analyzer (TN-1246 and TN-7200,
Tracor Northern USA). The system was size calibrated using polystyrene
microspheres (Duke Scientific, Palo Alto CA) and latex beads (Sigma Chemical Co.,
St. Louis MO).
39
Chlorophyll and phaeopigrnents
Chlorophyll was sampled along with cell
abundance and two replicates were analyzed daily. Concentrations of chlorophyll a
[chl a] and phaeopigments [phaeo] were determined fluorometrically (Turner Designs
AU-10, Sunnyvale CA) following standard techniques (Yentsch and Menzel, 1963).
Pigment content and distribution High pressure liquid chromatography was
used to determine the composition and concentration of photosynthetic and
photoprotective pigments. Samples were obtained concurrently with cell abundance,
filtered onto GF/F filters and frozen at 20°C. Samples were transferred to either a
40°C or a -80°C freezer for long term storage. Sample preparation followed the
methods of Wright and Jeffrey (1997) and canthaxanthin was added as an internal
standard. Samples were analyzed in a detector-pump system (detector: Thermo
Separation Products UV2000; pump: Perkin Elmer 400) previously calibrated for chl
a, chl
c1,
chl c2, fucoxanthin, diatoxanthan, diadinoxanthan and canthaxanthin. We
present only HPLC-derived [chi a] measurements in this paper, and treatment of the
observed variability in photosynthetic and photoprotective pigments will be covered in
a future publication.
Particulate nitrogen and carbon
PC and PN were sampled concurrently with
cell abundance. 10 mL of sample was filtered onto precombusted 21 or 25 mm GF/F
Whatman filters and frozen in precombusted tinfoil wrappers. Analysis was
performed on a CHIN analyzer (Carlo Erba NA 1500) calibrated against a L-cystine
standard.
Intensive sampling periods Intensive sampling was performed over two 24
hour periods in each experiment, one just before and one well after the shift from LL
to HL. During the LGR experiment, HPLC samples were taken every 90 minutes for
24 hours on days 257-8 and 269-70. During the HGR experiment, HPLC samples,
fluorometric chlorophyll, cell abundance and PC-PN were sampled at 90 minute
intervals on days 337-8 during LL conditions. HPLC, fluorometric chlorophyll and
cell abundance samples were collected at 90 minute intervals on day 348 during HL
conditions.
Terminology and notation
Natural fluorescence is detected using a
photomultiplier tube that observes a small and constant volume of the culture vessel.
The radiant power emitted by the phytoplankton culture in the natural fluorescence
wavelengths and measured by the photomultiplier will be referred to as the natural
fluorescence Fnat, in volts. Since we are interested only in relative changes to this
signal, we will not apply geometrical corrections to determine the total spherically
integrated radiance. Since we do not directly and continuously measure
phytoplankton absorption in the culture vessel, we cannot calculate the absolute
quantum yield of fluorescence
irradiance
(EPAR)
F
on a molar basis6. However, scalar ambient
is measured in the culture vessel and so for each Fnat we can
calculate Gordon's (1979) coefficient of fluorescence I(X, XE) as
Fnat
EPAR'. We
are not concerned with the spectral distribution of either irradiance or fluorescence,
and so we use a shorthand (ct) for this coefficient.
More chlorophyll in the optical viewing path will lead to a higher measured
Fnat
and vice versa. We define
as the ratio of c1 to chlorophyll concentration,
consistent with notation used to describe the photosynthesis-irradiance relationship
(e.g. pchI) and with the concept of an apparent quantum yield as described in Letelier et
al. (1997). There is no a priori reason to normalize for chlorophyll biomass, and so
we also define a cell specific coefficient of fluorescence cIF
as the ratio of 1 and
cell abundance. Although we present both parameters in this paper, we will primarily
discuss variability in
remote sensing data.
Fch1
FchI
because this property can be more easily constructed using
will be referred to as the apparent quantum yield of natural
fluorescence. Since {chl a] and cell abundance are typically measured once daily in
6
V and 4' will refer to the absolute quantum yield of photosynthesis (charge separation) and
fluorescence respectively. The symbol V will refer to the absolute quantum yield of carbon fixation.
41
these experiments, it is important to remember that
qFch1
and cI-'' normalize for
biomass effects on scales longer than 1 day.
3.4 Results
3.4.1 Culture Response to an Increase in Irradiance
Chlorophyll biomass, cell abundance, and chl a:cell are shown for the twenty
days during the LGR experiment bracketing the shift in irradiance on day 258 (Figure
3.1). Cell abundance for the 9 days preceding this increase are constant, suggesting
steady state population growth. Variability observed in the [chi a] measurements
under LL was initially attributed to differences in the chlorophyll extraction technique,
but later analysis of these data suggest that this variability may be in fact partly real.
After the shift to HL on day 258, sharp decreases in [chl a] and increases in cell
abundance are observed, and {chl a] per cell decreases.
Figure 3.2 shows twenty days during the HGR experiment surrounding the
increase in irradiance on day 341. Changes in cell abundance over the entire
experiment were consistent with logistic growth (apparent growth rate jt = 0.51 d', K
68000 cells m1', c = day 343.8) despite a constant dilution rate of 0.75
d1.
However, the general trend for the twenty days surrounding the shift in irradiance is
reasonably well described by simple exponential growth. Anomalous transient
responses in both cell abundance and [chl a] were observed for about 1 to 2 days
immediately following the shift to HL. The combined influence of these on [chl a] per
cell shows no such anomaly.
42
60000
I VU
140
50000
a)
120
a)
a)-
40000
a)
100
r0
80 o.
30000
0.
-J
60 o
20000
40
a)
0
10000
20
0
250
0
252
254
256
258
260
262
264
266
268
270
SDY
Figure 3.1 Cell abundance, [chi a] and cell specific [chi a] in the low growth rate
(LGR) culture. Twenty days bracketing the shift up in irradiance (predawn SDY258)
are shown, where symbols represent samples taken in the predawn.
250
I IJUUU
Q
60000
200 b
a)-
50000
a)
0
150
40000
(2)
a)
E
-c
(0
0
30000
100.
-J
03
20000
a)
50
-a)
10000
0
0
330
0
332
334
336
338
340
342
344
346
348
350
SDY
Figure 3.2 Cell abundance, [chl a] and cell specific [chl a] during the HGR
experiment. Twenty days bracketing the shift up in irradiance on day 341 are shown,
where symbols are as in Figure 3.1.
43
Figure 3.3 shows measurements of EPAR, Fnat,
FchI
and
Fce11
for the same
twenty days of the LGR experiment. In general, Fnat varies as EPAR but increases only
threefold as irradiance increases fivefold (Figure 3.3b). The apparent fluorescence
quantum yield
is regular and symmetric under LL conditions but exhibits a
significant midday depression coincident with the shift to HL.
As well, qFchI exhibits a transient increase for
9 days following the shift to
HL (Figure 3.3c). qFcelI decreases after the shift to HL, exhibits the same midday
depression, and exhibits a shorter transient character lasting only
5 days (Figure
3.3d). The different duration of the two transients suggests a decoupling between
changes in [chl a] biomass and population.
A similar plot is shown for days 330 to 350 during the HGR experiment
(Figure 3.4). A programming error caused growth lamp control to briefly fail during
the mornings of days 332, 338 and 340 and although this may have affected the long
term character of Fnat, we see no evidence to support this. Fnat increases gradually
during LL conditions and rises roughly by a factor of 7 in response to the fivefold
increase in irradiance (Figure 3.4a&b). Midday reductions in qFchI and
FceII
are
observed under UL conditions but are much weaker in magnitude than those observed
during nitrate limited conditions during the LGR experiment (Figure 3.4c and d).
Anomalous transient responses are observed in both
FchI
and
FceII
for two
days immediately following the shift to HL, probably driven mainly by the coincident
anomalies in [chi a] and cell abundance. These transients are considered anomalous
because they are not monotonic as a function of time, suggesting that the changes in
[chl a] and cell abudance which are evidenced after the shift to HL are most probably
nonlinear in nature.
- cuu
400
E
cc
c 300
cc
=
C-
0
a.
100
a.
W
0
250
252
254
256
252
254
256
252
254
256
258
260
262
264
266
268
270
258
260
262
264
266
268
270
I
I
I
I
V
250
x 10
5
4
-3
0
250
x 10
C
I
I
I
D
6I
r
250
1IH
252
254
256
258
260
262
264
266
268
2
SDY
Figure 3.3 frradiance and natural fluorescence parameters during the LGR culture
experiment: a) Irradiance (EPAR, tmol quanta m2 1) in the culture vessel, b) natural
fluorescence (Fnat, volts), c) ct (dimensionless) and d)
(dimensionless).
fcelI
45
I
I
I
I
I
I
I
A
400
4
E
300
.200
100
A A A
330
332
A1
334
2 C
A A1 A A1
A1 A
H
I
I
336
338
340
342
344
I
I
I
I
346
348
350
B
10
:3
A C\
I A
C
330
332
334
336
338
340
I
I
342
344
346
.1
348
350
x103
I
',
ii,
530
111
1I
332
334
336
338
332
334
336
338
340
342
344
346
H
348
350
340
342
344
346
348
350
fl
I
i
x 10
6
4
2
01
330
SDY
Figure 3.4 Irradiance and natural fluorescence during the HGR culture experiment.
Presentation is identical to Figure 3.3.
Photosynthetic physiology was determined from analysis of the variable
fluorescence measurements for these twenty days (Figure 3.5). Die! cycles can be
observed in all parameters almost always, except in the connectivity of reaction
centers (p) for about four days following the shift from LL to HL. The diurnal
character of Fv/Fm is driven primarily by Fm and exhibits a significant midday decrease
with the shift to I-IL (Figure 3.5a&b).
psii exhibits a significant diurnal decrease
during LL conditions, which is amplified after the shift to HL (Figure 3.5c). The
midday decreases in both FvfFm and
psii
indicate that nonphotochemical quenching of
fluorescence occurs daily in the antenna and reaction centers under both LL and HL
during LGR. Mean nocturnal apsii increases after the shift to HL, suggesting that the
light harvesting capacity for these cells increased. p often exhibits an approximate
30% diurnal enhancement during LL which during HL increases to over 50% (Figure
3.5d). The diel pattern oft is complex under LL but simplifies under I-IL conditions
to a near-sinusoid which increases over time (Figure 3.5e). Under LL conditions EK
well exceeds
EPAR
always, indicating light limited photosynthesis at all LL irradiances.
Under HL conditions
EPAR
surpasses EK around solar noon, indicating light saturated
photosynthesis during these periods.
Problems with the FRRF led to data loss during this period between days 338
and 341 and a decrease in the signal quality compared to the LGR experiment between
days 342 and 349. Because [chl a] levels were low initially during the LL period, a
higher instrument gain was required than that used in the LGR experiment. Automatic
gain switching can be seen between days 336 and 338. Corrections for gain changes
proved to be more nonlinear than expected and consequently were abandoned in this
analysis; only data collected at low gain (post day 337) will be directly compared in
magnitude to LGR values. Data collected at high gain will be examined only with
respect to diurnal and long term trends within this period.
The period prior to the shift up in irradiance is characterized by virtually no
diel variability in photosynthetic physiology. Since strong diel variability was
47
observed before day 330 and the loss of diel character coincides with the increase in
dilution rate on day 328, the absence of did patterns under HGR-LL is presumed to
reflect actual physiology and not measurement artifact. Cells growing rapidly in
nitrate replete conditions exhibit a slow long term increase in Fv/Fm and a slow long
term decrease in psii, p and 'r. The increase in irradiance to HL corresponds to the
onset of diel cycles in all parameters: Fv/Fm and
YPSII
exhibit characteristic decreases
around solar noon, and 'r gradually develops a midday increase.
exceeded EPAR at all times, was consistently below
EPAR
EK,
which under LL
during HL for a period
around solar noon. Similar to the LGR experiment, this indicates that photosynthesis
is always light limited under LL conditions and is light saturated only at high
irradiances under HL conditions.
3.4.2 Nitrate-Driven Responses under LL and HL
During these two experiments there were three periods when the relative
availability of nitrate on a per cell basis changed substantially in the culture vessel.
For each of these periods the photosynthetic parameters
F0, Fm, FvIFm,
psii, p and 'r
are plotted along with the coefficient of fluorescence c1 (Figure 3.7, Figure 3.8 and
Figure 39)7 The first period coincided with the first sixteen days of the LGR
experiment, after cells were inoculated on day 209 into an undiluted chemostat in
which dilution began on 0730 on SDT 213 (213.3, Figure 3.7). Under these
conditions, increases in cell abundance decreased the amount of nitrate available to
each cell. All physiological parameters show some degrees of adaptation over these
sixteen days and all eventually converge to the kinetics observed during LGR before
the shift to high irradiance (Figure 3.5). The primary changes in C1 during this period
were an approximate thirteen day increase in overall magnitude and an evolution in
diurnal shape from a quasi-linear form to one more symmetric.
Chlorophyll samples were not routinely taken during these periods and so
Fch
cannot be calculated.
A
-
E
LL
\
\
IL
0
2 0
252
254
I
I
258
256
I
260
264
262
I
I
266
268
I
I
I
B
0.6
S
,'
E
\A\
'
;.
0.5
U-
0.4
2 0
270
5,
\:\
252
254
256
258
260
262
264
266
268
270
252
254
256
258
260
262
264
266
268
270
500
b
400
300
2
0.4
0.3
0.2
0.1
2
2
n
250
F
\'
200
w
\/' i'.(
0
250
252
254
5'
t'
/
256
t/\
'S
.
.
.1
258
I
.1
260
262
.
'4
.1
264
.1.
266
.1
268
270
SDY
Figure 3.5 Photosynthetic physiological parameters measured by variable
fluorescence during the LGR culture experiment, a) and Fm (relative units), b)
FvIFm (dimensionless), c)
sii (A2 quanta'), d)p (dimensionless), e) 'r (ps, note shift
in scale), and I) EPAR and EK Qimol quanta m2 s, dashed and solid line respectively).
Data from SDY253-254 were lost during processing.
F0
330
0.6[-
332
334
336
...........................
338
340
342
346
348
350
.
I
E04
4
0
344
I
V.
'VV\
I
I
330
332
334
336
338
340
342
344
346
348
350
330
332
334
336
338
340
342
344
346
348
350
I
I
I
I
346
348
350
0.3
330
332
334
336
338
I
340
342
I
344
10000
Iç,III
;1/
E
°:
330
I
332
334
336
338
340
342
I
I
344
346
348
350
I
F
400
2o:
330
332
334
336
338
340
342
344
346
348
350
SDY
Figure 3.6 Photosynthetic physiological parameters measured by variable
fluorescence during the HGR culture experiment. Panels and data are as in Figure 3.5.
The shift in values on day 337 coincides with the change in instrument gain and data
from days 338-340 were lost during collection.
50
The second period of variable nitrate availability occurred during the first
fifteen days of the HGR experiment, when dilution rate was initially zero but was
increased eight days later on day 328 (Figure 3.8). Unlike the period following
inoculation in the LGR experiment,
Fv/Fm
and Ct' decreased steadily during the
undiluted period, indicating relative nitrate starvation and loss of biomass respectively.
Since the volume of inoculum was much greater in the HGR experiment, the
progressive decrease probably reflects conditions where cells experience relative
nitrate deficiency due to resource overutilization. The concurrent decreases in
photochemical competency (Figure 3.8b), increases in connectivity (Figure 3.8d) and
high diurnal increases in 'r support this relationship (Figure 3.8e). These responses are
apparent in 1 in that afternoon values are always lower than forenoon values (Figure
3.8g). As dilution commences on day 328, a reversal in the diurnal and long term
character of FvfFm,
cYpsii and EK
are all immediately apparent and consistent with
increases in relative nitrate availability. Particularly, 1 begins to develop the diurnal
quasi-linear pattern observed at the beginning of the LGR experiment when relative
nitrate was abundant (Figure 3.8g).
The third period of change in relative nitrate availability is associated with the
onset of nitrate starvation under HL conditions as dilution ceases on day 349 (Figure
3.9).
C1
in general decreases in amplitude and exhibits an enhancement in the
characteristic midday depression (Figure 3.9g), presumably corresponding to
decreasing chlorophyll biomass and increasing nitrate stress respectively.
Photochemical competency decreases (Figure 3.9b), presumably due to an increased
inability to repair damaged Dl reaction center protein under high light and low nitrate
(Han et al. 2000). To compensate for the loss of functional RCII relative to antenna,
cYpsii begins to increase after four days, but PSU electron turnover time increases
sooner (Figure 3.9c&e). The photosynthetic compensation irradiance EK
progressively decreases during most of the light period, indicating that the overall
capacity for photosynthesis becomes increasingly light saturated over time (Figure
3.90.
2
.
210
212
I
I
214
216
218
51
220
222
224
226
224
226
0.6[-
J\AJ+J
212
210
800
600
214
212
210
0.2
214
218
216
222
-
. 0 1
220
/
222
224
I
I
212
I
214
216
218
1
210
300
I
iook !\
212
214
216
I
I
.iVI.11
226
E
(
-'j
220
222
I
I
224
i\
J\
!\
1
.I
212
I
f. 'i
/11/I//i
/
214
1\
216
I
218
226
F
/
220
... .!\'fl
I
ii.J!.i
/
222
224
I
I
226
p
1
L L L
.
214
!\:.!
71 /
......
212
.
I
I
1
I
j
0
210
218
224
222
f
0.04
0.02
220
I
Ui
0
210
D
j.
I
210
226
.....
i.
'
C
-
......................................
I
j
200
220
r
.
400
218
216
216
218
220
222
224
226
SDY
Figure 3.7 Photosynthetic physiological parameters during a period when nitrate
availability per cell decreased under LL. Panels, data and units are as in Figure 3.6,
except for the addition of Ct in the fifth panel. FRR variable fluorescence data were
collected at high gain before day 214 and at low gain after day 217.
52
1.5
I
J__
0
\rI
IL
1
E
'-_'
/
IL05
,-.___
/. -...........
I
324
322
328
326
I
I
330
332
334
336
I
I
: :
0
I
I
322
324
600
i
,.:.
328
330
332
i!r
334
336
....................................
200 .
400
b
326
I
II
324
322
0.3
r
I
326
I
328
332
330
I
334
336
I
I
JI
i
t
'
0
I
I
324
x 1 o22
D
iFHø?
326
I
328
330
I
332
334
336
332
334
336
:. ................... ii.
;'/M
324
322
326
,
328
330
.....................
wI)(tA!AAAAA
200h
1.
f&$
324
322
326
328
330
'.
F
F
.AA .AAAI
332
334
336
G
eoolF-
'"
'
'
r-
L
0
322
324
326
328
330
332
334
336
SDY
Figure 3.8 Photosynthetic physiological parameters during a period when nitrate is
depleted by and then provided to HGR culture experiment under LL. Panels, data and
units are as in Figure 3.7.
53
;'jf
.t
E05
348
346
350
354
352
358
356
360
----I
0.6-
:B
EQ 4
-S
0 2
/
Q
I
-
I
348
346
1000
I
350
I
I
354
352
I
I
I
358
356
360
..
I
I
800
.
JI
r
b
348
46
354
352
350
0.6
I
358
356
I
I
360
I
:Dy
0.4
V
'-/
0
I
348
350
2
I
352
I
I
354
I
356
I
358
360
I
I
r -
I
346
348
I
wS4oQ
350
352
I
I
354
I
I
A
t
358
356
360
I
1
I,
r
200
8
346
352
356
358
Q.Q
002
I'
p
J
/I
..
0.01
i 4
346
348
- rr
350
r-'
,,
,,
-
,352
,
354
5356
358
360
SDY
Figure 3.9 Photosynthetic physiological parameters during a period when nitrate
delivery ceased to the HGR culture experiment under HL. Panels, data and units are
as in Figure 3.7.
3.4.3 Diel Variability during Intensive Sampling
Fchl
and [chl a] during the 24 hour intensive measurement periods during
LGR are shown in Figure 3. lOa-b for LL and ilL conditions, respectively. Under LL,
[chi a] shows a relatively constant diel pattern, but under ilL it declines steadily, each
response consistent with the volume specific trends observed in Figure 3.1. Diurnal
variability in
Fch1
appears to be uncorrelated with [chi a] under LL and somewhat
more correlated under HL, suggesting that some of the diurnal character in each can be
ascribed to changes in [chl a]. A similar presentation for intensive sampling during
HGR is shown in Figure 3.lOc-d where again diurnal [chl a] is consistent with the
longer term trends both under LL and HL. During HGR, diel
FchI
appears correlated
with [chl a] under LL conditions, but a relationship cannot be clearly seen under HL.
Cell abundance as well appears correlated with
FchI
during HGR-LL but again data
are inconclusive under ilL (Figure 3. lOe&f). Consequently, the observed AM-PM
asymmetry in 1 during low light and high nitrate appears to be primarily due to
population increases because cell specific [chl a] remains constant during the diurnal
period.
3.5 Discussion
3.5.1 Correlation between c1 and [chl a]
Rearranging the terms in Equation 3.1 illustrates the relationship between the
natural fluorescence per unit irradiance (i.e. 1), chlorophyll biomass, the chlorophyllspecific absorption coefficient and the quantum yield of natural fluorescence:
ct,=[chl].a*.c1Y
3.2
55
a
b
257[chla]&
140
'.A.
0.0002
C)
80
-.-- chla
0.00015
3
0.0001
0
24 28
20
tine of day (hour)
0.0003
4---chla
L._336
0.0002
o.000i
0
0
22
0.00005
0
14
9
d
0.0004
18
26
348(chla]&q
0.001
0.0009
0.0008
0.0007
la
0.0006
10
0
0.0005
4
8
16
12
20
tine of day (hour)
336 abundance & chi per cell
0.015
0.01
24
19
90
80
70
60
50
40
30
20
tirre of day (hour)
e
o.000i
tine of day (hour)
0.0005
14
0.00015
.--chla
4
0.0006
10
..i-"
348 abundance & chl per cell
3100
2900
0.002
61000
60500
2700
2500
0.0015
60000
59500
59000
58500
58000
57500
57000
2300
a
0.005
0
10
14
chVcell
i9oo
0000s
336 cells
1700
1500
0
18
22
tine of day (hour)
0.00025
0.0002
0
336[chla]&q
5
40
32
30
10
9
10
0
16
0.0003
50
30
0.00005
0
C
0.00035
20
257
20
12
0.0004
/
60
100
8
0.00045
70
o.0002s
120
3
269[chlaj&4
80
0.0003
160
26
a
4
8
ell
348 cells
12
16
20
tine of day (hour)
Figure 3.10 Sub-die! variability during the LGR and HGR experiments. Panels in the
left column represent LL conditions; right column panels represent Fit conditions. a)
[chi a] and FchI for LGR-LL, b) [chl a] and cJth1 for LGR-HL, c) [chl a] and Fchl for
HGR-LL, d) [chi a] and qFch1 for HGR-HL, e) cell specific [chl a] and cell abundance
for HGR-LL, f) cell specific [chl a] and cell abundance for HGR-HL.
56
Other workers have shown a clear correlation between parameters functionally similar
to
cI
and [chl a] (e.g. Kiefer et al. 1989; Cullen et al. 1997), and so we performed
Model H linear regressions on our data to determine the extent of this relationship in
these culture experiments. Since culture data were collected under well defined
irradiance and nitrate conditions, the strength of and changes in the relationship
between ct and [chl a] can be examined with respect to each of these factors. We
created two metrics to monitor day to day variability in natural fluorescence: the
absolute diurnal maximum c1 ("max t") and the value of c1 observed exactly two
hours before solar noon ("AM c1"). The latter is more appropriate to remote sensing
applications when the time of the absolute diurnal maximum cannot be directly
determined. We examined the correlations between these metrics and [chi a] for five
periods in our data: before, immediately after and well after the shift up in irradiance
during the LGR experiment, and before and well after the shift up in irradiance during
the HGR experiment (Figure 3.11).
In general, the relationship between c1 and [chl a] was very strong and only
one period exhibited poor correlation, during the LGR experiment well after the shift
up in irradiance (days 262-8, Figure 3.11c). In the LGR two periods where a strong
correlation was observed, the proportionality of these relationships were not similar,
indicating that a single relationship would be inadequate for predicting [chl a] from c1
when nitrate-limited cells are exposed to a shift up in irradiance. In contrast, during
the entirety of the HGR experiment the correlation between max c1 and AM t was
high under both low and high irradiances (Figure 3.1 ld&e). Further, the correlation
over all twenty days of HGR was equally strong
(r2
0.87), suggesting that the two
relationships noted independently under LL and HL are effectively the same during
nitrate replete conditions. From these results we can conclude that although 1 is well
coupled to chlorophyll a biomass under certain conditions, there are other situations
where physiological decoupling between
to estimate the latter.
Fnat
and [chl a] prevents the use of the former
57
correlations with [chi a]
140
I
120
A
r2=0.82
*
--_ -
A
................................
o
r=O.80
LGR LL 250-57
80
0.015
0.025
0.02
12C
0.03
I
B
r2=0.97
-
110
90
...
80
LGRHL25B-61
0.015
0.02
0.025
0.03
...
aol
I
I
...................
60
50
40
0.015
I
L_S.AI I
I
I
I
C'J1J1_I
0.02
25
I
I
I
0.025
0.03
I
I
0
20
*
15
HGR LL 332-7
10
I
2
4
I
I
8
6
10
12
14
16
xlO
80
70
? 60
50
0.025
0.03
0.035
0.04
max () and AM
(A) t
0.045
0.05
(relative units)
0.055
0.06
Figure 3.11 Diurnal maximum (*) and lOAM (Li) 't regressed against [chi a] for two
periods of steady state growth during LGR under LL (a) and HL (b), and three periods
of positive population growth: LGR shift to HL (c), HGR-LL (d) and HGR-HL (e).
There is no a priori reason to assume that Equation 3.2 would be universally
dominated on the right hand side by chlorophyll biomass. Our results in fact suggest
the opposite; that when nitrate limited cells are adapting to a high light environment,
variability in [chl a] can account for only about 40% of the variability in
ct.
In all
other situations resolved by these experiments, the strong correlations between ct and
[chl a] occur over relatively small ranges of [chl a], generally less than a factor of two.
In the correlations observed by other workers (Kiefer et al. 1989; Cullen et al. 1997)
the variability in chlorophyll biomass spanned at least two orders of magnitude and
thus probably dominated Equation 3.2, effectively masking variability in either a* or
F
The correlations of Kiefer et al. (1989) and Cullen et al. (1997) exhibit substantial
statistical power but may only reflect between-community relationships. When
variability in [chl a] is approximately the same order as variability in a* or
F
only
then do within-community relationships like those observed in our experiments
become apparent.
3.5.2 Nitrate-driven Variability in 't
For the three periods when per cell availability of nitrate varied substantially,
corresponding gradual changes in the diurnal character of 1 reflect physiological and
ecological responses to changes in these environmental properties. To determine if
variability in 4 can be used to track nitrate-driven responses, the ratio of afternoon to
forenoon
("PMIAM") was examined both before and after changes in nitrate
availability. When equally spaced before and after solar noon, changes between AM
and PM 1 reflects the degree of asymmetry between morning and afternoon t (Figure
3.12). When PMJAM is greater than one, afternoon t is larger than forenoon and
relative abundance of nitrate is expected, because the diurnal slope of c1 appears to be
correlated to [chl a] and cell abundance, but not to chlorophyll a per cell (Figure 3.10).
Values around unity suggest steady state growth (i.e. diurnal symmetry in c1) and
values less than one indicate conditions of relative nitrate depletion.
59
LL - dilution started on morning day 213
0
21
a-
SDY
LL - dilution started evening day 328
1.4
1.3
o 1.2
1.1
c11
0.9
320
325
330
335
HL - dilution ceased morning day 349
1.2
11
0
a-
0.9
346
348
350
352
354
356
358
360
S DY
Figure 3.12 Long term changes in the ratio of PM to AM 1. Six different ratios with
increased sampling between forenoon and afternoon are shown (1 to 6 hours).
The PM/AM ratio behaves as predicted under LL conditions. Cells introduced
into a nondiluted chemostat show PM/AM above 1, which decreases to unity as a slow
dilution rate is established and cells acclimate to steady state growth (Figure 3. 12a). A
clear response to an increase in available nitrate is evident in the first 15 days of the
HGR experiment when dilution rate was increased from zero in the evening of day
328 (Figure 3.12b). The initial introduction of culture to a nondiluted culture vessel
resulted in a gradual decrease in PM/AM below unity, and an increased supply of
nitrate immediately reversed this trend.
However, under IlL conditions this ratio does not appear to capture the
dynamics of nitrate starvation (Figure 3.12c), except possibly when the PM/AM
spacing interval is 6 hours. This is most likely because of the severe midday decrease
in
C1
during HL conditions. Although PM/AM may work well as an indicator of
physiological response to relative nitrate availability under LL, the poor performance
under HL conditions severely constrains the practical utility of this proxy in situations
where the daily irradiance profile is best characterized as "HL".
The improved quality of PM/AM at large intervals (>6 hrs) suggests that
c1
does contain information about gradual physiological responses to nitrate availability
(e.g. Figure 3.9g), but such a simple ratio is apparently not sophisticated enough to
identify the pertinent kinetics. Six other quantitative metrics derived from
measurements of 1 were created to emphasize the amplitude and timing of the midday
depression and thus better capture the diurnal variability in 'I under HL. Time series
of these metrics are shown in Figure 3.13 for the same Fit data over which PM/AM
performed poorly in Figure 3.12c. The first panel shows how specific aspects of
diurnal t under I-IL are quantified, and each metric is calculated from either the
magnitude or timing of the morning and afternoon t maxima (dashed lines) and the
diurnal minima (dot-dashed line).
61
Monotonic and quantifiable changes can be identified in five of the six metrics.
The timing of the morning and afternoon maxima (Figure 3.13c) shows opposite
trends which reflect the increasing width of the midday decrease and presumably the
increasing influence of nitrate starvation. Timing of the midday minimum appears
relatively insensitive and shows no identifiable long term kinetics (Figure 3.13e). The
time elapsed between morning and afternoon maxima shows four distinct phases
(Figure 3.13b), suggesting that up to four different responses may be contained in this
proxy. Similarly, the ratio of PM to AM maximum (Figure 3.13d) and the ratio of
diurnal minimum to maximum (Figure 3.130 show four identifiable phases, which do
not always coincide with those in Figure 3.13b.
These metrics convert complex variability in Fnat into relatively simpler
variability more easily interpreted using models familiar to most ecologists. The large
number of identifiable kinetics suggests that several different physiological and/or
ecological responses may potentially be recovered from time series of c1. The ratio in
Figure 3.13d is very similar to the PMJAM ratio discussed previously, and exhibits the
expected PM/AM response for the onset of nitrate starvation. Although the
relationship between c1 and cell abundance was not directly measured (although
implied from the correlation in Figure 3.1 lj), a logistic model can be applied to these
changes to estimate time constants of adaptation, percent changes in population and
apparent population growth rate.
3.5.3
Physiological Sources of Variability in Fnat
A great deal of variability was observed in the time series of photosynthetic
physiology collected using the FRRF during these experiments. As a first look into
how physiological changes drive natural fluorescence variability, we examined the
relationship between
FchI
and the light saturation irradiance
EK.
62
Defining simple metrics
time between maxima
9.5
0.015
9
\.
0.01
8.5
/
8
6,
&
7.5
:.
-C
7
0.005
6.5
6
0
0
10
5
15
350
345
20
355
360
time of day, hours
morning & afternoon max time
afternoon max I morning max
22
20,,
:4
-.-""-'
D
/
C
18
095
--.-''''
16
4
-C
0.9
14
4
12
10
345
350
355
345
360
350
355
360
355
360
day
time of minimum
mm/max
0.8
22
E
20
0.75
0.7
18
0.65
C
16
0.6
14
12-ic
345
0.55
.............
05
A A
350
355
day
360
350
day
Figure 3.13 Additional empirical metrics to examine II under HL conditions. a)
Definition of maxima (dashed lines), minima (dot-dashed lines) and solar noon
(circle), b) timing between maxima, c) timing of maxima, d) ratio of maxima, e) time
of minimum and ratio of minimum to morning maximum.
63
Two periods during HL conditions, one for 12 days during LGR and one for 7
days during LGR, were examined to determine the irradiance at which diurnal
ct'
maximal in both the forenoon and afternoon. For each of these days we also
determined the irradiances at which EPAR first increased above and finally fell below
EK.
These latter irradiances represent points at which photosynthesis switches
between light limited and light saturated. This photosynthetic threshold was then
plotted against the gross characteristics (i.e. AM and PM maxima) noted in 1 under
HL. The correlation between the two irradiances is shown for HL periods during LGR
(Figure 3. 14a) and HGR (Figure 3. 14b). The AM irradiance of maximal VthI was
better correlated with this photosynthetic threshold than the afternoon maximum
(Figure 3. 14a) as the afternoon maximum
transient period during HGR-HL,
FchI
FchI
tended to overpredict EK. After the
morning maxima were greater in absolute
value but still reasonably well correlated to EK. Afternoon qFchl maxima again were
not well correlated, tending to underpredict EK (Figure 3. 14b).
The comparatively poorer performance of qFch1 to predict EK in the afternoon
is most likely due to physiological hysteresis in PS11. Since EK is calculated from only
apsjj and t, the additional influence of Fv/Fm and/or p on
Ft may be the source of the
decoupling between qFch1 from EK in the afternoon. Under LGR conditions, p is only
slightly greater in the afternoon than in the morning (Ap <0.04) and would tend to
decrease afternoon qFchI compared to morning values. Consequently, diurnal
variability in p may be the source of the decoupling, because it drives qFch1 to
correspondingly lower values while not influencing EK. At the irradiance at which
FchI
is maximal in the afternoon, Fv/Fm is lower than morning values, which would
tend to increase afternoon qFchI relative to morning values. Consequently, p and not
Fv/Fm appears to be driving the overestimation of maximal EK by maximal
afternoon.
Fch1
in the
LGR-HL SDY258-269
300
250
200
I
150
100
50
0
50
100
200
150
Irradiance of AM & PM max
eFchl
250
(io1 quanta m2 s1)
HGR-HL SDY343-349
U)
2E
=
o 2(
U,
Dl
0
0
OL
0
50
100
150
Irradiance of AM & PM max
eFchI
200
(pmol quanta
250
300
m2 s)
Figure 3.14 The relationship between the irradiance of maximal qFchI and the
threshold at which photosynthesis switches between light limited and light saturated
(EK) in the forenoon (*) and the afternoon (x). The straight line in each plot represents
a 1:1 relationship.
3.6 Conclusions
Field measurements of total variability in natural fluorescence have been
correlated to volume specific properties such as chlorophyll concentration and carbon
fixation, but to the best of our knowledge these laboratory experiments are the first to
focus specifically on the physiologically-driven variability in natural fluorescence.
This physiological variability can be isolated using data which can be presently
retrieved using satellite remote sensors, and simple metrics appear to be robust enough
to monitor long term variability in 'I and
VchI.
Changes in either c1 and
FchI
can be
used to infer physiologically pertinent responses at the level of an individual
phytoplankter, e.g. changes in chlorophyll per cell, as well as ecologically pertinent
responses, e.g. changes in total population due to nutrient availability. Applications of
relationships similar to those presented in this paper will increase our ability to infer
physiological and ecological responses of phytoplankton from the natural fluorescence
signal, and will greatly extend our ability to interpret and apply remotely sensed
natural fluorescence data to the important global biogeochemical questions in
biological oceanography.
Ideally, physiological and ecological inferences based on natural fluorescence
could be applied to improve estimates of primary production based on remote sensing
data. By definition, the quantum yield J is proportional to
F
but the variability in
this relationship is a complex function of physiology, largely ignored due to the lack
of experimental data (Kiefer et al. 1989). Laboratory investigations like those
described here are critical to better understanding how these different quantum yields
are connected. In the single diatom species examined, the daily morning maximum in
Fch1
appears to be associated with the irradiance at which photosynthesis switches
from light limited to light saturation. If this relationship exists in natural
phytoplankton populations, estimated of EK from fluorescence radiance could be
applied to greatly improve basin-scale estimates of primary production, because the
irradiance dependent photosynthetic period could be identified and separated from the
irradiance independent period. Although one or two satellite sensors may not be
adequate to resolve this maximum accurately, higher resolution sensors which monitor
Fnat
quasi-continuously, such as on drifters and moorings, might be used to provide
continuous estimates of this switching irradiance using
FchI
Acknowledgments. We thank Claudia Mengelt and Lisa Eisner for advice and
assistance with processing and analysis of the HPLC samples. Nutrient samples were
analyzed by the Prahl lab at Oregon State University. This research was supported by
the National Aeronautics and Space Administration (NAS5-3 1360).
3.7 References
Abbott, M.R. and R.M. Letelier, 1997a: Bio-optical drifters Scales of variability of
chlorophyll and fluorescence. Society of Photo-Optical Instrumentation
Engineers 2963, 216-221.
Abbott, M.R., and R.M. Letelier, 1997b: Going with the Flow The Use of Optical
Drifters to Study Phytoplankton Dynamics. Monitoring Algal Blooms: New
Techniques for Detecting Large-Scale Environmental Changes. M. Kahru and
C.W. Brown, Eds., Landes Bioscience, pp. 143-168.
Chamberlin, W.S., C.R. Booth, D.A. Kiefer, J.H. Morrow, and R.C. Murphy, 1990:
Evidence for a simple relationship between natural fluorescence,
photosynthesis and chlorophyll in the sea. Deep-Sea Research 37, 95 1-973.
Chamberlin, S. and J. Marra, 1992: Estimation of photosynthetic rates from
measurements of natural fluorescence: analysis of the effects of light and
temperature. Deep-Sea Research 39, 1695-1706.
Cullen, J.J., A.M. Ciotti and R.F. Davis, 1997. "The relationship between nearsurface chlorophyll and solar-stimulated fluorescence: biological effects" in
Ocean Optics XIII, Steven G. Ackleson, Robert Frouin, Editors, Proc. SPIE
2963, 272-277.
Cullen, J.J. and M.R. Lewis, 1995: Biological processes and optical measurements
near the sea surface: Some issues relevant to remote sensing. J. Geophys. Res.
100, 13,255-13,266.
Eppley, R.W., R.W. Holmes, and J.D.H. Strickland, 1967: Sinking rates of marine
phytoplankton measured with a fluorometer. J. EXP. MAR. BIOL. ECOL. 1,
19 1-208.
Gordon, H.R., 1979: Diffuse reflectance of the ocean: the theory of its augmentation
by chlorophyll a fluorescence at 685 nm. Appi. Optics 18, 1161-1166.
Gower, J.F.R, and G. Borstad, 1981: Use of the in-vivo fluorescence line at 685 nm
for remote sensing surveys of surface chlorophyll a. In 'Oceanography from
Space'. (ed. J.F.R. Gower) pp. 329-338. (Plenum Press: New York.)
Han, B.-P., M. Virtanen, J. Koponen, and M. Straskraba, 2000: Effect of
photoinhibition on algal photosynthesis: a dynamic model. J. Plankton Res.
22, 865-885.
Kiefer, D.A., 1973: Fluorescence properties of natural phytoplankton populations.
Marine Biology 22, 263-269.
Kiefer, D.A., W.S. Chamberlin and C.R. Booth, 1989: Natural fluorescence of
chlorophyll a: relationship to photosynthesis and chlorophyll concentration in
the western South Pacific gyre. Limnology and Oceanography 34, 868-88 1.
Letelier, R.M., M.R. Abbott and D.M. Karl, 1997: Chlorophyll natural fluorescence
response to upwelling events in the Southern Ocean. Geophys. Res. Lets., 24,
409-412.
Stegmann, P.M., M.R. Lewis, C.O. Davis and J.J. Cullen, 1992: Primary production
estimates from recordings of solar-stimulated fluorescence in the Equatorial
Pacific at 150 degrees West. Journal of Geophysical Research 97C, 627 638.
Wright, S.W. and S.W. Jeffrey, 1997: High-resolution HPLC system for chlorophylls
and carotenoids of marine phytoplankton. In: Phytoplankton pigments in
oceanography: Guidelines to modern methods. Jeffrey, S.W., Mantoura,
R.F.C., Wright, S.W. (Eds.), UNESCO Publishing, Paris, pp. 327-341.
Yentsch, C.S. and Menzel, D.W. (1963). A method for the determination of
phytoplankton, chlorophyll, and phaeophytin by fluorescence. Deep-Sea Res.
10, 221-231.
4. Assessment of Reaction Center Connectivity in Phytoplankton
using Fast Repetition Rate Fluorometry
"Like Zen Buddhism, an understanding of primary production in natural waters may be achieved
through meditation as well as through reading conventional wisdom".
Paul Falkowski
4.1 Abstract
Energetic connectivity between Photosystem II (PS11) reaction centers can be
expressed as a probability (p) that energy arriving at photochemically closed reaction
centers will be redirected to other PS11. Such redirection can potentially enhance
photochemical quantum yield, and so the influence of p on PS11 electron flow rates
(Pe) was examined in numerical simulations and culture experiments. Simulations
identified two conditions where increased p enhanced Pe by a substantial amount:
when the proportion of competent reaction centers is decreased and when the PS11
population is densely packed. Using a Fast Repetition Rate fluorometer (FRRF), the
actual diel and long term variability in p was measured in continuous cultures of the
marine diatom Thalassiosira weissflogii (Bacillariophyceae). Connectivity and other
photosynthetic parameters were estimated from FRRF measurements using both the
commercial FRRF software and a more robust nonlinear fitting procedure. Estimates
differed between software packages, primarily around solar noon. Although input and
calibration data were identical, the selection of numerical fitting approach appears to
influence the accuracy of fitted parameters, especially p and the functional absorption
cross section of Photosystem II, cJpsii. Examination of the physiological model
currently used to interpret FRRF variable fluorescence measurements revealed a
numerical interdependence between p and cYpsij, which can partly explain the
comparatively poorer performance of the commercial software. When PS11 are
densely packed, underlying statistical assumptions of the physiological model may be
violated and in such cases FRRF-derived estimates of p may not reflect connectivity
per se
and may instead contain some indication as to the degree of packing. To better
understand how opsii and p together control
pe
in actual marine environments using
variable fluorescence methods, numerical fitting procedures must be carefully
examined and the current physiological theory of PS11 variable fluorescence must be
improved.
4.2 Introduction
4.2.1 Background
Almost all the chlorophyll a fluorescence ( 90%) observed at in vivo
temperatures emanates from Photosystem II (PS11), a photosynthetic structure
responsible for absorbing light energy from the ambient environment and converting
this energy into a biochemical form. Light energy absorbed by a PS11 migrates
through a complex of light harvesting pigments until it reaches a reaction center
(RCII), where the excited state energy is converted into biochemical energy in the
form of a reduced electron acceptor. This conversion is called charge separation and
is the first step in a long chain of events ultimately ending in the fixation of organic
carbon.
Certain experimental evidence suggests that in many algal species individual
PS11 do not act independently (Joliot and Joliot 1964; Ley and Mauzerall, 1986).
Energy appears sometimes to be transferred between PSII, which can increase the
overall efficiency of photochemistry by better sharing excitation energy among the
entire PSU population. The molecular basis for this connectivity is not well
understood, and a complete physiological explanation to explain this sharing has not
yet been developed. Instead, RCH connectivity is instead often described empirically
as a probability p that excitation energy, having once already visited a RCII busy
70
processing prior energy, will be redirected to another PS11 which may or may not be
so occupied. Fast Repetition Rate fluorometric techniques can be applied to examine
RCII connectivity in field samples,
in situ
and in real time. The theory and
methodology of FRRF measurement is discussed in detail elsewhere (Kolber et al.
1998) and describe how variable fluorescence kinetics can be analyzed to examine the
structure and function of the PSH population.
To recover photosynthetic information from FRRF measurements, a
physiological model of variable fluorescence () is required to relate observed
changes in
to the underlying physiological properties that lead to fluorescence in
PS11. The physiological model of Kolber et al. (1998) is represented mathematically
by Equations 4.1 and 4.2, where
ip
fn=Fo+(Fm_FoCni_CpJ
4.1
and
1cfl-1
C,,
cr11i
.
4.2
lcn-1p
f,,
represents the fluorescence yield corresponding to flashlet n, and F0 and Fm are the
initial and maximal observed fluorescence yields8. C,, is an intermediate term
describing the percentage of RCII closed due to charge separation at any point during
the measurement sequence, and 4, is the number of excitation photons delivered by
flashlet n in units of tmol photons m2 flashlef'. Initial conditions are defined such
that all RCII are open before the first flashlet (i.e., Co = 0). Figure 4.1 a & b show
8
We will use nomenclature and symbols from Kolber et al. (1998) throughout this paper.
71
data collected during a typical FRRF induction protocol, including the best fit curve of
the physiological model and the residuals of the fit.
Vanable fluorescence yield data and best fit
0.8
w
0
=
0.4
(0
0.2
A
0
50
100
150
200
250
300
350
400
450
350
400
450
time in s
0.06
0.04
c
0.02
:\j\
(0
0)
0.02
\
0.04
0.06
0
50
100
150
200
250
300
time in s
Figure 4.1 Typical FRRF variable fluorescence data. a) A typical variable
fluorescence observed during a FRRF ST flash sequence collected on dark adapted T.
weissflogii in continuous culture. Best fit parameter estimates are generated using our
v3 custom analysis package. b) Residuals between observed data and expected fit
values. The residuals are evenly distributed and show no apparent bias.
4.2.2 Motivation
Few experimental analyses of connectivity have been performed either in the
laboratory or on field samples, and so the photosynthetic and ecological importance of
connectivity in the highly variable marine environment is largely unknown. FRRF
72
methodology can be used to rapidly measure the connectivity parameter, but there has
not yet been a detailed examination of the experimental retrieval of p using these
methods. In order to better understand how variability in connectivity influences Pe in
oceanic phytoplankton, both the concept of connectivity and the methodological
retrieval of this parameter need to be closely examined.
Ideally, parameters used to construct any physiological model will be
independent of each other, but a linear approximation of Equation 4.2 for small p
reveals terms combining psii andp (see §4.7.1). When interdependent parameters are
presumed independent, the performance of algorithms used to fit models to data can
be degraded, leading to inaccurate parameter estimation and erroneous assignment of
variability, especially among associated parameters. Estimates of p are particularly
vulnerable in the present model because p is more difficult to fit than apsii, which is
easily illustrated using the data in Figure 4.la. Presuming that fitted
F0
and Fm are
"correct", Pearson's X2 statistic was calculated for repeated fits of these data over the
entire range of p and a proportionately smaller range of
sti (Figure 4.la). The
resulting surface shows quality of fit in Opsii-p space, and each contour of constant X2
exhibits much greater latitude in p than in
PSll.
This relationship appears to be
amplified when the input data come from samples collected around solar noon as
opposed to midnight (Figure 4. ib). Unless it is known a priori that p is zero, fitting
Equations 4.1 and 4.2 to FRRF variable fluorescence data may not correctly retrieve
either, due to the interdependence of p and psii. Sophisticated ("robust") algorithms
can generally improve parameter estimation in such cases, but numerical techniques
cannot rectify deficiencies caused by an improper model. Correct interpretation of the
physiological impact of changes in p and cYpsii requires accurate assessment of
photosynthetic parameters, and so we needed to carefully examine the theoretical and
numerical relationship between the two, especially p.
73
x2
surface: data around solar noon
B
500
450
400
g350
300
250
nn
0
0.1
0.2
0.3
0.4
0.5
p
0.6
0.7
0.8
0.9
Figure 4.2 Fit space for the current physiological model. a) Quality of fit (Pearson's
X2) in Opsii-p space surrounding the best-fit solution (asterisk) to the data in Figure
4.la. b) Similar plot for variable fluorescence data collected near solar noon on the
same day.
Several recent papers presenting analyses of FRRF data identify significant
variability in cYpsii, but they do not discuss variability in p (e.g. Behrenfeld and Kolber
1999, Strutton et aT. 1997, Behrenfeld et al. 1998, Gorbunov et aT. 1999). p may not
be reported because it is ignored or otherwise not evaluated, or because it is deemed
physiologically or ecologically irrelevant. Given the interdependence between these
two parameters and the comparative weakness of p. it is necessary to first examine the
numerical methods used to recover these properties from variable fluorescence
measurements. Only then can reliable estimates of p and
sii can be recovered from
74
field and laboratory FRRF measurements and the physiological and ecological
importance of RCH connectivity be examined.
4.3 Methods
4.3.1 PSI! Electron Flow Simulations
Numerical simulations were developed to examine how different physiological
conditions and different models of PS11 structure affect pe In these simulations, the
the probability of exciton transfer
functional cross section of each PS11
between PSII (p), the photon flux through the area per unit time (E), and the time
constant of QA reoxidation ('r) are specified.
n PSH
are randomly distributed within
an area L2, and a Monte Carlo process delivers photons into L2 with a random spatial
distribution at a fluence rate determined by E. The energetic fate of each photon is
individually tracked.
A fraction of n can be defined as photochemically incompetent. Photons
intersecting PS11 with open reaction centers contribute to electron flow, while photons
arriving at PSH with closed reaction centers have the probability p of being redirected
to another randomly chosen PS11. Photochemically closed RCH reopen according to
the time constant of QA reoxidation
t.
Each simulation is run for 500 ms of model
time (approximately 50t) for numerical and statistical stability. Each simulation
tracks the total number of absorbed photons and charge separations over a range of
intensities between 0 and 300 j.tmol photons m2 s1 and a range of connectivity
between 0 and 0.9. Photosynthetic quantum yield (cIv) is calculated as the ratio of
photons resulting in photochemical charge separation to the total number of photons
absorbed. Variability in thermal and chemical dissipation in PS11 is ignored in all
simulations.
75
4.3.2 Estimates of p in Continuous Cultures
Variable fluorescence was measured using a commercial FRRF (Fastracka,
Chelsea Instruments, UK) at one minute intervals during two 60 day continuous
culture experiments with the marine diatom Thalassiosira weissflogii
(Bacillariophyceae). These experiments encompassed a wide range of irradiance and
nitrate conditions. In each experiment measurements were performed in the FRRF
dark chamber within 90 seconds of removal from the culture vessel. Consequently the
measured photosynthetic properties reasonably reflect the degree of nonphotochemical
quenching experienced in the culture vessel.
Both the commercial FRRF software and a robust custom analysis package
were used to fit the physiological model of Equations 4.1 and 4.2 to the variable
fluorescence measurements. The commercially supplied software (FRS version 1.6
beta, Chelsea Instruments, UK) employs a minimization algorithm in which each
parameter is improved individually by stepping along maximum gradients in each
parameter dimension. We used the equations of Kolber et al. (1998) as a starting point
in developing the custom (v3) software9, which uses a trust region method (Coleman
and Li 1996; Coleman et al. 1999) to minimize parameters and an interior-reflective
Newton method to accommodate constraints on parameter values (Coleman and Li
1994). Best retrieval of model parameters is accomplished using a two-step
optimization process where F0, Fm and cYi'sii are first fit with p = 0, and then the
process is repeated holding the first three parameters constant and fitting for p.
Although the reason why the two step fitting process outperforms a single step, four
parameter fit is not understood, we take the qualitatively enhanced performance of v3
as further evidence that p is not an independent parameter in the current physiological
model. Quality of fit is calculated using Pearson's X2 statistic, and poor fitting
Although we discovered errors in Kolber et al. (1998) regarding the calculation of QA reoxidation
kinetics, we did not find any errors with the discussion of saturation kinetics (see §4.7.2).
76
solutions are rejected based on this statistic10. The software was written using the
most recent Optimization Toolbox in Matlab (Matlab Ri 1.1, The Math Works, Natick
MA).
Simulated variable fluorescence data were generated from Equations 4.1 and
4.2 to compare the inherent fitting performance of each software package. p was
varied over the range of [0 to 0.5]. Performance was first examined using ideal, noise
free data over a range of p from 0 to 0.5. Next, the simulated data were corrupted with
known levels of random noise ranging from 2% to 40% of the mean variable
fluorescence magnitude. These corrupted data were used to evaluate the robustness of
each software package with respect to data quality.
4.3.3 Numerical Simulations using Actual Culture Data
Fitting the physiological model to the FRRF continuous culture time series
using both software packages generates two time series of photosynthetic parameters.
The numerical simulation was modified to read in these time series and calculate PS11
electron flow at each point in time as a function of Fv/Fm,
YpSii, p, t
and EPAR. The
resulting modeled PC (with sample resolution of 1 min1) were then plotted as a
function of irradiance to produce pseudo P-I curves, which were used to examine how
actual variability in photosynthetic physiology, especially in p. controls gross
photosynthetic rates using electron flow as a proxy. Although the use of such proxies
is not well established in the plant physiology research community, the use of FRRF
methods is becoming increasingly popular in oceanographic studies and productivity
estimates derived from these methods, regardless of their accuracy, is gaining
acceptance given the unique requirements of oceanographic sampling.
Although FRSvI .6 (beta) also reports X2, we identified a computational error in the source code
where this statistic is calculated and could not use this parameter for evaluating FRS quality of fit.
77
4.4 Results
4.4.1
in Different Models of PSII Structure
Five different organizations of PSI! photosynthetic quantum yield were
modeled as a function of irradiance and p (Figure 4.3a-e). In all cases cJ' increases
with p and decreases with irradiance, even though each simulation differs with respect
to the organization and/or functionality of PSI!. The first simulation is the baseline
case where all reaction centers are functional (FvIFm = 1.0), self-shading within the
PS11 population was minimal (less than 0.05%) and PS11 acted as "connected units",
meaning that there are no restrictions on sharing between individual PSI! (Bemhardt
and Trissl 1999). In the baseline case (Figure 4.3a), an increase in p from 0 to 0.45 led
to enhancement in t' of over 50%, primarily between 50 and 150 .tmol photons m2 s
1
(Table 4.1). When 50% of the reaction centers are nonfunctional (Figure 4.3b)
absolute cI' drops accordingly, but enhancement of
irradiances,
50 tmol photons m2
11)
by p is still apparent at low
When FvIFm is 1.0 and self-shading is
increased by a factor of sixteen (Figure 4.3c), a 0 to 0.45 change in in p can enhance
1" by almost a factor of 2, primarily at higher irradiance levels.
Restriction on the sharing between reaction centers is examined using
"domain" models, where reaction centers can only share excitons within a "domain"
of PSI!, where the domain size is a small integer value. Domain models do not differ
appreciably from the connected units baseline case at p = 0 (Figure 4.3d-e, Table 4.1),
except that increases in p appear to have a lesser effect on enhancement of
41)
A
dimer model of PSI! connectivity (i.e., two PS11 per domain) appears to be slightly
more efficient than a domain model of size 5 as p increases.
Different cases of PSII model and influence of p
A) Baseline
0.5
I'
0
50
100
150
200
250
300
B) F/F = 0.5
0.5
I.'
0
50
100
150
200
250
300
C)x16 PSII packing
0.5
0
50
100
150
200
250
300
D) Domain size = 2
0.5
(I
I
0
50
I
I
100
I
150
I
200
I
250
300
I
E) Domn size = 5
0.5
I)
0
50
100
150
Irradiance (tiE m2 s1)
200
250
300
Figure 4.3 ct as a function of irradiance and p for five different numerical
simulations of a population of PSH. p in each panel ranges from 0 to 0.45 in
increments of 0.05, where p 0.55 is always the topmost curve, a) Tthe baseline
connected units case with FV/Fm = 1.0 and minimal self-shading. b) FvIFm = 0.5. c)
Significant self-shading. d) A dimer model with 2 PS11 per domain. e) A domain
model with 5 PS11 per domain.
79
Table 4.1 cI for five different simulations when p = 0 and when p = 0.4 (in
parentheses) at four selected irradiance intensities.
Simulation:
Baseline
Fv/Fm=0.5
10 jimol
photons m2 s1
83% (98%)
45%(66%)
84% (98%)
16x
50 pmol
photons m2 s
56% (86%)
28%(45%)
64% (93%)
photons m2
31% (44%)
16%(21%)
40% (70%)
56% (73%)
56% (69%)
31% (40%)
31% (35%)
150 j.tmo1
300 pmol
photons m2
19% (23%)
10%(11%)
26% (42%)
s1
packaging______________
Domain = 2
Domain = 5
83% (96%)
83% (96%)
19% (2 1%)
19% (20%)
4.4.2 Comparing Software Performance with Simulated Data
In both noise free and noise corrupted data, the v3 custom software
outperformed the FRSv1.6 commercial package in the estimation of photosynthetic
parameters, especially
sii and p (Figure 4.4). Traces in each panel track the mean
estimated parameter value forp from 0 to 0.5 (in increments of 0.1) for noise levels
between 2 and 40% (in increments of 2%). Error bars for each p at each noise
increment indicate ±1 standard deviation calculated from 100 randomly corrupted
simulated variable fluorescence data. As a comparison, the data in Figure 4.la
correspond roughly to noise corruption of 10% and are considered good for
unaveraged FRRF variable fluorescence measurements.
v3 estimated F0 within 2.5% of the correct values over the full range of p at all
noise levels and estimated Fm to within 1.5%, exhibiting a slight trend to
underestimate Fm at very low p (Figure 4.4b&d). v3 estimated cYp511 within 8% in all
cases, with greatest overestimation of
PSH
at low p (Figure 4.41). Its estimates of p
over the range 0 to 0.5 showed a sensitivity to increasing noise at low p.
overestimating p in the range between 0 and 0.05 (Figure 4.4h). Standard deviation of
all parameters increased monotonically with noise magnitude and deviation of the
mean from expected values was minimal in v3 results, except at very low p and in
0PSH at low p.
Given identical data, FRSv1.6 estimates of F0 and Fm deviated from the
expected values' by as much as 8% and in general deviated more as p increased
(Figure 4.4a&c). Estimates of 0psii were overestimated by at most 30%, with the
worst performance occurring when p was highest (Figure 4.4e). Retrieval of
connectivity was very poor, and the FRSv1.6 software appeared to converge on the
parameter boundary of 0.8 in a majority of the randomly corrupted data (Figure 4.4g).
The noise performance of FRSv1.6 was not monotonic or normally distributed,
indicating some degree of computational bias in the parameter retrieval algorithm.
4.4.3 Variable Fluorescence Time Series from Cultures
The FRRF variable fluorescence data from the continuous culture experiments were
processed using the commercial FRSv1.6 and the custom v3 analysis software, and
resulting best fit estimates for physiological parameters are presented in
Figure 4.5 for two 3 day periods of nitrate limited growth: one during growth under
low irradiance (LL, left) and one during growth under high irradiance (HL, right). v3
results are shown as solid lines, FRSv 1.6 results as dotted lines, and data are low pass
filtered at 30 minutes to better display long term trends.
FRSvI.6 internally scales and Fm by a factor of= 20 and so actual values in
Figure 4.4a and c appear larger in magnitude than the unscaled values reported by the v3 software.
F0
E;I]
v3 custom software
FRSV1 .6 (beta) corrimencal software
0.22
f4
0.21
O
0.2
III H
0.19
0.18
fl
10
20
30
40
0
10
20
30
40
400
0
10
20
30
40
20
30
percent added noEse
40
0
0.42
0.41
E
0.4
0.39
A '
650
800
600
700
rI
550
:j
0)
b-500
450
0.8
H
0.2
10
percent added noise
Figure 4.4 Best fit estimates of F0, Fm, sii and p using simulated, noise-corrupted
data. FRSv1.6 beta software results are shown (left) and v3 (right). Six different
simulated data sets are shown at each noise level corresponding to a range of p
between 0 and 0.5 (horizontal axis), staggered to the right for clearer presentation.
Error bars indicate one s.d. for 100 random noise replicates at each p at each noise
corruption level.
Low irradiance growth
High irradiance growth
500
100
400
300
A
255
A A A
256
257
258
/.
.......
.:
1.
200
100
0
/
B
265
..........................
i
/
266
\i
267
268
0.6
::
255
' H
256
..\
257
258
::"
265
266
267
268
600
500
=400
.......
0.5
E05
500
/
400
300
f
300
J
200
255
256
257
258
200
265
266
267
268
255
256
257
258
265
266
267
268
256
257
258
265
266
267
day of year
268
0.z
255
day of year
Figure 4.5 Results from processing variable fluorescence time series data using the
commercial FRSv1.6 (dotted lines) and the custom v3 software (solid lines). Data on
the left come from a short time series of cells grown under low irradiance conditions;
the right column shows a similar time series under high irradiance, seen in the
irradiance time series in the first row. The remaining panels show FvJFm (c&d), sii
(e&f), p (g&h), and the difference between p estimates (i&j). Physiological
parameters have been low pass filtered at 30 minutes.
Substantial differences are consistently observed between estimates around solar noon,
primarily in
sii
and p. Compared to the v3 results, FRSv1 .6 underestimates Fv/Fm
during LL and overestimates Fv/Fm during HL (Figure 4.5c&d). FRSv1.6
overestimates cNpsii during LL but underestimates cYpsij during HL (Figure 4.5e&f).
FRSv1 .6 in general overestimated p but the difference between estimates of p shows a
clear diel pattern emphasized around solar noon, which reverses trend after the shift
from LL to HL (Figure 4.5i&j). The light saturation parameter EK during the same
three day periods was calculated using both FRSv1.6 and v3 estimates of
sii and t
(Falkowski and Raven 1997), and these time series of EK are shown in Figure 4.6
along with the measured culture irradiance. Under LL conditions (left panels) neither
software package indicates light saturated photosynthesis but under HI.
conditions (right panels) light saturation occurs in a period around solar noon.
4.4.4 Assimilation of Time Series Results
The physiological time series in Figure 4.5 were fed into the numerical
simulation. Figure 4.7 shows the resulting relationships between Pe vs.
EPAR
for a
typical day under LL conditions (left panels) and under HL conditions (right panels,
two different irradiance axes). Under LL conditions, v3 estimates lead to a roughly
linear relationship between Pe and irradiance (Figure 4.7a), suggesting that gross
photosynthesis was predominantly light limited under these conditions. This result is
consonant with the p-independent assessment of EK in Figure 4.7a. In contrast, results
derived from FRSv1.6 parameter estimates lead to
pe
vs.
relationship which is
linear only in the range of 0 to 30 pmol photons m2 s' and which exhibits a reduction
in electron flow at higher irradiances (Figure 4.7d, dots). Interpreting these results as
indicative of light saturated photosynthesis is inconsistent with the independent
calculation of EK derived from FRSv1 .6 estimated cpsii and t.
Low irradiance conditions
High irradiance conditions
600
600
A
500
400
300
200
100
OL_
U.
255
>
256
257
258
265
600
600
500
500
400
400
300
300
200
200
100
100
266
267
268
267
day of year
268
Cl)
(I
255
256
257
day of year
258
0
265
266
Figure 4.6 EK and EPAR from LL conditions (left) and HL conditions (right) for the T.
weissflogii culture data. Top panels result from using v3 estimates of 0psii and t and
bottom panels from FRSv1.6. Light saturated photosynthesis is apparent when EK <
EPAR.
Pe calculated using v3 physiological data under HL growth conditions shows a
substantial decrease in Pe at 1OO tmol photons m2 s1 (Figure 4.7b&c, dots) which is
coincident with the irradiance threshold predicted solely by psii and 'r (Figure 4.6b).
PS11 electron flow rates calculated using FRSv1.6 results do not change substantially
with the fivefold increase in overall daily irradiance dosage and intensity (Figure
4.7e&f), which is inconsistent with the independently predicted EK in Figure 4.6d.
Since the estimation of 'r was not examined in this research, differences
between FRSv1.6 and v3 estimates oft might bias these simulation results. The
E1
FRSv1.6 data were reprocessed through the simulation using estimates of
t
from v3
instead and in this way differences between simulation results can only be due to
variability in
Fv/Fm,
sii
and p. This approach did not resolve the inconsistencies
between FRSv1 .6-based estimates of Pe and EK under either LL or HL conditions
(Figure 4.7d-f, star symbols), and so differences between v3 and FRSv1.6 simulation
results cannot be reconciled on the basis of t alone.
For an equal number of incident photons, predicted electron flow using
FRSv1.6 parameter estimates was only
60% of that predicted using v3 estimates.
Neglecting connectivity altogether underestimated PSH electron flow during LL and
HL by approximately 10% (Figure 4.7a-c, diamonds), with the greatest influence
apparent at low diurnal irradiance levels. This response is similar to that observed in
the theoretical simulations around 50 to 100
4.5
t
mol photons m2 s (Figure 4.3).
Discussion
4.5.1
Connectivity and the Analysis of Variable Fluorescence
Our analysis shows that connectivity, as defined in the physiological model of
Kolber et al. (1998), is not a trivial parameter to retrieve from FRRF variable
fluorescence data. The inclusion of p in the physiological model introduces numerical
artifacts which can lead to poor estimation and inappropriately assigned variability in
both p and cYpsil. This problem is exacerbated when inadequate numerical methods are
employed to analyze FRRF variable fluorescence data. In light of this, results from
field and laboratory experiments where FRRF methods identify large variability in
cYpsil but where estimates of p are not presented should be considered with caution.
v3:255
v3:265
v3: 265
ou'-J
B
A
600
600
600
a) 400
400
400
Cl)
a-
..
200
200
200
use p (dot)
p = 0 (dIamond)
IC 0
(1
'0
FRSv1 .6: 255
0
IC
0
0
200
400
FRSv1 .6: 265
FRSv1 .6: 265
800"
600
600
600
a 400
400
400
200
200
(I)
a-
200
0
50
EAR
100
0
50
EPAR
100
0
200
EPAR
40C
Figure 4.7 pe vs. EPAR curves from PS11 electron flow models. These results represent
physiological data in Figure 4.5 assimilated into the PS11 electron flow model for v3
(top) and FRSv 1.6 (bottom) estimates. pe vs. EPAR is shown for LL conditions (left)
and under HL conditions (middle & right). Stars represent FRSv1.6 assimilation using
pe
t from v3. Diamonds in the top row represent the
calculated by neglecting p.
It is important to clearly specify how these parameters are treated in the
analysis of variable fluorescence data when reporting variability in one or the other.
Unless p is explicitly being examined, it may be advisable to use a physiological
model where PSU are not connected, i.e. a "separate units" model where p is
effectively zero. If a connected units model must be used, robust estimation methods
are essential. The present commercial software, and presumably all other
noncommercial software based on a similar computational approach, encounters
substantial difficulty in properly retrieving all parameters, especially
psii
and p.
Measurement noise in the data further degrades the performance of the commercial
software in an unpredictable fashion.
Neglecting p when calculating PS11 electron flow appears to underestimate Pe
predominantly in an irradiance range around the light saturation intensity EK. The
magnitude of this effect can be more than a factor of two, but the practical significance
of this error will depend on the error in other terms used to calculate Pe. The
numerical simulations demonstrate enhancement in pe and so claims that PS11
connectivity has a negligible effect on photosynthetic quantum yield appear to be
incorrect.
Pe
calculated using actual physiology estimated from FRRF data is directly
affected by the accuracy of the parameter estimates, and poorly estimated parameters
lead to substantial inconsistencies in the relationship between irradiance and gross
photosynthesis.
4.5.2 Does p Truly Reflect PSII Connectivity?
The observed diel changes in p reflect physiological variability related to
phytoplankton photosynthesis and presumably are not numerical artifacts. However, it
is unclear that the apparent variability in p actually reflects changes in PS11
connectivity. Because of the interdependence of p and
sii , it is possible that
variability in p may in fact reflect changes in photosynthetic physiology not directly
related to connectivity per
Se.
The physiological model presented by Kolber et al.
(1998) is a modification to the cumulative one-hit Poisson function, which requires
that all PS11 targets have equal opportunity to service incoming photons. When PS11
are densely packaged and significantly self-shaded, individual PS11 do not have an
equal probability to absorb photons from the ambient environment, and a fundamental
assumption for using this model is violated. This violation can be propagated through
the data analysis process and instead of reflecting changes in connectivity, apparent
variability in p and cYpsii might reflect pigment packaging effects. Modification of the
current physiological model is required to better describe observed variable
fluorescence when PSU are densely packed. Until such expanded models are
available, apparent variability in p may in some cases reflect packaging effects and not
simply connectivity.
4.6 Conclusions
Increases in RCII connectivity can enhance PS11 electron flow rates under
certain physiological and environmental conditions, specifically when photochemical
competency is low or when PS11 are densely packed. Experimentally derived
estimates of p in phytoplankton cultures exhibit substantial variability over time scales
from hours to days, which in some instances may actually reflect enhancement of
photochemical efficiency by adjustment to PS11 connectivity. Other apparent
variability may be simply reflect connectivity-related reactions to environmental
changes or may be artifacts due to the application of inappropriate physiological
models of PS11 fluorescence.
Estimation of connectivity using FRRF methodology is hampered by the
difficulty of numerically retrieving p from variable fluorescence data. Examination of
the FRSv1.6 FRRF analysis software reveals performance and accuracy problems
likely related to the choice of parameter fitting technique. These problems may also
be partly due to the numerical behavior of the current physiological model for PS11
variable fluorescence, which exhibits an interdependent behavior between p and opsii.
Recently developed robust methods demonstrate considerably better convergence on
simulated variable fluorescence data and appear to better recover PSII physiological
parameters from actual variable fluorescence measurements.
Diurnal variability in reaction center connectivity may play a significant role in
maintaining optimal charge separation rates, but at present there is not enough
conclusive evidence to support this hypothesis. Better determination of the role and
variability of connectivity in marine environments requires robust numerical retrieval
methods and improvement in the theoretical model for PS11 variable fluorescence.
Greater sophistication in instrument calibration and signal clarity may be also
required, but this might be difficult using the present generation of commercially
available instruments. Connectivity between reaction centers is an interesting and
important aspect of photosynthesis and appears to be influenced by nitrate limitation.
Quantitative models of phytoplankton photosynthesis based on FRRF variable
fluorescence data cannot be trusted, however, until both the instrumentation and the
numerical approach used to estimate physiological parameters are improved.
Acknowledgements. We gratefully acknowledge Russ Desiderio for his
detailed analysis of the PS11 physiological model and Emmanuel Boss for advice
regarding optimization. We wish to thank John Atkins at Chelsea Instruments Ltd. for
critical comments on this manuscript and Chelsea Instruments Ltd. for encouraging
the continued development and improvement of FRRF analysis software. This
research was supported by the National Aeronautics and Space Administration
(NAS5-3 1360).
4.7 Appendices
4.7.1 Linear Approximation of Equation 4.2
p can lie on the interval [0,1) and so for small values of p, Equation 4.2 can be
linearly approximated as
c
i
(i
Xi + c1p)=
,
i
(i
C1 + C1p
C1p)
4.3
which is reasonably valid for experimentally observed values of p. The
interdependence of cYpsii and p is apparent to the first and second order of C, but since
p and C both range from 0 to 1, the contribution of the second order term is
comparatively minor. The interdependence of
ps
and p influences C by an order of
magnitude less than cps11 alone and so for p close to zero, terms containing both p and
Ypsii
contribute about ten percent to C.
4.7.2 Corrections to Kolber et al. (1998)
While reviewing Kolber et al. (1998) we discovered inconsistencies in
discussion and derivation of QA reoxidation kinetics. Specifically, the correction for
the n-exponential QA reoxidation kinetic used in the iterative calculation of C is
incorrect (Kolber et al. 1998, Equation 9). Although the published equation produces
terms of A,k correct in magnitude, dimensional analysis demonstrates that Equation 9
is inconsistent in units. Substituting the published equation into the prior as instructed
will further illustrate the inconsistencies in Equation 9. The adjustment to C1 due to
reoxidation kinetics of multiple factors operating in series is more appropriately
expressed as
Ak ak expi
I.
4.4
Tk )
The subscripting forA as a function of flash number (i.e. subscripts of n) should be
removed as well to make these two equations consistent, as the kinetics of QA
reoxidation are not a function of flashlet in the general case (Kolber et al. 1998,
Equation 8).
91
4.8 References:
Behrenfeld, M.J., 0. Prasil, Z.K. Kolber, M. Babin, and P.G. Falkowski, 1998:
Compensatory changes in Photosystem II electron turnover rates protect
photosynthesis from photoinhibition. Photosynthesis Res. 58, 259-268.
Behrenfeld, M.J. and Z.S. Kolber, 1999: Widespread Iron Limitation of
Phytoplankton in the South Pacific Ocean. Science 283, 840-843.
Bernhardt, K. and H.-W. Trissl, 1999: Theories for kinetics and yields of fluorescence
and photochemistry: how, if at all, can different models of antenna
organization be distinguished experimentally? Biochim. Biophys. Acta 1409,
125- 142.
Coleman, T., M.A. Branch and A. Grace, 1999: Optimization Toolbox Users' Guide,
Version 2. The Math Works, Inc., Natick MA, 305 pp.
Coleman, T.F. and Y. Li, 1994: On the convergence of interior-reflective Newton
methods for nonlinear minimization subject to bounds. Mathematical
Programming 67, 189-224.
Coleman, T.F. and Y. Li, 1996: An interior trust region approach for nonlinear
minimization subject to bounds. SIAM J. Optimization 6, 418-445.
Falkowski, P.G. and J.A. Raven, 1996: Aquatic Photosynthesis (Blackwell Science:
Oxford.), 375 pp.
Gorbunov, M.Y., Z.S. Kolber and P.G. Falkowski, 1999: Measuring photosynthetic
parameters in individual algal cells by Fast Repetition Rate fluorometery.
Photosynthesis Res. 62, 141-153.
Joliot, P. and A. Joliot, 1964: Etudes cinétique de la reaction photochimique libératn
l'oxygene au cours de Ia photosynthese. C. R. Acad. Sci. Paris 258, 46224625.
Kolber, Z.K., 0. Prasil and P.G. Falkowski, 1998: Measurements of variable
chlorophyll fluorescence using fast repetition rate techniques: defining
methodology and experimental protocols. Biochim. Biophys. Acta 1367, 88106.
Ley, A.C. and D.C. Mauzerall, 1986: The extent of energy transfer among
Photosystem II reaction centers in Chiorella. Biochim. Biophys. Acta 850,
234-248.
92
Strutton, P.G., J.G. Mitchell, J.S. Parsiow and R.M. Greene, 1997: Phytoplankton
patchiness: quantifying the biological contribution using Fast Repetition Rate
Fluorometry. Journal of Plankton Res. 19, 1265-1274.
Vassiliev, I.R., Z.S. Kolber, D. Mauzerall, V.K. Shukia, K. Wyman and P.G.
Falkowski, 1995: Effects of iron limitation on photosystem II composition and
energy trapping in Dunaliella tertiolecta. Plant Physiol. 109, 963-972.
93
5. SUMMARY
The manner in which natural fluorescence reflects phytoplankton adaptation to
environmental change is not random, but neither is it simple. Proper interpretation of
observed variability in natural fluorescence depends of course on an empirical and
mechanistic understanding of the physiological relationships between environmental
factors and Fnat. More importantly, however, proper interpretation depends on a solid
understanding of the ecosystem under study and the comparative dominance of
environmental variables. The experiments performed during the course of this
research show that significant correlation can be found between variability in natural
fluorescence and certain physiological and photosynthetic properties, but a typical
marine environment might reasonably contain several of these factors varying in
concert over a wide range of scales, complicating interpretation of the natural
fluorescence signal. We have only begun to mine the "rich and complex" variability
in natural fluorescence for physiological and ecological information, and experiments
like the ones presented in this thesis are a first and essential step toward applying
physiological variability in natural fluorescence to ecological questions in marine
ecosystems.
5.1 Practical Importance of this Research
The natural fluorescence measurements obtained with the NEC and presented
in this thesis are novel because they are, to the best of our knowledge, the first such
published measurements. Although preliminary data were collected over ten years
ago using prototype instrumentation (D.A. Kiefer, unpublished data), those data,
although promising, were inconclusive and went largely unexamined. The present
research reflects a high degree of instrument characterization and substantial care in
the measurement of natural fluorescence. The attention paid to the details of
measurement allows us to unequivocally assign apparent changes in natural
fluorescence to changes in phytoplankton physiology. To arrive at the point where
meaningful experiments could be performed, the prototype instrument had to be
redesigned and fabricated and its performance evaluated. This effort, documented in
the first research section of this thesis, has produced a workable system and
methodology which provides for the much needed laboratory investigation of
phytoplankton natural fluorescence.
During the course of this research, natural fluorescence data were collected
over more than 150 days with very high temporal resolution (Jamp
= 1
1)
These
measurements were collected on phytoplankton cultures experiencing a wide range of
physical and chemical conditions. The data set which formed the basis of the second
research section of this thesis is of sufficient quality to support detailed examination of
how environmental variability influences variability in natural fluorescence at the
physiological level. Because this data set is comprehensive and extensive, it serves as
an exploratory resource to direct future studies into more detailed aspects of natural
fluorescence.
Substantial emphasis was placed on researching and developing improved
software for analyzing FRRF variable fluorescence data. The FRRF is a valuable tool
for rapidly and noninvasively determining the photosynthetic properties of
phytoplankton, but the analysis tools provided with this instrument are not robust
enough to extract the physiological information we needed from raw variable
fluorescence data. We first identified unusual kinetics in the time series of
connectivity and functional cross section, and later investigation showed that the
problems with the current method of analysis are both theoretical and numerical. The
analysis tools developed during the course of this research will be made available on
the Internet within the next year in order to a) provide an improved analysis package
for those users limited at present to the commercial software and to b) provide
interested researchers with a valuable tool to explore the nuances in variable
fluorescence data.
95
5.2 Implications of the Observed Relationships in Fnat
We examined a very specific subset of environmental change in these
experiments, that of marine diatoms limited by nitrate exposed to different irradiance
histories. From these experiments, we can draw several firm conclusions given the
deliberate changes in the physical and chemical environment and the natural
fluorescence response observed. First, the diurnal shape of .t holds information about
the relative availability of nitrate in a qualitative way. If forenoon and afternoon
measurements of
are spaced far enough apart, it appears that we can infer the way
in which diatoms "perceive" the availability of nitrate in the environment. Second,
variability in
Ct'
is not always a function primarily of chlorophyll concentration. We
observed in these experiments that chlorophyll concentration was not always wellcorrelated with c1, and that the decoupling between these properties could not be
ascribed solely to the influence of high inadiance. Third, the proxies needed to
monitor C1 over the scale of days to weeks are easily derived from remote sensing
data. Only EPAR is are needed to construct C1; remote sensing products of irradiance
are contained in the latest version of SEADAS and presumably similar products will
be soon available from satellite platforms with onboard natural fluorescence sensors.
Fourth, and perhaps most importantly, there appears to be a relationship between gross
features in the natural fluorescence signal and the inherent photosynthetic capacity of
cells in these cultures. EK changes continuously throughout the day due to
physiological responses, but the irradiance at which
EPAR
surpasses EK appears to be
correlated with the inadiance at which 1 develops its midday depression. Although
the physiological explanation for this correlation has not yet been established, the
utility of this empirical result has significant potential impact with respect to global
productivity estimates.
Some degree of prior research has been performed on natural fluorescence in
the field, and the results presented here do not conflict with prior findings in any
substantive way. If anything, the current findings underscore the complexity of
natural fluorescence variability and provide explanations for the sources of variability
in field measurements of Fnat. Chlorophyll biomass and the ratio of natural
fluorescence to irradiance () did correlate well under low irradiance growth
conditions (see Kiefer et al. 1989) but there were periods where it was apparent that
the irradiance history, not the instantaneous and immediate irradiance, controlled the
physiological variability in 1 under high irradiance. Further, although qualitative
changes in c1 were correlated with the relative availability of a growth limiting
nutrient, the magnitude of J changed opposite to that observed in certain field studies
when phytoplankton are released from nutrient limitation (e.g. Letelier et al 1997). In
the field study, increased supply of (presumably) iron led to a decrease in a parameter
similar to ct, whereas increases in nitrate to these diatom cultures increased cJ
Although there is substantial theoretical speculation on how environmental factors
control natural fluorescence, laboratory experiments like these are an important
complement to field studies.
5.3 Natural Fluorescence and Photosynthetic Rate
Much has been made about the potential use of natural fluorescence to directly
measure primary production, and it is hypothesized that instantaneous rates of primary
production can be directly estimated from natural fluorescence data'2. Interpretation
of field data appear to support this presumed relationship. The connection between
natural fluorescence and carbon incorporation evolves from identities relating the rates
of each with the rate at which irradiance is absorbed by a cell (e.g. Kiefer et al. 1989;
12
Manuals supplied with certain commercial in situ "natural fluorometers" provide equations to
calculate carbon incorporation rates directly from natural fluorescence radiance.
97
Kiefer and Reynolds 1992; Chamberlin and Marra 1992). These identities identify the
absolute quantum yield of each process
and I such that
Ff= cI.Fa
5.1
Fc=Fa
5.2
and
where Fa represents the rate of photon absorption by phytoplankton, Fc represents the
rate of carbon assimilation, and Ff is the rate of fluorescence emission. Combining
these equations yields
Fc= t.t'j.Ff.
5.3
This apparently simple algebraic relationship has several important caveats pertinent
to the research described in this thesis:
First, the two quantum yields are empirically defined and do not have
physiological meaning. Given what is and has been known about the structure and
function of PS11, it is clear that these two quantum yields are not independent. Since a
quantum yield is essentially a probability, both ct and Ct can be represented as the
combined probability of any series of events through which an absorbed photon results
in fluorescence and fixed carbon, respectively. cl1can be rewritten as the probability
that an exciton will not dissipate while migrating through the antenna and the
probability that once at the reaction center, it will not be fluoresced, i.e.,
ctf
(1
tdissipate) . (1
tseparate).
5.4
If these physiologically-based yields determine the fate of an absorbed photon
reaching a reaction center, then clearly
cI
comprises both
dissipate and
separate,
albeit
in different algebraic combinations, in addition to a host of other yields describing the
remainder of the electron transport chain and the carbon fixation pathway. In this
research we applied variable fluorescence techniques to look at changes in PS11
physiology which control J?f and cIi, and using such an approach we observed a
complex pattern of physiological responses which differed in decay scales, irradiance
thresholds and potential magnitude of control. Even if cI and J? were fully
independent, or if the common probabilities cancel in Equation 5.3, continuing to
express the relationship between fluorescence and photosynthesis in quantum yields
established by definition precludes a mechanistic understanding. Ultimately,
physiological description of natural fluorescence will be the only way to separate the
complicated underlying physiological factors which control
ct
and
Second, Equation 5.3 does not and cannot guarantee that Fc calculated from Ff
will be accurate on temporal scales identical to that at which natural fluorescence is
measured. Dark reactions during the night represent the trivial case. Some carbon is
fixed in the absence of light although Equation 5.3 predicts no fixation. In the
nontrivial daytime case, physiological hysteresis will increase decoupling between
fluorescence and carbon fixation because changes in photosynthetic physiology which
influence natural fluorescence do not occur on the same scales or thresholds as those
which directly influence gross and/or net photosynthesis. The physical and temporal
separation of the light and dark reactions have been known for a very long time, and
these degrees of separation dictate a decoupling between the light-driven production of
ATP and NADP+ and the light-independent use of these products in the fixation of
carbon. No degree of oversampling natural fluorescence will improve estimates of
primary production if the two processes are poorly coupled. The "physiological"
scales of the two processes need to be matched in order to predict one from the other,
which is a much more difficult proposition than scale matching in the traditional
temporal or spatial sense.
5.4 Implications of Connectivity
The connectivity of reaction centers was examined during the course of this
research as a side issue, motivated by a desire to obtain more accurate data from FRRF
measurements. We observed that connectivity exhibited measurable diel variability
during the course of these experiments, and so numerical experiments were performed
to determine the theoretical influence of such variability on gross photosynthesis.
Two environmental conditions were identified where increased connectivity led to
enhanced photosynthesis, and both of these conditions are especially important in the
first optical depth from which satellite sensors will obtain measurements of
At
present, the theoretical models of PSH variable fluorescence are not robust enough to
determine if connectivity per
se
is varying in these measurements or if the observed
variability is due to an associated property numerically propagated through the fitting
routine. Determining the true variability in connectivity is an issue which plant
physiologists need to resolve before oceanographers can examine this property in the
field. Even when a perfect model including the effects of connectivity is available,
measurement noise may limit the extent to which variability in "connectivity" can be
explored given the currently available variable fluorescence instrumentation. Both
measurement and interpretation methods must be advanced before the influence of
variable connectivity on photosynthesis can be well explored in field populations of
marine phytoplankton.
5.5 Future Work
Because the instrumentation and protocols now exist to measure natural
fluorescence reliably in controlled laboratory cultures, a wide range of experimental
investigations can be performed. Such investigations can go into great detail about
specific aspects of the natural fluorescence response, but certain avenues of
100
investigation are particularly important. First, for the purposes of satellite remote
sensing, investigation of the natural fluorescence of picoplankton is warranted.
Although the experiments detailed in this thesis examined in detail the degree of
coupling between fluorescence, light and nitrate availability, in much of the ocean
nitrate availability does not vary much, and consequently ecological dynamics in such
regions involve primarily K-selected species, instead of the r-selected specie used in
this study. Macronutnent limitation is easily accomplished in the present
experimental apparatus, and the optical excitation system is sufficient for such an
investigation. Picoplankton employ different photosynthetic and photoprotective
structure and mechanisms which may lead to substantial differences between the
results observed in our work and from those observed in such an experiment. The
ecological consequences of variability in light and nitrate observed in this thesis may
not, and probably will not, be reflected in studies using picoplankton as a model
species.
Second, nitrate is not the ultimate limiting nutrient in all of the ocean. In
various different ecosystems iron, phosphorous or silicate have all been shown to exert
substantial control on phytoplankton physiology and ecology. Where this is the case,
the conclusions of this study based on nitrate limitation may not apply. Although
nitrate as an element is essential for light harvesting and processing, these other
elements are used in different parts of the photosynthetic process, and should produce
different natural fluorescence responses when light levels change. Inspection of the
influence of these other nutrients is an important avenue of future research to
determine the applicability of natural fluorescence monitoring to basin scale
investigation of primary production.
Third, closure is needed among the three potential fates of absorbed light
energy. Although we measured '1 and ctp over the course of this research, we did not
measure
cID,
the quantum yield of thermal dissipation in the light harvesting
complexes. In oceanographic science, photosynthesis is usually represented as a
101
function of ambient irradiance, not absorbed irradiance, because traditionally the latter
has been difficult to measure in situ. However, instruments are now available to
measure phytoplankton absorption in situ at spatial and temporal scales required to
monitor the types of physiological-based variability in absorption which are known to
occur in phytoplankton. Inclusion of the physiological mechanisms which control cI
is an essential part closing the fate of light which is absorbed by phytoplankton.
102
BIBLIOGRAPHY
Abbott, M.R., and R.M. Letelier, 1997a: Going with the Flow The Use of Optical
Drifters to Study Phytoplankton Dynamics. Monitoring Algal Blooms: New
Techniques for Detecting Large-Scale Environmental Changes. M. Kahru and
C.W. Brown, Eds., Landes Bioscience, pp. 143-168.
Abbott, M.R., and R.M. Letelier, 1997b: Bio-optical drifters Scales of variability of
chlorophyll and fluorescence. Society of Photo-Optical Instrumentation
Engineers 2963, 216-221.
Abbott, M.R., and R.M. Letelier, 1998: Decorrelation scales of chlorophyll as
observed from bio-optical drifters in the California Current. Deep-Sea
Research 1145, 1639-1667.
Bartlett, J.S., K.J. Voss, S. Sathyendranath, and A. Vodacek, 1998: Raman scattering
by pure water and seawater. Applied Optics 37, 3324-3332.
Behrenfeld, M.J., 0. Prasil, Z.K. Kolber, M. Babin, and P.G. Falkowski, 1998:
Compensatory changes in Photosystem II electron turnover rates protect
photosynthesis from photoinhibition. Photosynthesis Res. 58, 259-268.
Behrenfeld, M.J. and Z.S. Kolber, 1999: Widespread Iron Limitation of
Phytoplankton in the South Pacific Ocean. Science 283, 840-843.
Bernhardt, K. and H.-W. Trissi, 1999: Theories for kinetics and yields of fluorescence
and photochemistry: how, if at all, can different models of antenna
organization be distinguished experimentally? Biochim. Biophys. Acta 1409,
125- 142.
Chamberlin, W.S., C.R. Booth, D.A. Kiefer, J.H. Morrow, and R.C. Murphy, 1990:
Evidence for a simple relationship between natural fluorescence,
photosynthesis and chlorophyll in the sea. Deep-Sea Research 37, 95 1-973.
Chamberlin, S., and J. Marra, 1992: Estimation of photosynthetic rates from
measurements of natural fluorescence: analysis of the effects of light and
temperature. Deep-Sea Research 39, 1695-1706.
Coleman, T., M.A. Branch and A. Grace, 1999: Optimization Toolbox Users' Guide,
Version 2. The MathWorks, Inc., Natick MA, 305 pp.
103
Coleman, T.F. and Y. Li, 199: On the convergence of interior-reflective Newton
methods for nonlinear minimization subject to bounds. Mathematical
Programming 67, 189-224.
Coleman, T.F. and Y. Li, 1996: An interior trust region approach for nonlinear
minimization subject to bounds. SIAMJ. Optimization 6,418-445.
Cullen, J.J., A.M. Ciotti and R.F. Davis, 1997. "The relationship between nearsurface chlorophyll and solar-stimulated fluorescence: biological effects" in
Ocean Optics XIII, Steven G. Ackleson, Robert Frouin, Editors, Proc. SPIE
2963, 272-277.
Cullen, J.J., and M.R. Lewis, 1995: Biological processes and optical measurements
near the sea surface: Some issues relevant to remote sensing. J. Geophys. Res.
100, 13,255-13,266.
Doerffer, R., 1993: Estimation of primary production by observation of solarstimulated fluorescence. ICES Mar. Sci. Symp. 197, 104-113.
Eppley, R.W., R.W. Holmes, and J.D.H. Strickland, 1967: Sinking rates of marine
phytoplankton measured with a fluorometer. J. EXP. MAR. BIOL. ECOL. 1,
191-208.
Falkowski, P.G. and J.A. Raven, 1996: Aquatic Photosynthesis (Blackwell Science:
Oxford.), 375 pp.
Gorbunov, M.Y., Z.S. Kolber and P.G. Falkowski, 1999: Measuring photosynthetic
parameters in individual algal cells by Fast Repetition Rate fluorometery.
Photosynthesis Res. 62, 14 1-153.
Gordon, H.R., 1979: Diffuse reflectance of the ocean: the theory of its augmentation
by chlorophyll a fluorescence at 685 nm. Appi. Optics 18, 1161-1166.
Gordon, H.R., and A. Morel, 1983: Remote Assessment of Ocean Color for
Interpretation of Satellite Visible Imagery, a Review, Lecture Notes on
Coastal and Estuarine Studies, Volume 4, Springer Verlag.
Govindjee, 1995: Sixty-three years Since Kautsky: Chlorophyll a Fluroescence. Aust.
J. Plant Physiol. 22, 131-160.
Gower, J.F.R, and 0. Borstad, 1981: Use of the in-vivo fluorescence line at 685 nm
for remote sensing surveys of surface chlorophyll a. In 'Oceanography from
Space'. (ed. J.F.R. Gower) pp. 329-338. (Plenum Press: New York.)
104
Han, B.-P., M. Virtanen, J. Koponen, and M. Straskraba, 2000: Effect of
photoinhibition on algal photosynthesis: a dynamic model. J. Plankton Res.
22, 865-885.
Hu, C., and K.J. Voss, 1998: Measurement of solar-stimulated fluorescence in natural
waters. Limnology and Oceanography 43, 1198-1206.
Joliot, P. and A. Joliot, 1964: Etudes cinétique de la reaction photochimique libératn
l'oxygene au cours de la photosynthese. C. R. Acad. Sci. Paris 258, 46224625.
Kiefer, D.A., 1973: Fluorescence properties of natural phytoplankton populations.
Marine Biology 22, 263-269.
Kiefer, D.A., W.S. Chamberlin and C.R. Booth, 1989: Natural fluorescence of
chlorophyll a: relationship to photosynthesis and chlorophyll concentration in
the western South Pacific gyre. Limnology and Oceanography 34, 868-881.
Kiefer, D.A., and R.A. Reynolds, 1992: Advances in understanding phytoplankton
fluorescence and photosynthesis. Primary Productivity and Biogeochemical
Cycles in the Sea. P.G. Falkowski and A.D. Woodhead, Eds., Plenum Press,
pp. 155-174.
Kolber, Z.K., 0. Prasil and P.G. Falkowski, 1998: Measurements of variable
chlorophyll fluorescence using fast repetition rate techniques: defining
methodology and experimental protocols. Biochim. Biophys. Acta 1367, 88106.
Kubitschek, H. E., 1970: Introduction to Research with Continuous Cultures.
Prentice-Hall Inc., Englewood Cliffs, N.J.
Letelier, R.M., and M.A. Abbott, 1996: An Analysis of Chlorophyll Fluorescence
Algorithms for the Moderate Resolution Imaging Spectrometer (MODIS).
Remote Sens. Environ. 58, 215-223.
Letelier, R.M., M.R. Abbott, and D.M. Karl, 1996: Chlorophyll natural fluorescence
response to upwelling events in the Southern Ocean. Geophys. Res. Lets., 24,
409-412.
Ley, A.C. and D.C. Mauzerall, 1986: The extent of energy transfer among
Photosystem II reaction centers in Chlorella. Biochim. Biophys. Acta 850,
234-248.
Lorenzen, C.J., 1966: A method for the continuous measurement of in vivo
chlorophyll concentration. Deep-Sea Research 13, 223-227.
105
Mobley, C.D., 1994: Light and Water, Radiative Transfer in Natural Waters,
Academic Press, 594 pp.
Neville, R.A., and J.F.R. Gower, 1977: Passive Remote Sensing of Phytoplankton via
Chlorophyll a Fluorescence. Journal of Geophysical Research 82, 3487-3493.
Stegmann, P.M., M.R. Lewis, C.O. Davis and J.J. Cullen, 1992: Primary production
estimates from recordings of solar-stimulated fluorescence in the Equatorial
Pacific at 150 degrees West. Journal of Geophysical Research 97C, 627 638.
Strutton, P.G., J.G. Mitchell, J.S. Parsiow and R.M. Greene, 1997: Phytoplankton
patchiness: quantifying the biological contribution using Fast Repetition Rate
Fluorometry. Journal of Plankton Res. 19, 1265-1274.
Vassiliev, I.R., Z.S. Kolber, D. Mauzerall, V.K. Shukla, K. Wyman and P.G.
Falkowski, 1995: Effects of iron limitation on photosystem II composition and
energy trapping in Dunaliella tertiolecta. Plant Physiol. 109, 963-972.
Wright, S.W. and S.W. Jeffrey, 1997: High-resolution HPLC system for chlorophylls
and carotenoids of marine phytoplankton. In: Phytoplankton pigments in
oceanography: Guidelines to modern methods. Jeffrey, S.W., Mantoura,
R.F.C., Wright, S.W. (Eds.), UNESCO Publishing, Paris, pp. 327-341.
Yentsch, C.S. and Menzel, D.W. (1963). A method for the determination of
phytoplankton, chlorophyll, and phaeophytin by fluorescence. Deep-Sea Res.
10, 221-231.
Download