RSE-09413; No of Pages 10

Remote Sensing of Environment xxx (2015) xxxxxx

Contents lists available at ScienceDirect

Remote Sensing of Environment

journal homepage: www.elsevier.com/locate/rse

Characterization of phytoplankton variability in the Cariaco Basin using

spectral absorption, taxonomic and pigment data
L. Lorenzoni a,, G. Toro-Farmer a, R. Varela b, L. Guzman b, J. Rojas b, E. Montes a, F. Muller-Karger a
a
University of South Florida, College of Marine Science, St. Petersburg, FL 33701, United States
b
Fundacin La Salle de Ciencias Naturales, EDIMAR, Porlamar, Edo Nueva Esparta, Venezuela

a r t i c l e i n f o a b s t r a c t

Article history: The spectral absorption coefcient of marine phytoplankton provides information on phytoplankton community
Received 1 July 2014 structure, biomass, and general physiological conditions. These variables are necessary for understanding and
Received in revised form 5 May 2015 predicting ocean productivity, carbon uxes, underwater light propagation, water quality, and for assessing
Accepted 13 May 2015
marine photochemical processes. The Cariaco Basin, located on the continental shelf of Venezuela in the south-
Available online xxxx
eastern Caribbean Sea, is the site of the CARIACO Ocean Time-Series project. Since 1995, CARIACO has collected
Keywords:
bio-optical (hyperspectral inherent and apparent optical properties IOPs and AOPs, respectively), biogeochem-
Hyperspectral ical and ecological observations to characterize local ecosystem variations in response to regional and global
Phytoplankton absorption changes in climate. We examine phytoplankton taxonomic and pigment time series data collected by this
Cariaco Basin program between 2006 and 2012 to understand how seasonal changes in these parameters relate to bio-
HPLC optical data (i.e., absorption spectra). TChla and accessory pigments varied seasonally in response to changes
POC:TChla in the phytoplankton community composition, with higher concentrations of microphytoplankton (N 20 m;
45%) during upwelling (DecemberApril) than during the rainy season (16%; May/JuneOctober/November).
Picophytoplankton (b 2 m) dominated during the rainy season (66%). The absorption properties also exhibited
seasonal variations. Diagnostic pigments could not be identied in a quantitative way using derivative analysis of
phytoplankton absorption, likely because of overlapping of absorption spectra among the pigments present. The
POC:TChla ratio at CARIACO was variable and dependent on bulk carbon (not necessarily related to phytoplankton)
and the functional groups present at any given time, underscoring the fact that using a xed ratio of POC:Chla in
biogeochemical models can lead to large uncertainties in carbon budgets from coastal zones. Low POC:TChla was
associated with microphytoplankton size class (diatoms), while picophytoplankton (cyanobacteria) exhibited
higher ratios. These results contribute to furthering our understanding of coastal phytoplankton dynamics and
how they relate to optical signatures.
2015 Published by Elsevier Inc.

1. Introduction The biogeochemistry and optical properties of continental margin

Advances in bio-optical methods to study phytoplankton focus
on the retrieval of products such as pigment concentration, phyto-
plankton size classes, and functional types from bio-optical data
(see IOCCG, 2014 for a complete review). Resolving shifts in phyto-
planktonic communities is important to understand the impacts of
both natural and anthropogenic changes on marine systems (Le Quere
et al., 2005; Muller-Karger et al., 2014). There is a need to acquire
more information on phytoplankton community structure, pigment
distribution and bio-optical properties in order to further the develop-
ment of algorithms required for a true global characterization of phyto-
plankton biodiversity from remote sensing (Devred, Sathyendranath,
Stuart, & Platt, 2011; Devred et al., 2013).
Corresponding author.
E-mail address: laural@mail.usf.edu (L. Lorenzoni).
waters vary widely because of high biological productivity, inputs of
land-derived material, and resuspension of benthic material (D'Sa,
Miller, & Del Castillo, 2006; Hoepffner & Sathyendranath, 1992). Addi-
tionally, these regions are particularly sensitive to environmental shifts,
and their hydrographic and biogeochemical conditions can change
within short timescales; resolving shifts in phytoplanktonic populations
in these ecosystems is therefore important to better determine the
impacts of climate change on coastal marine systems (e.g., Le Quere
et al., 2005).
Features of the spectral phytoplankton absorption coefcient have
been used to infer phytoplankton size and taxonomic information
(Bricaud & Stramski, 1990; Bricaud, Claustre, Ras, & Oubelkheir, 2004;
Ciotti, Lewis, & Cullen, 2002; Vijayan & Somayajula, 2014). These tech-
niques are not as sensitive as direct pigment analyses, especially in
optically-complex waters with high spatial and temporal variability in
both phytoplankton and detritus concentration and composition
(Bricaud & Stramski, 1990; Bricaud et al., 2004; Ciotti et al., 2002).

http://dx.doi.org/10.1016/j.rse.2015.05.002
0034-4257/ 2015 Published by Elsevier Inc.

Please cite this article as: Lorenzoni, L., et al., Characterization of phytoplankton variability in the Cariaco Basin using spectral absorption,
taxonomic and pigment data, Remote Sensing of Environment (2015), http://dx.doi.org/10.1016/j.rse.2015.05.002
2 L. Lorenzoni et al. / Remote Sensing of Environment xxx (2015) xxxxxx
Nevertheless, hyperspectral particulate absorption coefcient measure-
ments may potentially provide useful information (Bidigare, Morrow, &
Kiefer, 1989; Moisan, Moisan, & Linkswiler, 2011; Organelli, Bricaud,
Antoine, & Uitz, 2013; Torrecilla, Piera, & Vilaseca, 2009; Torrecilla,
Stramski, Reynolds, Milln-Nez, & Piera, 2011).
Absorption coefcients can also be obtained from remote sensing
reectance through bio-optical models (e.g., Roesler and Perry, 1995;
Garver & Siegel, 1997; Lee & Carder, 2004). This can potentially enable
the use of satellite data to derive phytoplankton absorption spectra,
and provide information on size and composition. Hyperspectral
space-based sensors, such as the upcoming Hyperspectral Infrared
Imager (HyspIRI), the Pre-Aerosol, Cloud, and ocean Ecosystem Mission
(PACE), and the GEOstationary Coastal and Air Pollution Events (GEO-
CAPE), planned by the National Aeronautics and Space Administration
(NASA), will be able to provide remote sensing data and derivable ab-
sorption coefcients, to study phytoplankton dynamics in coastal envi-
ronments, at regional to global spatial scales, of unparalleled spectral
and spatialtemporal resolution. More specically, the goals of a
HyspIRI-class mission include providing observations needed to address
some key science questions identied in the National Research Council
Decadal Survey (NRC, 2007) and rened by the HyspIRI Science Study
Groups and research community (http://hyspiri.jpl.nasa.gov/science).
Such a mission would help address questions requiring a combination
of VSWIR and TIR data. This includes focusing on local and landscape-
scale changes in inland, coastal, and open ocean aquatic ecosystems.
Understanding changes in carbon cycling is also fundamental to
build accurate biogeochemical budgets for the ocean. More pressing is
the need to accurately quantify phytoplankton carbon remotely, and
understand how it changes through time. Phytoplankton pigment con-
centration is typically used as an index of phytoplankton biomass. In
particular, total chlorophyll a (TChla) has been used as a proxy for phy-
toplankton carbon (C) biomass. However, the relationship between C
and TChla is variable. It depends on a variety of factors such as nutrient
availability, light, temperature and taxononomic groups (Behrenfeld,
Boss, Siegel, & Shea, 2005; Longhurst & Harrison, 1989; Montagnes,
Berges, Harrison, & Taylor, 1994). In order to develop reliable methods
for estimating carbon-based biomass and primary production from sat-
ellite remote sensing observations, it is essential to accurately parame-
terize CTChla relationships.
We examined the relationship between phytoplankton taxonomic,
pigment concentration, and bio-optical data collected monthly by the
CARIACO Ocean Time-Series Project between 2006 and 2012 in the
Cariaco Basin, located in the southeastern Caribbean Sea. We explored
phytoplankton taxonomic and pigment time series data to understand
how seasonal changes in biological parameters relate to bio-optical
data, specically absorption coefcients. We also assessed the relation-
ship between carbon, chlorophyll a, and phytoplankton size classes with
the aim of understanding the potential retrieval of these by remote
sensing measurements.
2. Methods
2.1. Study area
The Cariaco Basin is a small tectonic basin located off eastern
Venezuela (Fig. 1). It contains two sub-basins, each approximately
1400 m deep and divided by a saddle of ~900 m (Schubert, 1982). The
Cariaco Basin is open to the north to the Caribbean Sea in the upper
~140 m, and to the south it features a wide, shallow platform. Most of
the sediment that enters the basin comes from the small local rivers
that drain directly onto the Unare Platform. The hydrology and
hydrography of the region are inuenced by the position of the Inter-
tropical Convergence Zone (ITCZ). During boreal winter (December
April), strong Trade Winds (N6 m s 1) induce wind-driven upwell-
ing in the southern Caribbean Sea. This stimulates high primary pro-
duction and associated vertical and horizontal organic matter export
Please cite this article as: Lorenzoni, L., et al., Characterization of phytoplankton variability in the Cariaco Basin using spectral absorption,
taxonomic and pigment data, Remote Sensing of Environment (2015), http://dx.doi.org/10.1016/j.rse.2015.05.002
(Muller-Karger et al., 2004; Mller-Karger et al., 2010). During boreal
summer (May/JuneOctober/November), when the ITCZ has a more
northward position, wind-driven upwelling becomes weaker and pre-
cipitation increases (Astor, Meri, & Muller Karger, 1998; Lorenzoni
et al., 2009). A short (b1 month) secondary upwelling event occurs reg-
ularly between June and July, linked to variations in the wind curl
(Rueda-Roa, 2012).
The CARIACO Ocean Time-Series Project has been collecting oceano-
graphic observations at the CARIACO station (1030 N, 6440 W),
within the Cariaco Basin, since 1995. These data have helped character-
ize the seasonality of biogeochemical processes in the basin (Astor et al.,
1998; Muller-Karger et al., 2004; Mller-Karger et al., 2010; Thunell,
Benitez-Nelson, Varela, Astor, & Mller-Karger, 2007). Between about
1995 and 1999, observations show regular cycles of strong upwelling.
During the 2000s, and particularly during 20042005, intensity of the
Trade Wind over the southern Caribbean decreased during the boreal
winter, resulting in a reduction in upwelling intensity. This led to
lower annual primary productivity and ecosystem shifts in the Cariaco
Basin (Taylor et al., 2012) with respect to average conditions during pre-
vious years.
2.2. Measurements and data analysis
A total of 68 monthly cruises to the CARIACO Ocean Time-Series sta-
tion were conducted between July 2006 and August 2012 using the R/V
Hermano Gins of the Fundacin la Salle de Ciencias Naturales (FLASA).
Only surface (1 m) data were used in this study. Salinity and tempera-
ture data were collected with a Seabird SBE25 CTD mounted on a 12
8-L bottle rosette. For phytoplankton taxonomic identication, 500 ml
of water were collected into HDPE bottles and xed with a 4% formalin
solution neutralized with sodium tetraborate. For high performance
liquid chromatography (HPLC) analyses, between 0.5 and 2 L of water
(depending on the biomass concentration present) were vacuum-
ltered through a 47 mm glass ber lter (Whatman GF/F, 0.7 m
pore size). Filters were stored folded in aluminum foil at 20 C at
sea. Once on land, the HPLC samples were moved to a 40 C freezer
until analyzed. Particulate organic carbon (POC) was sampled by
vacuum-ltering 2 L of water through a 47 mm glass ber lter
(Whatman GF/F, 0.7 m pore size); lters were refrigerated and sub-
sequently oven-dried at 60 C. POC concentrations were obtained by
running the lters through a Perkin Elmer 2400 elemental analyzer.
Phytoplankton taxonomy was analyzed at the Universidad de
Oriente (Boca de Ro), Venezuela, using the Utermhl technique
(Hasle, 1978) with 100 mL sedimentation chambers and a settling peri-
od of 48 h. Taxonomic identication was done through an inverted
microscope (Axiovert Carl Zeiss). HPLC analyses were conducted at
Horn Point Laboratory, Maryland (HPL; from July 2006 to July 2010),
and at the NASA Goddard Space Flight Center, Maryland (GSFC; From
August 2010 to August 2012) following the method described in Van
Heukelem and Thomas (2001) and Hooker et al. (2005). All data are
openly available online (CARIACO Home Page: http://imars.marine.
usf.edu/cariaco/cariaco-ocean-time-series-program; BCO-DMO; http://
www.bco-dmo.org/project/2047). The major HPLC pigments used in
this manuscript are shown in Table 1. Seven major diagnostic pigments
(DP) were used to calculate the proportions of micro-, nano-, and pico-
phytoplankton present at any given time, following Uitz, Claustre,
Morel, and Hooker (2006). Specically, the fraction of each pigment-
based size class was calculated as:
Microphytoplankton N20 m 1:41Fuco 1:41Peri=DP
 
Nanophytoplankton 2 to 20 m 0:60Allo 0:35 190 BF
 0 
1:27 19 H F =DP
Picophytoplankton b2 m 0:86Zea 1:01TChlb=DP
 L. Lorenzoni et al. / Remote Sensing of Environment xxx (2015) xxxxxx 3

Fig. 1. The Cariaco Basin showing the location of the CARIACO Ocean Time-Series station (open circle).

where pigment abbreviations are as in Table 1; DP is the weighted sum
of the seven diagnostic pigments, as:
   
DP 1:41Fuco 1:41Peri 0:60Allo 0:35 190 B F 1:27 190 H F
0:86Zea 1:01TChlb:
To further quantify and investigate variations in accessory pigments,
these were grouped into photosynthetic carotenoids (PSC composed
of fuco-, peri-, 19BF, 19HF, and prasino-xanthin) and photoprotective
(nonphotosynthetic) carotenoids (PPC zea-, diadino-, allo-,
carotenes, diato-xanthin, violaxanthin and luthein) following
Bidigare, Ondrusek, Morow, and Kiefer (1990) and Bricaud et al.
(2004).
Total particle absorption coefcient was measured using the lter
pad technique (Mitchell & Kiefer, 1988). Water was ltered through
25 mm Whatman GF/F lters using a vacuum pump. The volume
ltered varied between 1 and 2 L depending on the season; during the
upwelling season lters usually clogged with less than 2 L. Filters
were stored in petri dishes and frozen at 20 C until analysis within
a few days of collection. Spectral optical density (OD()) was measured
on the lters from 380 to 732 nm at 4 nm intervals, using a PR-650
spectroradiometer (Photoresearch, Inc.; July 2006November 2009)
and a PR-655 (Photoresearch, Inc.; December 2009August 2012).

Table 1
Summary of phytoplankton taxonomy and pigment data for the period 20062012. Relative percentage abundance or concentration is shown, accompanied by standard deviation. Num-
ber of observations (n) is given in parentheses.

Pigment (concentrations in mg m3) Abbreviation Upwelling season Rainy season

Total chlorophyll a (monovinyl chlorophyll a + divinyl-chlorophyll TChla 1.18 1.84 (n = 28) 0.19 0.21 (n = 41)
a + chlorophyllide a + chlorophyll a epimer)
Total chlorophyll b (chlorophyll b + divinyl chlorophyll b) TChlb 0.043 0.037 (n = 23) 0.023 0.037 (n = 34)
Total chlorophyll c (chl c1 + chl c2 (chl c1c2) + chl c3) TChlc 0.25 0.41 (n = 24) 0.025 0.025 (n = 34)
-(,) + (,) carotenes carotenes 0.036 0.039 (n = 28) 0.015 0.005 (n = 41)
19-Butanoyloxyfucoxanthin 19-BF 0.011 0.007 (n = 27) 0.004 0.003 (n = 39)
19-Hexanoyloxyfucoxanthin 19-HF 0.048 0.038 (n = 28) 0.021 0.17 (n = 41)
Alloxanthin Allox 0.009 0.009 (n = 21) 0.005 0.007 (n = 9)
Diadinoxanthin Diadino 0.034 0.046 (n = 28) 0.005 0.004 (n = 41)
Fucoxanthin Fuco 0.45 0.88 (n = 28) 0.018 0.024 (n = 41)
Peridinin Perid 0.011 0.008 (n = 26) 0.006 0.005 (n = 39)
Zeaxanthin Zeax 0.035 0.028 (n = 26) 0.060 0.021 (n = 41)
Prasinoxanthin + diatoxanthin + violaxanthin + luthein Others 0.025 0.018 (n = 19) 0.033 0.033 (n = 16)

Phytoplankton (given in relative percentage abundance)

Diatoms 37.8% 36 (n = 28) 44.3% 31 (n = 38)
Dinoagellates 18.4% 24 (n = 28) 16.2% 20 (n = 38)
Coccolithophorids 37.9% 37 (n = 28) 32.3% 31 (n = 38)
Cyanobacteria 4.1% 13 (n = 28) 6.84% 16 (n = 38)
Others 1.88% 4 (n = 28) 0.33% 1 (n = 38)
Indicates signicant difference between seasons.

Please cite this article as: Lorenzoni, L., et al., Characterization of phytoplankton variability in the Cariaco Basin using spectral absorption,
taxonomic and pigment data, Remote Sensing of Environment (2015), http://dx.doi.org/10.1016/j.rse.2015.05.002
4 L. Lorenzoni et al. / Remote Sensing of Environment xxx (2015) xxxxxx
This provided a measure of the total particulate absorption coefcient
(ap; in m1). ap was estimated based on the initial OD measurement fol-
lowing:
1
ap ln 10  OD  s  Vf   way ANOVA (F(1,16) = 2.44, p = 0.13 and F(1,17) = 1.05, p = 0.32,
where s is the clearance area of the lter, Vf is the volume of seawater
ltered, and () is the path elongation factor (Bricaud & Stramski,
1990; Mitchell & Kiefer, 1988). The lters were washed with hot
(60 C) methanol (Kishino, Takahashi, Okami, & Ichimura, 1985) and
re-scanned to obtain an estimate of the detrital absorption coefcient
(ad). The absorption coefcient of phytoplankton (aph) was calculated
by subtracting ad from ap. Every aph spectra was corrected by
subtracting the absorption at 732 nm. The chlorophyll-specic absorp-
tion coefcient (a*ph(), m2 (mg chla) 1) was obtained dividing
aph() by the corresponding HPLC Tchla value (Moisan and Mitchell,
1999).
Derivative analysis was used for separating absorption peaks corre-
sponding to specic pigments. The 4th derivative maximum provides
a means of indirectly assessing concentration of various pigments
(Bidigare et al., 1990; Torrecilla et al., 2009). The derivative analysis
was performed on each individual monthly a*ph() spectrum using
Matlab (Mathworks) by calculating differences between adjacent
waveband elements over the entire spectral range of the data. To
test the spectral resolution necessary to detect and estimate the con-
centration of individual phytoplankton pigments from absorption
data, two different interpolations bandwidths were applied: 7 nm
and 10 nm. In addition, to remove any possible noise in the spectra,
a SavitzkyGolay smoothing lter was applied to each spectra prior
to the derivation.
A hierarchical cluster analysis was used to classify and group the
optical (a*ph and 4th derivative) data. The hierarchical cluster tree (den-
drogram) grouped similar objects together. The Euclidean distance be-
tween the pairs of objects in each matrix was evaluated using Matlab
(Mathworks). The inner squared distance (minimum variance) was
used as the distance between clusters.
Carbon and chlorophyll relationships were assessed using regres-
sions; a quantile regression was used to determine the carbon associat-
ed with phytoplankton following Sathyendranath et al. (2009). Briey,
with this regression the lowest envelope of the carbon:chlorophyll re-
lationship is obtained; the assumption is that the lowest particulate car-
bon content present in the data corresponds to phytoplankton carbon,
as opposed to the total particulate organic carbon present. The percen-
tile chosen was the rst percentile (q = 0.01).
3. Results and discussion
3.1. Phytoplankton composition and temporal variability
Phytoplankton pigment concentration at the CARIACO site showed
marked seasonality. During the upwelling season (DecemberApril)
for the period 20062012, surface Chlorophyll-a concentration (TChla)
was 1.18 1.84 mg m3 (n = 28). The highest surface TChla concen-
tration measured was 8.54 mg m3 (March 2012). In the rainy season
(MayJune through OctoberNovember), the average TChla value was
0.196 0.121 mg m3 (n = 41), with the highest TChla concentration
measured in August 2009 (0.26 mg m3). Seasonal averages for the rst
decade of the time-series (19952005, not shown in this work) were
higher for both seasons (~ 1.62 mg m3 and 0.38 mg m 3 for the
upwelling and non-upwelling period, respectively; Lorenzoni et al.,
2011). In 2010, 2011, and 2012, surface TChla values returned to values
above 2 mg m 3 in FebruaryMarch, compared to those observed
between 2004 and 2009. TChla values measured in March 2012
were the highest measured since the beginning of the time series
in 1995. Mean surface temperature for March 2012 of 22.58 C was
below climatological values for the same month between 1995 and
Please cite this article as: Lorenzoni, L., et al., Characterization of phytoplankton variability in the Cariaco Basin using spectral absorption,
taxonomic and pigment data, Remote Sensing of Environment (2015), http://dx.doi.org/10.1016/j.rse.2015.05.002
2012 (24.12 1.62 C), which may suggest a strong upwelling. Con-
versely, salinity was slightly lower (36.69) than the mean climatology
for March 19952012 (36.83 0.095). Temperature and salinity were
not statistically difference from the climatology as determined by one-
respectively). Dissolved silicate was signicantly higher, as determined
by one-way ANOVA (2.15 M; March 19952012 climatology = 0.91
0.94 M; F(1,17) = 6.19, p = 0.02). POC and terrigenous matter ux
measured by moored sediment traps (see methods in Thunell et al.,
2007 for sediment trap data) were nearly twice as high as the climato-
logical average during March 2012 (POC = 2.99 g m 2 mo 1;
terrigenous = 12.92 g m2 mo 1; March 19952012 climatology:
POC = 1.48 g m2 mo1; terrigenous = 6.32 g m2 mo1). This sug-
gests that the event was related to coastal runoff transporting nutrients
and terrigenous matter to the time-series site, through plumes as those
described by Lorenzoni et al. (2009) for this location, rather than
upwelling.
All accessory pigments, except zeaxanthin and TChlb, were signi-
cantly higher (T-test p b 0.01) during the upwelling season between
2006 and 2012. Elevated fucoxanthin and peridinin concentration dur-
ing upwelling (mean values of 0.45 0.87 and 0.011 0.008, respec-
tively) were indicative of high diatom and dinoagellate abundance
(Fig. 2, Table 1). There was less seasonal variation in peridinin (indica-
tive of dinoagellates) than in fucoxanthin (indicative of diatoms).
Coccolithophorids, identied by 19-HF, often showed elevated num-
bers in both upwelling and rainy seasons. TChlb showed no signicant
difference between the rainy and upwelling seasons (Table 1).
TChlb is a diagnostic pigment of smaller phytoplankton taxa, such
as chlorophytes and prochlorophytes. Even though the presence of
these groups has not been recorded at the CARIACO site due to limita-
tions of our microscopy methods (Mutshinda, Troccoli-Ghinaglia,
Finkel, Muller-Karger, & Irwin, 2013; Pinckney et al., 2015), the occur-
rence of TChlb suggests a year-round presence. The absence of TChlb
for the rst year of this dataset (July 2006June 2007) may have been
an analytical problem (i.e., sampling bias). However, disappearance of
TChlb from the observations has been previously noted for reasons
that we still don't understand (Lorenzoni et al., 2011).
Zeaxanthin, a diagnostic pigment for cyanobacteria, was signicantly
lower (p b 0.01) during upwelling (Fig. 2, Table 1). The presence of the
cyanobacterium Trichodesmium thiebautii is observed at the CARIACO
site on a year-round basis through microscopy, but its abundance is
signicantly higher during the absence of upwelling, when nutrients in
surface waters are at detection limit as a result of strong water column
thermal stratication (Montes et al., 2013; Mutshinda et al., 2013)
(Fig. 3, Table 1). Another cyanobacterium reported for CARIACO,
Synechococcus sp. (Mutshinda et al., 2013), was generally more abun-
dant during upwelling in early CARIACO samples. This species has
not been reported in microscopic counts since 2003 because of
uncertainty in the classication of such small cells. The fraction of
photosynthetic carotenoid relative to TChla ([PSC]/[TChla]) was
also higher during upwelling (0.36 0.12; rainy season = 0.23
0.08). In contrast, photoprotective carotenoid concentration, relative
to TChla ([PPC]/[TChla]) was more abundant during the rainy season
(0.42 0.25; upwelling season = 0.32 0.21); this may be expected
when waters are thermally stratied due to high solar irradiance condi-
tions (Porra et al., 1997), and in particular for these samples that corre-
spond solely to surface waters. The difference between pigment/TChla
ratio observed between seasons is also attributable to changes in the
community composition.
It is important to note that there was a difference between the taxo-
nomic data and the pigment data; while the pigments, and resulting
pigment-based size class division, showed a clear seasonality, with
abundance of microplankton during the upwelling season and a domi-
nance of smaller taxa (picoplankton) during the rainy period, the taxon-
omy did not reect this. Diatoms and dinoagellates were clearly
identied as dominant for the periods of upwelling (Fig. 3), and were
 L. Lorenzoni et al. / Remote Sensing of Environment xxx (2015) xxxxxx 5

Fig. 2. Percentage contribution of HPLC accessory pigments to total accessory pigments measured between 2006 and 2012. Black overlaid curve represents TChla concentration.

present nearly year-round, but the dominant picoplankter, cyanobacteria,
was rarely noted, contrasting the pigment results (Fig. 4). This discrep-
ancy is likely a shortcoming of the microscopic technique employed in
the identication of phytoplankton taxonomy in CARIACO, which is
biased toward larger and easily identiable cells (refer also Pinckney
et al., 2015). Thus, it is likely that the results obtained using pigment
analysis reect more accurately the richness of phytoplankton groups
present in the surface waters of the Cariaco Basin.
The seasonal shift in accessory pigments was mirrored in the
phytoplankton size classes calculated utilizing diagnostic pigments
and supported by microscopy analyses (Fig. 4). The upwelling season
was dominated by microphytoplankton (average for the period
20062012 of 45%). These included primarily diatoms and dinoa-
gellates (Fig. 3). During 19952005, the phytoplankton had over 60%
microphytoplankton (Lorenzoni et al., 2011; Rondon, 2009). The
lower percentage calculated for the 20062012 period is consistent
with the reports of Taylor et al. (2012) and Mutshinda et al. (2013).
Conversely, the nano- (nanoagellates and coccolithophorids) and
picophytoplankton (cyanobacteria and potentially chlorophytes
and prochlorophytes) contributed ~ 22% and 33%, respectively,
during upwelling. These percentages are higher than that observed
during the decade of 19952005, concomitant with the decrease of
microphytoplankton (Lorenzoni et al., 2011) and with the docu-
mented shift in phytoplankton taxon dominance from diatoms and
dinoagellates to smaller taxa after 2004 (Taylor et al., 2012).
During the rainy season, the population was dominated by
picophytoplankton (N65%), with micro and nanophytoplankton con-
tributing roughly the same each (~ 17%). On occasion, larger phyto-
plankton, together with higher associated DP (e.g., fucoxanthin and
peridinin), were observed during JuneAugust (e.g., July 2007; August
2010; Figs. 3 and 4). During this time, a short secondary upwelling
event takes place along the entire southern Caribbean Sea, linked to var-
iations in the wind curl (Rueda-Roa and Muller-Karger, 2013). These
stimulate primary production and thus increase the percentage
of microphytoplankton present. For the rainy seasons during the
decade 19952005, picophytoplankton was ~ 45%, while micro and
nanophytoplankton were ~ 33% and ~ 22%, respectively (Lorenzoni
et al., 2011); this increase in picophytoplankton during the rainy season
can be related to less secondary upwelling events observed since 2005
in the Basin (Rueda-Roa and Muller-Karger, 2013), as well as to an in-
crease in surface water stratication in the last decade (Astor et al.,
2013).
The phytoplankton specic absorption coefcient (a* ph ) also
showed marked seasonal variability, with lowest spectra during the
upwelling season when TChla concentrations were highest (Fig. 5).
This inverse correlation with TChla has been related to changes in
cell size and accessory pigment composition, in addition to the in-
verse proportionality to chlorophyll concentration in a*ph (Bricaud
& Stramski, 1990; Bricaud et al., 1998, 2004). In the Cariaco Basin
in particular, most of the variability in the phytoplankton specic ab-
sorption is caused by changes in accessory pigment concentration,
rather than by changes in packaging effect (Lorenzoni et al., 2011).
In order to better understand the variability of the chlorophyll-
specic absorption coefcient spectra (a*ph) and its derivative, cluster
analyses were performed. Fig. 6 shows the hierarchical clusters of the
a*ph spectra. The absorption-based cluster tree (Fig. 6 top) shows two
main clusters containing 36 (red) and 30 (blue) groups corresponding
to different cruises. When the clusters are analyzed based on HPLC pig-
ments (excluding TChla), the dendrograms suggest that the differences
observed are due to the presence and concentration of certain pigments.
For example, cluster 1 (red) had more pigment variability and there was
a signicantly higher concentration (p b 0.05) of zeaxanthin as com-
pared to cluster 2. Cluster 2 (blue) was dominated by time periods
where fucoxanthin was signicantly more abundant (p b 0.001). Cluster
2 was largely composed of months that fell within the upwelling period,
while most of the data within cluster 1 corresponded to non-upwelling

Fig. 3. Percentage contribution of major phytoplankton groups to total phytoplankton counts as determined through microscopy observed at the CARIACO station between 2006 and 2012.

Please cite this article as: Lorenzoni, L., et al., Characterization of phytoplankton variability in the Cariaco Basin using spectral absorption,
taxonomic and pigment data, Remote Sensing of Environment (2015), http://dx.doi.org/10.1016/j.rse.2015.05.002
6 L. Lorenzoni et al. / Remote Sensing of Environment xxx (2015) xxxxxx

Fig. 4. Relative contribution of phytoplankton size classes (micro, nano and picophytoplankton) to total phytoplankton abundance derived from diagnostic HPLC pigments measured
between 2006 and 2012.

months. In summary, cluster 1 was dominated by picophytoplankton, spectra (a*ph) showed instead 3 separate clusters (Fig. 6 bottom). Cluster
while cluster 2 was more abundant of microphytoplankton. The clusters 1 (red) was composed of months where fucoxanthin was more abun-
of the 4th derivative of the chlorophyll-specic absorption coefcient dant and, similar to Cluster 2 in the a*ph dendrogram, corresponded to
months of upwelling. Clusters 2 and 3 were both dominated by zeaxan-
0.18
thin and not signicantly different pigment-wise. Some of the difference
between these clusters, as well as the difference in distance between
0.16 clusters 2, 3 and 1, may arise from the intensity of the original a*ph spec-
0.14
tra; a*ph of cluster 3 had a higher absorption spectra in the blue, as com-
pared to cluster 2, but the difference was not signicant; similarly,
a*ph(m2(mg chla)-1)

0.12 cluster 2 had more PPC than cluster 3, but again this difference was not
0.10

0.08

0.06

0.04

0.02

0.00
400 450 500 550 600 650 700

Wavelength (nm)

0.18

0.16

0.14
a*ph(m2(mg chla)-1)

0.12

0.10

0.08

0.06

0.04

0.02

0.00
400 450 500 550 600 650 700

Wavelength (nm)

Fig. 5. Chlorophyll specic absorption (a*ph) for the upwelling (top) and non-upwelling Fig. 6. Hierarchical cluster analysis of a*ph (top), and 4th derivative of a*ph (bottom) data.
seasons (bottom) at the CARIACO time-series station for the period 20062012. Average The main clusters are separated by color and indicated by labels. Refer to text for an expla-
for each season is indicated with the thick black line. nation of each cluster. X axis indicates individual arbitrary cluster group.

Please cite this article as: Lorenzoni, L., et al., Characterization of phytoplankton variability in the Cariaco Basin using spectral absorption,
taxonomic and pigment data, Remote Sensing of Environment (2015), http://dx.doi.org/10.1016/j.rse.2015.05.002
 L. Lorenzoni et al. / Remote Sensing of Environment xxx (2015) xxxxxx 7

signicant. We carried out seasonally separated clusters (i.e., upwelling
and non-upwelling) for both the a* ph and its 4th derivative, but
no new information resulted from these analyses. The clustering
appears to be dominated by the pigment concentration, mainly driv-
en by fucoxanthin and zeaxanthin (diatoms or microphytoplankton
and cyanobacteria or picophytoplankton).
The consistency of how the data is aggregated within the dendro-
grams shown in Fig. 6 suggests that the derivative analysis has some
applicability for discriminating differences in pigment assemblages in
Cariaco waters. We chose to perform all hyperspectral analyses with
a*ph since TChla is a strong seasonal driver of absorption. We tested
the derivative analysis on original hyperspectral data with 4 nm resolu-
tion as well as two sets (7 nm and 10 nm) of interpolated data. Fig. 7
shows the average 4th derivative of a*ph by season for the period
20062012 for the different spectral interpolations. Even with the
SavitzkyGolay smoothing lter we found that the variability in the
original data at 4 nm resolution masked the peaks and their correlation
to pigments, and noise can lead to false spectral features (e.g., Torrecilla
et al., 2009). The 7 nm and 10 nm derivatives were similar in shape, with
the 10 nm interpolation showing main peaks more clearly, sometimes
helping with the identication of the pigments and quantitative deter-
mination (Fig. 7). The optimal resolution of derivative spectra found
here (10 nm) is in agreement with what has been reported by others
(Organelli et al., 2013; Torrecilla et al., 2011).
The position of the inection points in the spectra derivatives
were similar between seasons (Fig. 7), but the amplitude varied.
This can be attributed to changes in pigment composition and/or differ-
ent interactions between pigments and proteins. The package effect
also affected absorption efciency (Bidigare et al., 1989; Bricaud et al.,
TChla
TChla
0.002
Phyco/
Chl c Zeax? Fuco?
0.001
4th Derivative
2004; Lorenzoni et al., 2011). For example, the seasonal change in
amplitude of the maximum around 490500 nm could be attributed
to a higher concentration of cyanobacteria in stratied waters, corre-
sponding to the light absorption peak of zeaxanthin, lower during
upwelling and higher during the rainy season. The secondary maxima
in the 480550 nm were likely caused by accessory pigments. Speci-
cally, fucoxanthin absorbs in the 521530 nm region, while TChlb ab-
sorbs both at around 480 nm and 650 nm. A peak at 650 nm
corresponding to TChlb was observed intermittently and is not present
in the average derivative spectra. Small peaks in the 560630 nm region
could be attributed to degradation products and Chl c (Fig. 7; Bidigare
et al., 1989; Hoepffner & Sathyendranath, 1991).
We examined the relationship between HPLC pigment concentra-
tions and the 4th derivative values of the a*ph spectra at wavelengths as-
sociated with pigment absorption maxima (Bidigare et al., 1989;
Torrecilla et al., 2009). Because overall the relationships between pig-
ment and derivative values were very weak, we divided the a*ph spectra
rst by season (upwelling and rainy) and then using the three clusters
obtained in Fig. 6. The best correlations were observed when the spectra
were categorized by clusters; a relationship was found between photo-
synthetic carotenoids (PSC), normalized by [TChla], and the 4th deriva-
tive value at 540 nm in cluster 1 (red group, where fucoxanthin was
more abundant; R2 = 0.45, n = 35; p b 0.001); similarly, normalized
photoprotective carotenoids (PPC/TChla) were signicantly correlated
with the derivative at 490 nm (R2 = 0.45, n = 35; p b 0.001). PSC, com-
posed of fucoxanthin, peridinin, 19-HF and 19-BF, have absorption
peaks between 500 and 550 nm and are used in the transfer of energy
during photosynthesis. PPC, on the other hand, help protect the cell
from excess light, and are largely dominated, in this case, by zeaxanthin
which has a peak at 490 nm. While Cluster 2 (blue) did not show a
relationship with HPLC pigments, zeaxanthin in Cluster 3 correlated sig-
nicantly with the derivative at 490 nm (R2 = 0.52, n = 16; p b 0.001).
This cluster was dominated by phytoplankton with high zeaxanthin
concentrations. No other signicant correlations were found in these
data.
The correlation between aph 4th derivative and HPLC pigment con-

centrations suggested that there was a potential for retrieval of some in-
formation about phytoplankton groups and size classes from the
0.000
absorption spectra, specically at 490 nm and 540 nm. However, the
lack of a strong correlation between the 4th derivative peaks at other
wavelengths and the concentration of specic pigments typically asso-
-0.001 Accessory Pigments Degradation
products? ciated with those particular absorption features puts in question the
applicability of this type of analysis in coastal waters such as those of
the Cariaco Basin, where there is a mixture of phytoplankton popula-
-0.002
400 450 500 550 600 650 700
tions which can respond quickly to changes in the hydrography and bio-
geochemistry of the water column. It is also possible that the lack of
correlation between specic wavelengths and corresponding pigments
may be due to overlapping of absorption spectra among these pigments.
0.002 7nm interpolation
10nm interpolation Regardless, the derivative technique was able to detect the dominant
4nm data seasonal changes, which for some types of analyses may sufce.
0.001
3.2. Organic carbon and TChla
4th Derivative

Total particulate organic carbon to Chlorophyll (POC:Chla) ratios in

0.000
the CARIACO time-series varied between 65 and ~980. The lowest values
were observed during periods of high biomass (TChla N 0.5 mg m 3)
when phytoplankton blooms were dominated by diatoms (Taylor,
-0.001
Geider, & Gilbert, 1997). The POC:Chla tends to increase under con-
ditions of high irradiance and low nutrient concentration due to
the presence of smaller cells with low intracellular Chla (Taylor
-0.002
400 450 500 550 600 650 700
et al., 1997; MacIntyre et al., 2002; Garibotti, Vernet, Kozlowski, &
Ferrario, 2003; Maran, 2005). The high POC:Chla ratios reported
Wavelength (nm)
here are above what has been cited in the literature for a variety of en-
Fig. 7. Comparison between 4 nm, 7 nm and 10 nm interpolated 4th derivatives. Top:
vironments and cultures (20300; e.g., de Jonge, 1980; Longhurst &
Upwelling season. Bottom: Rainy season. The arrows and brackets indicate the derivative Harrison, 1989; Cloern, Grenz, & Vidergar-Lucas, 1995; Taylor et al.,
positions for the major phytoplankton pigments. 1997; Maran, 2005; Sathyendranath et al., 2009). However, it has

Please cite this article as: Lorenzoni, L., et al., Characterization of phytoplankton variability in the Cariaco Basin using spectral absorption,
taxonomic and pigment data, Remote Sensing of Environment (2015), http://dx.doi.org/10.1016/j.rse.2015.05.002
8 L. Lorenzoni et al. / Remote Sensing of Environment xxx (2015) xxxxxx
also been shown that for any given Chla there is signicant variation in
the associated particulate organic carbon, hence the scatter observed in
the regression in Fig. 8. This may be due to the presence of high and
varying detrital material and bacterial biomass, as well as variability of
the phytoplankton carbon (Maran, 2005; Stramski et al., 2008;
Sathyendranath et al., 2009). POC and TChla can vary separately with
environmental factors, including nutrient availability, ambient light,
and temperature (Longhurst & Harrison, 1989; Montagnes & Franklin,
2001; Montagnes et al., 1994). Growth stage can also inuence a cell's
chemical composition (Menden-Deuer & Lessard, 2000). Detrital matter
and bacterial biomass can further confound the correlation between
Chla and POC concentrations (Behrenfeld & Boss, 2003; Garibotti
et al., 2003; Maran, 2005). In addition, taxonomic groups also con-
tribute differently to the POC:Chla ratio. Carbon density decreases
with increasing volume; specically, the POC:Chla ratio for diatoms is
lower than that of other phytoplankton (Menden-Deuer & Lessard,
2000; Montagnes & Franklin, 2001; Montagnes et al., 1994; Strathmann,
1967).
Utilizing the pigment and taxonomic information, it was possible to
obtain more information regarding the type of phytoplankton associated
with TChla and carbon concentrations (e.g., Sathyendranath et al., 2009).
Months with diatom (representative of microphytoplankton) domi-
nance (N50%) generally fell in the upper right hand portion of the line,
corresponding to higher chlorophyll concentrations, while waters with
higher cyanobacteria (representative of picophytoplankton) abundance
(N 50%) were conned in the lower left corner, corresponding to lower
TChla concentrations (Fig. 8). Coccoliths (nanoplankton) did not have a
clear dominance period. Microphytoplankton-dominated periods
(except for 2 data points) consistently had TChla N 0.3 mg m3 and
POC concentrations of 150700 mg m3, while picophytoplankton was
below ~0.25 mg m3 TChla, with POC concentrations between 40 and
200 mg m3. To test whether this separation was valid for other time
periods as well, we added TChla, POC and pigment data between
1995 and 2001. While there was more overlap between diatoms
and cyanobacteria in the mid-concentration ranges, the separation
between groups was still clear (Fig. 8).
Using the quantile regression, a minimum value for particulate
carbon directly associated with phytoplankton was obtained from the
relationship between POC and TChla (Fig. 8). The equation for this esti-
mate is also shown in Fig. 8, and can be used to infer carbon from a
specic TChla concentration. This relationship indicated that diatoms
1000
y = 223x0.48
POC (mg m-3)
100
y = 114x0.41
10
0.01 0.1 1 10 100
-3
TChla (mg m )
Fig. 8. Particulate organic carbon (POC) vs. TChla for the CARIACO time-series
(19952012). Least squares t, log transformed, is shown in black. Minimum carbon, as
estimated using quantile regression, is shown in green. Corresponding equations for
each are also shown. Samples dominated by diatoms (microphytoplankton) are identied
in blue; samples dominated by cyanobacteria (picophytoplankton) are shown in yellow.
(For interpretation of the references to color in this gure legend, the reader is referred
to the web version of this article.)
Please cite this article as: Lorenzoni, L., et al., Characterization of phytoplankton variability in the Cariaco Basin using spectral absorption,
taxonomic and pigment data, Remote Sensing of Environment (2015), http://dx.doi.org/10.1016/j.rse.2015.05.002
(microphytoplankton) had a carbon to TChla ratio of 71 (range =
32233), while the ratio for cyanobacteria (picophytoplankton) was
331 (range = 250490). Because of the separation that exists between
the two size classes, it should be possible, as a rst approximation, to
estimate the fraction of carbon corresponding to the microphyto-
plankton and picophytoplankton size classes remotely.
4. Conclusion
We examined phytoplankton taxonomy and pigment composition
observations collected between 2006 and 2012 in the Cariaco Basin
under the framework of the CARIACO Ocean Time-Series Project, and
analyzed how seasonal changes in these parameters related to absorp-
tion spectra and concentrations of organic carbon. The results presented
here underline the importance of understanding local ecological and
hydrographic dynamics, and having high temporal resolution data,
especially in variable environments such as continental shelves, to test
and validate biogeochemical models.
While no signicant correlation was found between 4th derivative
peaks of a*ph and specic pigments typically associated with the corre-
sponding absorption features, there was a signicant relationship
between PSC/TChla and the 4th derivative value at 540 nm, and be-
tween PPC/TChla and the derivative at 490 nm. This suggested that,
while the technique may not be applicable for the retrieval of individual
pigments in these types of coastal waters typically showing a mixture of
populations that change relatively quickly on a seasonal basis, it can
identify seasonal variations and separate large size classes (i.e., micro-
phytoplankton and pico-phytoplankton). In order to detect more
specic changes in the phytoplankton community composition, it
may be necessary to further test other techniques, such as a coupled
smoothing/higher-order derivative computation of the absorption
spectra. While this was attempted here it was not fully realized.
Most of the seasonal changes observed in a*ph were driven by pres-
ence and concentration of certain pigments, with zeaxanthin
(picophytoplankton) dominating during the rainy season and fucoxan-
thin (i.e., diatoms) during the upwelling. The seasonality of accessory
pigment composition responded to changes in the phytoplankton com-
munity composition (Mutshinda et al., 2013). The difference between
the taxonomic data and the pigment data suggests that the microscopy
taxonomy may not capture the variability in the pico-phytoplankton.
Since the beginning of the time-series in 1995, there has been a change
in the abundance of phytoplankton size classes based on diagnostic pig-
ments: micro-phytoplankton has decreased year-round, while pico-
phytoplankton has increased, particularly during the rainy season.
This is consistent with the decrease in larger phytoplankton previously
reported for the region.
The POC:TChla ratio at CARIACO was variable and dependent on en-
vironmental carbon (not only related to phytoplankton) and the func-
tional groups present at any given time, underscoring the fact that
using a xed ratio of C:Chla in biogeochemical models can lead to
large uncertainties in carbon budgets from coastal zones. Low
POC:TChla values were associated with microphytoplankton size class
(diatoms), while picophytoplankton (cyanobacteria) exhibited higher
ratios. There was a clear separation between these groups according
to their POC:TChla ratio, mostly dependent on TChla concentrations, en-
abling the estimate of a minimum POC:TChla for each size class.
This work complements previously published results for the area
and showed the potential application of high resolution spectral data
of absorption coefcient (~ 10 nm) for discriminating different
phytoplankton pigment assemblages. This is especially important for
developing valid retrieval algorithms of algal biomass and size structure
from spaceborne hyperspectral instruments. Planned hyperspectral
resolution missions, such as HyspIRI, will help study phytoplankton-
dynamics in coastal areas. It is expected that the reectance measured
with HyspIRI (assuming enough signal-to-noise ratio [SNR]) will
enable the discrimination of marker pigments to separate broader
 L. Lorenzoni et al. / Remote Sensing of Environment xxx (2015) xxxxxx
phytoplankton communities (Devred et al., 2013). HyspIRI's temporal
resolution, with a revisit period of 19 days, may not be appropriate for
studies of rapid changes in phytoplankton communities, including
harmful algal blooms (Devred et al., 2013). Nevertheless, this revisit
cycle can be sufcient for studies of more persistent events as well as
seasonal changes in phytoplankton communities as those observed at
CARIACO. It is important to note that, in order to implement the
approach described here to satellite-based data, it will be necessary to
perform accurate inversion of the spectral reectance to total absorp-
tion, and from there partition the absorption coefcient into phyto-
plankton and non-phytoplankton contributions. In CARIACO, the
detrital fraction is relatively small, but the absorption coefcient due
to CDOM can be comparable to that of phytoplankton, and in some
cases dominate light absorption (e.g., Lorenzoni et al., 2011). This will
require testing of different models.
Time series efforts that can provide coupled optical and pigment
datasets, such as CARIACO, present a unique opportunity to test the
applicability of hyperspectral data for remotely retrieving optical and
biological properties. Most global biogeochemical time-series are
coastal (e.g., http://igmets.net/), thereby offering the possibility to as-
sess the validity of this approach in optically complex waters. We sug-
gest utilizing these time-series platforms to test the applicability of
hyperspectral data for the retrieval of phytoplankton assemblages. Fu-
ture work should continue testing the performance of such retrieval
models on data from coastal waters where a mix of populations is
often present, as well as detrital matter, so as to constrain the limitations
of data collected with hyperspectral instruments designed to globally
observe coastal and inland waters (e.g., HyspIRI; Devred et al., 2013).
Acknowledgments
This work was supported by the National Science Foundation (NSF,
Grants OCE-0752139, OCE-9216626, OCE-9729284, OCE-9401537,
OCE-9729697, OCE-9415790, OCE-9711318, and OCE0963028), NASA
(NASA Grants NAG5-6448, NAS5-97128, and NNX13AE85G) and
the Venezuelan Fondo Nacional de Ciencia y Tecnologa (FONACIT,
Venezuela, Grant 2000001702). We thank Crystal Thomas for the pro-
cessing of all HPLC samples used here, and Luis Troccoli for the phyto-
plankton taxonomic identication. We are indebted to the personnel of
the Fundacin La Salle de Ciencias Naturales, Estacin de Investigaciones
Marinas, Isla de Margarita (FLASA/EDIMAR) who have been responsible
for the collection, analysis, and quality control of data presented here,
and to the crew of the R/V Hermano Gins for their professional support.
We would also like to acknowledge the three reviewers that carefully re-
vised this manuscript; their revisions and suggestions improved our
work signicantly. This is IMaRS contribution #168.
References
Astor, Y., Meri, J., & Muller Karger, F. (1998). Variabilidad estacional hidrogrca en la
Fosa de Cariaco. Memoria de la Sociedad de Ciencias Naturales La Salle, 63(149), 6172.
Astor, Y., Lorenzoni, L., Thunell, R.C., Varela, R., Muller-Karger, F. E., Troccoli, L., et al.
(2013). Interannual variability in sea surface temperature and fCO2 changes in the
Cariaco Basin. Deep Sea Research II, 93, 3343. http://dx.doi.org/10.1016/j.dsr2.2013.
01.002.
Behrenfeld, M.J., & Boss, E. (2003). The beam attenuation to chlorophyll ratio: An optical
index of phytoplankton physiology in the surface ocean? Deep-Sea Research Part I, 50,
15371549. http://dx.doi.org/10.1016/j.dsr.2003.09.002.
Behrenfeld, M.J., Boss, E., Siegel, D.A., & Shea, D.M. (2005). Carbon-based ocean productiv-
ity and phytoplankton physiology from space. Global Biogeochemical Cycles, 19,
GB1006. http://dx.doi.org/10.1029/2004GB002299.
Bidigare, R.R., Morrow, J.H., & Kiefer, D.A. (1989). Derivative analysis of spectral absorp-
tion by photosynthetic pigments in the western Sargasso Sea. Journal of Marine
Research, 47(2), 323341. http://dx.doi.org/10.1357/002224089785076325.
Bidigare, R.R., Ondrusek, M.E., Morow, M.E., & Kiefer, D.A. (1990). In vivo absorption prop-
erties of algal pigments. Proceedings from the SPIE ocean optics conference X. 1302. (pp.
290297).
Bricaud, A., Claustre, H., Ras, J., & Oubelkheir, K. (2004). Natural variability of
phytoplankton absorption in oceanic waters: Inuence of the size structure of algal
populations. Journal of Geophysical Research, 109, C11010. http://dx.doi.org/10.1029/
2004JC002419.
Please cite this article as: Lorenzoni, L., et al., Characterization of phytoplankton variability in the Cariaco Basin using spectral absorption,
taxonomic and pigment data, Remote Sensing of Environment (2015), http://dx.doi.org/10.1016/j.rse.2015.05.002
9
Bricaud, A., & Stramski, D. (1990). Spectral absorption coefcients of living phytoplankton
and nonalgal biogenous matter: A comparison between the Peru upwelling area and
the Sargasso Sea. Limnology and Oceanography, 35, 562582.
Bricaud, A., Morel, A., Babin, M., Allali, K., & Claustre, H. (1998). Variations of light absorp-
tion by suspended particles with the chlorophyll a concentration in oceanic (Case 1)
waters: analysis and implications for bio-optical models. Journal of Geophysical
Research, 103, 3103331044.
Ciotti, A.M., Lewis, M.R., & Cullen, J.J. (2002). Assessment of the relationships between
dominant cell size in natural phytoplankton communities and the spectral shape of
the absorption coefcient. Limnology and Oceanography, 47(2), 404417.
Cloern, J.E., Grenz, C., & Vidergar-Lucas, L. (1995). An empirical model of the phytoplank-
ton chlorophyll: Carbon ratio The conversion factor between productivity and
growth rate. Limnology and Oceanography, 40(7), 13131321.
D'Sa, E., Miller, R.L., & Del Castillo, C. (2006). Bio-optical properties and ocean color algo-
rithms for coastal waters inuenced by the Mississippi River during a cold front.
Applied Optics, 45, 74107428.
de Jonge, V.N. (1980). Fluctuations in the organic carbon to chlorophyll a ratios for estu-
arine benthic diatom populations. Marine Ecology: Progress Series, 2, 345353.
Devred, E., Sathyendranath, S., Stuart, S., & Platt, T. (2011). Absorption-derived phyto-
plankton cell size: Application to satellite ocean-colour data in the Northwest Atlan-
tic. Remote Sensing of Environment, 115, 22552266.
Devred, E., Turpie, K.R., Moses, W., Klemas, V.V., Moisan, T., Babin, M., et al. (2013). Future
retrievals of water column bio-optical properties using the Hyperspectral Infrared
Imager (HyspIRI). Remote Sensing, 5, 68126837.
Garibotti, I., Vernet, M., Kozlowski, W.A., & Ferrario, M.E. (2003). Composition and bio-
mass of phytoplankton assemblages in coastal Antarctic waters: A comparison of che-
motaxonomic and microscopic analyses. Marine Ecology: Progress Series, 247, 2742.
Garver, S.A., & Siegel, D.A. (1997). Inherent optical property inversion of ocean color spectra
and its biogeochemical interpretation: 1. Time series from the Sargasso Sea. Journal of
Geophysical Research, 102(C8), 1860718625. http://dx.doi.org/10.1029/96JC03243.
Hasle, G.R. (1978). The inverted microscope method. In A. Sournia (Ed.), Phytoplankton
manual (pp. 8896). Paris: SCOR-UNESCO.
Hoepffner, N., & Sathyendranath, S. (1991). Effect of pigment composition on absorption
properties of phytoplankton. Marine Ecology: Progress Series, 73, 1123.
Hoepffner, N., & Sathyendranath, S. (1992). Bio-optical characteristics of coastal waters:
Absorption spectra of phytoplankton and pigment distribution in the western
North Atlantic. Limnology and Oceanography, 37, 16601679.
Hooker, S., Van Heukelem, L., Thomas, C.S., Claustre, H., Ras, J., Barlow, R., et al. (2005). The
second SeaWiFS HPLC analysis round-robin experiment (SeaHARRE-2). NASA techni-
cal memorandum NASA/TM-2005-212785.
IOCCG (2014). Phytoplankton functional types from space. In S. Sathyendranath (Ed.),
Reports of the International Ocean-Colour Coordinating Group, no. 15. Dartmouth,
Canada: IOCCG.
Kishino, M., Takahashi, M., Okami, N., & Ichimura, S. (1985). Estimation of the spectral
absorption coefcients of phytoplankton in the sea. Bulletin of Marine Science, 37,
634642.
Le Quere, C., Harrison, S.P., Prentice, I.C., Buitenhuis, E.T., Aumont, O., Bopp, L., et al.
(2005). Ecosystem dynamics based on plankton functional types for global ocean bio-
geochemistry models. Global Change Biology, 11, 125.
Lee, Z., & Carder, K.L. (2004). Absorption spectrum of phytoplankton pigments derived
from hyperspectral remote-sensing reectance. Remote Sensing of Environment, 89,
361368.
Longhurst, A.R., & Harrison, W.G. (1989). The biological pump: Proles of plankton
production and consumption in the upper ocean. Progress in Oceanography, 22,
47123.
Lorenzoni, L., Hu, C., Varela, R., Arias, G., Guzman, L., & Muller-Karger, F. (2011). Bio-
optical characteristics of Cariaco Basin (Caribbean Sea) waters. Continental Shelf
Research, 31(6), 582593.
Lorenzoni, L.R.C., Thunell, C., Benitez-Nelson, D., Hollander, N., Martinez, E., Tappa, R., et al.
(2009). The importance of subsurface nepheloid layers in transport and delivery of
sediments to the Eastern Cariaco Basin, Venezuela. Deep-Sea Research Part I, 56,
22492262.
MacIntyre, H.L., Kana, T.M., Anning, T., & Geider, R. (2002). Photoacclimation of photosyn-
thesis irradiance response curves and photosynthetic pigments in microalgae and
cyanobacteria. Journal of Phycology, 38, 1738.
Maran, E. (2005). Phytoplankton growth rates in the Atlantic subtropical gyres.
Limnology and Oceanography, 50, 299310.
Menden-Deuer, S., & Lessard, E.J. (2000). Carbon to volume relationships for dinoagel-
lates, diatoms, and other protest plankton. Limnology and Oceanography, 45, 569579.
Mitchell, G., & Kiefer, D.A. (1988). Chlorophyll a specic absorption and uorescence
excitation spectra for light-limited phytoplankton. Deep-Sea Research, 35(5),
639663.
Moisan, T.A., & Mitchell, B.G. (1999). Photophysiological acclimatation of Phaeocystis
antartica Karsten under light limitation. Limnology and Oceanography, 44(2), 247258.
Moisan, J.R., Moisan, T.A.H., & Linkswiler, M.A. (2011). An inverse modeling approach to
estimating phytoplankton pigment concentrations from phytoplankton absorption
spectra. Journal of Geophysical Research, 116, C09018.
Montagnes, D.J., Berges, J.A., Harrison, P.J., & Taylor, F.J.R. (1994). Estimation of carbon, ni-
trogen, protein, and chlorophyll a from volume in marine phytoplankton. Limnology
and Oceanography, 39, 10441060.
Montagnes, D.J.S., & Franklin, D.J. (2001). Effect of temperature on diatom volume, growth
rate, and carbon and nitrogen content: Reconsidering some paradigms. Limnology
and Oceanography, 46, 20082018.
Montes, E., Altabet, M., Muller-Karger, F.E., Scranton, M.I., Thunell, R.C., Benitez-Nelson,
C.R., et al. (2013). Biogenic nitrogen gas production at the oxicanoxic interface in
the Cariaco Basin, Venezuela. Biogeosciences, 10, 267279.
10 L. Lorenzoni et al. / Remote Sensing of Environment xxx (2015) xxxxxx
Muller-Karger, F.E., Kavanaugh, M.T., Montes, E., Balch, W.M., Breitbart, M., Chavez, F.P.,
et al. (2014). A framework for a marine biodiversity observing network within
changing continental shelf seascapes. Oceanography, 27(2), 1823. http://dx.doi.org/
10.5670/oceanog.2014.56.
Muller-Karger, F.E., Varela, R., Thunell, R., Astor, Y., Zhang, H., & Hu, C. (2004). Processes of
coastal upwelling and carbon ux in the Cariaco Basin. In D.A. Siegel, A.C. Thomas, & J.
Marra (Eds.), Special issue: Views of ocean processes from the Sea-viewing Wide Field-
of-view Sensor (SeaWiFS) mission: Volume 2. Deep-Sea Research II, Vol. 51/1011. (pp.
927943).
Mller-Karger, F., Varela, R., Thunell, R.C., Scranton, M.I., Taylor, G.T., Astor, Y., et al.
(2010). CARIACO: A time series of primary production and vertical export in the
Cariaco Basin. In Kon-Kee Liu, Larry Atkinson, Renato Quinones, & Liana Talaue-
McManus (Eds.), Carbon and nutrient uxes in continental margins: A global synthesis.
JGOFS Continental Margins Task Team (CMTT). Berlin/Heidelberg: Springer-Verlag.
Mutshinda, Crispin M., Troccoli-Ghinaglia, Luis, Finkel, Zoe V., Muller-Karger, Frank E., &
Irwin, Andrew J. (2013). Environmental control of the dominant phytoplankton in
the Cariaco basin: A hierarchical Bayesian approach. Marine Biology Research, 9,
247261. http://dx.doi.org/10.1080/17451000.2012.731693.
NRC (2007). Earth Science and Applications from Space: National Imperatives for the Next
Decade and Beyond. National Academies: Press 456 pp.
Organelli, E., Bricaud, A., Antoine, D., & Uitz, J. (2013). Multivariate approach for the
retrieval of phytoplankton size structure from measured light absorption spectra in
the Mediterranean Sea (BOUSSOLE site). Applied Optics, 52(11), 22572273. http://
dx.doi.org/10.1364/AO.52.002257.
Pinckney, J.L., Benitez-Nelson, C.R., Thunell, R.C., Muller-Karger, F., Lorenzoni, L., Troccoli,
L., et al. (2015). Phytoplankton community structure and depth distribution changes
in the Cariaco Basin between 1996 and 2010. Deep-Sea Research Part I, 101, 2737.
http://dx.doi.org/10.1016/j.dsr.2015.03.004.
Porra, R.J., Pfndel, E.E., & Engel, N. (1997). Metabolism and function of photosynthetic
pigments. In S. W. Jeffrey, R. F. C. Mantoura, & S. W. Wright (Eds.), Phytoplankton pig-
ments in oceanography: Guidelines to modern methods (pp. 85126). Paris: UNESCO.
Roesler, C.S., & Perry, M.J. (1995). In situ phytoplankton absorption, uorescence emis-
sion, and particulate backscattering spectra determined from reectance. Journal of
Geophysical Research, 100(C7), 1327913294.
Rondon, A.N. (2009). Seasonal and interannual variability in phytoplankton species compo-
sition in the cariaco basin, Venezuela. (Master's Thesis) University of Dalhousie.
Rueda-Roa, D. (2012). On the spatial and temporal variability of upwelling in the southern
Caribbean Sea and its inuence on the ecology of phytoplankton and of the Spanish sar-
dine (Sardinella aurita). (PhD Dissertation) University of South Florida.
Please cite this article as: Lorenzoni, L., et al., Characterization of phytoplankton variability in the Cariaco Basin using spectral absorption,
taxonomic and pigment data, Remote Sensing of Environment (2015), http://dx.doi.org/10.1016/j.rse.2015.05.002
Rueda-Roa, D., & Muller-Karger, F. (2013). The southern Caribbean upwelling system: Sea
surface temperature, wind forcing and chlorophyll concentration patterns. Deep-Sea
Research Part I, 78, 102114. http://dx.doi.org/10.1016/j.dsr.2013.04.008.
Sathyendranath, S., Stuart, V., Nair, A., Oka, K., Nakane, T., Bouman, et al. (2009). Carbon-
to-chlorophyll ratio and growth rate of phytoplankton in the sea. Marine Ecology
Progress Series, 383, 7384.
Schubert, C. (1982). Origin of Cariaco Basin, southern Caribbean Sea. Marine Geology, 47,
345360.
Stramski, D., Reynolds, R.A., Babin, M., Kaczmarek, S., Lewis, M.R., Rottgers, R., et al.
(2008). Relationships between the surface concentration of particulate organic car-
bon and optical properties in the eastern South Pacic and eastern Atlantic Oceans.
Biogeosciences, 5, 171201.
Strathmann, R.R. (1967). Estimating the organic carbon content of phytoplankton from
cell volume or plasma volume. Limnology and Oceanography, 12, 411418.
Taylor, A.H., Geider, R.J., & Gilbert, F.J.H. (1997). Seasonal and latitudinal dependencies of
phytoplankton carbon-to-chlorophyll ratios: Results of a modelling study. Marine
Ecology: Progress Series, 152, 5166.
Taylor, G.T., Muller-Karger, F., Thunell, R.C., Scranton, M.I., Astor, Y., Varela, R., et al.
(2012). Ecosystem response to global climate change in the southern Caribbean
Sea. Proceedings of the National Academy of Sciences (USA), 109(47), 1931519320.
Thunell, R., Benitez-Nelson, C., Varela, R., Astor, Y., & Mller-Karger, F. (2007). Particulate
organic carbon uxes along upwelling-dominated continental margins: Rates and
mechanisms. Global Biogeochemical Cycles, 21, GB1022. http://dx.doi.org/10.1029/
2006GB002793.
Torrecilla, E., Piera, J., & Vilaseca, M. (2009). Derivative analysis of oceanographic
hyperspectral data. In G. Jedlovec (Ed.), Advances in geoscience and remote sensing
(pp. 597619). Vienna: InTech.
Torrecilla, E., Stramski, D., Reynolds, R.A., Milln-Nez, E., & Piera, J. (2011). Cluster anal-
ysis of hyperspectral optical data for discriminating phytoplankton pigment assem-
blages in the open ocean. Remote Sensing of Environment, 115, 25782593.
Uitz, J., Claustre, H., Morel, A., & Hooker, S. (2006). Vertical distribution of phytoplankton
communities in open-ocean: An assessment based on surface chlorophyll. Journal of
Geophysical Research, 111, CO8005. http://dx.doi.org/10.1029/2005JC003207.
Van Heukelem, L., & Thomas, C.S. (2001). Computer-assisted high-performance liquid
chromatography method development with applications to the isolation and analysis
of phytoplankton pigments. Journal of Chromatography A, 910, 3149.
Vijayan, A.K., & Somayajula, S.A. (2014). Effect of accessory pigment composition on the
absorption characteristics of a dinoagellate bloom in a coastal embayment.
Oceanologia, 56(1), 107124. http://dx.doi.org/10.5697/oc.56-1.107.