Download - The identification of phytoplankton pigments from ...ajit/Papers/aguirre... · pigments. The spectra used are modelled phytoplankton, spectrophotometric measurements of algal cultures

int j remote sensing 2001 vol 22 no 2 amp 3 315ndash338

The identi cation of phytoplankton pigments from absorption spectra

R AGUIRRE-GOMEZdagger A R WEEKSDaggersect and S R BOXALL d

daggerInstituto de Geograa Universidad Nacional Autonoma de MexicoCircuito Exterior CP 4510 Coyoacan DF MexicoDaggerMaritime Faculty Southampton Institute East Park Terrace SouthamptonSO14 OYN England UKd University of Southampton Department of Oceanography SouthamptonOceanography Centre Waterfront Campus European WaySouthampton SO17 1BJ England UK

(Received 30 September 1998 in nal form 1 July 1999)

Abstract The absorption spectra of photosynthetic algae are characterized bya continuous envelope which is a result of the overlapping spectra of the indivualpigments This feature makes it di cult to estimate the contribution of eachpigment to the total absorption spectra Derivative analysis is an objective toolfor isolating absorption peaks in phytoplankton Theoretically electrons and ionsof interacting molecules can be regarded as simple harmonic oscillators in anelectromagnetic eld which result in a Lorentzian shape However when meas-ured by an optical spectrophotometer the signal is transformed into a Gaussiancurve Thus a combination of both types of curve provides a realistic approachto the decomposition of absorption spectra In this study derivative analysis isperformed on absorption spectra in order to prove that the method can besuccessfully used to identify the individual absorption spectra of componentpigments The spectra used are modelled phytoplankton spectrophotometricmeasurements of algal cultures and samples from natural waters A combinationof GaussianndashLorentzian shaped curves centred on the identi ed peaks werecompared with the original spectra and showed good agreement

1 IntroductionOcean colour sensors mounted on satellites and aircraft have allowed the estima-

tion of phytoplankton biomass using ratios of water-leaving radiance at diŒerentspectral bandwidths (see eg Gordon et al 1983) The launch of higher resolutionsensors (eg Medium Resolution Imaging Spectrometer (MERIS)) will provideopportunities to gain more information about the light absorbing properties ofphytoplankton If major diŒerences in taxonomic groups can be identi ed by theirvariation in light absorbing properties these new sensors will provide excitingopportunities to diŒerentiate between phytoplankton populations on a global scaleAlthough it is widely recognized that diŒerentiation of mixtures of phytoplanktontaxonomic groups may prove di cult where monospeci c algal blooms develophigh-resolution optics are considered a useful tool (JeŒrey 1997) Another approach

sectTo whom correspondence should be addressed

International Journal of Remote SensingISSN 0143-1161 printISSN 1366-5901 online copy 2001 Taylor amp Francis Ltd

httpwwwtandfcoukjournals

R Aguirre-Gomez et al316

is to identify spectral peaks associated with particular pigments or optical markersof particular phytoplankton taxonomic groups (Barlow et al 1993)

Since the absorption spectra of phytoplankton are complex due to the mixturesof pigments and associated proteins it is di cult to identify these components byoptical analysis Some studies have focused on separating the absorption of chloro-phyll-a from that of accessory pigments Among these studies the derivative analysisof spectral curves has proven to be a powerful method (Bidigare et al 1989)Derivative analysis is useful for resolving secondary peaks and shoulders producedby pigments present in the algae within the overlapping absorption regions (Butlerand Hopkins 1970) Absorption maxima found using this method occur near orat the correct wavelength position where absorption peaks may be attributed toparticular photosynthetic pigments

Over the last 25 years there have been considerable eŒorts to identify the compon-ents of algal absorption using mathematical techniques French et al (1972) decon-volved red absorption spectra of chlorophyll-a in green plants using a Gaussiancurve- tting approximation Their results were in good agreement with the conclu-sions of both room-temperature rst-derivative spectral analyses (Gulyayev andLitvin 1970) and low-temperature fourth-derivative spectroscopic analyses (Butlerand Hopkins 1970) HoepŒner and Sathyendranath (1991) applied a Gaussianapproach to decompose in vivo absorption spectra of diŒerent algal cultures

Another approach utilizes one of the simplest theoretical models of absorptionproposed by H A Lorentz Near the beginning of the twentieth century Lorentzdeveloped a classical theory of optical properties in which the electrons and ions ofmatter were treated as simple harmonic oscillators subject to the driving force ofapplied electromagnetic elds (Bohren and HuŒman 1983) The Lorentz model hasbeen applied in molecular biophysics to explain the electronic excitation energytravelling in a periodic structure this is the so-called exciton theory (Dexter andKnox 1965) Pigment molecules provide excellent examples of the usefulness of theanalysis of excitation interactions particularly their observation in absorption spectra(Sauer 1975) Bricaud and Morel (1986) successfully applied the Lorentzian modelto decompose the absorption spectrum of the alga Platymonas suecica

However when considering the shape of the absorption spectra it is necessaryto take into account the distortion caused by the dispersive slit When the bandwidthandor the componentsrsquo bands are of the same order as the slit width a distortionmay occur In this case it is di cult to determine the true band shapes from theobserved ones There are however two methods of tackling this problem (1) Whenthe true shape of the spectrum is unknown but the instrument distortion is analytic-ally known or can be easily measured a deconvolution can be applied using Fouriermethods (Boldeskul et al 1971) (2) When the true band shape and the slit functionare known the observed shape can be calculated by convolution Thus when thetrue band shape is known as a function of the band parameters (position half-widthand weight) the convoluted curve can be calculated and the parameters varied untila good t between the convoluted function and the data is obtained This approachis used in the present study

For a modern monochromator the slit function is very similar to a Gaussian atleast for small slit widths (Sundius 1973) Thus if the true band shape can beapproximated by a Lorentzian shape the observed pro le will be the convolutionof a Gaussian and a Lorentzian This type of convolution has long been used inastrophysical spectroscopy (Mitchell and Zemansky 1934) but to our knowledge

Algal blooms detection monitoring and prediction 317

there has been no attempt to combine Lorentzian and Gaussian models for thedecomposition of absorption spectra of diŒerent alga This approach appears tobe comprehensive as the theoretical Lorentzian absorption spectral shape issmeared with a Gaussian deconvolution which is caused by the optical design of aspectrophotometer

In this study the derivative method will be used to identify the major absorptionpeaks of algae Firstly theoretical spectra for a number of taxonomic groups willbe constructed and the derivative method will be used to identify the peaks If thesepeaks are identi ed as those of the individual pigments used to construct the modelthe method can be explored further with monocultures and nally with naturalpopulations The absorption spectra of a number of algal cultures will then beanalysed using both in vivo samples and samples in which the pigments have beenextracted from the cells using solvents This will be followed by analysis of theabsorption spectra from a number of natural populations The peaks identi ed fromthe solvent-extracted and natural samples will be compared with those expectedfrom the component pigments and those measured by High Performance LiquidChromatography (HPLC) A comparison will be made between the measured spectraand simulated spectra which are constructed from individual pigment absorptionspectra Finally a method to simulate absorbance spectra using a combination ofGaussianLorentzian curves centred on the derived spectral peaks will be used tomaximize the t of a modelled spectrum to a measured one

2 Methodology21 Derivative analysis

When attempting to analyse the absorption spectrum of phytoplankton it isnecessary to detect and determine quantitatively the position and the intensity of aweak absorption band which may be overlaid or obscured by another absorptionband Derivative analysis aims to provide a mathematical basis for identifying themain bands and hence the components of the absorption spectrum It providesinformation regarding the convexity and concavity of a given absorption curve andis useful for separating the secondary absorption peaks and shoulders produced byalgal pigments in regions of overlapping absorption

The derivative of a curve is simply its slope at a given point In calculus notationthe derivative of the spectral trace is dAdl where A is the absorbance and l is thewavelength The derivative is thus a plot of the slope dAdl against l However ifthe diŒerentiation interval is very small then the diŒerence between the successivevalues may be small compared with the random noise and a noisy derivative spectrummay be obtained On the other hand a larger diŒerentiating interval will reduce thenoise and maximize the signal but sharp features may be lost Butler and Hopkins(1970) found that the optimum diŒerentiating interval depends on the level of noisein the data and the spectral bandwidth of the signal

The derivative plot has some advantages over a direct record For example itindicates more precisely the centre of each absorption peak Figures 1(a) and 1(b)show respectively a schematic diagram of the simulated Gaussian and Lorentzianabsorbance curve and four derivatives (dnAdln n 5 1234) At the centre of a peakthe curve is horizontal hence dAdl shows an in ection point parallel to the axis lThese in ection points indicate the presence of either a maximum or a minimum inthe curve (see the rst derivative in both gures) The use of derivative analysis toidentify small changes in the slope of spectra can be observed in these gures which


(a)

Figure 1 (a) Simulated absorption spectrum using ve Gaussian curves and a step-by-stepderivative process (b) Simulated absorption spectrum using ve Lorentzian curvesand a step-by-step derivative process

show the sum of ve simulated absorption curves In gure 1(a) the individual curvesare Gaussian-shaped whilst in gure 1(b) they are Lorentzian-shaped These curveshave their central wavelengths at 425 450 470 640 and 660nm and intensities of 124 16 06 and 05 relative units respectively The half-width is 30 nm for all thecurves in both cases These values are rather arbitrary but simply aim to provide atypical simulated absorption curve

In order to determine whether the in ection points represent a minimum or amaximum the second derivative is calculated The maxima are found by looking atthe negative minima in the second-derivative spectrum It is however important tonote that not all the peaks have been resolved with the second derivative particularlyin the case of the Gaussian curves ( gure 1(a)) Since there are still unresolved peaksit is necessary to continue with the derivative analysis The third derivative providessimilar information to that of the rst derivative and by looking at the in ection


(b)

points on this curve a new set of maxima and minima becomes apparent The third-derivative curve is almost an inverse image of the rst derivative but has theadvantage of enhancing subtle features This is evident when comparing both curvesat every bandwidth marked in the grid With this information it becomes apparentthat the positive maxima of the fourth derivative correspond to the in ection pointsof the third derivative Again the fourth-derivative curve is almost the inverse of thesecond-derivative curve except that the former shows new features which wereunresolved in the latter In the case of the Lorentzian curves the second derivativewas su cient to resolve the peaks of the ve component spectral curves This canbe observed in gure 1(b) However to resolve all the Gaussian peaks it was necessaryto continue up to the fourth derivative

22 GaussianndashL orentzian deconvolutionTo understand the absorption characteristics of phytoplankton it is important to

analyse the physical and instrumental mechanisms involved in the process The shapeof absorption spectra is conformed largely by both factors These factors can be


represented by well known mathematical functions The two most commonlyobserved shapes are the Lorentzian and Gaussian These curves are analyticallyrepresented by the following expressions

Gaussian function

Y 5 Y0exp C Otilde ln 2A2(X Otilde X

0)

DX1 2

BD 2(1)

Lorentzian function

Y 5 Y0 C1 1 A2(X Otilde X

0)

DX1 2

B2D Otilde 1(2)

where Y0

represents the band height X represents the abscissa coordinate of thepeak X

0and DX

1 2is the bandwidth at half-height

Among other physical reasons electrons and ions of interacting molecules cantheoretically be regarded as simple harmonic oscillations in an electromagnetic eldresulting in a Lorentzian shape However when measured by an optical spectro-photometer the signal is transformed into a Gaussian curve Thus a combinationof both types of curve provides a realistic approach for analysis of absorptionspectra Thus in any attempt to reconstruct the absorption coe cient a

p h(l) of

phytoplankton all of these factors have to be taken into account

23 Physical factors of spectral absorption lines broadeningMolecules have three main types of energy A molecule may possess rotational

energy due to a bodily rotation about its centre of gravity it may have vibrationalenergy due to the periodic displacements of its atoms from their equilibriumposition and it may have electronic energy since the electrons associated with eachatom or bond are in constant motion (Rees 1990)

The absorption spectrum of any pigment is a signature of the energy levels mosteŒectively absorbed and re ects the electronic state characteristics of the nuclearstructure of the molecule (Sauer 1975) The nal spectrum however is not a simpleonemdashit can appear broadened by three main eŒects Doppler broadeningHeisenbergrsquos uncertainty principle collision broadening (Schanda 1986)

231 Doppler broadeningAbsorption spectra can have a Gaussian shape caused by Doppler shift eŒects

which makes the absorption spectral lines broader than might be expected (Schanda1986) It has been found that in the visible region at room temperature the Doppler(Gaussian) width is comparable to the resolution limit of a large grating spectro-photometer which may account for up to a maximum of 1 nm (Thorne 1974)

232 Natural line shape (Heisenbergrsquos uncertainty principle)The energyndashtime uncertainty relation shows that a state with a nite lifetime

does not have a precisely de ned energy rather its energy has a spread or uncertaintywhich increases with decreasing lifetime Apart from the ground state all otherenergy states show spontaneous emission Thus an excited state does not show asharply de ned energy The line shape of the energy distribution is a typical exampleof a Lorentzian curve The classical natural width in the visible (400ndash700nm) hasbeen estimated as 16 Ouml 10 Otilde 5 nm constant for all wavelengths (Thorne 1974) Thusthe contribution of this mechanism to band broadening is small enough to bedetected by optical instruments


233 Collision broadeningMolecules of condensed matter ( liquids and gases) are in continuous motion and

frequently collide These collisions produce some deformation of the particles andperturb the energies of at least the outer electrons in each particle This fact accountsfor the width of visible spectral lines since these deal largely with transitions betweenouter electronic shells Because of these collisions the temperature of the mediummay increase So an equilibrium must be reached between the thermal energy andthe electrical energy of the molecular oscillations This fact indicates that the spectralproperties of the absorption lines re ect the kinetic temperature of the medium Itis not straightforward to give a typical value of the broadening created by thismechanism as many physical assumptions are involved in the process However thecontribution due to this broadening eŒect may be from 2 to 10 nm mainly dependingon the type of molecular collision involved Collision broadening is known to adopta Lorentzian shape (Thorne 1974)

24 Instrumental factors for spectral absorption line broadeningA number of studies have shown that in analysing the shape of absorption spectra

it is necessary to consider the distortion caused by the dispersive slit of a spectro-photometer Basically a double beam UVndashvisible spectrophotometer consists of asource of light an optical arrangement and a photosensor

The physical width of the beam falling on the photosensor can be limited by theprovision of an exit slit just in front of the detector entrance However since it isimpossible to make either of the slits in nitely narrow a range of wavelengths ratherthan a single one falls on the photosensor This results in a broadening of theabsorbance peaks of the substance It has been found that this eŒect smears thesignal into a Gaussian shape Banwell (1972) provides a detailed description ofthis process

The physical and instrumental factors discussed above confer the shape ofGaussianndashLorentzian curves to the absorption spectra of pigments The convolutionof these mathematical curves is known as the Voigt function

25 Voigt pro le of the absorption coeYcient of phytoplanktonThere are two important consequences when considering the broadening eŒects

and the pigment composition expression Firstly the absorption coe cient of eachpigment can now be estimated by decomposition of a

p h(l) into several

GaussianndashLorentzian bands simulating the absorption properties of these pigmentsThis result means that the absorption coe cient of phytoplankton can beexpressed as

ap h

(l) 5 ni=1

Ci(l

m i) a

iG 1 b

iL (3)

where

G 5 expAl Otilde lm i

)2

wi

B (4)

is the Gaussian contribution to the phytoplankton absorption spectra

L 5 C1 1 4Al Otilde lm i

wiBD Otilde 1

(5)


is the Lorentzian contribution to the phytoplankton absorption spectra lm i

is theposition of maximum absorption for the ith GaussianndashLorentzian contribution w

idetermines the width of the peak a

iis the mean speci c absorption coe cient for

the ith Gaussian absorption band at lm i

bi

is the mean speci c absorption coe cientfor the ith Lorentzian absorption band at l

m i and C

iis the concentration of the

pigment responsible for the ith absorption band This expression contains the phys-ical and instrumental elements for the spectral reconstruction of the absorptioncoe cient of phytoplankton a

p h(l) through a GaussianndashLorentzian deconvolution

Consequently the original or the reconstructed GaussianndashLorentzian absorptionspectra of phytoplankton can be used indistinctly but the latter approach has theadvantage of being suitable for a mathematical treatment

The second consequence of spectral broadening is that absorption spectra ofbiological material becomes a rather complicated mixture of absorption bands thatvery often are di cult to resolve into their individual components Improvementsin resolving power can be achieved not only by using better electro-optical designswhich can be very expensive but also by applying mathematical techniques to theexperimental data In this respect Butler and Hopkins (1970) have shown thatderivative analysis is useful for resolving overlapping absorption regions The nextsection shows how through this method the position of the maximum absorptionpeak l

m ican be objectively located

26 Voigt curves ttingIn order to determine the optimum combination of Voigt curves the curve tting

program RESOL initially described by French et al (1969) was slightly updatedand adapted to the requirements of this study The program requires informationabout wavelength peaks heights and half-widths The rst two parameters wereprovided by derivative analysis of the absorption spectra of the algae the third wasestimated based on a literature review (French et al 1972 HoepŒner andSathyendranath 1991) The procedure was applied to the absorption spectra ofacetone-extracted and in vivo samples of phytoplankton and also to the naturalpopulations

The input speci cations for each absorption spectrum were (a) both Gaussianand Lorentzian shapes for each curve- tting analysis (b) selected half-widths (c) noweighting of the data (ie the program was allowed to work out the best combinationof weights) (d) a maximum of 20 iterations (e) band centres (derivative analysis)and hald-bandwidth changes of no more than 1 nm per iteration

The program then optimized these parameters in each band to provide the best t with the absorbency curve Because all parameters are adjusted for each iterationthe convergence was rapid At the end the program gave the values of four parametersthe central peak the Gaussian weight the Lorentzian weight and the half-width

Section 3 shows how this method was used to locate the position of the maximumabsorption peak l

m i

3 Results31 Modelled absorption spectra

The derivative method was applied to two modelled absorption spectra of diatomsand green algae these containing phytoplankton with diŒerent pigment assemblagesand concentrations and hence diŒerent optical characteristics

The spectral curves were constructed in the following way Initially the individual


pigment spectra for each pigment (taken from Prezelin and Boczar (1986) Kirk(1983) and JeŒrey (1980)) were digitized for computer handling and then normalizedfor standardization purposes These spectra were then multiplied by a proportionalityfactor and each of the component pigments summed to obtaining the absorptionspectra The proportions of the diŒerent pigments present in each alga were takenfrom the works of Kirk (1983) and JeŒrey (1980) Peak positions for the mostrepresentative pigments were compared with those reported in the literature(eg Bidigare et al 1989 Prezelin and Boczar 1986 HoepŒner and Sathyendranath1991) and are summarized in table 1

311 DiatomsThis kind of algae usually has a yellowndashbrown colouration produced by the

masking eŒects of fucoxanthin pigments over the chlorophylls (Boney 1976) Thesealgae contain photosynthetic pigments of chlorophyll-a chlorophyll-c

1 chlorophyll-

c2 fucoxanthin and a low amount of b-carotene (Prezelin and Boczar 1986)

The proportion of chlorophyll-c is on a molar basis about the same as chloro-phyll-b has in green algae and higher plants (Kirk 1983) The mean value of themolar ratio of chlorophyll-a to chlorophyll-c is about three (considering c 5 c

11 c

2)

JeŒrey (1976) stated that the relative amount of chlorophyll-c1

to c2

has been foundto be one in most cases Gugliemelli et al (1981) found a molar ratio of carotenoidto chlorophyll-(a 1 c) in the diatom Phaeodactylum of about 105 Here carotenoidincludes both fucoxanthin and b-carotene in the proportion 11 The relative propor-tions of chlorophyll-a chlorophyll-c

1 chlorophyll-c

2 fucoxanthin and b-carotene

were used for constructing the absorption spectrum for a diatom (D) as follows

D 5 05[3 Chl-a 1 Chl-(c1

1 c2)] 1 (fucoxanthin 1 b-carotene) (6)

312 Green algaeThese phytoplankton contain pigmentation closely resembling the chlorophyll-

andashchlorophyll-bndashcarotenoid system characteristic of higher plants (Prezelin andBoczar 1986) In higher plants and in fresh water green algae the molar ratio ofchlorophyll-a to chlorophyll-b is about 31 (Kirk 1983) However several marinespecies of green algae are characterized by low chlorophyll-ab ratios lying in therange 10 to 23 (Yokohama and Misonou 1980) Kirk (1983) reports that in higherplants and green algae the molar ratio of carotenoids to chlorophyll-(a 1 b) is about1 3 Although not the predominant carotenoid b-carotene pigment is present inalmost all algal groups (Kirk and Tilney-Bassett 1978) it was therefore chosen as

Table 1 Position of peaks of pigments and those found for the modelled theoretical curvesthrough the derivative analysis Shoulders are marked with the symbol (~ )

Casepigment Peak position (nm)

Diatom 431 460 487 531 577 617 634 661Green algae 430 457 488 512 533 562 579 612 643 662Chlorophyll-a 430 ~ 610 660Chlorophyll-b ~ 430 453 645Chlorophyll-c

1446 580 628

Chlorophyll-c2

449 580 628b-carotene 450 480Fucoxanthin 460 480


the representative carotenoid pigment used in this test These observations havebeen used to construct the absorption spectrum of green algae (GA) using thefollowing relationship

GA 5 3(16 Chl-a 1 Chl-b) 1 b-carotene (7)

the value of 16 corresponds to the mean of the Chl-aChl-b ratio for marine species

32 Derivative analysis of the modelled absorption spectraThe derivative method was applied to both constructed theoretical curves

(equations (6) and (7))

321 DiatomsThe peaks found through use of the fourth-derivative algorithm were located at

the following wavelengths 431 460 487 531 577 617 634 and 661nm Those peaksthat are attributable to light absorbing pigments are shown in bold A comparisonof these values with those in table 1 shows that the peaks located at 431 and 661nmapproximately coincide with the expected peaks of chlorophyll-a The peak at 617nmis probably due to the presence of chlorophyll-a but it has been slightly reinforcedby the presence of both types of chlorophyll-c pigments The peak appearing at460nm may also be due to the mixture of both types of chlorophyll-c while thepeak located at 634nm is probably entirely caused by the presence of chlorophyll-

c2 Finally the peak at 487nm is produced by a combination of fucoxanthin and

b-caroteneDue to the fact that both types of chlorophyll-c overlap in the blue region with

chlorophyll-c1

completely masking chlorophyll-c2 the peaks are unresolved

Likewise fucoxanthin and b-carotene have either very close absorption peaks orpeaks at the same wavelengths this results in a reinforced absorption peak ratherthan in separated peaks Figure 2(a) shows the diatom absorption spectra with itscomponent pigments (upper panel ) and the fourth-derivative curve ( lower panel )

322 Green algaeAbsorption peaks for the theoretical curve of green algae were found at the

following wavelengths 430 457 488 512 533 562 579 612 643 and 662nm Itshould be noted that the peaks appearing in the green region (500ndash600nm) are smallcompared to those located in the blue and red regions (bold numbers) which havea clear biological signi cance A comparison of these values with those in table 1shows that the peaks located at 430 612 and 662nm coincide with the expectedpeaks of chlorophyll-a Peaks located at 457 and 643nm may be attributed tochlorophyll-b The b-carotene pigment matches the peak at 488nm

It should be noted that some peaks may appear slightly shifted from theirexpected positions especially when two absorption peaks are very close Duringsummation the curves combine and the peaks are shifted to one side or the other ofthe original curve In these cases the resultant absorption peak would appear shifted

Figure 2 (a) Top theoretical absorption spectrum of a diatom (mdashmdash) based upon the relativeproportions of chlorophyll-a (- middot-) chlorophyll-c

11 c

2(u ) b-carotene () and fucoxan-

thin (- -) (see text) Bottom peaks determined by the fourth-derivativeanalysis (b) Toptheoretical absorption spectrum of green algae (mdashmdash) based upon the relative propor-tions of chlorophyll-a (- middot-) chlorophyll-b () and b-carotene (- -) (see text) Bottompeaks determined by the fourth-derivative analysis


(b)

(a)


(b)

(a)

Figure 3 Absorption spectrum of in vivo monocultures of (a) Phaeodactylum and(b) T etraselmis full curves reconstructed absorption curve dotted curve using aGaussianndashLorentzian convolution (see text) The component curves calculated by theconvolution and the GaussianLorentzian shapes are also shown The bottom panelshows the magnitude of error between the two curves

towards the major pigment according to the magnitude of the absorbing pigmentFor example two chlorophyll-a peaks should exist at the blue region of the spectrumbut as both chlorophyll-b and b-carotene spectra have a monotonic and ascendantresponse in the blue they compensate for the chlorophyll-a depression between thepeaks at 410 and 430nm Figure 2(b) shows the original green algae spectrum withits component pigments (upper panel ) and the fourth-derivative curve ( lower panel )


These results show that by reconstrucing spectra from component pigmentsderivative analysis can be used to determine original spectrum peaks

33 Cultured samplesThe next stage was to determine the major spectral peaks in a number of algal

cultures and to consider the pigments these represent particularly considering thepresence of biomarkers for particular taxonomic groups Mantoura and Llewellyn(1983) have developed methods for identifying and determining the concentrationof the major light absorbing pigments using HPLC Following this method a numberof algal groups were analysed in order to determine their pigment characterization(table 2) This shows for example zeaxanthin is a biomarker for Synechococcuswhile peridinin is a biomarker for dino agellates Similarly diatoms do not containperidinin and zeaxanthin but they do contain fucoxanthin

There are however a number of problems with this approach The HPLCtechnique requires that the pigments are isolated from the cell and for the majorityof pigments this is most eŒective when solvents are used The absorption propertiesof intact phytoplankton diŒer from the extracted samples due to the fact thatpigments are bound with protein complexes Additionally the package eŒect mayproduce diŒerent absorption characteristics in whole cells and this is particularlyimportant in larger cells (Morel and Bricaud 1981) Therefore comparison of spectralpeaks with HPLC derived pigments can only be carried out eŒectively for absorptionspectra of solvent-extracted phytoplankton Comparisons were made for in vivomonocultures and natural populations with consideration given to spectral shifts

34 Spectrophotometric measurements on in vivo samplesSpecies in culture included a diatom (Phaeodactylum tricornotum) and a chloro-

phyte (T etraselmis suecica) selected for their diŒerent pigment composition andtherefore diŒerent colour Values of the absorption coe cient of phytoplankton weretaken from a direct measurement of whole suspended particles according to thetechnique presented by Lewis et al (1985) and Sathyendranath et al (1987) Thesamples of water containing intact phytoplankton cells were located just at theentrance of the spectrophotometerrsquo s photomultiplier in order to increase the scat-tering reception angle Bricaud et al (1983) using angular scattering fucntionsrepresentative of phytoplankton calculated that in this way more than 995 of thescattered light can be captured The samples were prepared by obtaining three 5 ml

Table 2 Pigments detected by HPLC and their respective concentrations (mg l Otilde 1 ) (afterMantoura and Llewelyn (1983))

Phaeodactylum Rhizosolenia Synechococcus AmphidiniumPigment tricornotum delicatula sp carterae

Chl-c3

040 308646 18392 030Chl-c

11 -c

2263724 278719 119140

Fucoxanthin 385072 392125 48084Peridinin 92684Diadinoxanthin 219985 142106 516724Chl-a 311911 96968 145190 709596b-carotene 16739 9286 156999 58366Zeaxanthin 108018


units from an aseptic 25 ml sample of each alga Absorption spectra measurementswere calibrated by placing two samples of ltered seawater at the entrance of thephotomultiplier The response was a at line that was then zeroed to yield abackground correction factor This process was repeated before each new measure-ment The absorption spectrum of each unit was measured from 350 to 750nmusing ltered seawater as a blank in a Hitachi Model U-2000 double beam spectro-photometer with a spectral resolution of 2 nm The path length of the sample cellswas 1 cm The mean of these spectra was taken as representative with a standarderror of less than 25 between replicates

35 Derivative analysis of measured absorption spectra of algal culturesThe derivative method was applied to the absorption spectra of the diatom

Phaeodactylum tricornotum and the chlorophyte T etraselmis suecica and the resultsare shown in tables and gures 3a and 3b respectively The major light absorbingpigments namely chlorophyll-a and carotenoids are the cause of these peaks Sincechlorophyll-a is common to both kinds of algae it is discussed separately the otherpigments are discussed as a diŒerent group

Table 3a Absorption peaks (nm) for in vivo cultures of Phaeodactylum tricornotum fromthe fourth-derivative method and curve- tting procedure including proportions ofGaussian and Lorentzian weights

No Original peak (nm) Adjusted peak (nm) Gaussian weight Lorentzian weight

1 418 41472 05214 000002 442 43713 01587 000003 466 46688 04666 000964 500 50079 03456 0012856 544 54431 02151 0000078 586 58411 03158 000009 616 61654 02623 00062

10 638 64262 02101 0019311 680 68148 03455 01990

Table 3b Absorption peaks for in vivo cultures of T etraselmis suecica from the fourth-derivative method and curve- tting procedure including proportions of Gaussian andLorentzian weights


1 414 41140 04741 000002 440 43050 01583 000003 466 46193 02885 002004 500 50808 01465 00132

8 590 57599 01956 000119 620 62017 01737 00000

10 640 64312 01112 0014211 680 67880 02845 00010


351 Chlorophyll-aConsideration of the other results suggests that absorption peaks 1 2 9 and 11

with their maxima at 415ndash425 440ndash455 620 and 675nm respectively correspond toin vivo absorption by chlorophyll-a Goedheer (1970) observed that the in vivoabsorption spectra of the brown algae L aminaria digitata had chlorophyll-a peaksat 418 437 618 and 673nm Prezelin (1980) found absorption peaks of Glenodiniumsp at 419 437 618 and 675nm More recently HoepŒner and Sathyendranath(1991) reported averaged values of chlorophyll-a absorption peaks of three diŒerentgroups of algae located at 412 435 623 and 675nm These values are similar tothose found here

352 CarotenoidsIn relation to carotenoid-like pigments the range of absorption maxima is wider

than that of chlorophyll-a However similar results to those found here have beenreported Mann and Myers (1968) assigned the range 425ndash530nm to carotenoidswith the range 460ndash530 nm due to fucoxanthin and that of 425ndash500nm due to b-carotene Similarly Bidigare et al (1989) located the fucoxanthin absorption rangeat between 521 and 531nm HoepŒner and Sathyendranath (1991) concluded thatthe peaks found at 489 and 536nm may correspond to a mixture of carotenes andxanthophylls rather than to the separate groups Xanthophylls diŒer from carotenesby the presence of oxygenated groups in their molecules but this alone is notsu cient to distinguish them by their spectra

Peak 6 located between 545ndash555nm may also be attributed to fucoxanthin Insupport of this Goedheer (1970) detected fucoxanthin in the in vivo absorptionspectrum of L aminaria digitata in the interval 530ndash545nm Finally peak 8 locatedat 590ndash600nm may be attributed to chlorophylls For example a small feature at585nm due to chlorophyll-c was observed in the in vivo absorption spectrum of

L aminaria digitata (Goedheer 1970) However the same feature was observed insome absorption spectrum of chlorophyll-andashchlorophyll-cndashprotein complexes(Prezelin and Boczar 1986)

There are a number of peaks that are characteristic of pigments only found insome of the algae In the case of Phaeodactylum tricornotum apart from chlorophyll-

a peaks (415 440 620 and 675nm) there are two peaks (470 640nm) caused by thepresence of chlorophyll-(c

11 c

2) and in the green region four absorption peaks (505

530 545 and 590nm) due to carotenoid-like pigments In relation to both types ofchlorophyll-c several authors have reported similar results For instance HoepŒnerand Sathyendranath (1991) averaging over a number of diatoms found two in vivoabsorption peaks of chlorophyll-c at 460 and 642nm Bidigare et al (1989) alsoreported two absorption peaks of chlorophyll-c located at 467 and 630nm SimilarlyMann and Myers (1968) found absorption maxima at 460 and 640nm

In the case of T etraselmis suecica peaks attributable to chlorophyll-a were locatedat 425 455 620 and 675nm Chlorophyll-b peaks appeared at 480 and 640nmCarotenoidndashxanthophyll pigment peaks were found in the green region ranging from505 to 575nm The peak located at 600nm may be attributed to the in uence of thechlorophyll-andashchlorophyll-b complex Chlorophyll-b absorption peaks have beenfound in a range of positions in the blue region of green algae spectra For instanceGulyayev and Litvin (1970) assumed that peaks appearing at 442ndash446 470ndash474and 641ndash650nm are due to this pigment Bidigare et al (1989) found absorptionpeaks at 483 and 650nm HoepŒner and Sathyendranath (1991) interpreted that on


average the absorption peaks at 463 and 654nm are due to chlorophyll-b JeŒrey(1980) found that in the in vivo spectra of three marine species of green algaechlorophyll-b contributes with a shoulder sometimes a peak at about 650nm andadds to the strong blue light absorption around 440ndash460nm

36 Spectrophotometric measurements on acetone-extracted samplesIn the laboratory two aseptic 5 ml samples of the diatom Phaeodactylum tri-

cornotum and the bluendashgreen alga Synechococcus sp were ltered through a 47 mmglass bre lter (Whatman GFF) using a Millipore vacuum pump The lters werefolded in half with the algal material inside placed in a plastic bag wrapped inaluminium foil labelled and stored at Otilde 20 szlig C The lters were subsequently manuallyground using a glass rod and then homogenized with a motorized homogenizer(Gallenkamp) in 6 ml of 90 of acetone Homogenization continued until the lterswere well ground The homogenized algal material was placed in a para lm-coveredtube and centrifuged at 3000 rpm for ten minutes (Mistral MSE 2000) A smallamount of the extract was then used for HPLC pigment analysis and a 5 ml samplefor absorption measurements using the spectrophotometer Absorption spectra meas-urements were calibrated and corrected for background noise by placing two samplesof acetone in the cuvettes and measuring the spectrum The process was repeatedbefore each measurement Absorption spectra were recorded from 350 to 750nm intriplicate using 90 acetone as a blank The path length of the sample cells was1 cm The mean spectrum of each sample was taken as representative The standarderror between replicates was less than 1 Samples were measured using a HitachiModel U-2000 double beam spectrophotometer with a spectral resolution of 2 nmand an instrumental error of Ocirc 1 nm

37 Derivative analysis of acetone-extracted cultured samplesDerivative analysis was performed and the results are shown in tables 4a and 4b

Peaks were found both in the blue and in the red regions with minor peaks locatedin the green part of the spectrum More speci cally for the diatom Phaeodactylum(table 4a) peaks 1 2 6 and 7 can be attributed to chlorophyll-a while peak 3 canbe attributed to chlorophyll-c and peaks 4 to 5 to fucoxanthin and to carotenoidcomplexes respectively For the green alga Synechococcus (table 4b) peaks 1 2 6 and7 are attributable to chlorophyll-a peak 3 to b-carotene 5 to fucoxanthin and 4 tocarotenoid complexes These peaks were compared with the results from HPLCanalysis which showed the presence of a number of pigments and their concentrations(table 4c) Fucoxanthin was found in Phaeodactylum a biomarker for diatomsZeaxanthin a biomarker for Synechococcus was present in the blue green alga Since

Synechococcusrsquo biliproteins are water soluble they were not considered in this analysisFigures 4(a) and 4(c) show the absorption of Phaeodactylum tricornotum and

Synechococcus sp respectively Once the pigments had been identi ed the individualabsorption spectra were combined to construct theoretical absorption spectra foreach alga However since the pigments were extracted using a number of diŒerentsolvents (which maximized the e ciency of extraction) diŒerent absorption peakpositions and magnitudes are expected Bearing this in mind the spectra werereconstructed following the same procedure as for the theoretical spectra of diatomsand green algae Thus since the component pigments had been isolated usingdiŒerent solvents a normalization procedure was carried out to avoid magnitudeproblems A normalized value of one was given to the concentration of the major


Table 4a Absorption peaks for solvent-extractedcultures of Phaeodactylum tricornotum fromfourth-derivative method and the curve- tting procedure including proportions ofGaussian and Lorentzian weights


1 408 40700 01578 00002 434 43412 03391 00003 480 47900 01605 00004 538 53700 00151 000055 578 57900 00177 000016 622 62300 00304 000057 662 66300 01094 00036

Table 4b Absorption peaks for solvent-extracted cultures of Synechococcus sp from thefourth-derivative method and curve- tting procedure including proportions ofGaussian and Lorentzian weights


1 408 40700 00001 000002 430 42900 00700 000003 458 45700 00437 000004 484 48500 00300 000005 542 54100 00065 000006 576 57641 00022 000007 616 61500 00076 000008 662 66100 00380 00000

Table 4c Pigment analysis (HPLC) showing pigments their concentration in diŒerent algalgroups and their retention times

Pigment Concentration (mg l Otilde 1 ) Retention time (min)

Phaeodactylum tricornotumChlorophyll-c

33020 0373

Chlorophyll-c1

1 c2

8496 0575Fucoxanthin 32831 0990Diadinoxanthin 10721 1081Chlorophyll-a 79420 1306b-carotene 882 1473

Synechococcus spChlorophyll-c

3634 0312

Fucoxanthin 4100 0987Zeaxanthin 5115 1024

pigment Then based on that a weighting factor was calculated for the remainingcomponent pigments as the relative proportion in relation to the norm Finally aweighted sum according to the relative concentration of the pigments was the basisfor plotting a reconstructed absorption curve The reconstructed spectra of

Phaeodactylum tricornotum and Synechococcus sp are shown in gures 4(b) and4(d ) respectively along with the original measured spectra and the spectra of thecomponent pigments


(b)

(a)

Figure 4 Absorption spectrum of solvent-extracted pigments of (a) Phaeodactylummdashthisincludes the reconstructed spectrum (dotted curve) by the GaussianndashLorentzian convo-lution the bottom panel shows the magnitude of the error between the two curves(b) the measured absorption spectrum and the reconstructed spectrum ofPhaeodactylum calculated by normalizing the pigment spectra (c) Synechococcusmdashthisincludes the reconstructed spectrum (dotted curve) by a GaussianndashLorentzian convolu-tion the bottom panel shows the magnitude of the error between the two curves(d ) the measured absorption and the reconstructed spectrum of Synechococcuscalculated by normalizing the pigment spectra

38 Natural populationsNatural populations were collected over several days during the early summer

of 1995 at a site 12 km oŒthe Plymouth coast (50 szlig 10 frac34 N 4 szlig 10 frac34 W) The examplespresented here were collected on 9 May when two litres of seawater samples werecollected onboard the research vessel Squilla The samples were later analysed at the


(d)

(c)

Plymouth Marine Laboratory (PML) Absorption spectra collected in the eld weremeasured at PML using a double beam Perkin-Elmer lambda 2 spectrophotometerwith a spectral resolution of 1 nm with an associated instrumental error of Ocirc 05nmThe method used to measure the absorption coe cient of phytoplankton was themodi ed opal glass technique which measures the absorption of whole phytoplank-ton on a lter pad It is possible to measure the absorption of the pigments containedin the phytoplankton by subtracting the absorption spectra of the lter paper afterprocessing in methanol (Kishino et al 1985)

The fourth-derivative method was applied to the absorption spectra of the naturalsamples resulting in peaks at 12 wavelengths (see the second column of table 5a)Pigment analysis (HPLC) showed the presence of a number of diŒerent biomarkers(table 5b) The group of pigments chlorophyll-a fucoxanthin chlorophyll-c and


diadinoxanthin indicate the presence of diatoms andor coccolithophorideschlorophyll-b means green algae peridinin denotes the presence of photosyntheticdino agellates and zeaxanthin marks the presence of bluendashgreen algae (JeŒrey 1980Kirk 1983 Prezelin and Boczar 1986)

Band 1 centred at 400nm may be due to either chlorophyll-a derivatives or tonon-photosynthetic pigments The peaks appearing at 428 and 674nm are causedby the presence of chlorophyll-a Peaks at 454 588 and 630nm can be assigned tochlorophyll-c while the peak at 468nm denotes the presence of chlorophyll-b(Bidigare et al 1990)

Carotenoids completely dominate the region from 482ndash556nm (peaks 5ndash9)Absorption due to zeaxanthin occurs at the peak centred at 482nm Fucoxanthinand its derivatives are responsible for the peaks appearing at 514 530 and 556nmHowever the fuxcoxanthin derivative pigment butoxyfucoxanthin (a photosyntheticactive carotenoid) causes the large absorption at 494nm (Rowan 1989 Bidigareet al 1990 HoepŒner and Sathyendranath 1993)

Table 5a Absorption peaks of mixed natural populations from 9 May 1995 derived fromthe fourth-derivative method and curve- tting procedure including proportions ofGaussian and Lorentzian weights


1 400 39086 00426 000162 428 43015 00696 000813 454 45473 00379 001024 468 47059 00348 001065 482 48800 00272 000266 494 49666 00217 000007 514 51499 00148 000308 530 53479 00138 000549 556 55378 00098 00042

10 588 58746 00115 0001311 630 63449 00143 0001312 674 67531 00336 00073

Table 5b Pigment analysis (HPLC) showing pigments their concentrations and retentiontimes


Chlorophyll-c3

1478 01446Peridinin 0223 04225Butoxyfucoxanthin 32394 05850Fucoxanthin 1045 06446Hexoxyfucoxanthin 1028 07013Diadinoxanthin 7141 07892Alloxanthin 6445 08621Zeaxanthin 0226 09208Cholorphyll-b 3244 11638Chlorophyll-a 110330 12583


39 GaussianndashL orentzian curvesThe method for the Voigt approximation described earlier was applied to the

three types of absorption spectramdashthe in vivo monocultures the solvent-extractedmonocultures and the natural populations (in vivo)mdashusing the spectral peaksobtained by the fourth-derivative analysis The adjusted peaks and the Gaussian andLorentzian weightings of in vivo algal groups are shown in tables 3a and 3b resultsfor acetone-extracted samples are shown in tables 4a and 4b and results for naturalpopulations are presented in table 5a

Figures 3(a) and 3(b) 4(a) and 4(c) and 5 show examples of the convolutedspectra compared to the measured spectra and also show the shape of the componentspectra for each peak The gures show a good t for all the examples whether thesamples were in vivo monocultures (r 5 0979 Phaeodactylum tricornotum r 5 0989

T etraselmis suecica) or solvent-extracted monocultures (r 5 0995 Phaeodactylumtricornotum r 5 0999 Synechococcus sp) or in vivo natural populations (r 5 0998)

These results show the eŒectiveness of the Voigt approach to simulate absorptionspectra from known spectral peaks It is important to note that the iterative processresults in changes in the peak location This is particularly true for the in vivo samples(see tables 3a and 5a and gures 3(a) and 3(b) and 5) but less so for the solvent-extracted samples

4 DiscussionThe results presented here demonstrate the use of derivative analysis to identify

accurately spectral peaks caused by light absorbing pigments They also show thatit is more di cult to obtain information from the absorption spectra of intactphytoplankton than from solvent-extracted pigments This is because the opticalsignal is masked by the light absorbing and scattering properties of the cell wall andcontents

Figure 5 Absorption spectrum of natural populations collected on 9 May 1995 close toPlymouth and the GaussianndashLorentzian reconstruction the bottom panel shows themagnitude of the error between the two curves


The method was tested using theoretical absorption spectra constructed usingthe combined spectra of a number of pigments whose spectral absorption wasknown resulting in accurate identi cation of the pigments by attributing them tothe obtained peaks In this study typical spectra for two taxonomic groups diatomsand green algae were used

The derivative analysis described can be used to identify spectral peaks in algalmonocultures for both acetone-extracted and in vivo samples The individual absorp-tion spectra of the component pigments analysed by HPLC were combined tomodel the absorption spectrum of the algal culture There was good agreementbetween the modelled spectrum and the measured spectrum for the acetone-extractedsamples This can be partly explained by the fact that when component pigmentabsorption spectra are measured in acetone or other solvent there is a spectral shiftfrom the absorption characteristics in water However it is also the case thatbiliproteins were not extracted and so do not appear in the modelled results althoughthey are present in Synechococcus sp

When the derivative method is simply used to identify peaks and the ratio ofthe Gaussian or Lorentzian t of the individual peaks is maximized there is a good t between the measured and the modelled spectra There is some indication thatthe Gaussian t is more appropriate at shorter wavelengths and the Lorentzian t

improves relatively at longer wavelengths This seems appropriate since the signalfrom pigments in the red part of the spectrum is more de ned than that from theblue (French et al 1972)

The work on natural populations also shows encouraging results The fourth-derivative method identi es the spectral peaks and curve tting using theGaussianndashLorentzian technique provides a close match to the original absorptionspectrum However the information from the pigments is more di cult to matchwith the absorption spectrum This is partly due to the mixture of phytoplanktonstudied This has also previously been observed by JeŒrey (1997) who noted thatoptical techniques using hyperspectral data to identify phytoplankton taxonomicgroups would be more useful in a monospeci c algal bloom than in mixtures ofphytoplankton

Since information at the taxonomic level is useful the characterization of biomar-kers that are speci c to particular groups of phytoplankton will be a valuable toolFor example as we have shown here chlorophyll-b can be used as a marker forchlorophytes fucoxanthin can be used as a marker for diatoms and chlorophyll-ccan be used as a marker for diatoms and prymnesiophytes The relative proportionsof these biomarkers may also be used as a means of identi cation This will beparticularly true when satellite and airborne ocean colour sensors can provide dataof higher spectral resolution (eg Moderate Resolution Imaging Spectroradiometer

(MODIS) and MERIS)

AcknowledgmentsWe thank Dr Nicholas HoepŒner for providing us with a copy of the program

used for determining the Gaussian or Lorentzian characteristics of absorptionspectra and Dr Luis Proenca for his assistance in the HPLC analysis Dr J Aikenand Mr Guy Westbrook are gratefully acknowledged for their assistance in eldworkcarried out at Plymouth Marine Laboratory We also thank Kate Davis for preparingthe gures


References

Banwell C N 1972 Fundamentals of Molecular Spectroscopy Third Edition ( londonMcGraw-Hill)

Barlow R G Mantoura R F C Gough M A and Fileman T W 1993 Pigmentsignatures of the phytoplankton composition in the northeastern Atlantic during the1990 spring bloom Deep-Sea Research 40 459ndash477

Bidigare R R Morrow J H and Kiefer D A 1989 Derivative analysis of spectralabsorption by photosynthetic pigments in the western Sargasso Sea Journal of MarineResearch 47 323ndash341

Bidigare R R Ondrusek M E Morrow J H and Kiefer D A 1990 In vivo absorptionproperties of algal pigments Proceedings SPIE Ocean Optics X 1302 290ndash302

Bohren C F and Huffman D R 1983 Absorption and Scattering of L ight by SmallParticles (New York Wiley)

Boldeskul A E Buyan G P Kondilenko I I Novichenko V N and PogorelovV E 1971 Optical Spectroscopy 31 306ndash308

Boney A D 1976 Phytoplankton Studies in Biology 52 (London Edward Arnold)Bricaud A and Morel A 1986 Light attenuation and scattering of phytoplanktonic cells

a theoretical modelling Applied Optics 25 571ndash580Bricaud A Morel A and Prieur L 1983 Optical e ciency factors for some phytoplank-

ters L imnology and Oceanography 28 816ndash832Butler W L and Hopkins D W 1970 Higher derivative analysis of complex absorption

spectra Photochemistry and Photobiology 12 439ndash450Dexter D L and Knox R S 1965 Excitons (New York Wiley Interscience)French C S Brown J S and Lawrence M C 1972 Four universal forms of chlorophyll-

a Plant Physiology 49 421ndash429French C S Brown J S Prager L and Lawrence M C 1969 Analysis of spectra

of natural chlorophyll complexes Carnegie Institute Washington Yearbook 67536ndash546

Goedheer J C 1970 On the pigment system of brown algae Photosynthetica 4 97ndash106Gordon H R Clark D K Brown J W Brown O B Evans R H and Broenkow

W W 1983 Phytoplankton pigment concentrations in the Middle Atlantic Bightcomparison of ship determinations and CZCS estimates Applied Optics 22 20ndash36

Gugliemelli L A Dutton H J Juursinic P A and Siegelman H W 1981 Energytransfer in a light-harvesting carotenoidndashchlorophyll cndashchlorophyll a protein ofPhaeodactylum tricornotum Photochemistry and Photobiology 33 903ndash907

Gulyayev B A and Litvin F F 1970 First and second derivatives of the absorptionspectrum of chlorophyll and associated pigments in cells of higher plants and algaeat 20 szlig C Bioreg zika 15 701ndash712

Hoepffner N and Sathyendranath S 1991 EŒect of pigment composition on absorptionproperties of phytoplankton Marine Ecology Progress Series 73 11ndash23

Hoepffner N and Sathyendranath S 1993 Determination of the major groups of phyto-plankton pigments from the absorption spectra of total particulate matter Journal ofGeophysical Research 98 22789ndash22803

Jeffrey S W 1976 The occurrence of chlorophyll c1

and c2

in algae Journal of Phycology12 349ndash354

Jeffrey S W 1980 Algal pigments systems In Primary Production in the Sea edited byP Falkowski (New York Plenum) pp 33ndash58

Jeffrey S W 1997 Applications of pigment methods to oceanography In PhytoplanktonPigments in Oceanography edited by S W JeŒrey R F C Mantoura and S WWright (Paris UNESCO SCOR)

Kirk J T O 1983 L ight and Photosynthesis in Aquatic Ecosystems (Cambridge CambridgeUniversity Press)

Kirk J T O and Tilney-Bassett R A E 1978 T he Plastids (Amsterdam Elsevier)Kishino M Takahashi M Okami N and Ichimura S 1985 Estimation of the spectral

absorption coe cients of phytoplankton in the sea Bulletin of Marine Science 37634ndash642

Lewis M R Warnock R E and Platt T 1985 Absorption and photosynthetic action

Algal blooms detection monitoring and prediction338

spectra for natural phytoplankton populations Implications for production in theopen ocean L imnology and Oceanography 30 794ndash806

Mann J E and Myers M 1968 On pigments growth and photosynthesis of Phaeodactylumtricornotum Journal of Phycology 4 349ndash355

MantouraR F C and Llewellyn C A 1983The rapid determinationof algal chlorophylland carotenoid pigments and their breakdown products in natural waters by reverse-phase high-performanceliquid chromatographyAnalytica Chimica Acta 151 297ndash314

Mitchell A C and Zemansky M W 1934 Resonance Radiation and Excited Atoms(Cambridge Cambridge University Press)

Morel A and Bricaud A 1981 Theoretical results concerning light absorption in a discretemedium and application to speci c absorption of phytoplankton Deep-Sea Research28 1375ndash1393

PreAacute zelin B B 1980 Light reactions in photosynthesis Canadian Bulletin of Fisheries andAquatic Science 210 1ndash43

PreAacute zelin B B and BoAacute czar A A 1986 Molecular bases of cell absorption and uorescencein phytoplankton potential applications to studies in optical oceanography InProgress in PhycologicalResearch Volume 4 edited by F E Round and D J Chapman(Bristol Biopress) pp 349ndash464

Rees W G 1990 Physical Principles of Remote Sensing (Cambridge Cambridge UniversityPress)

Rowan K S 1989 Photosynthetic Pigments of Algae (Cambridge Cambridge UniversityPress)

Sathyendranath S Lazzara L and Prieur L 1987 Variations in the spectral values ofspeci c absorption of phytoplankton L imnology and Oceanography 32 403ndash415

Sauer K 1975 Primary events and the trapping of energy In Bioenergetics of Photosynthesisedited by Govindjee (New York Academic Press)

Schanda E 1986 Physical Fundamentals of Remote Sensing (Berlin Springer-Verlag)Sundius T 1973 Computer tting of Voigt pro les to Raman lines Journal of Raman

Spectroscopy 1 471ndash488Thorne A A 1974 Spectrophysics (London Chapman and Hall)Yokohama Y and Misonou T 1980 Chlorophyll a b ratios in marine benthic green algae

Japanese Journal of Phycology 28 219ndash223


is to identify spectral peaks associated with particular pigments or optical markersof particular phytoplankton taxonomic groups (Barlow et al 1993)

Since the absorption spectra of phytoplankton are complex due to the mixturesof pigments and associated proteins it is di cult to identify these components byoptical analysis Some studies have focused on separating the absorption of chloro-phyll-a from that of accessory pigments Among these studies the derivative analysisof spectral curves has proven to be a powerful method (Bidigare et al 1989)Derivative analysis is useful for resolving secondary peaks and shoulders producedby pigments present in the algae within the overlapping absorption regions (Butlerand Hopkins 1970) Absorption maxima found using this method occur near orat the correct wavelength position where absorption peaks may be attributed toparticular photosynthetic pigments

Over the last 25 years there have been considerable eŒorts to identify the compon-ents of algal absorption using mathematical techniques French et al (1972) decon-volved red absorption spectra of chlorophyll-a in green plants using a Gaussiancurve- tting approximation Their results were in good agreement with the conclu-sions of both room-temperature rst-derivative spectral analyses (Gulyayev andLitvin 1970) and low-temperature fourth-derivative spectroscopic analyses (Butlerand Hopkins 1970) HoepŒner and Sathyendranath (1991) applied a Gaussianapproach to decompose in vivo absorption spectra of diŒerent algal cultures

Another approach utilizes one of the simplest theoretical models of absorptionproposed by H A Lorentz Near the beginning of the twentieth century Lorentzdeveloped a classical theory of optical properties in which the electrons and ions ofmatter were treated as simple harmonic oscillators subject to the driving force ofapplied electromagnetic elds (Bohren and HuŒman 1983) The Lorentz model hasbeen applied in molecular biophysics to explain the electronic excitation energytravelling in a periodic structure this is the so-called exciton theory (Dexter andKnox 1965) Pigment molecules provide excellent examples of the usefulness of theanalysis of excitation interactions particularly their observation in absorption spectra(Sauer 1975) Bricaud and Morel (1986) successfully applied the Lorentzian modelto decompose the absorption spectrum of the alga Platymonas suecica

However when considering the shape of the absorption spectra it is necessaryto take into account the distortion caused by the dispersive slit When the bandwidthandor the componentsrsquo bands are of the same order as the slit width a distortionmay occur In this case it is di cult to determine the true band shapes from theobserved ones There are however two methods of tackling this problem (1) Whenthe true shape of the spectrum is unknown but the instrument distortion is analytic-ally known or can be easily measured a deconvolution can be applied using Fouriermethods (Boldeskul et al 1971) (2) When the true band shape and the slit functionare known the observed shape can be calculated by convolution Thus when thetrue band shape is known as a function of the band parameters (position half-widthand weight) the convoluted curve can be calculated and the parameters varied untila good t between the convoluted function and the data is obtained This approachis used in the present study

For a modern monochromator the slit function is very similar to a Gaussian atleast for small slit widths (Sundius 1973) Thus if the true band shape can beapproximated by a Lorentzian shape the observed pro le will be the convolutionof a Gaussian and a Lorentzian This type of convolution has long been used inastrophysical spectroscopy (Mitchell and Zemansky 1934) but to our knowledge









(a)





(b)






Gaussian function


0)

DX1 2

BD 2(1)

Lorentzian function


0)

DX1 2

B2D Otilde 1(2)

where Y0


0and DX



p h(l) of


















p h(l) into several


ap h

(l) 5 ni=1

Ci(l

m i) a

iG 1 b

iL (3)

where


)2

wi

B (4)



wiBD Otilde 1

(5)







bi


m i and C












m i








1 chlorophyll-



11 c

2)


to c2


1 chlorophyll-c



D 5 05[3 Chl-a 1 Chl-(c1







1446 580 628

Chlorophyll-c2












chlorophyll-c1







11 c




(b)

(a)


(b)

(a)












Chl-c3

040 308646 18392 030Chl-c

11 -c

2263724 278719 119140








1 418 41472 05214 000002 442 43713 01587 000003 466 46688 04666 000964 500 50079 03456 0012856 544 54431 02151 0000078 586 58411 03158 000009 616 61654 02623 00062

10 638 64262 02101 0019311 680 68148 03455 01990



1 414 41140 04741 000002 440 43050 01583 000003 466 46193 02885 002004 500 50808 01465 00132

8 590 57599 01956 000119 620 62017 01737 00000

10 640 64312 01112 0014211 680 67880 02845 00010










11 c















1 408 40700 01578 00002 434 43412 03391 00003 480 47900 01605 00004 538 53700 00151 000055 578 57900 00177 000016 622 62300 00304 000057 662 66300 01094 00036



1 408 40700 00001 000002 430 42900 00700 000003 458 45700 00437 000004 484 48500 00300 000005 542 54100 00065 000006 576 57641 00022 000007 616 61500 00076 000008 662 66100 00380 00000




33020 0373

Chlorophyll-c1

1 c2



3634 0312





(b)

(a)





(d)

(c)









1 400 39086 00426 000162 428 43015 00696 000813 454 45473 00379 001024 468 47059 00348 001065 482 48800 00272 000266 494 49666 00217 000007 514 51499 00148 000308 530 53479 00138 000549 556 55378 00098 00042

10 588 58746 00115 0001311 630 63449 00143 0001312 674 67531 00336 00073



Chlorophyll-c3


















(MODIS) and MERIS)




References




















and c2































(a)





(b)






Gaussian function


0)

DX1 2

BD 2(1)

Lorentzian function


0)

DX1 2

B2D Otilde 1(2)

where Y0


0and DX



p h(l) of


















p h(l) into several


ap h

(l) 5 ni=1

Ci(l

m i) a

iG 1 b

iL (3)

where


)2

wi

B (4)



wiBD Otilde 1

(5)







bi


m i and C












m i








1 chlorophyll-



11 c

2)


to c2


1 chlorophyll-c



D 5 05[3 Chl-a 1 Chl-(c1







1446 580 628

Chlorophyll-c2












chlorophyll-c1







11 c




(b)

(a)


(b)

(a)












Chl-c3

040 308646 18392 030Chl-c

11 -c

2263724 278719 119140








1 418 41472 05214 000002 442 43713 01587 000003 466 46688 04666 000964 500 50079 03456 0012856 544 54431 02151 0000078 586 58411 03158 000009 616 61654 02623 00062

10 638 64262 02101 0019311 680 68148 03455 01990



1 414 41140 04741 000002 440 43050 01583 000003 466 46193 02885 002004 500 50808 01465 00132

8 590 57599 01956 000119 620 62017 01737 00000

10 640 64312 01112 0014211 680 67880 02845 00010










11 c















1 408 40700 01578 00002 434 43412 03391 00003 480 47900 01605 00004 538 53700 00151 000055 578 57900 00177 000016 622 62300 00304 000057 662 66300 01094 00036



1 408 40700 00001 000002 430 42900 00700 000003 458 45700 00437 000004 484 48500 00300 000005 542 54100 00065 000006 576 57641 00022 000007 616 61500 00076 000008 662 66100 00380 00000




33020 0373

Chlorophyll-c1

1 c2



3634 0312





(b)

(a)





(d)

(c)









1 400 39086 00426 000162 428 43015 00696 000813 454 45473 00379 001024 468 47059 00348 001065 482 48800 00272 000266 494 49666 00217 000007 514 51499 00148 000308 530 53479 00138 000549 556 55378 00098 00042

10 588 58746 00115 0001311 630 63449 00143 0001312 674 67531 00336 00073



Chlorophyll-c3


















(MODIS) and MERIS)




References




















and c2
























(b)






Gaussian function


0)

DX1 2

BD 2(1)

Lorentzian function


0)

DX1 2

B2D Otilde 1(2)

where Y0


0and DX



p h(l) of


















p h(l) into several


ap h

(l) 5 ni=1

Ci(l

m i) a

iG 1 b

iL (3)

where


)2

wi

B (4)



wiBD Otilde 1

(5)







bi


m i and C












m i








1 chlorophyll-



11 c

2)


to c2


1 chlorophyll-c



D 5 05[3 Chl-a 1 Chl-(c1







1446 580 628

Chlorophyll-c2












chlorophyll-c1







11 c




(b)

(a)


(b)

(a)












Chl-c3

040 308646 18392 030Chl-c

11 -c

2263724 278719 119140








1 418 41472 05214 000002 442 43713 01587 000003 466 46688 04666 000964 500 50079 03456 0012856 544 54431 02151 0000078 586 58411 03158 000009 616 61654 02623 00062

10 638 64262 02101 0019311 680 68148 03455 01990



1 414 41140 04741 000002 440 43050 01583 000003 466 46193 02885 002004 500 50808 01465 00132

8 590 57599 01956 000119 620 62017 01737 00000

10 640 64312 01112 0014211 680 67880 02845 00010










11 c















1 408 40700 01578 00002 434 43412 03391 00003 480 47900 01605 00004 538 53700 00151 000055 578 57900 00177 000016 622 62300 00304 000057 662 66300 01094 00036



1 408 40700 00001 000002 430 42900 00700 000003 458 45700 00437 000004 484 48500 00300 000005 542 54100 00065 000006 576 57641 00022 000007 616 61500 00076 000008 662 66100 00380 00000




33020 0373

Chlorophyll-c1

1 c2



3634 0312





(b)

(a)





(d)

(c)









1 400 39086 00426 000162 428 43015 00696 000813 454 45473 00379 001024 468 47059 00348 001065 482 48800 00272 000266 494 49666 00217 000007 514 51499 00148 000308 530 53479 00138 000549 556 55378 00098 00042

10 588 58746 00115 0001311 630 63449 00143 0001312 674 67531 00336 00073



Chlorophyll-c3


















(MODIS) and MERIS)




References




















and c2

























Gaussian function


0)

DX1 2

BD 2(1)

Lorentzian function


0)

DX1 2

B2D Otilde 1(2)

where Y0


0and DX



p h(l) of


















p h(l) into several


ap h

(l) 5 ni=1

Ci(l

m i) a

iG 1 b

iL (3)

where


)2

wi

B (4)



wiBD Otilde 1

(5)







bi


m i and C












m i








1 chlorophyll-



11 c

2)


to c2


1 chlorophyll-c



D 5 05[3 Chl-a 1 Chl-(c1







1446 580 628

Chlorophyll-c2












chlorophyll-c1







11 c




(b)

(a)


(b)

(a)












Chl-c3

040 308646 18392 030Chl-c

11 -c

2263724 278719 119140








1 418 41472 05214 000002 442 43713 01587 000003 466 46688 04666 000964 500 50079 03456 0012856 544 54431 02151 0000078 586 58411 03158 000009 616 61654 02623 00062

10 638 64262 02101 0019311 680 68148 03455 01990



1 414 41140 04741 000002 440 43050 01583 000003 466 46193 02885 002004 500 50808 01465 00132

8 590 57599 01956 000119 620 62017 01737 00000

10 640 64312 01112 0014211 680 67880 02845 00010










11 c















1 408 40700 01578 00002 434 43412 03391 00003 480 47900 01605 00004 538 53700 00151 000055 578 57900 00177 000016 622 62300 00304 000057 662 66300 01094 00036



1 408 40700 00001 000002 430 42900 00700 000003 458 45700 00437 000004 484 48500 00300 000005 542 54100 00065 000006 576 57641 00022 000007 616 61500 00076 000008 662 66100 00380 00000




33020 0373

Chlorophyll-c1

1 c2



3634 0312





(b)

(a)





(d)

(c)









1 400 39086 00426 000162 428 43015 00696 000813 454 45473 00379 001024 468 47059 00348 001065 482 48800 00272 000266 494 49666 00217 000007 514 51499 00148 000308 530 53479 00138 000549 556 55378 00098 00042

10 588 58746 00115 0001311 630 63449 00143 0001312 674 67531 00336 00073



Chlorophyll-c3


















(MODIS) and MERIS)




References




















and c2
































p h(l) into several


ap h

(l) 5 ni=1

Ci(l

m i) a

iG 1 b

iL (3)

where


)2

wi

B (4)



wiBD Otilde 1

(5)







bi


m i and C












m i








1 chlorophyll-



11 c

2)


to c2


1 chlorophyll-c



D 5 05[3 Chl-a 1 Chl-(c1







1446 580 628

Chlorophyll-c2












chlorophyll-c1







11 c




(b)

(a)


(b)

(a)












Chl-c3

040 308646 18392 030Chl-c

11 -c

2263724 278719 119140








1 418 41472 05214 000002 442 43713 01587 000003 466 46688 04666 000964 500 50079 03456 0012856 544 54431 02151 0000078 586 58411 03158 000009 616 61654 02623 00062

10 638 64262 02101 0019311 680 68148 03455 01990



1 414 41140 04741 000002 440 43050 01583 000003 466 46193 02885 002004 500 50808 01465 00132

8 590 57599 01956 000119 620 62017 01737 00000

10 640 64312 01112 0014211 680 67880 02845 00010










11 c















1 408 40700 01578 00002 434 43412 03391 00003 480 47900 01605 00004 538 53700 00151 000055 578 57900 00177 000016 622 62300 00304 000057 662 66300 01094 00036



1 408 40700 00001 000002 430 42900 00700 000003 458 45700 00437 000004 484 48500 00300 000005 542 54100 00065 000006 576 57641 00022 000007 616 61500 00076 000008 662 66100 00380 00000




33020 0373

Chlorophyll-c1

1 c2



3634 0312





(b)

(a)





(d)

(c)









1 400 39086 00426 000162 428 43015 00696 000813 454 45473 00379 001024 468 47059 00348 001065 482 48800 00272 000266 494 49666 00217 000007 514 51499 00148 000308 530 53479 00138 000549 556 55378 00098 00042

10 588 58746 00115 0001311 630 63449 00143 0001312 674 67531 00336 00073



Chlorophyll-c3


















(MODIS) and MERIS)




References




















and c2





























bi


m i and C












m i








1 chlorophyll-



11 c

2)


to c2


1 chlorophyll-c



D 5 05[3 Chl-a 1 Chl-(c1







1446 580 628

Chlorophyll-c2












chlorophyll-c1







11 c




(b)

(a)


(b)

(a)












Chl-c3

040 308646 18392 030Chl-c

11 -c

2263724 278719 119140








1 418 41472 05214 000002 442 43713 01587 000003 466 46688 04666 000964 500 50079 03456 0012856 544 54431 02151 0000078 586 58411 03158 000009 616 61654 02623 00062

10 638 64262 02101 0019311 680 68148 03455 01990



1 414 41140 04741 000002 440 43050 01583 000003 466 46193 02885 002004 500 50808 01465 00132

8 590 57599 01956 000119 620 62017 01737 00000

10 640 64312 01112 0014211 680 67880 02845 00010










11 c















1 408 40700 01578 00002 434 43412 03391 00003 480 47900 01605 00004 538 53700 00151 000055 578 57900 00177 000016 622 62300 00304 000057 662 66300 01094 00036



1 408 40700 00001 000002 430 42900 00700 000003 458 45700 00437 000004 484 48500 00300 000005 542 54100 00065 000006 576 57641 00022 000007 616 61500 00076 000008 662 66100 00380 00000




33020 0373

Chlorophyll-c1

1 c2



3634 0312





(b)

(a)





(d)

(c)









1 400 39086 00426 000162 428 43015 00696 000813 454 45473 00379 001024 468 47059 00348 001065 482 48800 00272 000266 494 49666 00217 000007 514 51499 00148 000308 530 53479 00138 000549 556 55378 00098 00042

10 588 58746 00115 0001311 630 63449 00143 0001312 674 67531 00336 00073



Chlorophyll-c3


















(MODIS) and MERIS)




References




















and c2



























1 chlorophyll-



11 c

2)


to c2


1 chlorophyll-c



D 5 05[3 Chl-a 1 Chl-(c1







1446 580 628

Chlorophyll-c2












chlorophyll-c1







11 c




(b)

(a)


(b)

(a)












Chl-c3

040 308646 18392 030Chl-c

11 -c

2263724 278719 119140








1 418 41472 05214 000002 442 43713 01587 000003 466 46688 04666 000964 500 50079 03456 0012856 544 54431 02151 0000078 586 58411 03158 000009 616 61654 02623 00062

10 638 64262 02101 0019311 680 68148 03455 01990



1 414 41140 04741 000002 440 43050 01583 000003 466 46193 02885 002004 500 50808 01465 00132

8 590 57599 01956 000119 620 62017 01737 00000

10 640 64312 01112 0014211 680 67880 02845 00010










11 c















1 408 40700 01578 00002 434 43412 03391 00003 480 47900 01605 00004 538 53700 00151 000055 578 57900 00177 000016 622 62300 00304 000057 662 66300 01094 00036



1 408 40700 00001 000002 430 42900 00700 000003 458 45700 00437 000004 484 48500 00300 000005 542 54100 00065 000006 576 57641 00022 000007 616 61500 00076 000008 662 66100 00380 00000




33020 0373

Chlorophyll-c1

1 c2



3634 0312





(b)

(a)





(d)

(c)









1 400 39086 00426 000162 428 43015 00696 000813 454 45473 00379 001024 468 47059 00348 001065 482 48800 00272 000266 494 49666 00217 000007 514 51499 00148 000308 530 53479 00138 000549 556 55378 00098 00042

10 588 58746 00115 0001311 630 63449 00143 0001312 674 67531 00336 00073



Chlorophyll-c3


















(MODIS) and MERIS)




References




















and c2

































chlorophyll-c1







11 c




(b)

(a)


(b)

(a)












Chl-c3

040 308646 18392 030Chl-c

11 -c

2263724 278719 119140








1 418 41472 05214 000002 442 43713 01587 000003 466 46688 04666 000964 500 50079 03456 0012856 544 54431 02151 0000078 586 58411 03158 000009 616 61654 02623 00062

10 638 64262 02101 0019311 680 68148 03455 01990



1 414 41140 04741 000002 440 43050 01583 000003 466 46193 02885 002004 500 50808 01465 00132

8 590 57599 01956 000119 620 62017 01737 00000

10 640 64312 01112 0014211 680 67880 02845 00010










11 c















1 408 40700 01578 00002 434 43412 03391 00003 480 47900 01605 00004 538 53700 00151 000055 578 57900 00177 000016 622 62300 00304 000057 662 66300 01094 00036



1 408 40700 00001 000002 430 42900 00700 000003 458 45700 00437 000004 484 48500 00300 000005 542 54100 00065 000006 576 57641 00022 000007 616 61500 00076 000008 662 66100 00380 00000




33020 0373

Chlorophyll-c1

1 c2



3634 0312





(b)

(a)





(d)

(c)









1 400 39086 00426 000162 428 43015 00696 000813 454 45473 00379 001024 468 47059 00348 001065 482 48800 00272 000266 494 49666 00217 000007 514 51499 00148 000308 530 53479 00138 000549 556 55378 00098 00042

10 588 58746 00115 0001311 630 63449 00143 0001312 674 67531 00336 00073



Chlorophyll-c3


















(MODIS) and MERIS)




References




















and c2
























(b)

(a)


(b)

(a)












Chl-c3

040 308646 18392 030Chl-c

11 -c

2263724 278719 119140








1 418 41472 05214 000002 442 43713 01587 000003 466 46688 04666 000964 500 50079 03456 0012856 544 54431 02151 0000078 586 58411 03158 000009 616 61654 02623 00062

10 638 64262 02101 0019311 680 68148 03455 01990



1 414 41140 04741 000002 440 43050 01583 000003 466 46193 02885 002004 500 50808 01465 00132

8 590 57599 01956 000119 620 62017 01737 00000

10 640 64312 01112 0014211 680 67880 02845 00010










11 c















1 408 40700 01578 00002 434 43412 03391 00003 480 47900 01605 00004 538 53700 00151 000055 578 57900 00177 000016 622 62300 00304 000057 662 66300 01094 00036



1 408 40700 00001 000002 430 42900 00700 000003 458 45700 00437 000004 484 48500 00300 000005 542 54100 00065 000006 576 57641 00022 000007 616 61500 00076 000008 662 66100 00380 00000




33020 0373

Chlorophyll-c1

1 c2



3634 0312





(b)

(a)





(d)

(c)









1 400 39086 00426 000162 428 43015 00696 000813 454 45473 00379 001024 468 47059 00348 001065 482 48800 00272 000266 494 49666 00217 000007 514 51499 00148 000308 530 53479 00138 000549 556 55378 00098 00042

10 588 58746 00115 0001311 630 63449 00143 0001312 674 67531 00336 00073



Chlorophyll-c3


















(MODIS) and MERIS)




References




















and c2
























(b)

(a)












Chl-c3

040 308646 18392 030Chl-c

11 -c

2263724 278719 119140








1 418 41472 05214 000002 442 43713 01587 000003 466 46688 04666 000964 500 50079 03456 0012856 544 54431 02151 0000078 586 58411 03158 000009 616 61654 02623 00062

10 638 64262 02101 0019311 680 68148 03455 01990



1 414 41140 04741 000002 440 43050 01583 000003 466 46193 02885 002004 500 50808 01465 00132

8 590 57599 01956 000119 620 62017 01737 00000

10 640 64312 01112 0014211 680 67880 02845 00010










11 c















1 408 40700 01578 00002 434 43412 03391 00003 480 47900 01605 00004 538 53700 00151 000055 578 57900 00177 000016 622 62300 00304 000057 662 66300 01094 00036



1 408 40700 00001 000002 430 42900 00700 000003 458 45700 00437 000004 484 48500 00300 000005 542 54100 00065 000006 576 57641 00022 000007 616 61500 00076 000008 662 66100 00380 00000




33020 0373

Chlorophyll-c1

1 c2



3634 0312





(b)

(a)





(d)

(c)









1 400 39086 00426 000162 428 43015 00696 000813 454 45473 00379 001024 468 47059 00348 001065 482 48800 00272 000266 494 49666 00217 000007 514 51499 00148 000308 530 53479 00138 000549 556 55378 00098 00042

10 588 58746 00115 0001311 630 63449 00143 0001312 674 67531 00336 00073



Chlorophyll-c3


















(MODIS) and MERIS)




References




















and c2
































Chl-c3

040 308646 18392 030Chl-c

11 -c

2263724 278719 119140








1 418 41472 05214 000002 442 43713 01587 000003 466 46688 04666 000964 500 50079 03456 0012856 544 54431 02151 0000078 586 58411 03158 000009 616 61654 02623 00062

10 638 64262 02101 0019311 680 68148 03455 01990



1 414 41140 04741 000002 440 43050 01583 000003 466 46193 02885 002004 500 50808 01465 00132

8 590 57599 01956 000119 620 62017 01737 00000

10 640 64312 01112 0014211 680 67880 02845 00010










11 c















1 408 40700 01578 00002 434 43412 03391 00003 480 47900 01605 00004 538 53700 00151 000055 578 57900 00177 000016 622 62300 00304 000057 662 66300 01094 00036



1 408 40700 00001 000002 430 42900 00700 000003 458 45700 00437 000004 484 48500 00300 000005 542 54100 00065 000006 576 57641 00022 000007 616 61500 00076 000008 662 66100 00380 00000




33020 0373

Chlorophyll-c1

1 c2



3634 0312





(b)

(a)





(d)

(c)









1 400 39086 00426 000162 428 43015 00696 000813 454 45473 00379 001024 468 47059 00348 001065 482 48800 00272 000266 494 49666 00217 000007 514 51499 00148 000308 530 53479 00138 000549 556 55378 00098 00042

10 588 58746 00115 0001311 630 63449 00143 0001312 674 67531 00336 00073



Chlorophyll-c3


















(MODIS) and MERIS)




References




















and c2





























1 418 41472 05214 000002 442 43713 01587 000003 466 46688 04666 000964 500 50079 03456 0012856 544 54431 02151 0000078 586 58411 03158 000009 616 61654 02623 00062

10 638 64262 02101 0019311 680 68148 03455 01990



1 414 41140 04741 000002 440 43050 01583 000003 466 46193 02885 002004 500 50808 01465 00132

8 590 57599 01956 000119 620 62017 01737 00000

10 640 64312 01112 0014211 680 67880 02845 00010










11 c















1 408 40700 01578 00002 434 43412 03391 00003 480 47900 01605 00004 538 53700 00151 000055 578 57900 00177 000016 622 62300 00304 000057 662 66300 01094 00036



1 408 40700 00001 000002 430 42900 00700 000003 458 45700 00437 000004 484 48500 00300 000005 542 54100 00065 000006 576 57641 00022 000007 616 61500 00076 000008 662 66100 00380 00000




33020 0373

Chlorophyll-c1

1 c2



3634 0312





(b)

(a)





(d)

(c)









1 400 39086 00426 000162 428 43015 00696 000813 454 45473 00379 001024 468 47059 00348 001065 482 48800 00272 000266 494 49666 00217 000007 514 51499 00148 000308 530 53479 00138 000549 556 55378 00098 00042

10 588 58746 00115 0001311 630 63449 00143 0001312 674 67531 00336 00073



Chlorophyll-c3


















(MODIS) and MERIS)




References




















and c2
































11 c















1 408 40700 01578 00002 434 43412 03391 00003 480 47900 01605 00004 538 53700 00151 000055 578 57900 00177 000016 622 62300 00304 000057 662 66300 01094 00036



1 408 40700 00001 000002 430 42900 00700 000003 458 45700 00437 000004 484 48500 00300 000005 542 54100 00065 000006 576 57641 00022 000007 616 61500 00076 000008 662 66100 00380 00000




33020 0373

Chlorophyll-c1

1 c2



3634 0312





(b)

(a)





(d)

(c)









1 400 39086 00426 000162 428 43015 00696 000813 454 45473 00379 001024 468 47059 00348 001065 482 48800 00272 000266 494 49666 00217 000007 514 51499 00148 000308 530 53479 00138 000549 556 55378 00098 00042

10 588 58746 00115 0001311 630 63449 00143 0001312 674 67531 00336 00073



Chlorophyll-c3


















(MODIS) and MERIS)




References




















and c2


































1 408 40700 01578 00002 434 43412 03391 00003 480 47900 01605 00004 538 53700 00151 000055 578 57900 00177 000016 622 62300 00304 000057 662 66300 01094 00036



1 408 40700 00001 000002 430 42900 00700 000003 458 45700 00437 000004 484 48500 00300 000005 542 54100 00065 000006 576 57641 00022 000007 616 61500 00076 000008 662 66100 00380 00000




33020 0373

Chlorophyll-c1

1 c2



3634 0312





(b)

(a)





(d)

(c)









1 400 39086 00426 000162 428 43015 00696 000813 454 45473 00379 001024 468 47059 00348 001065 482 48800 00272 000266 494 49666 00217 000007 514 51499 00148 000308 530 53479 00138 000549 556 55378 00098 00042

10 588 58746 00115 0001311 630 63449 00143 0001312 674 67531 00336 00073



Chlorophyll-c3


















(MODIS) and MERIS)




References




















and c2
























(b)

(a)





(d)

(c)









1 400 39086 00426 000162 428 43015 00696 000813 454 45473 00379 001024 468 47059 00348 001065 482 48800 00272 000266 494 49666 00217 000007 514 51499 00148 000308 530 53479 00138 000549 556 55378 00098 00042

10 588 58746 00115 0001311 630 63449 00143 0001312 674 67531 00336 00073



Chlorophyll-c3


















(MODIS) and MERIS)




References




















and c2
























(d)

(c)









1 400 39086 00426 000162 428 43015 00696 000813 454 45473 00379 001024 468 47059 00348 001065 482 48800 00272 000266 494 49666 00217 000007 514 51499 00148 000308 530 53479 00138 000549 556 55378 00098 00042

10 588 58746 00115 0001311 630 63449 00143 0001312 674 67531 00336 00073



Chlorophyll-c3


















(MODIS) and MERIS)




References




















and c2





























1 400 39086 00426 000162 428 43015 00696 000813 454 45473 00379 001024 468 47059 00348 001065 482 48800 00272 000266 494 49666 00217 000007 514 51499 00148 000308 530 53479 00138 000549 556 55378 00098 00042

10 588 58746 00115 0001311 630 63449 00143 0001312 674 67531 00336 00073



Chlorophyll-c3


















(MODIS) and MERIS)




References




















and c2







































(MODIS) and MERIS)




References




















and c2
























References




















and c2























Download - The identification of phytoplankton pigments from ...ajit/Papers/aguirre... · pigments. The spectra used are modelled phytoplankton, spectrophotometric measurements of algal cultures

Top Related