Journal of Plankton Research Vol.20 no.2 pp.187-217, 1998
Pigments and species composition of natural phytoplankton
populations: effect on the absorption spectra
Venetia Stuart1, Shubha Sathyendranath'^ Trevor Platt2, Heidi Maass1 and
Brian D.Irwin2
1
Department of Oceanography, Dalhousie University, Halifax, Nova Scotia B3H
4J1 and 2Ocean Sciences Division, Bedford Institute of Oceanography, Box 1006,
Dartmouth, Nova Scotia B2Y4A2, Canada
Abstract Data from three cruises (Arabesque 1 and 2 cruises in the Arabian Sea and the Vancouver
Island cruise) were examined to assess the importance of species composition and accessory pigments
in modifying specific absorption coefficients. The three cruises differed widely in their phytoplankton
assemblages with small cells dominating the Arabesque 2 cruise and large diatoms the Vancouver
Island cruise. Absorption spectra from each cruise were decomposed into 13 Gaussian bands representing absorption by the major chlorophylls and accessory pigments. The maximum specific peak
height (p£) for each Gaussian band was obtained by regressing Gaussian peak heights against the
concentration of the pigment responsible for the absorption band. This relationship was generally
non-linear and was fitted with a rectangular hyperbolic function. Changes in the maximum specific
peak heights between cruises reflect changes in the packaging effect, which were most apparent at
Gaussian bands of high absorption (400-490 nm), but were close to zero at the Gaussian band
centered around 623 nm, associated with chlorophyll a (Chi a). The maximum specific peak height of
this Gaussian band, p£(623), may be used to obtain reliable estimates of Chi a for any phytoplankton
assemblage, unaffected by variations caused by the package effect. Comparing the Arabesque 2 peak
heights with the Vancouver data, an apparent flattening effect of 0.42 at 440 nm and 0.62 at 676 nm
was found for the Vancouver data relative to the Arabesque 2 data (assuming a zero flattening effect
for the latter). Multiple linear regression analysis suggested that 29-42% of the variability in the
specific absorption coefficient of phytoplankton at 440 nm was due to changes in pigment composition, while the remaining 58-71% was due to changes in the package effect. An inverse relationship
was found between the proportion of non-photosynthetic pigments (NPC) and ambient Chi a concentration, suggesting that small cells (generally found in oligotrophic waters) had higher proportions of
photoprotective pigments.
Introduction
In the surface waters of the open ocean, light absorption by photosynthetic pigments of phytoplankton cells usually dominates the optical variability of the
water column, and contributes to the attenuation of light with depth. Absorption
by phytoplankton populations is known to vary widely from one water body to
another and estimation of absorption coefficients of natural phytoplankton cells
has recently received a great deal of attention. The upsurge in interest may be
attributed to the increasing use of remote sensing of ocean color to estimate algal
biomass (Gordon and Morel, 1983) and to model primary production in the ocean
(Platt and Sathyendranath, 1988; Morel and Andre, 1991; Platt et al, 1995;
Sathyendranath et al., 1995). Particulate absorption properties form an integral
part of these models, and the spatial and temporal variability in the absorption
characteristics of natural phytoplankton populations must be understood if the
models are to be optimized for a particular region or season.
The absorption coefficient of phytoplankton, aph, is commonly taken to be a
function of the concentration of chlorophyll a (Chi a), the major pigment in all
© Oxford University Press
187
VStuart et al
species of phytoplankton. It is important to note, however, that absorption by
phytoplankton is not due to Chi a alone, but also to auxiliary pigments. The
absorption coefficient of phytoplankton at wavelength X can thus be written as:
aph{\) = ZflXAOQ
(1)
where af (X) is the in vivo specific absorption coefficient of the ith pigment, C, is
the concentration of this pigment and n represents the total number of pigments
considered.
To account for the influence of auxiliary pigments in modifying the absorption
spectra of phytoplankton, we need to examine the variability of in vivo absorption coefficients of individual pigments. Hoepffner and Sathyendranath (1991)
developed a method for estimating in vivo specific absorption coefficients of the
major phytoplankton pigments, by decomposing the absorption spectra into
several Gaussian bands simulating the absorption properties of the individual pigments. An alternative method to estimate the in vivo absorption coefficient of
phytoplankton, and to account for the influence of accessory pigments (Bidigare
et al., 1987,1990), assumes that in vitro specific absorption coefficients obtained
from extracted pigments are representative of in vivo coefficients of unextracted
pigments bound in pigment-protein complexes. This assumption may not always
hold (Johnsen et al., 1994a) and there is some evidence that this reconstruction
technique is not suitable for natural phytoplankton populations with a high diversity of sizes and shapes (Arbones et al., 1996). Therefore, we have used the
decomposition method of Hoepffner and Sathyendranath (1991) to estimate the
specific absorption coefficients of individual pigments.
Laboratory experiments have shown that changes in pigment package effects
and differences in the composition of the auxiliary pigments are the primary
sources of variability in specific absorption coefficients of phytoplankton cells
(Sathyendranath et al., 1987; Mitchell and Kiefer, 1988; Babin et al, 1993). The
package effect (also known as the flattening effect) describes the decreased
absorption of pigments in cells compared with the absorption potential for the
same amount of pigment in solution (Duysens, 1956; Kirk, 1975,1994; Morel and
Bricaud, 1981). An increase in pigment packaging occurs either as cell size
increases or as the internal concentration of pigments increases (Kirk, 1976;
Morel and Bricaud, 1981; Sosik and Mitchell, 1994). If there were no package
effect, the in vivo specific absorption coefficients of individual pigments should
be independent of species, provided there are no species-dependent differences
in the absorption efficiencies of pigment-protein complexes. In fact, Hoepffner
and Sathyendranath (1991) noted that there was no difference in specific absorption coefficients for individual pigments in three different classes of cultured
algae, which had been corrected for the package effect. More recent studies,
however, have suggested that there might be some species dependence in the
absorption characteristics of pigment-protein complexes (Johnsen et al., 1994a).
The aim of the present study is to examine the effects of species composition
on the absorption characteristics of the four major pigment groups (Chi a, b and
c, and carotenoids) and to assess whether estimates of in vivo specific absorption
188
Pigments, species composition and absorption spectra
coefficients of these pigments can be applied to a variety of natural algal
populations. Pigment-specific absorption coefficients may be used to reconstruct
the in vivo absorption spectrum of mixed algal populations if the concentration
of the pigments is known (see Hoepffher and Sathyendranath, 1991) or, conversely, they can be used to estimate pigment concentrations from decomposed
absorption spectra.
Method
To cover a wide range of phytoplankton types and a variety of oceanographic conditions, we used data from three different cruises. The first two cruises were conducted in the NW Indian Ocean in different seasons: the Arabesque 1 cruise
sampled the region during the vigorous SW monsoons (August-September, 1994)
and the Arabesque 2 cruise was undertaken during the inter-monsoon period
(November-December, 1994) (Sathyendranath et al, 1998). These two cruises
sampled the inshore phytoplankton populations as well as the oligotrophic offshore populations. Samples from the third cruise were collected in March 1996,
in the shallow eutrophic waters of Saanich Inlet and among the Gulf Islands off
Vancouver Island, British Columbia. Sampling depths varied from 1 to 75 m.
Pigment analysis
Chlorophyll a concentrations were measured on board with a Turner Design
fluorometer, while samples for the detailed analysis of pigment composition were
filtered onto GF/F filters and stored at -70°C until analyzed. Pigment concentrations were quantified with reverse-phase, high-performance liquid chromatography (HPLC) as described in Head and Home (1993). Frozen filters were
homogenized in 1.5 ml of 90% acetone, centrifuged and diluted with 0.5 M
ammonium acetate buffer at a ratio of 2:1 before injection. Pigments were identified by comparing their retention times and absorption characteristics with
known standards. Divinyl Chi a and b (Chi a2 and Chi b2), which are characteristic of prochlorophytes, were detected by acidifying samples with 1 N HC1 and
noting the presence of divinyl phaeophytin-like pigments. Samples from the Vancouver Island cruise were not acidified since only minute quantities of zeaxanthin
(characteristic of prochlorophytes) were detected.
Light absorption by phytoplankton
Particulate samples were collected on GF/F glass fiber filters, and stored at -70°C
until processed according to the filter technique of Yentsch (1962), as modified
by Mitchell and Kiefer (1984,1988). Filters were placed on a drop of filtered seawater and the absorption of the total particulate material [ap(A.)], relative to a
blank filter saturated with seawater, was measured on a dual-beam Shimadzu
UV-2101 PC scanning spectrophotometer equipped with an integrating sphere.
Absorption by the detrital component, ad{k), of the sample was estimated according to the method of Kishino et al. (1985) with some minor modifications. Firstly,
189
\StuartetaL
phytoplankton pigments were extracted from the niters using 20 ml of a mixture
of 90% acetone and dimethyl sulfoxide (DMSO) (6:4 v:v) which was allowed to
flow passively through the sample filter, followed by 10 ml of filtered seawater to
remove any residual solvents. Next, absorption of the decolorized filters was
measured as before, ensuring that the niters were in exactly the same orientation
in the spectrophotometer as for the total particulate absorption. However, many
samples contained large numbers of cyanobacteria, which possess water-soluble
phycobiliproteins that are not readily extracted by the acetone/DMSO solution.
Besides, several extracted samples (mainly those with high Chi a concentrations)
also contained residual Chi a and/or phaeopigments (the phaeopigments may
have been formed during the extraction process). To correct for these unextracted
pigments, which would lead to an underestimate in the phytoplankton absorption, the detrital absorption spectrum was decomposed into an exponential curve
plus several Gaussian curves superimposed at the wavelengths of absorption of
the non-extracted pigments (usually around 420 and 666 nm for phaeopigments,
and 510, 550 and 590 nm for the biliproteins) (see Sathyendranath et al, 1998).
Curves were fitted using the Marquardt-Levenberg algorithm to determine the
parameters that minimize the sum of squares of differences between the dependent variable and the observations (Lawson and Hanson, 1974). There was usually
an excellent correlation between measured data and the sum of the constituent
curves (r2 around 0.99) and the fitted exponential curve was then used as the
measure of the detrital absorption for the sample.
Optical density measurements were converted to absorption coefficients by
subtracting the optical density at 750 nm from all other wavelengths, dividing by
the geometrical path length (volume filtered divided by clearance area of the
filter), and adjusting for path length amplification (P-correction algorithm) due
to scattering by the filter. The choice of the p-correction algorithm may profoundly influence the interpretation of absorption results and is discussed in detail
below. Absorption by photosynthetic pigments, aph(X), was then calculated as the
difference between ap(K) and the absorption by detritus, ad(X). The specific
absorption coefficient of phytoplankton at 440 nm [flp/,(440)] was obtained by
dividing aph(440) by the Chi a concentration (either the sum of Chi a plus divinyl
Chi a from HPLC analysis, or Chi a estimated using the Turner fluorometer). Two
samples from the Vancouver Island cruise had anomalously high phaeopigment
concentrations and were eliminated from the data set.
Choice of ^-correction algorithm
According to a recent study by Moore et al. (1995) on cultures of Prochlorococcus and Synechococcus, many of the currently used P-correction algorithms may
overestimate the absorption coefficient by up to a factor of two in waters dominated by prokaryotes. Since most stations from the Arabesque 2 cruise, and
several from the Arabesque 1 cruise, were dominated by these small cells, the
choice of the P-correction algorithm may be critical to the interpretation of our
results. We tried three different approaches to correct for the path length
190
Pigments, species composition and absorption spectra
amplification factor in an attempt to find the most suitable method for our highly
diverse data set.
In the first approach, absorption spectra from all three cruises were corrected
for the effects of path length amplification using the quadratic equation proposed
by Hoepffner and Sathyendranath (1992, 1993), (OD, = 0.31ODf+0.57[ODf]2),
which is very similar to the equation proposed by Mitchell (1990). This algorithm
was developed using laboratory cultures of relatively large Prymnesiophyceae
and Bacillariophyceae. In the second approach, we attempted to correct for the
presence of small prochlorophytes, which have a significantly different P-correction algorithm compared to the larger cells. In this approach, we used the P-correction algorithm for Prochlorococcus marinus from Moore et al. (1995) for all
samples containing divinyl Chi a (indicating the presence of prochlorophytes),
while the remaining samples were corrected using the algorithm of Hoepffner and
Sathyendranath (1992, 1993). Since many samples contained only a small percentage of divinyl Chi a, additional errors may be introduced by assuming a single
P-correction algorithm for the entire sample. In the third approach, we attempted
to alleviate this problem by estimating the proportion of Chi a that may be attributed to large or small cells in each sample, and applying the relevant P-correction
algorithm to each fraction of the sample. We used the percentage of divinyl Chi
a to estimate the proportion of total HPLC Chi a that may be attributed to small
prochlorophytes and corrected this fraction of the sample with the P-correction
algorithm for P.marinus (Moore et al., 1995), while the remaining fraction of the
chlorophyll was attributed to large cells and corrected using the quadratic equation of Hoepffner and Sathyendranath (1992,1993). This approach is similar to
that used by Kyewalyanga etal. (1997) for samples from the North Atlantic ocean.
One of the disadvantages of this approach is that the presence of small cyanobacteria, which were abundant in both Arabesque cruises, was not taken into
account. However, since there is only a relatively small difference between the Pcorrection algorithm for Synechococcus (Moore et al., 1995) and that of
Hoepffner and Sathyendranath (1992,1993) (-10% at an optical density of 0.4),
we assumed that the resultant errors would be minimal.
Decomposition of absorption spectra
Each absorption spectrum of photosynthetic pigments was decomposed into 13
Gaussian bands representing absorption by the major phytoplankton pigments
(Chi a, b and c, and carotenoids), as described in detail by Hoepffner and
Sathyendranath (1991, 1993). The initial centers and half-widths (nm) of each
Gaussian band were taken from Hoepffner and Sathyendranath (1991,1993), but
the initial peak heights of each band \p(k) in nr 1 ] were adjusted according to the
magnitude of the absorption at 440 nm. Gaussian bands nominally centered
around 384, 413, 435, 623, 676 and 700 nm correspond to absorption by Chi a,
those at 464 and 655 nm to Chi b, those at 461,583 and 644 nm to Chi c, and those
at 490 and 532 nm to total carotenoids (note: band 583 nm is also related to Chi
a). For most Gaussian bands, the relationship between the peak height and the
191
VStwutetaL
corresponding pigment concentration, C, (mg m~3), was non-linear with the slope
being steep at low pigment concentrations and gradually decreasing as pigment
concentration increases. One possible explanation for this type of non-linear
relationship is that the flattening of the curve reflects an increased package effect
with increasing pigment concentrations. This, in turn, may be attributed to either
an increase in cell size or in intracellular pigment concentration at higher ambient
concentrations of the pigment. A rectangular hyperbolic function (Lutz et ai,
1996) was used to describe this relationship:
where Pm(X) is the maximum specific peak height at wavelength X, and pm(k) is
the maximum asymptotic value of the peak height at that wavelength. Intercepts
were not used since absorption should be zero when pigment concentration
equals zero. The parameter p%,(\) describes the initial slope of the curve at low
pigment concentrations and represents the maximum specific absorption for the
p(k)-Ci pairs for each Gaussian band. In theory, the maximum specific absorption should approach the specific absorption in the absence of flattening, assuming that small cells (with minimal package effects) are found at low ambient
chlorophyll concentrations (Yentsch and Phinney, 1989). Moreover, unlike the
slope of a linear fit, this parameter is not affected by data points at the high end
of the range, since the slope of the curve approaches a maximum at the origin.
It should be noted here that, for many of the Gaussian bands, a linear regression also provided a good fit to the data at the low range of pigment concentrations, and the curve only became noticeably non-linear at higher pigment
concentrations. The parameter pm(k) is the theoretical value of maximum peak
height, which is always attained outside the range of measurements. The value
of this parameter approaches infinity as the curve tends towards linearity. In
those cases where pm(X) reached an unrealistically high value, the data were
fitted with a Model I linear regression equation, resulting in very similar r2
values. For each of the three cruises, the rectangular hyperbolic function was
fitted to the p(\)-Ci pairs from each Gaussian band. It should be noted that the
applicability of equation (2) is confined to the range of data values for which it
was established.
Non-linearity between absorption (Gaussian band height) and pigment
concentration implies variable specific absorption coefficients within the data set,
which may be attributed to changes in the package effect (i.e. changes in cell size
or shape, or changes in the intracellular pigment concentration). Similarly, differences in slopes [pm(A.)] between data sets imply different packaging effects for the
data sets, even for low pigment concentrations. Another possibility that should
not be ruled out is that variations in p£,(k) reflect species-dependent changes in
the absorption efficiency of pigment-protein complexes. A linear relationship
does not necessarily imply no packaging effect, but merely the same degree of
packaging within the data set.
192
Pigments, species composition and absorption spectra
Errors in measurement
In an attempt to estimate how much of the scatter around trend lines could be
attributed to random errors in pigment determination and light absorption
measurements, we performed a series of reproducibility experiments. Twenty
replicate samples of a Chaetoceros sp. culture (5 ml each) and a seawater sample
(500 ml each) were filtered onto GF/F filters. Half the samples from each batch
were analyzed for HPLC pigment composition, while the remaining 10 samples
from each batch were used for measurement of absorption spectra and subsequent decomposition into Gaussian bands, as described above. The standard
deviation, a, of each set of measurements was used as an estimate of the error
associated with that parameter and expressed as a percentage of the mean value.
Errors associated with parameters obtained through division (i.e. specific absorption coefficients or specific peak height coefficients) were estimated from the sum
chl-b
chl-c
Vancouver
diadinox an thin
chl-b
chl-c
fucoxanthin
19-but T
diadinoxanthin
zeaxanthin
chl-c
Arabesque 2
19-hex
19-but \x~^ fucoxanthin
diadinoxanthin
Fig. L Relative composition of major accessory pigments for phytoplankton samples from the Vancouver, Arabesque 1 and Arabesque 2 cruises.
193
VStnart et aL
of averages of squares of deviations (see Bevington, 1969) and were also
expressed as a percentage of the mean value.
Results
Phytoplankton composition
Within the Arabesque cruises, regional and seasonal variations in the composition of the phytoplankton assemblages were observed (see Sathyendranath et aL,
1996,1998; Tarran et aL, 1998), which were also evident in the composition and
relative abundance of the predominant accessory pigments (Figure 1). Comparing the mean values from the three cruises, the most notable differences were in
the relative proportions of fucoxanthin and Chi b, as well as the absence of 19'butanoyloxyfucoxanthin, 19'-hexanoyloxyfucoxanthin and zeaxanthin from the
Vancouver cruise. The latter cruise was characterized by high concentrations of
fucoxanthin (2.18 ± 3.05 mg nr 3 ) and Chi (c, + c2) (0.76 ± 1.03 mg nr 3 ) as well as
the presence of diadinoxanthin (Figure 2), suggesting a phytoplankton assemblage dominated almost entirely by diatoms. Microscopic identification confirmed this, and revealed that the three dominant species were large-celled
Skeletonema costatum, Thalassiosira nordenskioldii and Thalassiosira aestivalis.
Diatoms were also present at several stations during the Arabesque 1 cruise and
a few stations in the Gulf of Oman during the Arabesque 2 cruise. Significant
quantities of 19'-butanoyloxyfucoxanthin (up to 38% of total carotenoids) and 19'hexanoyloxyfucoxanthin (up to 63% of total carotenoids) were found at many
stations during both the Arabesque cruises (Figure 2), suggesting the presence of
prymnesiophytes and chrysophytes. In addition, high concentrations of divinyl Chi
a (up to 74% of the total Chi a) and divinyl Chi b (up to 72% of total Chi b) were
found at most stations during the Arabesque 2 cruise and also in the oligotrophic
open-ocean stations of the Arabesque 1 cruise, suggesting the widespread distribution of small-celled prochlorophytes (Figures 1 and 2). Mean total Chi 6:Chl a
ratios were 0.15 ± 0.07 for thefirstcruise and 0.18 ± 0.09 for the second Arabesque
cruise. Finally, high concentrations of zeaxanthin at many stations in both
Arabesque cruises (maximum of 0.24 mg m"3, representing up to 82% of the total
carotenoids) suggest an abundance of small-celled cyanobacteria. This is in agreement with the flow cytometric results of Burkill and Tarran (1998) who noted high
densities (>106 cells H) of Synechococcus sp. in the surface mixed layer of the Gulf
of Oman and the Arabian Sea during these cruises. Based on these phytoplankton
assemblages, the cruises can be ranked for presumed cell size: the Vancouver
Island cruise, dominated by large-celled chain-forming diatoms, would have the
largest overall cell sizes, followed by the Arabesque 1 cruise (mixed assemblage of
diatoms, prymnesiophytes and other cells), and finally the Arabesque 2 cruise with
an abundance of small-celled prochlorophytes and cyanobacteria.
Choice of ^-correction algorithm
Absorption coefficients for the Arabesque 2 cruise were initially estimated using
the P-correction algorithm proposed by Hoepffner and Sathyendranath (1992,
194
Pigments, species composition and absorption spectra
0.000 25
0.009
Arabesque 2 acidified
g
0.006 H
0.003 89
0.000
5
10
IS
25
Time (minutes)
Fig. 2. HPLC absorbance chromatograms (a, b and c) of representative samples from each of the three
cruises as well as afluorescencechromatogram (d) of the sample from the Arabesque 2 cruise which
has been acidified to show the separation of divinyl chlorophylls a and b. Peak identification: (1) Chi
c3; (2) chlorophyllide a; (3) Chi (c, + c2); (4) peridinin; (5) a-type phaeopigment; (6) 19'-butanoyloxyfucoxanthin; (7) fucoxanthin; (8, 9) phaeoporphyrin c-like pigments; (10) 19'-hexanoyloxyfucoxanthin; (11) pyrophaeophorbide a; (12) diadinoxanthin; (13) alloxanthin; (14) zeaxanthin; (15) Chi
6; (16) Chi a (+ isomers); (17) phaeophytin b; (18) divinyl phaeophytin b; (19) phaeophytin a; (20)
divinyl phaeophytin a; (21) p^carotene.
1993). This approach resulted in relatively high absorption values (Figure 3)
which were attributed to the abundance of small-celled prochlorophytes and
cyanobacteria (see also Sathyendranath etal, 1998). In the second approach, the
P-correction algorithm was based upon the presence or absence of divinyl Chi a,
and a strong non-linear relationship was found between absorption and chlorophyll concentration (Figure 3), implying that the absorption values of samples
with high Chi a concentrations (>1 mg nr 3 ) were being underestimated. Notably,
195
VStnart et al
0.00
0.2
0.4
0.6
0.8
1.0
1.2
3
Chl-a + DV chl-a (mg m )
Fig. 3. Absorption at 440 nm (a^440) as a function of Chi a plus divinyl Chi a for the Arabesque 2
cruise, calculated using three different {^correction algorithms (see the text for details). The plotted
curves are hyperbolic tangent functions which have the general form y = mnxl(n + mx) [see equation
(2)] and the coefficients for m and n are given for each line. Line (1), ^-correction algorithm of
Hoepffner and Sathyendranath (1992,1993) (dotted line, open diamonds), m = 0.125, n = 0.595; line
(2), P-correction algorithm based on the presence or absence of divinyl Chi a (dashed line, open triangles), m = 0.088, n = 0.171; line (3), p*-correction algorithm based on % divinyl Chi a (solid line,
filled diamonds), m = 0.105, n = 0.845.
these high Chi a samples contained very low proportions of divinyl Chi a (<5%
of total Chi a), as well as relatively high concentrations of 19'-butanoyloxyfucoxanthin, 19'-hexanoyloxyfucoxanthin and fucoxanthin, suggesting a mixed
assemblage of crysophytes, prymnesiophytes, diatoms and prochlorophytes. In
cases such as these, the application of a single P-correction algorithm for an entire
sample is clearly not appropriate. This problem was minimized by using two
different P-correction algorithms for different fractions of the sample, as
described earlier. When comparing these results to the first approach, lower
absorption coefficients were noted at low Chi a concentrations (oligotrophic
stations comprised predominantly of small cells), but virtually no difference was
found for samples with relatively high Chi a concentrations (Figure 3). We feel
that this latter approach provides more accurate absorption estimates than the
other two approaches, especially for samples with diverse phytoplankton populations. Therefore, samples from all three cruises were corrected for path length
amplification using a p-correction algorithm based on the percentage of divinyl
Chi a present in the sample.
Variability in specific absorption
Specific absorption coefficients [a^h(440)] varied widely throughout the three
cruises, but the magnitude was dependent upon the method of normalization.
Traditionally, specific absorption coefficients were obtained by normalizing
196
Pigments, species composition and absorption spectra
Table L Mean phytoplankton specific absorption coefficients at 440 and 676 nm [a*h(X) m2 (mg Chi
a)"1 ± 1 SD], for samples from three different cruises (Vancouver, Arabesque 1 and Arabesque 2),
normalized to the sum of Oil a plus divinyl Chi a measured by high-performance liquid
chromatography (HPLC Chi a) or normalized to Chi a measured fluorometrically with a Turner
Design fluorometer (Fluor. Chi a). Results are also presented for samples normalized to the sum of
Chi a plus phaeopigments measured by HPLC or Turner fluorometer
a*h440 (HPLC Chi a)
<$,440(HPLCChla + phaeo)
a*k 440 (Fluor. Chi a)
a?A440(Fluor.Chla + phaeo)
a;* 676 (HPLC Chi a)
a;,,676(HPLCChla + phaeo)
<$, 676 (Fluor. Chi a)
<$, 676 (Fluor. Chla + phaeo)
Vancouver
Arabesque 1
Arabesque 2
0.046 ±0.013
0.041 ±0.011
0.037 ±0.008
0.028 ±0.005
0.028 ±0.054
0.024 ±0.005
0.022 ±0.003
0.017 ±0.002
0.068 ±0.015
0.063 ±0.014
0.045 ±0.012
0.037 ±0.009
0.029 ±0.006
0.028 ±0.005
0.020 ±0.005
0.016 ±0.003
0.099 ±0.017
0.095 ±0.016
0.089 ±0.014
0.068 ±0.012
0.039 ±0.007
0.037 ±0.006
0.035 ±0.005
0.027 ±0.004
absorption to fluorometrically determined Chi a as this is routinely measured on
most oceanographic cruises. Several authors use normalization to the sum of Chi
a and phaeopigments (Babin et ai, 1993; Sosik and Mitchell, 1995), since both
pigments are extracted by the solvents and contribute to aph(k). Standard fluorometric techniques, however, are known to over- or underestimate Chi a, depending on the presence of Chi b or c or phaeopigments. High concentrations of Chi
c, associated with diatoms, have been shown to cause overestimates in the Chi a
concentration (Lorenzen, 1981; Bianchi et al, 1995), while the presence of Chi b
causes an underestimate in Chi a and an overestimate in phaeopigment concentrations (Gibbs, 1979; Vernet and Lorenzen, 1987). In this study, we have normalized absorption values to HPLC-derived Chi a, but have presented results for
the other techniques (Table I) for comparative purposes. In general, the lowest
specific absorption estimates were obtained by normalizing absorption to fluorometrically derived Chi a, presumably because fluorometric methods overestimate
Chi a in the presence of diatoms, resulting in lower specific absorption coefficients. The highest specific absorption estimates were obtained by normalizing to
HPLC-derived Chi a. Relative differences were smallest for the Arabesque 2
cruise, which also had fewer diatoms. Including phaeopigments in the normalization further decreased the specific absorption coefficients. These variations
should be borne in mind when making comparisons with values from the literature. Regardless of the method of normalization, the Vancouver Island cruise had
the lowest mean a*A(440) and the Arabesque 2 cruise the highest, which is consistent with the presumed cell-size distribution between cruises. Smaller cells,
such as found in the Arabesque 2 cruise, would be expected to have a higher
specific absorption coefficient than the larger diatoms found in the Vancouver
Island cruise, because of lower pigment packaging effects.
Decomposition of phytoplankton absorption spectra
The parameters of the rectangular hyperbolic function [equation (2)] relating
peak height at the center of each Gaussian band to the corresponding pigment
197
VStnart et aL
Table IL Values of the parameters pm(X) (maximum value of absorption in nr') and p£(X) [maximum
specific peak height in m2 (mg Chi a)-1] for the rectangular hyperbola equation [equation (2)] relating
the peak height of the major Gaussian bands associated with chlorophylls a, b and c and total
carotenoids to the respective pigment concentration. Data are presented for the three cruises
(Vancouver, n = 34; Arabesque 1, n = 109; Arabesque 2, n = 95). The coefficient of determination (r2)
is also given
Band
center
384 (Chi a)
413 (Chi a)
440 (Chi a)
623 (Chi a)
676 (Chi a)
700 (Chi a)
464 (Chi b)
655 (Chi b)
461 (Chi c)
583 (Chi c, Chi a)
644 (Chic)
490 (carot)
532 (carot)
Vancouver
Pm
PX.
r*
Pm
Pi,
5.403
0.421
1.365
0.0321
0.0200
0.0338
0.0061
0.0239
0.0024
0.2050
0.1168
0.0938
0.0337
0.0281
0.0285
0.0199
0.90
0.90
0.92
0.94
0.95
0.83
0.82
0.76
0.97
0.97
0.99
0.95
0.95
1.260
0.193
0.888
0.0474
0.0274
0.0524
0.0059
0.0275
0.0023
0.0774
0.0338
0.1289
0.0265
0.0244
0.0368
0.0183
-
2.043
0.037
_
-
0.912
—
6.777
6.600
-
Arabesque 2
Arabesque 1
_
0.396
0.133
_
_
0.108
0.035
0.030
0.302
0.077
Pm
0.91
0.92
0.93
0.97
0.97
0.88
0.55
0.54
0.81
0.90
0.88
0.94
0.94
P%
r*
8.529 0.0616 0.80
0.543 0.0358 0.75
0.538 0.0825 0.88
0.0066 0.82
0.0370 0.92
0.008 0.0025 0.62
0.096 0.1077 0.73
_
0.0373 0.86
0.088 0.2980 0.65
0.042 0.0497 0.73
0.029 0.0523 0.77
0.090 0.0814 0.90
0.129 0.0285 0.81
Range of Chi a concentrations: Vancouver cruise, 0-15 mg nr 3 ; Arabesque 1 cruise, 0-4.4 mg nr 3 ;
Arabesque 2 cruise, 0-1.7 mg nr 3 .
concentration are given in Table II, and the results of the four major bands associated with HPLC-derived Chi a plus divinyl Chi a are presented in Figure 4. For
samples with extremely high pm(X) values (virtually no saturation), a linear
regression (Model I) was used to fit the data. In all these cases, the r2 value of the
linear function was virtually identical to that of the hyperbolic tangent function.
Non-linearity between absorbance and Chi a, as well as differences in slopes
between cruises, may be ascribed to packaging effects. For Chi a, the package
effect was clearly evident at 440 nm, and to a lesser extent at 413 and 676 rim,
with the Arabesque 2 cruise always having the highest specific absorption
coefficient (least package effect) and the Vancouver Island cruise the lowest. The
most striking observation, however, was the apparent lack of any packaging
effects at 623 nm. At this Gaussian band, a linear regression was fitted to the data
from all three cruises due to unrealistically high saturation levels for the hyperbolic tangent curves. The resulting slopes [p£(623) values] of all three cruises
were very similar, indicating little variation in the specific absorption coefficients.
Data from all three cruises were combined to obtain an overall estimate of the in
vivo specific absorption coefficient for Chi a at 623 nm, yielding a maximum
specific absorption coefficient (p* (623)] of 0.0061 (r2 = 0.96; n = 237). This specific
absorption coefficient, which is applicable to a wide range of phytoplankton
assemblages, could be used to predict accurately the Chi a concentration in all
three cruises from the peak height of the Gaussian band alone (Figure 5). The
calculated Chi a concentrations all fell remarkably close to the 1:1 line (Y =
1.0011AT; r2 = 0.97; n = 235). Since the 95% confidence limits of the slope
(0.9779-1.0234) encompass the 1:1 line (see Sokal and Rohlf, 1969), this implies
198
Pigments, species composition and absorption spectra
0.08
0.04
•
•
'
•
1
•
•
•
• ,r
/
0.06 --
0.03
y
6
X 0.04
0.02
a
-
0.02
0.01
.
0.00
0.00
0.0
0.5
1.0
0.0
1.5
0.5
.
.
.
1
1.0
.
.
.
.
I
1.5
chl-a + DV chl-a (mg m'3)
chl-a + DV chl-a (mg m"3)
0.04
0.03 -
0.02 -
0.01 -
0.00
0.0
chl-a + DV chl-a (mg m )
05
1.0
1.5
chl-a +DV chl-a (mg m"3)
Fig. 4. Peak height [p(X)] of the four major Gaussian bands associated with Chi a, plotted as a function of the sum of Chi a plus divinyl Chi a for the three different cruises. Curves are fitted to the data
using a rectangular hyperbola function (see Table II for values of the parameters). Vancouver cruise,
open squares, long dashed line; Arabesque 1 cruise, solid circles, solid line; Arabesque 2 cruise, open
triangles, short dashed line.
that our calculated Chi a values neither over- or underestimate the 'true' (HPLC)
Chi a concentration.
Specific absorption coefficients for Chi b did not show the anticipated trends
in pigment packaging, since the highest specific absorption coefficients were
found for the Vancouver Island cruise and the lowest for the Arabesque cruises
(Table II). Since there was a high degree of overlap in the Gaussian bands of Chi
b and c, one might expect a poor estimate of Chi b in samples with very high concentrations of Chi c (such as found in the Vancouver Island cruise). This may be
due to co-varying absorption by more than one pigment which is not adequately
separated by the decomposition procedure. In addition, poor fits were generally
obtained for Chi b in the Arabesque cruises, so these results should be treated
with caution.
For all three Gaussian bands associated with Chi c, the highest p%,(\) values
were obtained for the Arabesque 2 cruise, indicating low package effects, while
199
VStuart et aL
0.5
1.0
1.5
2.0
chl-a + DV chl-a (mg m"3)
Fig. 5. Calculated Chi a versus measured (HPLC) Chi a (sum of Chi a plus divinyl Chi a) for the combined data set (n = 237). Chi a was calculated from the peak height of the Gaussian band centered
around 623 nm, using a specific peak height pX,(623) = 0.0061. The 1:1 relationship is indicated with
a dashed line. Vancouver cruise, open squares; Arabesque 1 cruise, solid circles; Arabesque 2 cruise,
open triangles.
the Arabesque 1 and Vancouver Island cruises showed a similar degree of packaging (Table II). Comparable trends in packaging were also found at the two
carotenoid absorption bands, centered around 490 and 532 nm (Table II). It
should be noted that the Gaussian band centered around 583 nm may be associated with both Chi a and Chi c (Hoepffner and Sathyendranath, 1991), so variations in this band may also be attributed to changes in the relative proportions of
these two pigments.
Estimation of the magnitude of the package effect
Relative changes in the package effect with wavelength can be computed from
the difference in the maximum specific absorption coefficients (Table II) between
the Arabesque 2 and Vancouver Island samples at each Gaussian band (Figure
6). Comparing these two cruises, it is obvious that package effects were more
pronounced in the blue region of the spectrum corresponding to Gaussian bands
of maximal absorption (400-500 nm) and were lowest for the Gaussian bands of
minimal absorption (623 and 700 nm).
In Figure 7, we have plotted specific peak height measurements of Chi a in the
blue and red regions of the spectrum (both normalized to the peak height at 623
nm) for all three cruises. Samples from the three cruises were clearly separated
on the basis of the blue and red Chi a-specific peak heights, which can be related
to changes in the package effect with changes in cell size. The Vancouver Island
200
Pigments, species composition and absorption spectra
400
450
500
550
600
650
700
750
Wavelength (nm)
Fig. 6. Relative change in the maximum specific absorption coefficients between the Arabesque 2 and
Vancouver cruises, reflecting the amount of flattening at each Gaussian band.
5 -
2
3
4
5
6
7
8
9
p(676)/p(623)
Fig. 7. Peak height of the Gaussian band centered around 440 nm plotted against that at 676 run (both
normalized to the peak height at 623 nm). Superimposed are two curves representing the theoretical
flattening effect of particles of different sizes in suspension (0.5-50 um in diameter). The dotted line
was calculated using a blue-to-red ratio of 2.0, p%, = 0.10 and pm = 2 X 105 (nr1)- The solid line was
calculated using a blue-to-red ratio of 2.5, p?o/ = 0.084 and pcm = 2 X 105 (nr1)- Vancouver cruise, open
squares; Arabesque 1 cruise, solid circles; Arabesque 2 cruise, open triangles.
samples, composed of large diatoms, had a low specific absorption due to high
package effects, in contrast to the small cells found in the Arabesque 2 samples
which had a high specific absorption and minimal package effects.
To examine whether these variations can indeed be ascribed to the package effect
from a theoretical point of view, we used the equations of Duysens (1956) and van
de Hulst (1957) which were developed for homogeneous spherical particles (see
also Kirk, 1975; Morel and Bricaud, 1981; Sathyendranath et al., 1987). According
201
VStnart et aL
to van de Hulst (1957), the dimensionless efficiency factor for light absorption of
the cell, Qa(X) (which is defined as the ratio of energy absorbed to energy impinging on the geometrical cross-section of a particle), can be calculated as:
P2
(3)
where p' is the product of the absorption coefficient of the cellular material, acm,
and the cell diameter, d:
P' = dacm
(4)
The package effect, denned as Q*, can be computed from the ratio of the specific
absorption coefficient of particles in suspension, a*h, to the specific absorption
coefficient of the same cellular material dispersed in solution, a%{:
(5)
asol
Q*a is always less than 1 and tends toward 1 (no package effect) when p' is small
(i.e. the particles are small, or a^, is small) and towards zero when p' is large
(Sathyendranath et aL, 1987). By substituting specific peak height coefficients for
specific absorption coefficients in equations (3)^(5), we can calculate the theoretical peak heights at 440 and 676 nm for a range of particle sizes (0.5-50 urn). We
used the following notations for peak height coefficients instead of absorption coefficients: p?o/(X) (the specific peak height coefficient of Chi a in solution) instead of
afoh PanQC) (the specific peak height coefficient of cellular material) instead of a^,
andp*(X) (specific peak height coefficient of particles in suspension) instead ota*h.
We can thus define the specific peak heights at 440 and 676 nm as follows:
P*(440) = e*(440)pjU440)
(6)
and
„,_
p*(676) =
eg(440)p*K440)
((440)/PK676))
(7)
where pJo/(440)/pTO/(676) is the blue-to-red ratio of Chi a peak heights in solution.
For our calculations, we tentatively set p ^ , = 2 x 105 (nr 1 ), which is within the
range given by Morel and Bricaud (1981). A theoretical curve was then generated by calculating p*(440) and p*(676) values for a range of particle diameters
(0.5-50 nm) with p*o/ = 0.084 and pJO/(440)/p,oX676) = 2.0. A similar curve was
also generated withp*,, = 0.10 andpro/(440)/p*,/(676) = 2.5. These two theoretical
curves have been superimposed on the data in Figure 7. Note that the two curves
envelop most of the data.
202
Pigments, species composition and absorption spectra
The values we assigned were somewhat speculative and similar curves can be
derived with a variety of combinations of p%i, /?<„, and psoi(,440)lpsoi(676) values.
It is interesting to note that all the Arabesque 1 samples that fell outside the
theoretical curve on the left-hand side of the plot were from the Gulf of Oman
region. These samples appear to have an enhanced blue absorption compared to
that in the red region of the spectrum, which may be attributed to the high proportion of phaeopigments found at many stations in this area. In addition, high
concentrations of cyanobacteria, which harvest light using phycobiliproteins,
were found in the Gulf of Oman during the Arabesque 1 cruise (Burkill and
Tarran, 1998). Recent studies have shown that Synechococcus spp. are efficient
absorbers of blue-green light, especially strains containing phycourobilin, even
though the absorption maximum of the biliprotein is located at longer wavelengths (Ikeya et ai, 1994). The presence of large numbers of cyanobacteria at
these stations may thus also contribute to the enhanced blue absorption.
Some of the scatter in our data may be due to methodological errors, such as
incomplete separation of peaks by the Gaussian decomposition method, random
errors in the measurement of light absorption and pigment concentration (see
below), or differences in the P-correction factor for various phytoplankton assemblages. In addition, some variation may be due to species-specific differences in
absorption characteristics. For example, the blue-to-red ratio of pure divinyl Chi
a in solvent (diethyl ether) is higher than that for Chi a (Moore et al, 1995) so
the absorption characteristics of prochlorophytes would be different from those
of other phytoplankton species.
Since the decomposition method effectively separates out the effects of various
pigments, we can estimate the actual magnitude of the package effect on the Chi
a absorption peaks by comparing the mean peak heights of the Arabesque 2 and
the Vancouver cruises. Any difference in peak heights may be attributed to
changes in the packaging effect, since absorption by other pigments has been systematically removed. The average peak heights of the two major Chi a Gaussian
curves (normalized to the peak height at 623 nm) are shown in Figure 8 for each
400
Wavelength (nm)
Fig. 8. Mean Gaussian curves of the Chi a bands centered around 440, 623 and 676 nm, normalized
to the peak height of the Gaussian band centered around 623 nm, for each of the three cruises. Vancouver cruise, long dashed line; Arabesque 1 cruise, solid line; Arabesque 2 cruise, short dashed line.
203
VStuart et aL
of the three cruises. Package effects were most pronounced for samples from the
eutrophic waters around Vancouver Island and were minimal for the oligotrophic
Arabesque 2 cruise. Similar trends in packaging effects have also been observed
by Bricaud et al. (1995).
Comparing the Arabesque 2 cruise to the Vancouver Island cruise, we estimated that there was a 58% difference in average peak heights of Chi a at 440
nm (i.e. Q* = 0.42 assuming zero flattening for the Arabesque 2 cruise) and a 38%
difference at 676 nm (Q* = 0.62 assuming zero flattening for the Arabesque 2
cruise). Package effects of this magnitude can be attained theoretically using
equations (3)-(7), assuming a p*o, value of 0.1, a BIR ratio of 2 and pcm = 2 X 10s
(m -1 ). According to these calculations, a spherical cell of 14.5 um in diameter
would have a package effect (£?$) of 0.42 at 440 nm and 0.62 at 676 nm, which
matches our estimates of the package effect using peak height measurements of
Chi a. In other words, flattening alone can theoretically account for the observed
differences in peak heights between the cruises. Changes in pigment composition,
however, will be reflected in variations in the total absorption coefficient.
Errors associated with measurements
Errors associated with HPLC pigment determination and light absorption
measurements are shown in Table III (standard deviation expressed as a percentage of the mean value). In general, errors associated with all these measurements were <10%, suggesting that at least 90% of the observed variation in
absorption and peak height measurements can be explained by changes in the
magnitude of the package effect or changes in pigment composition. Note that
the difference in mean specific absorption coefficients between the Arabesque 2
Table 00. Estimates of the errors (standard deviation, a) associated with 10 replicate measurements
of HPLC pigment analysis, phytoplankton absorption and specific absorption, Gaussian peak height
and specific peak height determinations. Experiments were carried out using a diatom culture
(Chaetoceros sp.) and a seawater sample. Sample standard deviations have been expressed as a
percentage of the mean value (see the text for details)
Measurement
HPLC pigment
Chla
Chic
Phaeopigments
Fucoxanthin
Absorption
opA(440)
a *(676)
Specific absorption
aJ/,(440)
a*/,(676)
Peak height
p(440)
p(676)
Specific peak height
p*(440)
p*(676)
204
Chaetoceros % a
Seawater % a
5.57
4.30
8.44
6.23
6.74
3.02
5.75
8.30
4.88
3.41
7.05
5.27
7.41
6.53
9.73
8.44
5.68
3J8
7.39
4.88
7.95
6.51
9.97
8.27
Pigments, species composition and absorption spectra
and Vancouver cruises (Table I) varied by a factor of 2.15, which is considerably
greater than the potential error in measurements.
Pigment versus package effects
Since packaging effects were close to zero at 623 nm, the relative changes in
absorption between cruises can be observed by normalizing the absorption spectra
to aph(623) (Figure 9). As anticipated, the Vancouver Island cruise, comprised
mainly of large diatoms, had the lowest absorption after normalization at 623 nm,
while the Arabesque 2 cruise (small cells) had the highest. There were also some
small, but systematic, differences between cruises: both Arabesque cruises exhibited higher absorption around 650 nm, as a result of absorption by Chi b plus
divinyl Chi b, while the Vancouver Island cruise showed an increased absorption
around 545 nm, associated with the high fucoxanthin concentrations (Anderson
and Barrett, 1986). In addition, the blue absorption maximum in the Arabesque 2
cruise was somewhat shifted towards the red, as a result of high concentrations of
divinyl Chi a. Differences between the cruises were greatest in the blue region of
the spectrum and somewhat less in the red, which is in accordance with the changes
in p*, values (see Figure 6). These differences represent the combined effects of
packaging as well as differences in pigment composition between cruises.
To estimate the proportion of the variability of a*£440) that could be attributed
to changes in pigment composition alone, we performed a multiple regression
analysis on the combined data set, taking a*h(440) as the dependent variable and
the concentrations of all the major pigments normalized to Chi a as the independent variables (see also Sathyendranath et al, 1987). We found that a maximum
of 42% of the variability in the specific absorption coefficient could be explained
by the following pigments (all per unit concentration of Chi a): Chi b, 19'-butanoyloxyfucoxanthin and total non-photosynthetic carotenoids (NPC). To avoid biased
data, we repeated this analysis using samples with a Chi a concentration <1.5 mg
400
430
500
550
600
650
700
Wavelength (nm)
Fig. 9. Average absorption spectra for each cruise normalized to 623 nm. Vancouver cruise, long
dashed line; Arabesque 1 cruise, solid line; Arabesque 2 cruise, short dashed line.
205
VSrnart et al
m~3, and found that these pigments could explain a maximum of 29% of the variation in the specific absorption coefficient at 440 nm. Since we know that the combined effects of spectral flattening and variation in pigment composition can
account for almost all (-96%) of the variation in absorption coefficients (Sathyendranath et al., 1987), we can assume that the remaining 58-71% of the variability
in our samples must be attributed to changes in the packaging effect. Another way
of estimating the relative contribution of accessory pigments to absorption at 440
nm is from the difference between the phytoplankton absorption at 440 nm and
the Gaussian peak height of Chi a absorption at 440 nm. For all three cruises combined, accessory pigments were found to be responsible for a mean of 22.6% (±
2.6%) of the total absorption at 440 nm (range 14-31%), which is slightly lower
than the previous estimate, suggesting an even greater contribution to the package
effect at this wavelength. The role of accessory pigments, however, is more apparent at wavelengths associated with absorption by carotenoids i.e. p(490) and
/?(532) (see below).
Samples from the three cruises were further separated into two groups, representing predominantly small and large cells. For our data set, the large cells were
generally diatoms, with high fucoxanthin:Chl a ratios (usually >0.2). We used this
ratio to distinguish between samples comprised predominantly of large diatoms
(fucoxanthin:Chl a > 0.2) and those containing other (smaller) cells (fucoxanthin:Chl a < 0.2). Despite the fact that fucoxanthin was also found in other
classes of phytoplankton such as prymnesiophytes and crysophytes, this distinction appeared to separate the data adequately into two groups, as can be seen from
the example for the carotenoid Gaussian band (centered around 490 nm) shown
in Figure 10. Specific absorption coefficients for both size groups are presented in
0.04
0.00
0.2
0.4
0.6
0.8
1.0
1.2
3
Total carotenoids (mg m* )
Fig. 10. Peak height of the Gaussian band centered around 490 nm [p(490)] plotted as a function of
the total carotenoid concentration for the three different cruises separated on the basis of the fucoxanthin:Chl a ratio. Open symbols and dashed line represent samples with a ratio < 0.2 (i.e. small
cells) and solid symbols and solid line represent samples with a ratio > 0.2 (i.e. large cells). Curves are
fitted to the data using a rectangular hyperbola function (see Table IV for values of the parameters).
Vancouver cruise, squares; Arabesque 1 cruise, circles; Arabesque 2 cruise, triangles.
206
Pigments, species composition and absorption spectra
Table IV. These specific absorption coefficients are thus representative of the two
extremes of natural phytoplankton populations, i.e. small cells exhibiting limited
pigment packaging effects or large cells a high degree of pigment packaging. For
the low fucoxanthin samples, a linear regression provided the best fit to the data
for all the Gaussian bands associated with Chi a (Table IV), suggesting either that
the flattening effect had been removed, or that all samples had the same degree
of flattening. This also implies that our criteria for separating samples into two size
groups were effective. Differences in the maximum specific peak heights (p^Jj)
between these two data sets also give some indication of the relative degree of flattening at the various Gaussian bands. The greatest relative differences occurred
at the carotenoid Gaussian band centered around 490 nm, and also for the Chi c
Gaussian band centered around 461 nm, which may be related, in part, to differences in the intracellular absorption properties of the pigments within a Gaussian
band. This is examined in more detail in the following section.
Variations in carotenoid composition
In principle, variations in absorption characteristics due to changes in pigment
composition are not an issue when we are dealing with specific absorption coefficients of individual pigments. However, small changes in absorption characteristics within a Gaussian band as a result of variations in pigment composition
should not be overruled, especially in those bands assigned to absorption by
carotenoids. A wide variety of carotenoids are found amongst the algae, each
with different absorbing properties, depending on their chemical structure.
Carotenoids serve two distinct functions within the cell: some are involved in
Table IV. Values of the parameterspm(X) (maximum value of absorption in nr 1 ) and pm(\) [maximum
specific peak height in m2 (mg Chi a)-'] for the rectangular hyperbola equation [equation (2)] relating
the peak height of each Gaussian band to the corresponding pigment concentration. Data are
presented for samples with a fucoxanthin:Chl a ratio < 0.2 (representing small cells, n = 160) and those
with a ratio > 0.2 (representing larger diatom cells, n = 78). The coefficient of determination (r2) is also
given
Band center
384 (Chi a)
413 (Chi a)
440 (Chi a)
623 (Chi a)
676 (Chi a)
700 (Chi a)
464 (Chi b)
655 (Chi b)
461 (Chi c)
583 (Chi c, Chi a)
644 (Chi c)
490 (carot)
532 (carot)
fuco:Chl a < 0.2
Pm
Pt,
_
-
0.057
0.032
0.067
0.006
0.033
0.002
0.119
0.040
0.222
0.037
0.038
0.074
0.023
0.025
0.019
0.189
0.020
0.043
0.117
-
fuco:Chl a>0.2
0.73
0.75
0.76
0.84
0.82
0.70
0.50
0.82
0.59
0.59
0.64
0.73
0.72
Pm
PS.
1.809
0.354
0.811
_
1.759
0.037
_
_
0.268
0.110
0.089
0.508
0.284
0.037
0.022
0.042
0.006
0.024
0.002
0.130
0.068
0.117
0.036
0.031
0.031
0.021
0.90
0.92
0.91
0.95
0.96
0.88
0.62
0.43
0.84
0.71
0.742
0.93
0.79
Range of Chi a concentrations: Vancouver cruise, 0-15 mg nr 3 ; Arabesque 1 cruise, 0-4.4 mg nr 3 ;
Arabesque 2 cruise, 0-1.7 mg nr 3 .
207
V.Sruart el aL
harvesting light (i.e. photosynthetic carotenoids), while others are involved in
protecting the cells against photo-oxidation (i.e. non-photosynthetic or photoprotective carotenoids) (Rowan, 1989). For our data set, there was a much higher
proportion of NPC at low ambient Chi a concentrations than in chlorophyll-rich
waters (Figure 11), as has also been observed by Bricaud et al. (1995) (we have
assumed that zeaxanthin, alloxanthin, P-carotene, diadinoxanthin and diatoxanthin are NPC, and that fucoxanthin, peridinin, 19'-butanoyloxyfucoxanthin, 19'hexanoyloxyfucoxanthin, prasinoxanthin and a-carotene are photosynthetic
carotenoids, as in Bricaud etaL, 1995). The converse also holds true: samples from
eutrophic waters have a relatively high proportion of photosynthetic carotenoids.
Approximately 90% of the total carotenoids from the Vancouver Island samples
(mainly fucoxanthin) were involved in harvesting light.
To examine the influence of the major carotenoids on the peak height of the
Gaussian band centered around 490 run b(490)], we used a multiple linear regression equation on the combined data set, omitting samples with Chi a concentrations >1.5 mg m~3 to avoid biased data. We found that the concentrations of
fucoxanthin (fuco), zeaxanthin (zea) and 19'-hexanoyloxyfucoxanthin (hex) were
able to explain 68% of the variance in the Gaussian peak height centered around
490 nm [p(490) = 0.003 + 0.028fuco + 0.096zea + 0.050hex, r2 = 0.68, n = 213].
Similar results were obtained for the carotenoid absorption band centered
around 532 nm \p(532) = 0.001 + 0.021fuco + 0.043zea + 0.015hex, r2 = 0.70, n =
213]. At both Gaussian bands, the coefficient for zeaxanthin (a photoprotective
pigment) was much higher than that of the photosynthetic carotenoids, fucoxanthin and 19'-hexanoyloxyfucoxanthin, suggesting that it may also have a much
higher specific absorption. This is in agreement with the known extinction coefficients of these carotenoids in ethanol, where zeaxanthin has a higher extinction
coefficient (254 1 gr1 cm"1) than fucoxanthin(145 1 g"1 cm"1) (Rowan, 1989). This
would help to account for the high specific absorption coefficients obtained for
the Arabesque 2 samples, which were comprised predominantly of small cells
chl-a + DV chl-a (mg m'3)
Fig. LL Ratio of non-photosynthetic carotenoids (NPC) to total carotenoids as a function of total Chi
a (sum of Chi a and divinyl Chi a) for the three different cruises. Vancouver cruise, open squares;
Arabesque 1 cruise, solid circles; Arabesque 2 cruise, open triangles.
208
Pigments, species composition and absorption spectra
with high proportions of zeaxanthin, and would also explain the large difference
in p£(490) values between the low and high fucoxanthin data sets (Table IV).
Changes in the composition of the carotenoids, or more specifically, in the relative
proportions of non-photosynthetic and photosynthetic carotenoids, were thus
responsible for some of the variations in the specific absorption coefficients at this
Gaussian band. However, it should be borne in mind that since there is a strong
(inverse) relationship between the proportion of NPC and Chi a concentration
(Figure 11), and between chlorophyll concentration and cell size (large cells are
generally found in eutrophic waters; Yentsch and Phinney, 1989), there is also the
underlying influence of cell size (and thus packaging effect) on the relative proportions of photosynthetic and non-photosynthetic pigments.
Variations in the Chi c absorption band
Chlorophyll c is composed of a mixture of Chi (cj+ c2), c3 and Mg-D (Mg 2,4divinyl phaeoporphyrin a5 monomethyl ester). Our HPLC method was not
capable of separating Chi (cj + c2) from Mg-D, but the latter has only been found
in a few prasinophytes, so was not considered to be important. Chlorophyll c is
commonly found in prymnesiophytes and crysophytes (and a few species of
diatoms), while one or both forms of Chi q or c2 are found in all classes of the
Chromophyta (except the Eustigmatophyceae) (Rowan, 1989). When comparing
absorption spectra of pigments eluting from the HPLC column (using a Beckman
UV/Vis 167 detector), we found that the absorption maximum of Chi c3 in the
blue was red-shifted by -10 nm when compared to that of Chi (q + c2), indicating that these pigments have different spectral characteristics which may be
reflected in differences in the absorption bands assigned to Chi c.
We used a multiple linear regression equation on the combined data set (omitting samples with high Chi a concentrations, >1.5 mg m~3, to avoid biased data)
to assess the relative importance of Chi (ci+ c2) and c3 on the peak height at 461
nm, and found that changes in the concentration of these two pigments were able
to explain 45% of the variation in peak height [p(461) = 0.008 + 0.057Chl ci+2 +
O.llOChl c3, n = 213; r2 = 0.45], with similar results being obtained for the Chi c
Gaussian band centered around 644 nm. The higher coefficient for Chi c3 suggests
a higher specific absorption than for Chi (ci + c2), and implies that it may have a
significant effect on the in vivo absorption characteristics, even when present in
small amounts. However, if we omitted Chi c3 from the multiple linear regression, we found that the r2 values were only reduced by 2-4%, implying that the
role of Chi c3 in influencing the absorption coefficients was minimal.
Discussion
Pigment-specific absorption coefficients
Pigment-specific absorption coefficients for Chi a [/?m(A.)] were relatively high for
samples from the Arabesque 2 cruise and for samples with low fuco:Chl a ratios
(Tables II and IV). However, the discrepancies caused by different methods of
measuring and normalizing to Chi a (Table I) should be borne in mind. We
209
VStuart et al
normalized our values to the concentration of HPLC-derived Chi a, but if we
included phaeopigments in this normalization our values were somewhat
reduced, suggesting that phaeopigments were partially responsible for these high
specific absorption values. Our values were nevertheless in the same range as the
Pm(X) values recorded by Lutz et al. (1996) for samples from the NW Atlantic,
using a similar decomposition method, and were also comparable to specific
absorption coefficients of small phytoplankton cells in the literature. Stramski
and Morel (1990) recorded a*^440) values of 0.04-0.18 m2 mg-1 Chi a for the
cyanobacteria Synechocystis, while Morel et al. (1993) noted that the specific
absorption of phytoplankton at 678 nm could reach -0.04 m2 mg"1 for Synechococcus strains. These high values were attributed to the influence of accessory
pigments (carotenoids and biliproteins) and it was suggested that, for these small
cells, pigment composition plays a more important role in determining the
specific absorption coefficients than pigment packaging effects. Moore et al.
(1995) reached a similar conclusion when examining the influence of light levels
on absorptive properties oiSynechococcus and Prochlorococcus. They attributed
the observed variability in the specific absorption coefficients to photoacclimative changes in pigment ratios and not to pigment package effects (which are
minimal for these small cells). Package effects are, however, more important in
large cells, such as the diatoms in the Vancouver Island cruise.
Pigmentation effects
Multiple linear regression analysis suggested that changes in the composition of
accessory pigments were responsible for only 29-42% of the variability in the
total absorption by phytoplankton at 440 nm, with packaging effects playing a
dominant role in this region of the spectrum. Furthermore, Gaussian decomposition attributed an even smaller role to pigment composition. Only 14-31% of
the total phytoplankton absorption at 440 nm was due to pigments other than Chi
a. However, the contribution of accessory pigments was more apparent at Gaussian bands associated with absorption by carotenoids [p(490) andp(532)]. At these
wavelengths, three major carotenoids (fucoxanthin, zeaxanthin and 19'-hexanoyloxyfucoxanthin) were able to account for -70% of the variation in peak
heights at these Gaussian bands. These carotenoids are characteristic of diatoms,
prokaryotes and prymnesiophytes, respectively, and are also representative of the
dominant classes of phytoplankton cells found in the Vancouver Island cruise,
Arabesque 2 cruise and Arabesque 1 cruise, respectively.
Variations in specific absorption coefficients at the carotenoid absorption
bands were also attributed to changes in the relative proportion of photosynthetic
and non-photosynthetic pigments. High specific absorption coefficients were
found to be associated with samples from the oligotrophic waters of the
Arabesque 2 cruise, which contained high proportions of non-photosynthetic
carotenoids pigments such as zeaxanthin. However, we should not rule out the
possibility that there may also be species-dependent differences in absorption
coefficients for various pigment-protein complexes. It is known that the
210
Pigments, species composition and absorption spectra
composition of the chlorophyll-protein-carotenoid complexes varies considerably amongst the various classes of algae (Kirk, 1994), thus it is not surprising that
pigment-specific absorption coefficients varied by up to 3.5-fold at the carotenoid
absorption band. Coefficients obtained for one phytoplankton assemblage cannot
necessarily be applied to another population without introducing large errors.
Package effects
Laboratory experiments have shown that changes in a*h(X) are caused by two main
factors: (i) variations in the package effect and (ii) changes in the relative abundance of chlorophyll and accessory pigments (Sathyendranath et al., 1987; Bricaud
et al, 1988; Berner et al, 1989; Sosik and Mitchell, 1994). Variations in the package
effect, in turn, are also caused by two main factors: (i) changes in the size, shape or
morphology of the cells, chloroplasts and thylakoid membranes; and (ii) variations
in the intracellular pigment concentration, frequently caused by photoadaptive
responses (Morel and Bricaud, 1981; Falkowski and LaRoche, 1991; Johnsen et al.,
1994a,b; Kirk, 1994). In principle, spectral regions with high absorption (400-490
nm) are the regions with the greatest package effect, while wavelengths of low
absorption have minimal package effects. When comparing the Arabesque 2 and
the Vancouver Island cruises, we recorded a 58% difference in the mean peak
height of Chi a at 440 nm (representing a package effect of 0.42) and a 38% difference in peak heights of Chi a at 676 nm (representing a package effect of 0.62). In
addition, there was virtually no change in the maximum specific absorption for the
Gaussian band centered around 623 nm, implying a lack of a package effect for this
region of the spectrum (Table II). These results are similar to those obtained by
Geider and Osbome (1987) for a cultured Thalassiosira sp. They noted that the
package effect could reduce light absorption efficiency by -50% at the blue absorption maximum (435 nm) and by 30% at the red Chi a peak (670 nm), with the
absence of any significant flattening effect at the chlorophyll minimum (600 nm).
The apparent lack of a package effect at the absorption minimum allowed us
to use the Chi a-specific peak height at 623 nm as an effective tool for accurately
predicting Chi a concentrations using the phytoplankton absorption spectrum
alone. This may be useful when developing algorithms for estimating phytoplankton biomass from remotely sensed ocean-color data, since the optical effects
of variations in cell size can be ignored. On a practical basis, however, collecting
measurements at the red end of the spectrum is difficult because of the high attenuation of light at these wavelengths by seawater. In addition, the impact of
Raman scattering on the distribution of light may become significant at wavelengths >550 nm (Mobley, 1994) since Raman emission is a possible major contributor to the water leaving radiance at the red end of the spectrum (Mueller
and Austin, 1991). Furthermore, several species of cyanobacteria, particularly the
freshwater species such as Microcystis sp., have an absorption maximum around
620 nm associated with absorption by phycocyanin (Jupp et al., 1994) and in these
cases the use of the Gaussian band centered around 623 nm to measure Chi a
concentration would not be appropriate. For our data set, the dominant
biliprotein for the cyanobacteria appeared to be phycoerythrobilin, which has an
211
VStuart et al
absorption maximum near 545 nm (Sathyendranath et al, 1998); theoretically this
would not interfere with the use of 623 nm for Chi a calculations.
We estimated that changes in pigment composition were responsible for
29-42% of the variability in absorption at 440 nm, while package effects were
responsible for the remaining 58-71%. Several laboratory studies have indicated
that package effects are responsible for between 43 and 57% of the variability in
absorption (Sathyendranath et al, 1987; Berner et al, 1989; Sosik and Mitchell,
1994). Since our study incorporates a wide range of cell sizes, from minute
prochlorophytes (0.5 urn in diameter; Tarran et al, 1998) to large diatoms {T.aestivalis up to 45 urn in length), one would expect that the role of flattening would
be of major importance in the variability of a*/,(440). Relative changes in the red
part of the spectrum were much smaller than in the blue (Figure 6) and, since
there was only a small contribution by other pigments to the total absorption at
this Gaussian band [mean p(676)/a(676) for all three cruises was 0.92 ± 0.03 or
8% contribution by other pigments], the influence of accessory pigments is likely
to be minimal in this region of the spectrum. Sosik and Mitchell (1994) concluded
that the package effect was responsible for nearly all the variability in a^h{61€),
based on a laboratory study of cultured Dunaliella tertiolecta.
We have assumed that package effects were predominantly caused by changes
in cell size of the phytoplankton populations between the cruises, although this
may not necessarily be the case. Package effects may also be attributed to variations in the intracellular pigment concentration, often through photoacclimation
responses to shade. This is frequently accompanied by changes in cell volume and
the number and density of the thylakoid membranes (Falkowski, 1983; Falkowski
and LaRoche, 1991). In general, high-light-acclimated cells have high specific
absorption coefficients of Chi a caused by low intracellular pigment concentrations and a low package effect, in contrast to low-light-acclimated cells (Morel
and Bricaud, 1981; Johnsen et al, 1994a,b). In a study on photoadaptation and
the package effect in cultured D.tertiolecta, Berner et al. (1989) found that
changes in cell size did not significantly affect packaging, but that thylakoid stacking and the transparency of the thylakoid membranes were important factors. It
is likely that both cell size, as well as changes in intracellular pigment concentration, played a role in the observed variations in specific absorption coefficients
between cruises, although the influence of cell size probably dominated the flattening effect since there was such a large change in cell sizes between the small
prokaryotes found in the Arabesque 2 cruise and the large chain-forming diatoms
found in the Vancouver Island samples.
Implications of optical variability
In this study, we observed pronounced species-specific differences in the biooptical characteristics of phytoplankton cells from three different cruises, as well
as a high degree of variability within cruises. Such variability in the optical characteristics of phytoplankton assemblages seems to be the rule rather than the exception (Hoepffner and Sathyendranath, 1992; Bricaud era/., 1995; Sosik and Mitchell,
1995; Lutz et al, 1996; Sathyendranath et aL, 1996,1998). However, many current
212
Pigments, species composition and absorption spectra
algorithms for the retrieval of Chi a concentrations from remotely sensed ocean
color data assume constant optical properties for phytoplankton populations. This
variability in optical properties may thus be a source of noise which limits the accuracy of these algorithms. It also highlights the need for improving optical models
to incorporate some of the variations in optical characteristics of natural phytoplankton populations, such as incorporating information on changes in the package
effect or changes in the relative abundance of accessory pigments. Sosik and
Mitchell (1995) proposed the use of a bio-optical model based only on the photosynthetically active component [ap,(X)] rather than total phytoplankton absorption
[a*A(X)]. They found a smaller variance in a*s(\) since it does not include absorption by photoprotective carotenoids, which are an added source of variability. Likewise, Babin et al. (1996) found that the variations in the light absorption by
non-photosynthetic pigments were responsible for a 3-fold change in the maximum
quantum yield of carbon fixation at three sites in the NE tropical Atlantic.
The importance of variations in the optical characteristics of phytoplankton
assemblages may have further implications. For example, Sathyendranath et al.
(1998) found that the large differences in the Chl-specific absorption coefficients
between the intermonsoon (Arabesque 2) and monsoon (Arabesque 1) cruises
were responsible for substantially altering the maximum quantum yield for
photosynthesis during the monsoon season and could explain the occurrence of
the phytoplankton blooms observed in the Arabian Sea following the start of the
monsoon season. Changes in the P-I parameters alone (<xB and P%,) could not
account for the incidence of these blooms.
Concluding remarks
We have developed a method for accurately deriving Chi a concentrations from
the maximum specific peak height (p*,) of the Gaussian band centered around
623 nm. At this waveband, package effects were found to be close to zero; thus,
estimates of Chi a concentration were unaffected by variations in the absorption
coefficient caused by changes in phytoplankton composition. This information
could be used in the development of future algorithms for estimating Chi a
concentration from remotely sensed data, collected either by satellite or by biooptical sensors attached to in situ moorings or drifters (Smith et al., 1991; Abbott
et al., 1995; Cullen and Lewis, 1995). Bio-optical measurements from in situ
systems are frequently used to complement data from satellite observations and
may also provide a potential tool for testing the feasibility of using wavebands
centered around 623 nm to estimate Chi a concentration. In theory, it should be
possible to relate spectral downwelling irradiance and upwelling radiance
measurements at 623 nm to pigment concentration. Since package effects are not
a problem at this waveband, the resulting signal should not be influenced by
changes in the bio-optical properties of the phytoplankton assemblage. Besides,
the significance of colored dissolved organic matter (CDOM) would be much less
at these wavelengths than in the blue part of the spectrum.
Package effects were estimated to account for ~58% of the variability in
absorption at 440 nm and -38% at 676 nm, the remaining variability being
213
\StasatetaL
attributed to changes in pigment composition or to random errors in the measurement of pigments or absorption (5-10%). At the carotenoid absorption bands,
however, a greater part of the variation in specific peak heights (490 and 532 nm)
could be explained by changes in the relative proportions of photosynthetic
carotenoids and non-photosynthetic carotenoids. Three major carotenoids, fucoxanthin, 19'-hexanoyloxyfucoxanthin (photosynthetic carotenoids) and zeaxanthin (a photoprotective carotenoid) were found to be responsible for up to 68%
of the variability in absorption at the waveband centered around 490 nm.
Our findings highlight the need for improving bio-optical algorithms used for
Chi a retrieval from remotely sensed ocean color data. Ideally, changes in biooptical characteristics caused by variations in the package effect or changes in the
relative abundance of accessory pigments should be incorporated into these algorithms. This would decrease errors frequently associated with spatial and temporal changes in phytoplankton composition.
Acknowledgements
We thank Dr Erica Head for helpful advice with HPLC pigment determinations,
as well as two anonymous referees for their constructive comments on the
manuscript. The work presented in this paper was supported, in part, by the
Office of Naval Research, USA, the National Aeronautics and Space Administration, USA, and the Department of Fisheries and Oceans, Canada. Additional
support was provided by the Natural Sciences and Engineering Research
Council through operating grants to S.S and T.P. This work was carried out as
part of the Canadian contribution to the Joint Global Ocean Flux Study
(JGOFS). Data from the Arabesque cruises were collected during the British
cruises D 210 and D 212 to the Arabian Sea, and we thank the chief scientists
F.Mantoura and P.H.Burkill, as well as the staff and crew of RRS 'Discovery'.
Data from the Vancouver Island 'cruise' were collected from the launch
'Revisor' (Institute of Ocean Sciences, BC), and funded by the Ocean Sciences
Programs Fund - 9023. Phytoplankton samples from this cruise were identified
to species by Rowan Haigh.
References
Abbott,M.R., Brink^CH., Booth.C.R., BlascoJD., SwensonJM.S., Davis.C.O. and Codispoti,L.A.
(1995) Scales of variability of bio-optical properties as observed from near-surface drifters. /.
Geophys. Res., 100,13345-13367.
AndersonJ.M. and BarrettJ. (1986) Light-harvesting pigment-protein complexes of algae. In Photosynthesis III. Photosynthetic Membranes and Light Harvesting Systems. Encyclopedia of Plant
Physiology. Springer-Verlag, Berlin, Vol. 19, pp. 269-285.
ArbonesJ}., Figueiras,F.G. and Zapata.M. (1996) Determination of phytoplankton absorption
coefficient in natural seawater samples: Evidence of a unique equation to correct the pathlength
amplification on glass-fiber filters. Mar. EcoL Prog. Ser., 137,293-304.
Babin.M., Therriault,J.-C., Legendre.L. and Condal,A. (1993) Variations in the specific absorption
coefficient for natural phytoplankton assemblages: Impact on estimates of primary production.
LimnoL Oceanogr., 38,154-177.
214
Pigments, species composition and absorption spectra
Babin,M., Morel^A., Claustre,H., Bricaud,A., Kolber,Z. and Falkowski,P.G. (1996) Nitrogen- and
irradiance-dependent variations of the maximum quantum yield of carbon fixation in eutrophic,
mesotrophic and oligotrophic marine systems. Deep-Sea Res. I,43,1241-1272.
Berner.T., Dubinsky,Z., WymanJC. and FalkowskiJP.G. (1989) Photoadaptation and the 'package'
effect in Dunaliella tertiolecta (Chlorophyceae). /. Phycoi, 25,70-78.
Bevington,P.R. (1969) Data Reduction and Error Analysis for the Physical Sciences. McGraw-Hill,
New York.
Bianchi.T.S., Lambert,C. and Biggs,D.C (1995) Distribution of chlorophyll a and phaeopigments in
the northwestern Gulf of Mexico: a comparison between ftuorometric and high-performance liquid
chromatography measurements. Bull. Mar. Set, 56, 25-32.
Bidigare,R.R., Smith,R.C., Baker.K.S. and MarraJ. (1987) Oceanic primary production estimates
from measurements of spectral irradiance and pigment concentrations. Global Biogeochem. Cycles,
1,171-186.
Bidigare,R.R., Ondrusek,M.E., MorrowJ.H. and Kiefer,D.A. (1990) In vivo absorption properties of
algal pigments. Proc. SPIE Ocean Opt. X, 1302, 290-302.
Bricaud,A., Bedhomme,A.-L. and MorelA- (1988) Optical properties of diverse phytoplanktonic
species: experimental results and theoretical interpretation. J. Plankton Res., 10, 851-873.
Bricaud,A., Babin.M., Morel^A. and Claustre ,H. (1995) Variability in the chlorophyll-specific absorption coefficients of natural phytoplankton: analysis and parameterization. / Geophys. Res., 100,
13321-13332.
BurkillJ'.H. and Tarran.G.A. (1998) Flow cytometric analysis of microzooplankton grazing on
autotrophic picoplankton in the Arabian Sea. Deep-Sea Res., in press.
CullenJ.J. and Lewis,M.R. (1995) Biological processes and optical measurements near the sea
surface: some issues relevant to remote sensing. / Geophys. Res., 100,13255-13266.
Duysens.L.N.M. (1956) The flattening of the absorption spectrum of suspensions, as compared to that
of solutions. Biochim. Biophys. Acta, 19,1-12.
Falkowski,P.G. (1983) Light-shade adaptation and vertical mixing of marine phytoplankton: A
comparative field study. J. Mar. Res., 41, 215-237.
Falkowski J*.G. and LaRoche J. (1991) Acclimation to spectral irradiance in algae. /. Phycoi., 27,8-14.
Geider^RJ. and Osborne3-A. (1987) Light absorption by a marine diatom: experimental observations and theoretical calculations of the package effect in a small Thalassiosira species. Mar. Biol.,
96,299-308.
Gibbs.C.F. (1979) Chlorophyll b interference in the fluorometric determination of chlorophyll a and
'phaeopigments'. Aust. J. Mar. Freshwater Res., 30, 597-606.
GordonJH.R. and MorelA- (1983) Remote assessment of ocean color for interpretation of satellite
visible imagery. A review. In Barber,R.T., Mooers,N.K., Bowman,M.J. and Zeitzschel.B- (eds),
Lecture Notes on Coastal and Estuarine Studies. Springer-Verlag, New York.
HeadJE J.H. and Home, E.P.W. (1993) Pigment transformation and vertical flux in an area of convergence in the North Atlantic. Deep-Sea Res. 1,40, 329-346.
Hoepffner,N. and Sathyendranath.S. (1991) Effect of pigment composition on absorption properties
of phytoplankton. Mar. Ecol. Prog. Ser., 73,11-23.
Hoepffner,N. and Sathyendranath.S. (1992) Bio-optical characteristics of coastal waters: Absorption
spectra of phytoplankton and pigment distribution in the western North Atlantic. Limnol.
Oceanogr., 37,1660-1679.
Hoepffner.N. and Sathyendranath.S. (1993) Determination of the major groups of phytoplankton
pigments from the absorption spectra of total paniculate matter. / Geophys. Res., 98,22789-22803.
Ikeya.T, OhkiJC., Takahashi,M. and Fujita.Y. (1994) Photosynthetic characteristics of marine Synechococcus spp. with special reference to light environments near the bottom of the euphotic zone
of the open ocean. Mar. Biol., 118,215-221.
Johnsen,G., Nelson,N.B., JovineJR.V.M. and Prezelin,B.B. (1994a) Chromoprotein- and pigmentdependent modeling of spectral light absorption in two dinoflagellates, Prorocentrum minimum and
Heterocapsa pygmaea. Mar. EcoL Prog. Ser., 114,245-258.
Johnsen.G., Samset.O., GranskogJ-. and Sakshaug,E. (1994b) In vivo absorption characteristics in 10
classes of bloom-forming phytoplankton: taxonomic characteristics and responses to photoadaptation by means of discriminant and HPLC analysis. Mar. EcoL Prog. Ser., 105,149-157.
Jupp.D.L.B., KirkJ.T.O. and Harris.G.P. (1994) Detection, identification and mapping of cyanobacteria—Using remote sensing to measure the optical quality of turbid inland waters. Aust. J. Mar.
Freshwater Res., 45, 801-828.
KirkJ.T.O. (1975) A theoretical analysis of the contributions of algal cells to the attenuation of light
within natural waters. II. Spherical cells. New Phytol., 75, 21-36.
215
V-Sruart et al
KirkJ.T.O. (1976) A theoretical analysis of the contribution of algal cells to the attenuation of light
within natural waters. III. Cylindrical and spheroidal cells. New Phytol., TJ, 341-358.
KirkJ.T.O. (1994) Light and Photosynthesis in Aquatic Ecosystems. Cambridge University Press,
Cambridge.
Kishino.M, Takahashi,M., Okami,N. and Ichimura.S. (1985) Estimation of the spectral absorption
coefficients of phytoplankton in the sea. Bull. Mar. ScL, 37,634-642.
Kyewalyanga,M.K, Platt.T, Sathyendranath.S., Lutz,V.A. and Stuart.V. (1998) Seasonal variations in
physiological parameters of phytoplankton across the North Atlantic. J. Plankton Res., 20,17-42.
Lawson.C.L. and Hanson.RJ. (1974) Solving Least Squares Problems. Prentice-Hall Inc., New Jersey.
Lorenzen.CJ. (1981) Chlorophyll b in the eastern North Pacific Ocean. Deep-Sea Res., 28A,
1049-1056.
Lutz,V.A., Sathyendranath,S. and Head^J.H. (1996) Absorption coefficient of phytoplankton:
Regional variations in the North Atlantic. Mar. Ecol. Prog. Ser, 135,197-213.
Mitchell3-G. (1990) Algorithms for determining the absorption coefficient of aquatic particulates
using the quantitative filter technique (QFT). Proc. SPIE Ocean Opt. X, 1302,137-148.
Mitchell3-G. and Kiefer,D.A. (1984) Determination of absorption and fluorescence excitation
spectra for phytoplankton. In Holm-Hansen.O., BoIis.L. and Giles JR.. (eds), Marine Phytoplankton
and Productivity. Springer-Verlag, Berlin, pp. 157-169.
Mitchell3-G. and KieferJ3.A. (1988) Chlorophyll a specific absorption and fluorescence excitation
spectra for light-limited phytoplankton. Deep-Sea Res. I,35, 639-663.
Mobley.C.D. (1994) Light and Water; Radiative Transfer in Natural Waters. Academic Press, San
Diego, CA.
Moore,L.R., Goericke,R. and Chisholm,S.W. (1995) Comparative physiology of Synechococcus and
Prochlorococcus: influence of light and temperature on growth, pigments,fluorescenceand absorptive properties. Mar. Ecol Prog. Ser., 116, 259-275.
Morel,A. and Andr6,J.-M. (1991) Pigment distribution and primary production in the western
Mediterranean as derived and modeled from Coastal Zone Color Scanner observations. /.
Ceophys. Res., 96,12685-12698.
Morel.A. and Bricaud,A. (1981) Theoretical results concerning light absorption in a discrete medium,
and application to specific absorption of phytoplankton. Deep-Sea Res. 1,28A, 1375-1393.
Morel^A., Ahn,Y.-H., PartenskyJ., VaulotJD. and CIaustre,H. (1993) Prochlorococcus and Synechococcus: A comparative study of their optical properties in relation to their size and pigmentation. /. Mar. Res., 51,617-649.
Mueller,! and Austin,R. (1991) Proceedings of the NASA/GSFC SeaWiFS Project Workshop on
Ocean Optical and Other Measurements for SeaWiFS Radiometric Validation and Bio-Optical
Algorithm Development. Naval Postgraduate School, Monterey, CA.
Platt.T. and Sathyendranath.S. (1988) Oceanic primary production: Estimation by remote sensing at
local and regional scales. Science, 241,1613-1620.
Platt.T., Sathyendranath.S. and Longhurst^A. (1995) Remote sensing of primary production in the
ocean: Promise and fulfilment. Philos. Trans. R. Soc. London Ser. B, 348,191-202.
Rowan.K.S. (1989) Photosynthetic Pigments of Algae. Cambridge University Press, Cambridge.
Sathyendranath.S., LazzaraJL. and Prieur.L. (1987) Variations in the spectral values of specific absorption of phytoplankton. Limnol. Oceanogr., 32,403-415.
Sathyendranath.S., Longhurst^A., Caverhill.C.M. and Platt.T. (1995) Regionally and seasonally differentiated primary production in the North Atlantic. Deep-Sea Res. J, 42,1773-1802.
Sathyendranath,S., Platt.T., Stuart.V., Irwin3.D., Veldhuis,M.J.W., Kraay.G.W. and Harrison.W.G.
(1996) Some bio-optical characteristics of phytoplankton in the N.W. Indian Ocean. Mar. EcoL
Prog. Ser., 132, 299-311.
Sathyendranath,S., Stuart.V., Irwin3.D., Maass,H., Savidge.G., Gilpin X~ and Platt,T. (1998) Seasonal
variations in bio-optical properties of phytoplankton in the Arabian Sea. Deep-Sea Res., in press.
Smith,R.C., WatersjCJ. and Baker,K.S. (1991) Optical variability and pigment biomass in the
Sargasso Sea as determined using deep sea optical mooring data. / Ceophys. Res., 96, 8665-8686.
Sokal,R.R. and Rohlf, FJ. (1969) Biometry. The Principles and Practice of Statistics in Biological
Research. W.H.Freeman and Co., San Francisco, CA.
Sosik,H.M. and Mitchell3-G. (1994) Effects of temperature on growth, light absorption, and quantum
yield in Dunaliella tertiolecta (Chlorophyceae). /. Phycol., 30, 833-840.
SosikJrLM. and MitcheI13-G. (1995) light absorption by phytoplankton, photosynthetic pigments
and detritus in the California Current System. Deep-Sea Res. 1,42,1717-1748.
StramskiJ). and Morel^A. (1990) Optical properties of photosynthetic picoplankton in different
physiological states as affected by growth irradiance. Deep-Sea Res. 1,37,245-266.
216
Pigments, species composition and absorption spectra
Tarran,G.A., Burkill,P.H. and Edwards,E.S. (1998) Phytoplankton community structure in the
Arabian Sea, during and after the southwest monsoon, 1994. Deep-Sea Res., in press,
van de HulstJH.C. (1957) Light Scattering by Small Particles. Wiley, New York.
Vernet,M. and Lorenzen.CJ. (1987) The presence of chlorophyll b and the estimation of phaeopigments in marine phytoplankton. /. Plankton Res., 9, 255-265.
Yentsch,C.S. (1962) Measurement of visible light absorption by particulate matter in the ocean.
Limnol Oceanogr., 7,207-217.
Yentsch,C.S. and Phinney,D.A. (1989) A bridge between ocean optics and microbial ecology. Limnol.
Oceanogr., 34,1694-1705.
Received on April 4,1997; accepted on September 26,1997
111