Click
Here
for
Full
Article
Phytoplankton photocompensation from space‐based
fluorescence measurements
J. Ruairidh Morrison1 and Deborah S. Goodwin2
Received 25 November 2009; revised 10 February 2010; accepted 17 February 2010; published 25 March 2010.
[1] Recent satellite‐derived observations linked global scale
phytoplankton fluorescence variability with iron stress and
hinted at photophysiological responses associated with
changing light levels. These photocompensation reactions,
the sum of photoacclimation and photoadaptation, were
examined with climatological data for the Gulf of Maine.
Significant seasonal variability was observed in the
fluorescence quantum yield that was unrelated to patterns
of biomass. Up to 89% of the variability in the fluorescence
quantum yield was explained by a physiology‐based
photocompensation model. Spatial variability in seasonal
patterns was associated with differing hydrodynamic
regimes. This variability in the quantum yield demonstrates
that satellite‐based fluorescence is inappropriate for
phytoplankton biomass determinations. More importantly,
the work presented here provides the modeling foundation
for fluorescence‐based investigations of temporal and
spatial variability in phytoplankton physiology associated
with growth irradiance. These space‐based physiological
observations have the potential to decrease uncertainties in
future ocean color derived primary productivity estimates.
Citation: Morrison, J. R., and D. S. Goodwin (2010), Phyto-
plankton photocompensation from space‐based fluorescence mea-
surements, Geophys. Res. Lett., 37, L06603, doi:10.1029/
2009GL041799.
1. Introduction
[2] Phytoplankton alter their photosynthetic apparatus in
order to optimize the balance between capturing light for
photochemistry and potentially damaging excess energy.
Natural phytoplankton populations react to changing irra-
diance with both photoacclimation, a phenotypic response
limited by the plasticity of an individual species’ genome, and
photoadaptation, a range of genotypic photophysiological
states resulting from evolutionary adaptation [Falkowski and
Laroche, 1991]. Ultimately, the interaction between photo-
adaptation and environmental conditions is one determinant
of the phytoplankton assemblage composition [e.g., Cullen
and MacIntyre, 1998; Richardson et al., 1983].
[3] Two basic strategies for photoacclimation are observed
in algal taxa: increasing the absorption cross‐section of
photosystem II (sPSII) or increasing the overall number of
PSII reaction centers [Falkowski and Owens, 1980]. Indi-
vidual species may also vary their maximum carbon fixation
capacity, the amount of chlorophyll in each reaction center,
and the quantity of accessory pigments in the PSII antenna
[Falkowski and Raven, 2007]. Resulting changes in the ratio
of chlorophyll to carbon biomass have been detected from
space [Behrenfeld et al., 2005]; however, differentiating
between photoacclimation and photoadaptation is complex,
especially with remotely sensed data. The term photo-
compensation is used herein to describe the sum of both
phenotypic and genotypic responses.
[4] Phytoplankton in vivo fluorescence is derived almost
entirely from PSII [Falkowski and Kiefer, 1985]. A number
of processes act to alter the fluorescence quantum yield
(ratio of photons fluoresced to photons absorbed; ’f). Pho-
tochemical quenching (qP) occurs at low light levels and
decreases as irradiance increases, with reaction centers
progressively becoming closed to photochemical work
[Kiefer and Reynolds, 1992]. At higher irradiances, a variety
of mechanisms, together known as non‐photochemical
quenching (qN), attempt to regulate exciton energy transfer
to PSII reaction centers in order to match photochemical
utilization and prevent oxidative damage resulting in pho-
toinhibition [Horton et al., 1996; Krause and Weis, 1991].
Rapid (minutes) energy‐dependent quenching (qE) reduces
sPSII by thermal dissipation in the antenna via the xantho-
phyll cycle [Demmig‐Adams, 1990] and regulation of a
chlorophyll‐binding protein [Li et al., 2000]. In addition to
antennae quenching (qE), thermal dissipation of excess
exciton energy through reaction center quenching (qI) is also
invoked, which responds to incident light changes on longer
time scales through the reversible down‐regulation of PSII
reaction centers [Horton et al., 1996; Morrison, 2003].
[5] Sun‐induced chlorophyll fluorescence (SICF) is detect-
able in both in‐water and water‐leaving light fields and with
space‐based sensors [Abbott and Letelier, 1999; Neville and
Gower, 1977]. Two uses for SICF have been suggested: a
biomass indicator, which in turn requires a well‐constrained
yet often ignored knowledge of ’f variability [Letelier and
Abbott, 1996; Neville and Gower, 1977], or a physiological
indicator utilizing ’f variability [Behrenfeld et al., 2009;
Letelier et al., 1997; Morrison, 2003; Schallenberg et al.,
2008]. A global scale analysis of the MODIS fluorescence
signal found that increased ’f correlated with model‐
predicted areas of iron stress, as well as suggested the
importance of photocompensation [Behrenfeld et al., 2009].
Field studies have inferred increases in ’f with macro‐
nutrient stress [Letelier et al., 1997; Schallenberg et al., 2008]
but this limitation has also been postulated to be associated
with decreased yields [Babin et al., 1996].
[6] Morrison [2003] modeled SICF‐derived ’f in terms of
qP and qN with a constant fraction of undamaged reaction
centers at high irradiances. Schallenberg et al. [2008] found
no evidence of this constant fraction and the removal of this
1
Northeastern Regional Association of Coastal Ocean Observing
Systems, Rye, New Hampshire, USA.
2
Ocean Process Analysis Laboratory, University of New Hampshire,
Durham, New Hampshire, USA.
Copyright 2010 by the American Geophysical Union.
0094‐8276/10/2009GL041799
GEOPHYSICAL RESEARCH LETTERS, VOL. 37, L06603, doi:10.1029/2009GL041799, 2010
L06603 1 of 5

term from the original model had little effect on the overall
relationship. The present model of ’f without the constant
fraction is given by

f
¼ q
I
q
E
ð Þ
|fflffl{zfflffl}
q
N

f minAþ f maxð1 AÞ
 
|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
q
P
ð1Þ
where ’fmin and ’fmax are the minimum and maximum
quantum yields, respectively [Kiefer and Reynolds, 1992]
and A is the fraction of open and viable PSII reaction cen-
ters. In this model, both qE and A are parameterized using
target theory and are equal to the exponent of the negative
ratio of the irradiance and a light saturation parameter [see
Morrison, 2003].
[7] Two additional properties related to ’f are Ek and ET,
the light saturation parameters for photochemical and energy‐
dependent quenching, respectively. Ek is a function of
1/(sPSIItPSII), where 1/tPSII is the maximum turnover rate
of PSII under light saturation [Falkowski and Raven,
2007]. However, tPSII as defined above is not a constant
or predictable number and varies with photoacclimational
state [Behrenfeld et al., 2004, 1998]. Similar to Ek, ET can be
thought of as varying with 1/sPSII in the absence of other
physiological changes. If photocompensation is simply
modeled as varying sPSII (and therefore both Ek and ET), the
irradiance‐’f relationship is shifted to higher or lower irra-
diances as suggested by Behrenfeld et al. [2009].
[8] Decreasing values of qI from 1 increases down‐
regulation of PSII. An alternative way to look at qI in this
model is as a description of the relative probability of a
photon absorbed by pigment molecules in either PSII or
PSI being delivered to a PSII reaction center. Photo-
compensatory responses that would increase this proba-
bility, such as increasing the ratio of PSII to PSI units as
occurs under iron stress conditions, can also be modeled
by altering qI. More details of the parameterization of ’f
may be found in the auxiliary material.1
2. Methods
[9] The derivation of ’f from SICF measurements has
been documented many times [e.g., Huot et al., 2005; Kiefer
et al., 1989; Morrison, 2003]. These parameterizations are
all very similar and involve correcting the upwelling radi-
ance from fluorescence at one wavelength for isotropic
emission, the fraction of fluorescence emitted at other
wavelengths, vertical attenuation of excitation and fluo-
resced energies, the absorption of phytoplankton, the loss of
fluorescence light during crossing of the sea‐air interface,
and the magnitude of incident irradiance. Moderate Reso-
lution Imaging Spectroradiometer (MODIS) Aqua Fluores-
cence Line Height (FLH) at 678 nm was calculated from the
normalized water leaving radiances with linear baseline
interpolation between the 667 and 748 nm bands [Abbott
and Letelier, 1999]. The fluorescence radiance emanating
at the sea surface was calculated by dividing the FLH by the
solar constant and multiplying by the irradiance just above
the ocean surface [Behrenfeld et al., 2009]. Above‐water
irradiance was calculated from the MODIS/Aqua instanta-
neous Photosythentically Available Radiation (iPAR) product
using a regionally‐developed relationship from 3806 surface
irradiance measurements at the Great Bay Coastal Buoy in
Great Bay, NH (6.87 × 10−2 × iPAR, r2 = 0.999). A modified
version ofHuot et al.’s [2005] algorithmwas used to calculate
’f with MODIS/Aqua standard products used for iPAR,
chlorophyll‐a (chlor_a, OC3M) [O’Reilly et al., 2000] and
in‐water attenuation at 490 nm (K_490). This algorithm
accounted for fluorescence reabsorption within the cell and
the effects of pigment packaging. Two modifications were
made to Huot et al.’s [2005] algorithm, changing the pro-
portionality factor that parameterized the spectral nature of
emission (29.69, based on Gaussian emission centered at 683
nm with a 10.6 nm standard deviation) and removing the
sensor viewing angle correction, which was unnecessary
when working with normalized water leaving radiances.
[10] Four years (2004 to 2007) of MODIS/Aqua L2 data
were obtained through the SeaDAS processing environment
(version 5.1) with default atmospheric correction settings for
a region surrounding the Gulf of Maine. Individual granules
were reprojected to a standard conic projection encom-
passing 39.5077° to 46.3730°N and 62.5119° to 73.8880°
W. Monthly climatologies were calculated as the median of
individual satellite products. Seasonal climatologies for
winter (Dec–Feb), spring (Mar–May), summer (Jun–Aug),
and fall (Sept–Nov) were obtained by further concatenation
of monthly climatologies.
[11] In situ hydrographic information was obtained for the
same time period from the University of New Hampshire
Coastal Observing Center’s monthlyWilkinson Basin transects
(www.cooa.unh.edu). For transect station WB7 (42.8614°N,
Figure 1. Seasonal ’f climatologies for winter (Dec–Feb),
spring (Mar–May), summer (Jun–Aug), and fall (Sept–
Nov). Climatologies were constructed as the median values
of individual monthly climatologies derived from four years
of MODIS/Aqua observations (2004–2007). Contour lines
represent bathymetry. Locations in the summer plot corre-
spond to those sites used in subsequent regional analyses:
1, Wikinson Basin station (WB7); 2, Buoy M in Jordan
Basin (BuoyM); 3, Georges Bank (GB); and 4, south of Grand
Manan Island (GM).
1Auxiliary materials are available in the HTML. doi:10.1029/
2009GL041799.
MORRISON AND GOODWIN: PHOTOCOMPENSATION FROM SPACE L06603L06603
2 of 5