Emiliania huxleyi population growth rate response to light and temperature: A synthesis

Abstract

The relationship between the maximum specific growth rate (μ, d-1) of the coccolithophore Emiliania huxleyi and photon flux density (PFD, μmol photons m-2s-1) was quantified using a combination of quantile regression and culture experiment data from the literature (n = 1387). This relationship, used in ecosystem models incorporating E. huxleyi or coccolithophores as a functional group, is often assumed to follow a Monod function although values for the model parameters vary greatly. In this analysis, a Monod function was compared with other models to determine the model which best fit E. huxleyi growth rate data. Analysis showed that a Monod model of μ = 1.858 [PFD/(23.91 + PFD)] best described E. huxleyi maximum growth rate as a function of PFD. In addition, an expression combining the Monod function (this study) and the power function relating growth rate to temperature (Fielding 2013; Limnol Oceanogr 58:663-666) was calculated: when both temperature (T, °C) and PFD are known, the resulting expression μ = (0.199 × T0.716) × [PFD/(14.2 + PFD)] predicts maximum E. huxleyi specific growth rate. Current literature models either overestimate or underestimate maximum growth rate by up to 3-fold over a wide range of PFDs. The use of the Monod function and the combined expression presented here is therefore recommended for future models incorporating the growth rate of E. huxleyi when either light or both temperature and light are known.

Full text

AQUATIC MICROBIAL ECOLOGY
Aquat Microb Ecol
Vol. 73: 163–170, 2014
doi: 10.3354/ame01707 Published online October 2
INTRODUCTION
Coccolithophore algae are the major calcium car-
bonate producers in the ocean and are therefore fun-
damental for driving the inorganic carbon pump from
the ocean’s surface waters to the deep sea (Rost &
Riebesell 2004). Coccolithophore calcium carbonate
production also influences alkalinity and dissolved
inorganic carbon concentration in the upper ocean
(Zeebe & Wolf-Gladrow 2001). Additionally, cocco-
lithophores, as members of the haptophyte algae,
produce higher levels of intracellular dimethylsulfo-
niopropionate than most other algal groups (Keller
1988). Therefore, modelling and predicting the growth
of marine coccolithophore algae is key to understand-
ing them as a component of biogeochemical cycles
(e.g. organic carbon, calcium carbonate, di methyl -
sulfide) and aquatic food webs.
In the present-day ocean, Emiliania huxleyi is
the most numerically important species of cocco-
lithophore (Paasche 2001). E. huxleyi has a near
ubiquitous distribution, being found in environ-
ments ranging from estuarine to open ocean, and
from ~81°N (Hegseth & Sundfjord 2008, Sukhanova
et al. 2009) to ~61°S (Findlay & Giraudeau 2000,
Bollmann et al. 2009). Additionally, E. huxleyi
forms large blooms at temperate latitudes (Tyrrell &
Merico 2004).
Recent attempts to determine factors controlling
patterns of E. huxleyi (and other coccolithophore)
distribution and ecology in the present-day ocean
calculate growth rate as a function of major external
bottom-up limiting factors, such as temperature, light
and nutrients. This ap proach first relies on determin-
ing the model variables describing the maximum
growth rate response to individual limiting factors.
© Inter-Research 2014 · www.int-res.com*Corresponding author: s.r.fielding@outlook.com
Emiliania huxleyi population growth rate response
to light and temperature: a synthesis
Samuel R. Fielding*
School of Environmental Sciences, 4 Brownlow Street, University of Liverpool, Liverpool L69 3GP, UK
ABSTRACT: The relationship between the maximum specific growth rate (μ, d−1) of the coccolitho-
phore Emiliania huxleyi and photon flux density (PFD, µmol photons m−2 s−1) was quantified using a
combination of quantile regression and culture experiment data from the literature (n = 1387). This
relationship, used in ecosystem models incorporating E. huxleyi or coccolithophores as a functional
group, is often assumed to follow a Monod function although values for the model parameters vary
greatly. In this analysis, a Monod function was compared with other models to determine the model
which best fit E. huxleyi growth rate data. Analysis showed that a Monod model of μ= 1.858
[PFD/(23.91 + PFD)] best described E. huxleyi maximum growth rate as a function of PFD. In addi-
tion, an expression combining the Monod function (this study) and the power function relating
growth rate to temperature (Fielding 2013; Limnol Oceanogr 58: 663 – 666) was calculated: when
both temperature (T, °C) and PFD are known, the resulting expression μ= (0.199 × T0.716) ×
[PFD/(14.2 + PFD)] predicts maximum E. huxleyi specific growth rate. Current literature models ei-
ther overestimate or underestimate maximum growth rate by up to 3-fold over a wide range of
PFDs. The use of the Monod function and the combined expression presented here is therefore rec-
ommended for future models incorporating the growth rate of E. huxleyi when either light or both
temperature and light are known.
KEY WORDS: Coccolithophore · Photon flux density · PFD · Temperature · Monod function ·
Quantile regression · Calcium carbonate · Literature review
Resale or republication not permitted without written consent of the publisher

Aquat Microb Ecol 73: 163–170, 2014
These responses can then be combined to give an
expression describing growth rate variation in com-
plex environmental scenarios.
Previously, the effect of temperature on marine
phytoplankton maximum growth rate (Eppley 1972)
has been objectively defined using quantile regres-
sion (Koenker & Bassett 1978) on a large hetero -
geneous dataset derived from culture studies
(Bissinger et al. 2008). The effect of temperature on
maximum potential growth rate has also been
quantified for E. huxleyi using the same method
(Fielding 2013). The majority (>90%) of the E. hux-
leyi dataset used by Fielding (2013) was derived
from nutrient-replete cultures, making it unsuitable
for determining the effect of nutrient limitation on
growth rate in a similar way. However, these cul-
ture data were measured under a wide range of
light intensities, allowing the effect of light on max-
imum growth rate to be estimated using quantile
regression.
Light intensity in the ocean varies significantly
with depth below the sea-surface. Incident photosyn-
thetically active radiation (λ= 400 to 700 nm) meas-
ured as photon flux density (PFD) at the sea-surface
varies both spatially and temporally but can be at
least 2000 µmol photons m−2 s−1 (Kirk 1994, Frouin &
Murakami 2007). Light is then attenuated with depth
depending on factors such as turbidity. E. huxleyi
grows in ocean environments with highly variable
light intensity, with PFDs ranging from <10 µmol
photons m−2 s−1 (Egge & Heimdal 1994, Cortes et al.
2001) up to surface irradiance values.
Model calculations of E. huxleyi and coccolitho-
phore maximum growth rate as a function of PFD
usually use the rectangular hyperbolic function after
Monod (1949), where the maximum attainable growth
rate (μ) at each PFD is described by
(1)
where μopt is the maximum growth rate at ∞PFD and
Kis the light half-saturation constant.
Sometimes a function incorporating inhibition of
growth rate at high light intensities after Steele
(1962) is used, where μis described by
(2)
where μopt is the maximum growth rate at the PFD
which is optimal for growth rate and kis the initial
slope of the curve at low light. The parameter values
of these models are largely based on data from indi-
vidual strains or from cultures grown under differing
conditions even though models attempt to predict
responses of natural populations which comprise
diverse strain assemblages (Medlin et al. 1996). How-
ever, there is considerable intraspecific variation of E.
huxleyi growth rate in relation to light intensity
(Paasche 1999). Therefore, the suitability of current
models for describing the maximum potential growth
rate of E. huxleyi as a function of PFD in the ocean is
not certain.
To better estimate parameters for models describ-
ing growth rate as a function of PFD for E. huxleyi,
I applied quantile regression (Koenker & Bassett
1978) to literature growth rate data comprising
numerous individual E. huxleyi strains and experi-
ments. In addition, I applied quantile regression to
the data using the combined best-fit models for
growth rate as a function of PFD and growth rate as a
function of temperature (Fielding 2013).
MATERIALS AND METHODS
Data collection
Emiliania huxleyi growth rate data (n = 1387) were
obtained from literature culture experiments where
light intensity data were also recorded. Data that
were only presented graphically were extracted
using Engauge Digitizer v.4.1. All growth rates were
converted to cell-specific growth rate (μ, d−1). Non-
PFD light measurements (e.g. lumen ft−1, Langleys
min−1, W m−2) were converted to lux (lumen m−2) and
then converted to PFD (µmol photons m−2 s−1) using
the conversion factors in Table 1.
Literature-based E. huxleyi culture growth rate
data were derived from 95 publications, detailing 213
strains from 67 different locations (Fig. 1). Literature
culture experiments reported PFDs ranging from 3 to
1160 µmol photons m−2 s−1, day lengths from 10 to
24 h, temperatures from 2 to 30°C, and salinities from
12 to 45. Around 7% of the data were from nutrient-
limited cultures.
K
PFD
PFD
opt
μ=μ×+
μ=×μ××
×
PFD e
opt 1– PFD
k
k
164
Light source Factor
Cool white fluorescent 0.013
Daylight fluorescent 0.014
Unspecified fluorescent 0.0135
Halogen 0.019
Table 1. Light conversion factors used to convert literature
light intensity data expressed in lux (lumen m−2) to µmol
photons m−2 s−1 for different light sources. Units were multi-
plied by their respective conversion factors to obtain values
in µmol photons m−2 s−1

Fielding: Emiliania growth and light
Modelling and statistical analysis
Quantile re gression can be used to determine the
relationship between environmental variables and
growth rate for the upper edge of a scatterplot. This is
especially useful for datasets where variables other
than the one which has been measured are limiting to
growth (Cade & Noon 2003) and has previously been
used to solve similar problems (Bissinger et al. 2008,
Fielding 2013). To reliably define the upper edge of
the dataset a 99th quantile regression must be used,
as calculating the 100th quantile does not generate
parameter confidence intervals (Cade et al. 1999). In
this paper, 99th quantile regression was used to infer
the relationship between light intensity and maximum
growth rate for a highly heterogeneous set of E. hux-
leyi strains and environmental conditions where
growth was often limited by factors other than light. It
should be noted that a 99th quantile re-
gression is calculated using all data
and not just from the points at the ex-
treme upper edge of the dataset.
Quantile regression was used to esti-
mate the upper edge of the dataset (99th
quantile) using Rv.2.15.3 with the
quantreg v.4.96 package for a selection
of models commonly used to relate al-
gal growth or photosynthesis to light
intensity (Table 2). Model deviance can
be corrected to take into account the
number of para meters in the model,
with a higher number of parameters in-
curring a heavier penalty, using Akaike’s
information criterion (AIC; Akaike
1974), AIC = 2p− 2Lm, where pis the
number of parameters in the model
and Lmis the maximised log likelihood (−2Lmis equiv-
alent to the deviance of the model fit). However, all
the models used in this study use the same number of
independently varied parameters (p = 2), making AIC
redundant. Therefore, model fits are subsequently
compared using only their deviances.
RESULTS
Maximum Emiliania huxleyi specific growth rate
for individual measurements was 1.96 d−1 at approx.
600 µmol photons m−2 s−1 (Fig. 2). Maximum growth
rates appear to fall sharply below ~100 µmol photons
m−2 s−1. Above 600 µmol photons m−2 s−1, maximum
growth rates also appeared to be suppressed al -
though this is likely due to a lack of data (n = 13) in
this PFD range.
165
Fig. 1. Geographical locations of Emiliania huxleyi strains used in the litera-
ture compilation in this study (see Table S1 in the Supplement at www. int-res.
com/ articles/ suppl/ a073p163 _ supp .pdf for literature sources, strains used, and
geographical location of strain origin)
Model Deviance Equation μopt Kor k
Monod (1949) 11.18 1.858 ± 0.032 23.91 ± 5.608
Smith (1936) 12.81 1.795 ± 0.021 0.058 ± 0.016
Webb et al. (1974) 12.86 1.785 ± 0.022 0.069 ± 0.019
tanh (Jassby & Platt 1976) 13.28 1.780 ± 0.016 0.059 ± 0.020
Steele (1962) 24.34 2.939 ± 0.207 0.004 ± 0.000
K
PFD
PFD
opt
μ=μ×+
μ×+
μ+×
PFD
( PFD)
opt
22
k
k
opt
()
μ×−×μ
1e
opt – PFD/ opt
k
μ××
μ
⎛
⎝
⎜⎞
⎠
⎟
tanh PFD
opt
opt
k
×μ××
×
PFD e
opt 1– PFD
kk
Table 2. Model parameters with 95% confidence intervals and deviances for the models fit to the data in Fig. 2. Lower de-
viances indicate better relative model fits. PFD is photon flux density (µmol photons m−2 s−1), μopt is the modelled maximum
specific growth rate (d−1) across the entire PFD range, Kis the light half-saturation constant and kis the initial slope of the
curve at low light

Aquat Microb Ecol 73: 163–170, 2014
The rectangular hyperbolic Monod model (Eq. 1)
(3)
best described the 99th quantile of the specific growth
rate response to PFD (Fig. 2, Table 2). In contrast, the
Steele model (Eq. 2) had a relatively high deviance
and visibly overestimated maximum growth rate by
almost double at intermediate PFDs while marginally
underestimating maximum growth rate at low PFDs.
The other models (Smith 1936, Webb et al. 1974; and
the tanh model from Jassby & Platt 1976) had inter-
mediate deviances. While these models were nearly
identical to the Monod model at PFDs above ~400 µmol
photons m−2 s−1, they visibly overestimated maximum
growth rate at lower PFDs.
As the shapes of the responses of E. huxleyi maxi-
mum growth rate to both PFD (this study) and tem-
perature (Fielding 2013) have been quantified, it is
possible to formulate a combined expression by sub-
stituting μopt (from the Monod model) with the
power function describing the maximum attainable
growth rate as a function of temperature (Fielding
2013), as:
(4)
where Kis the light half-saturation constant, Tis the
temperature (°C), and aand bare the slope and the
power component of the growth rate versus tempera-
ture response, respectively. Quantile regression of
this combined PFD plus temperature model through
the 99th quantile of the data results in:
(5)
as shown in Fig. 3.
DISCUSSION
Growth rate response to light
intensity
This study represents the first syn-
thesis of literature-based Emiliania
hux leyi growth rate data as a function
of light intensity. Previous studies have
quantified models describing growth
rate response only from individual strains
or from a small number of strains, de-
spite there being considerable intra -
specific variation of E. huxleyi growth
rate in relation to light intensity (Paasche
2001). As a result, the use of these mod-
els incorporating coccolithophores such as E. huxleyi
(e.g. Merico et al. 2004) may not be straightforward.
However, this study provides an estimate of maximum
growth rate as a function of light intensity from a
multi-strain dataset which may be reasonably as-
sumed to more accurately represent a natural, geneti-
cally diverse global E. huxleyi population.
Potential biases
The dataset used in this study is larger (n = 1387)
than the minimum of n = 500 recommended by
Rogers (1992) for calculating the 99th quantile. How-
()
()
μ=××
+
0.199 PFD
14.2 PFD
0.716
T
aT K
bPFD
PFD
()
()
μ=× × +
μ=×+
1.858 PFD
23.91 PFD
166
Fig. 2. Emiliania huxleyi specific growth rate versus photon flux density. Models
(see Table 2) are fitted to the 99th quantile of the data
Fig. 3. 99th quantile regression of Emiliania huxleyi growth
rate response to both photon flux density and temperature.
Note reduced axes compared with Fig. 2. Colour scale: specific
growth rates (blue–red: low– high)

Fielding: Emiliania growth and light
ever, determining the best-fit model for the upper
edge of the dataset may be complicated by sampling
bias. Quantile regression is based on least absolute
deviations and is therefore less sensitive to extreme
values and outliers than ordinary least squares re -
gression (Bissinger et al. 2008). Nevertheless, the
lack of data (<1 %) and lower measured maximum
growth rates above 600 µmol photons m−2 s−1 may
affect model parameter estimates.
To test the effect of data above 600 µmol photons
m−2 s−1 on the model, these data were removed and
the models were re-run. Removal of these data did
not alter the order of model goodness of fit but did
slightly alter parameters for some of the models by
<0.7%, although these differences were not notice-
able when plotted. Nevertheless, the use of the Monod
model to describe E. huxleyi growth rate response to
light intensity above 600 µmol photons m−2 s−1 should
be made with this caveat until further data from
these high PFDs can be included in the dataset. How-
ever, E. huxleyi photosynthesis is notable for not dis-
playing inhibition at high light intensities of up to
2500 µmol photons m−2 s−1 (Paasche 2001). Therefore,
it is possible that maximum growth rates at high light
intensities are similarly uninhibited.
As was highlighted by Fielding (2013), only a small
proportion of the data are from strains isolated from
the Southern Hemisphere and far from continental
land masses (Fig. 1). Although it is not anticipated
that strains from different regions will have different
light-dependent growth rate responses, the applica-
tion of the recommended growth rate−PFD model
should be made with this caveat in mind.
Ideally, the growth rate response to both light and
temperature would be modelled for each individual
geographic or climatic region. This would serve both
to elicit any differences between regional-level pop-
ulations of E. huxleyi and to test whether the global
growth rate response models for light and tempera-
ture were universally applicable. However, dividing
the current dataset into regional subsets would de -
tract from the power of the combined global growth
rate response curve. A regional subset of the current
dataset would not be derived from as large a range of
culture variables due to the smaller number of exper-
iments carried out on that subset —for example, only
temperatures between 15 and 20°C or only PFDs
below 50 µmol photons m−2 s−1.
A further problem with dividing the current dataset
into regional subsets is that many regions (e.g. South
Africa, N Pacific) only have data derived from a small
number of strains or in some cases from a single
strain. There appears to be almost as much intraspe-
cific diversity within discrete geographic populations
of E. huxleyi than there is between populations from
different regions (Iglesias-Rodríguez et al. 2006).
Therefore, any differences observed in growth rate
response curves between subsets may not reflect a
true regional difference but may simply be a result of
under-sampling of the genetic diversity in each spe-
cific region.
The division of the current dataset into regional
subsets at the present stage may therefore be a little
premature, and more comprehensive data coverage
of individual regions is likely necessary before any
such analysis is made. The responses presented here
and in Fielding (2013) are, as such, still only rela-
tively blunt tools with which to parameterise E. hux-
leyi growth rate models.
Further bias may be introduced into the dataset by
the use of varying light sources in different literature
studies. Photosynthetically active radiation as meas-
ured by PFD includes all wavelengths between 400
and 700 nm. However, although 2 data points with the
same PFD would have the same quantity of photons
passing through this broad spectral band every sec-
ond, the spectral quality of the 2 data may be different.
For the collated literature studies presented here, the
light source was specified for 78% (n = 1082) of the
data. Around 75% of the entire dataset are from cul-
tures grown using fluorescent tubes, of which 52 %
were cool white, 26% were daylight and 22 % were
unspecified. Around 3% of the dataset were from cul-
tures grown using halogen lamps, of which one datum
was from a study published in 1992 (Balch et al. 1992)
and the remainder were from a study published in
1967 (Paasche 1967). These data all fell well below the
upper edge of the scatterplot and any resultant differ-
ences in spectral quality are not likely to influence the
results of this study. The 22% of the dataset where the
light source was unspecified were all from studies
published in 1995 or after, of which 82% were pub-
lished in 2000 or after. As the last recorded usage of
halogen lamps in this dataset was from 1992, it is
likely that data from cultures with unspecified light
sources were also grown using fluorescent tubes.
Combined growth rate response to light
and temperature
In addition to determining its response to PFD, this
study represents the first attempt at simultaneously
quantifying the response of E. huxleyi maximum spe-
cific growth rate to both light and temperature. This
combined model (see Results) can now be compared
167

Aquat Microb Ecol 73: 163–170, 2014
to other models used to estimate maximum specific
growth rate for E. huxleyi and for coccolithophores
as a functional group from temperature and light
intensity.
Four of the literature models for E. huxleyi use a
rectangular hyperbola (Monod 1949), while one uses
a hyperbolic tangent (tanh; Jassby & Platt 1976) func-
tion to describe growth rate versus PFD. In addition,
2 of the models use an exponential function after
Eppley (1972) to describe growth rate response to
temperature, while 3 use an exponential Q10 function
(van’t Hoff 1884). The combined expressions used in
these studies to calculate specific growth rate from
both temperature and PFD are detailed in Table 3.
In comparison with the combined model presented
in this study, all 5 literature models (Fig. 4A–E) over-
estimate E. huxleyi maximum growth rate by > 300%
across a wide range of PFDs at low temperatures,
while all 5 models overestimate E. huxleyi maximum
growth rate to a lesser extent across a wide range of
PFDs and temperatures, with the exception of Find-
lay et al. (2008), where growth rate is underestimated
over the majority of the PFD and temperature range
(Fig. 4A). The model used by Joassin et al. (2011)
overestimated maximum growth rate across the
entire PFD and temperature range (Fig. 4B).
The widespread overestimation of maximum E.
huxleyi growth rate by literature models is largely
due to the use of overly high values for μopt (i.e. the
maximum growth rate across all temperatures and
light intensities). For example, Merico et al. (2006)
give a maximum specific growth rate of 1.15 d−1 at
0°C. However, extrapolation using the specified tem-
perature−growth rate function results in high maxi-
mum specific growth rates of 4.05 d−1 at 20°C and an
even higher 7.61 d−1 at 30°C, much higher than that
observed in E. huxleyi culture experiments.
In addition to overly high μopt in literature models,
the literature values for the initial slope of the growth
rate response to PFD, with the exception of those
used by Oguz & Merico (2006) and Joassin et al.
(2011), are also higher than that presented in this
study. This results in generally shallower PFD−
growth rate slopes up to the growth optima in litera-
ture models, and therefore in the underestimation of
growth rate at low PFD values across all or the major-
ity of the temperature range for these studies
(Fig. 4A,C,E). Hence, previously used parameters for
both PFD and temperature components of growth
rate models appear to be inappropriate for E. huxleyi
and it is recommended that the combined model
parameters presented in this study should be used in
future.
In addition to modelling the coccolithophore E.
huxleyi as a discrete ecosystem component, there are
some studies which model all coccolithophore spe-
168
Literature model Tmodel PFD model Equation
E. huxleyi
Findlay et al. (2008) Eppley Monod
Joassin et al. (2011) Q10 Monod
Merico et al. (2004, 2006)a Eppley Monod
Oguz & Merico (2006) Q10 tanh
Tyrrell & Taylor (1996) Q10 Monod
Coccolithophores
Gregg et al. (2003) Eppley Monod
Gregg & Casey (2007) Eppley Monod
aSensitivity analysis run
()
()
××
+
×
0.5 e PFD
205 PFD
0.063 T
()
()
×+
2.64 / 1.5 PFD
20 PFD
(20– )/10T
()
()
××
+
×
1.5 e PFD
205 PFD
0.063 T
()
()
××
2.2 / 1.5 tanh 0.026 PFD
2.2 / 1.5
(20– )/10
(20– )/10
T
T
()
()
×+
1.8 / 2 PFD
100 PFD
(16 – )/ 10T
()
()
××
+
×
0.228 e PFD
71.2 PFD
0.063 T
()
()
××
+
×
0.321 e PFD
71.2 PFD
0.063 T
Table 3. Models used in literature studies to predict maximum specific growth rate (μ) from temperature (T, °C) and photon
flux density (PFD, µmol photons m−2 s−1). The differences between these models and the combined temperature – photon flux
density model calculated in this study are shown in Fig. 4

Fielding: Emiliania growth and light
cies as a combined functional group (Gregg et al.
2003, Le Quere et al. 2005, Gregg & Casey 2007).
When compared to the combined PFD and tempera-
ture model presented in this study (Fig. 4F – I), cocco-
lithophore functional group models described by
Gregg et al. (2003) and Gregg & Casey (2007) have a
lower maximum growth rate over almost the entire
PFD range due to lower μopt values.
A multi-species coccolithophore population is
indeed likely to have a lower combined maximum
growth rate than a monospecific E. huxleyi popula-
tion due to the inclusion of larger, slower growing
species such as Coccolithus spp. and Calcidiscus spp.
Existing functional group models likely somewhat
reflect this lower multi-species coccolithophore com-
munity growth rate. However, E. huxleyi is one of the
smallest (if not the smallest) species of coccolitho-
phore (Buitenhuis et al. 2008) and is therefore likely
to be the fastest growing as predicted by the meta-
bolic theory of ecology (Brown et al. 2004). As such
the 99th quantile E. huxleyi growth rate envelope
presented in this study likely represents the maxi-
mum potential growth rate for all coccolithophore
species as a functional group, and will become
increasingly more appropriate as E. huxleyi starts to
dominate the coccolithophore assemblage, for exam-
ple in an E. huxleyi bloom scenario.
CONCLUSION
The Monod model presented in this study repre-
sents a first step towards quantifying the maximum
specific growth rate response of E. huxleyi to light. It
is recommended that the function μ= 1.858[PFD/
(23.91 + PFD)], and not a Steele equation, be used in
future models incorporating E. huxleyi growth rate.
However, data from PFDs above 600 µmol photons
m−2 s−1 and from a more geographically diverse set of
culture strains than make up the current dataset will
allow for a future reappraisal of this relationship.
Where both temperature and PFD are known, maxi-
mum E. huxleyi growth rate should be calculated
from the combined expression μ= (0.199 × T0.716) ×
[PFD/(14.2 + PFD)].
Acknowledgements. Thanks to 3 anonymous reviewers for
advice and comments on the manuscript.
169
Fig. 4. The percentage which literature
models (Table 3) underestimate (negative
values) or overestimate (positive values)
maximum specific growth rate compared to
the combined temperature–photon flux den-
sity model presented in this study (Fig. 3).
0% equals no difference between the litera-
ture model and the model presented in this
study. Contours are at 50% intervals al-
though no data are shown above 300%.
Colour scale: specific growth rates (blue–
red: low– high). Models for Emiliania hux-
leyi: (A) Findlay et al. (2008), (B) Joassin et
al. (2011), (C) Merico et al. (2004, 2006) (D)
Oguz & Merico (2006) and (E) Tyrrell & Tay-
lor (1996); and for coccolithophores as a
functional group: (F) Gregg et al. (2003) low
light model, (G) Gregg et al. (2003) high
light model, (H) Gregg & Casey (2007) low
light model and (I) Gregg & Casey (2007)
high light model

Aquat Microb Ecol 73: 163–170, 2014
LITERATURE CITED
Akaike H (1974) A new look at statistical-model identifica-
tion. IEEE Trans Automat Contr 19: 716−723
Balch WM, Holligan PM, Kilpatrick KA (1992) Calcification,
photosynthesis and growth of the bloom-forming cocco-
lithophore Emiliania huxleyi. Cont Shelf Res 12: 1353−1374
Bissinger JE, Montagnes DJS, Sharples J, Atkinson D (2008)
Predicting marine phytoplankton maximum growth rates
from temperature: improving on the Eppley curve using
quantile regression. Limnol Oceanogr 53: 487−493
Bollmann J, Herrle JO, Cortes MY, Fielding SR (2009) The
effect of sea water salinity on the morphology of Emilia-
nia huxleyi in plankton and sediment samples. Earth
Planet Sci Lett 284: 320−328
Brown JH, Gillooly JF, Allen AP, Savage VM, West GB
(2004) Toward a metabolic theory of ecology. Ecology 85:
1771−1789
Buitenhuis ET, Pangerc T, Franklin DJ, Le Quere C, Malin G
(2008) Growth rates of six coccolithophorid strains as a
function of temperature. Limnol Oceanogr 53: 1181−1185
Cade BS, Noon BR (2003) A gentle introduction to quantile
regression for ecologists. Front Ecol Environ 1: 412−420
Cade BS, Terrell JW, Schroeder RL (1999) Estimating effects
of limiting factors with regression quantiles. Ecology 80:
311−323
Cortes MY, Bollmann J, Thierstein HR (2001) Coccolithophore
ecology at the HOT station ALOHA, Hawaii. Deep-Sea
Res II 48: 1957−1981
Egge JK, Heimdal BR (1994) Blooms of phytoplankton
including Emiliania huxleyi (Haptophyta). Effects of
nutrient supply in different N: P ratios. Sarsia 79: 333−348
Eppley R (1972) Temperature and phytoplankton growth in
the sea. Fish Bull 70: 1063−1085
Fielding SR (2013) Emiliania huxleyi specific growth rate de -
pendence on temperature. Limnol Oceanogr 58: 663−666
Findlay CS, Giraudeau J (2000) Extant calcareous nanno-
plankton in the Australian Sector of the Southern Ocean
(austral summers 1994 and 1995). Mar Micropaleontol
40: 417−439
Findlay HS, Tyrrell T, Bellerby RGJ, Merico A, Skjelvan I
(2008) Carbon and nutrient mixed layer dynamics in the
Norwegian Sea. Biogeosciences 5: 1395−1410
Frouin R, Murakami H (2007) Estimating photosynthetically
available radiation at the ocean surface from ADEOS-II
global imager data. J Oceanogr 63: 493−503
Gregg WW, Casey NW (2007) Modeling coccolithophores in
the global oceans. Deep-Sea Res II 54: 447−477
Gregg WW, Ginoux P, Schopf PS, Casey NW (2003) Phyto-
plankton and iron: validation of a global three-dimen-
sional ocean biogeochemical model. Deep-Sea Res II 50:
3143−3169
Hegseth EN, Sundfjord A (2008) Intrusion and blooming of
Atlantic phytoplankton species in the high Arctic. J Mar
Syst 74: 108−119
Iglesias-Rodríguez MD, Schofield OM, Batley J, Medlin LK,
Hayes PK (2006) Intraspecific genetic diversity in the
marine coccolithophore Emiliania huxleyi (Prymnesio-
phyceae): the use of microsatellite analysis in marine
phytoplankton population studies. J Phycol 42: 526−536
Jassby AD, Platt T (1976) Mathematical formulation of the
relationship between photosynthesis and light for phyto-
plankton. Limnol Oceanogr 21: 540−547
Joassin P, Delille B, Soetaert K, Harlay J and others (2011)
Carbon and nitrogen flows during a bloom of the cocco-
lithophore Emiliania huxleyi: modelling a mesocosm
experiment. J Mar Syst 85: 71−85
Keller MD (1988) Dimethyl sulfide production and marine
phytoplankton: the importance of species composition
and cell size. Biol Oceanogr 6: 375−382
Kirk J (1994) Light and photosynthesis in aquatic ecosys-
tems. Cambridge University Press, Cambridge
Koenker R, Bassett G (1978) Regression quantiles. Econo-
metrica 46: 33−50
Le Quere C, Harrison SP, Prentice IC, Buitenhuis ET and
others (2005) Ecosystem dynamics based on plankton
functional types for global ocean biogeochemistry mod-
els. Glob Change Biol 11: 2016−2040
Medlin LK, Barker GLA, Campbell L, Green JC and others
(1996) Genetic characterisation of Emiliania huxleyi
(Haptophyta). J Mar Syst 9: 13−31
Merico A, Tyrrell T, Lessard EJ, Oguz T, Stabeno PJ, Zee-
man SI, Whitledge TE (2004) Modelling phytoplankton
succession on the Bering Sea shelf: role of climate influ-
ences and trophic interactions in generating Emiliania
huxleyi blooms 1997–2000. Deep-Sea Res I 51: 1803−1826
Merico A, Tyrrell T, Cokacar T (2006) Is there any relation-
ship between phytoplankton seasonal dynamics and the
carbonate system? J Mar Syst 59: 120−142
Monod J (1949) The growth of bacterial cultures. Annu Rev
Microbiol 3: 371−394
Oguz T, Merico A (2006) Factors controlling the summer
Emiliania huxleyi bloom in the Black Sea: a modeling
study. J Mar Syst 59: 173−188
Paasche E (1967) Marine plankton algae grown with light-dark
cycles. I. Coccolithus huxleyi. Physiol Plant 20: 946−956
Paasche E (1999) Reduced coccolith calcite production
under light-limited growth: a comparative study of three
clones of Emiliania huxleyi (Prymnesiophyceae). Phyco -
logia 38: 508−516
Paasche E (2001) A review of the coccolithophorid Emiliania
huxleyi (Prymnesiophyceae), with particular reference to
growth, coccolith formation, and calcification-photosyn-
thesis interactions. Phycologia 40: 503−529
Rogers WH (1992) Quantile regression standard errors.
Stata Tech Bull 9: 16−19
Rost B, Riebesell U (2004) Coccolithophores and the biologi-
cal pump: responses to environmental changes. In: Thier-
stein HR, Young JR (eds) Coccolithophores: from molec-
ular processes to global impact. Springer, Berlin, p 99– 125
Smith EL (1936) Photosynthesis in relation to light and car-
bon dioxide. Proc Natl Acad Sci USA 22: 504−511
Steele JH (1962) Environmental control of photosynthesis in
the sea. Limnol Oceanogr 7: 137−150
Sukhanova IN, Flint MV, Pautova LA, Stockwell DA, Greb-
meier JM, Sergeeva VM (2009) Phytoplankton of the
western Arctic in the spring and summer of 2002: structure
and seasonal changes. Deep-Sea Res II 56: 1223−1236
Tyrrell T, Merico A (2004) Emiliania huxleyi: bloom observa-
tions and the conditions that induce them. In: Thierstein
HR, Young JR (eds) Coccolithophores: from molecular
processes to global impact. Springer, Berlin, p 75– 97
Tyrrell T, Taylor AH (1996) A modelling study of Emiliania
huxleyi in the NE Atlantic. J Mar Syst 9: 83−112
van’t Hoff MJH (1884) Etudes de dynamique chimique. Recl
Trav Chim Pays-Bas 3: 333−336
Webb WL, Newton M, Starr D (1974) Carbon dioxide ex -
change of Alnus rubra. A mathematical model. Oecolo-
gia 17: 281−291
Zeebe RE, Wolf-Gladrow DA (2001) CO2in seawater: equi-
librium, kinetics, isotopes. Elsevier Oceanography Series,
Vol 65. Elsevier, Amsterdam
170
Editorial responsibility: Hugh MacIntyre,
Halifax, Nova Scotia, Canada
Submitted: April 4, 2013; Accepted: June 6, 2014
Proofs received from author(s): August 28, 2014
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤