Limnol. Oceanogr., 52(4), 2007, 1645–1664
2007, by the American Society of Limnology and Oceanography, Inc.
E
Composition and degradation of marine particles with different settling velocities in the
northwestern Mediterranean Sea
Madeleine Goutx
Centre d’Océanologie de Marseille, Laboratoire de Microbiologie Marine, UMR6117, Campus de Luminy, Case 907,
13 288 Marseille Cedex 9, France
Stuart G. Wakeham
Skidaway Institute of Oceanography, 10 Ocean Science Circle, Savannah, Georgia 31411
Cindy Lee
Marine Sciences Research Center, Stony Brook University, Stony Brook, New York 11794-5000
Marie Duflos and Catherine Guigue
Centre d’Océanologie de Marseille, Laboratoire de Microbiologie Marine, UMR6117, Campus de Luminy, Case 907,
13 288 Marseille Cedex 9, France
Zhanfei Liu
Marine Sciences Research Center, Stony Brook University, Stony Brook, New York 11794-5000
Brivaëla Moriceau
UMR 6539, Institut Universitaire Européen de la Mer, Site du Technopole Brest-Iroise, Place Nicolas Copernic, 29280
Plouzané, France
Richard Sempéré and Marc Tedetti
Centre d’Océanologie de Marseille, Laboratoire de Microbiologie Marine, UMR6117, Campus de Luminy, Case 907,
13 288 Marseille Cedex 9, France
Jianhong Xue
Marine Sciences Research Center, Stony Brook University, Stony Brook, New York 11794-5000
Abstract
Settling particles were collected from the Ligurian Sea in the northwestern Mediterranean Sea in May 2003 and
separated by elutriation into different settling velocity classes (.230, 115–230, 58–115, and ,58 m d21). Particles
of the different classes were incubated for 5 d to study their biodegradability. Particulate opal content and organic
compound composition (amino acids, pigments, lipids, and carbohydrates) were analyzed initially and at regular
time intervals during the incubation period. Most particles (48–67% of total mass) sank at greater than 230 m d21
and were dominated by large diatom-derived aggregates produced during the spring bloom period. The initial
organic composition and the biological lability of these particles varied with settling velocity. The strong
phytoplankton signal was visible in all settling velocity classes, while slower settling particles carried with them
a greater zooplankton and bacterial signature. As the different class particles decomposed, their compositions
changed and became more similar with time, with a dominance of compounds that suggests a more degraded
state: the amino acids c-aminobutyric acid and b-alanine, the pigments pyropheophorbide and pheophytin, the
deoxysugars fucose and rhamnose, and lipid metabolites (diglycerides and monoglycerides, alcohols, and free
fatty acids). Biogenic opal in the particles dissolved faster in more degraded particles than in fresher particles,
suggesting that loss of organic matter may expose opal to dissolution. The coupling of settling velocity and
decomposition rate measurements shows quantitatively that slower settling particles are quickly degraded and
Acknowledgments
This research was part of the MedFlux and PECHE (Production and Export of Carbon: Control by Heterotrophs at small temporal
scale) programs and was supported by the U.S. National Science Foundation Chemical Oceanography Program (OCE-0136370, OCE0136318, and OCE-0113687) and the French CNRS (Centre National de la Recherche Scientifique), respectively. Participation of B.M.
was funded by ORFOIS (Origin and Fate of Biogetic Particle Fluxes in the ocean) (EVK2-CT2001-00100). We thank Michael Peterson,
Lynn Abramson, Jenni Szlosek, Meaghan Askea, and Isabell Putnam for shipboard and laboratory help; David Hirschberg and Michael
Peterson for CHN analysis; Claude Mante for help with statistical data treatment; and the captain and crew of the RV Seward Johnson II.
We wish to acknowledge the associate editor and two anonymous reviewers for very helpful comments and suggestions on the
manuscript. This is MedFlux contribution 7 and MSRC contribution 1318.
1645
1646
Goutx et al.
faster settling particles retain their original biological signal to a greater degree. Greater preservation of faster
settling particles testifies to the importance of these particles as food resources for bathypelagic and surface
sediment communities.
Settling particles are the major vehicle for transporting
organic matter (OM) from the sea surface to the deep ocean
and sea floor (Boyd and Trull 2006). How rapidly large
particles sink out of the upper ocean and how rapidly they
decompose or dissolve determines the depth of CO2
remineralization and release of dissolved organic carbon
(DOC) (Armstrong et al. 2002). This depth in turn
determines whether the CO2 will be returned quickly to
the atmosphere or be sequestered over longer time periods
in the deep sea.
The fundamental properties determining particle (aggregate) settling velocity are size, shape, and density. Organic
matter has only a slight excess density over that of
seawater, and thus most of the purely organic particles
will sink very slowly, if at all. Some type of ballast is
required to carry OM to depth (Ittekkot and Haake 1990).
Biogenic silica (BSiO2) and calcium carbonate (CaCO3) are
the major mineral components in marine plankton and
provide the excess density necessary for less-dense organic
matter to sink. Minerals may also protect organic matter
from degradation, allowing it to penetrate deeper into the
ocean (e.g., Hedges et al. 2001) or providing a matrix that
holds particles together in larger aggregates (Lee et al.
2004). Armstrong et al. (2002) showed that ratios of
particulate organic carbon (POC) to mineral ballast
converge to a nearly constant value at depth, showing the
strong relationship between organic carbon and mineral
ballasts. Klaas and Archer (2002) further demonstrated
that the variability in POC flux data compiled from deep
sediment traps from 52 locations around the globe could
largely be explained by the chemical composition of the
ballast (i.e., opal vs. carbonate vs. alumino-silicate dust,
indicative of biogenic silica of diatoms, calcareous shells of
coccolithophores, and lithogenic inputs, respectively).
Degradation of particulate organic matter in the pelagic
ocean is well documented (Wakeham and Lee 1993).
Decreasing POC concentrations and fluxes as a function
of depth in the water column show great spatial and
temporal variability in relation to the flux regime and the
pelagic food webs that produce settling particles (Karl et al.
1988; Antia et al. 2001; Sheridan et al. 2002). Degradation
rate constants have been estimated in the laboratory for
various components of the particle flux, such as phytoplankton (e.g., Harvey et al. 1995; Harvey and Macko
1997), fecal pellets (Jacobson and Azam 1984), and
laboratory-produced aggregates and natural large particles
(Sempéré et al. 2000; Panagiotopoulos et al. 2002; Engel et
al. unpubl. data), and show a large range of values (0.001–
0.10 d21). OM degradation may in turn have a strong effect
on mineral dissolution because the fate of organic matter
and ballast minerals are intimately intertwined (Hedges et
al. 2001; Ingalls et al. 2003), and the rate at which ballast
dissolution occurs is a critical control on particulate
organic matter decomposition.
In some studies, organic mater–mineral interaction has
considerably decreased phytoplankton OM bioavailability
in dinoflagellates, at least on a timescale of weeks
(Arnarson and Keil 2000). Calcified coccolithophores with
their calcium carbonate tests made more refractory
particles than naked coccolithophores (Engel et al. unpubl.
data). At the same time, OM can protect minerals from
dissolution at cellular and particulate scales. Bacterial
activity enhanced diatom frustule dissolution through
ectoenzymatic hydrolysis of the protein membrane surrounding the diatom frustule (Bidle and Azam 1999, 2001);
however, in the bathypelagic layers, hydrostatic pressure
slows down this process through inhibition of bacterial
ectohydrolase (Tamburini et al. 2006). At the scale of the
particle, when diatoms are embedded inside aggregates in
a matrix of degraded OM, dissolution of biogenic silica
(BSiO2) decreases (Moriceau et al. 2007).
Owing to their ballast function and their role in organic
matter biodegradation, organic matter–mineral interactions have strong implications for OM cycling. However,
relationships between mineral ballast, particle settling,
mineral dissolution, and organic matter decomposition
processes in the marine water column have rarely been
addressed simultaneously. The MedFlux biogeochemistry
program aimed to examine the role of minerals that are
produced by organisms or introduced into the surface
ocean by winds, on the magnitude and dynamics of organic
carbon export to the deep ocean and sediments. Our
specific objective was to explore relationships between
settling velocity, organic carbon composition, opal content,
and dissolution and degradation of settling particles. We
used a newly developed NetTrap/elutriator system (Peterson et al. 2005) that enables fractionation according to
settling velocity of the various particle classes that make up
the carbon flux; we coupled this unique particle collection
technology with microbial decomposition incubations.
Methods
Sampling strategy and incubation experiment—Samples
were collected at the French time-series DYFAMED
(Dynamique des Flux Atmosphériques en Mediterranée)
station in the Ligurian Sea, 52 km off Nice, France, at
43u259N, 07u529E during the MedFlux Period 1 cruise (6–
15 May 2003) using a NetTrap, a floating plankton net
used in a settling particle collection mode (Peterson et al.
2005). The NetTrap was deployed at 200 m during three
sampling periods (7–8, 8–10, and 10–14 May for 15.7 h,
64.3 h, and 74.8 h, respectively) near the end of the annual
spring diatom bloom (e.g., Marty et al. 2002). After each
sampling period, the contents of the NetTrap cod end were
resuspended in filtered (precleaned glass fiber filters, 0.7mm nominal porosity) seawater from 200 m depth. The
particle/seawater mixture was transferred into a four-stage
Degradation of marine particles
elutriator system, and particles were sorted by their settling
velocity against a counterflow of seawater (Peterson et al.
2005). Settling velocity (SV) thresholds of elutriated
samples were calculated based on the seawater flow rate
and the cross-sectional area of the individual elutriator
tubes. Particles with four different SVs (A, .230 m d21; B,
115–230 m d21; C, 58–115 m d21; D, 29–58 m d21) were
obtained from the first two NetTrap deployments (NT1-A,
NT1-B, NT1-C, NT1-D and NT2-A, NT2-B, NT2-C, NT2D, respectively) and were used in incubation experiments.
For the third deployment (NetTrap 3), the four samples
(NT3-A, NT3-B, NT3-C, and NT3-D) were analyzed for
organic composition but not incubated. NetTrap 3 samples
were split using a particle splitter (MacLane WSD-10 wet
sample divider) into samples for the analysis of CHN,
amino acids, chloropigments, and lipid biomarkers. Samples were filtered onto combusted glass fiber filters and
frozen until analysis.
Elutriated NetTrap 1 and 2 particles were incubated in
the dark for 120 h at 13uC, the in situ temperature at
200 m. Incubations began within 3–6 h of trap recovery.
Particle suspensions from each settling velocity class were
split into 10 equal fractions using the MacLane WSD-10
particle splitter and placed into precombusted 500-mL
incubation flasks. The incubation medium was brought to
450 mL with 0.7-mm filtered seawater containing natural
bacterial communities. One flask was poisoned with HgCl2
(10 mg L21 final concentration) to act as a control with no
biological activity. Concentrations of dissolved oxygen are
175–180 mmol kg21 at 200 m depth at the DYFAMED site
according to Copin-Montégut and Bégovic (2002). Moreover, an air headspace (170 mL) above the incubation
medium enabled gas exchange between the air and the
aqueous phase in the bottles, which were gently rolled once
a day in the incubation chamber. Under these conditions,
we assume that oxic conditions were maintained in the
flasks during the 5-d particle incubation.
Batches were stopped after different incubation times, 0,
6, 12, 24, 48 (plus a duplicate), and 120 h, and samples were
split into aliquots (,40 mL) using a Perimatic liquid
dispenser (Jencons Scientific). The aliquots were filtered
(glass fiber filters for organic carbon and polycarbonate
membrane for biogenic silica) so that biogenic silica
(BSiO2) and silicic acid (DSi), organic carbon, amino acid,
pigment, lipid, and carbohydrate contents could be
measured. Bacteria were enumerated (DAPI [49,6-diamidino-2-phenylindole] positive) over the time-course of the
NetTrap 2 incubation. Bacterial carbon was computed
using 20 fg C per cell conversion factor.
A list of the analyzed compounds and their abbreviations is shown in Table 1.
Chloropigment analysis—Chloropigments (chlorophyll +
pheopigments) and fucoxanthin were measured by reversephase high-performance liquid chromatography (HPLC)
(Mantoura and Llewellyn 1983; Bidigare et al. 1985) as
described in Lee et al. (2000). Chloropigments measured
include chlorophyll a, pheophorbide a, pyropheophorbide
a, and pheophytin a; monovinyl and divinyl chlorophylls
(Bidigare and Ondrusek 1996) were not separated. Filters
1647
were sonicated in 100% acetone to extract pigments.
Acetone extracts were filtered through 0.2-mm Zetapor
membrane filters, diluted 20% with MilliQ water, and
injected onto a 5-mm Adsorbosphere C-18 column.
Chloropigments were identified with a fluorescence detector (excitation l 5 440 nm, emission l 5 660 nm), and
fucoxanthin was identified with an absorbance detector (l
5 446 nm). Chloropigment and fucoxanthin concentrations were determined by comparison of sample peaks and
pigment standards. Chloropigment standards were either
purchased (chlorophyll a from Turner Design and pheophorbide a from Porphyrin Products) or prepared (pyropheophorbide a and pheophytin a were synthesized from
purified Chl a, and their concentrations determined
spectrophotometrically using known extinction coefficients; King 1993). Fucoxanthin was prepared from
extracted Phaeodactylum tricornutum cultures and quantified by comparison with a standard from Horn Point
Laboratories (University of Maryland). Duplicate analyses
of the same extract agreed within 10%.
Amino acid analysis—Total hydrolyzed amino acids
(THAA) were analyzed in all samples by HPLC using
precolumn o-phthaldialdehyde (OPA) derivatization after
acid hydrolysis as described in Lee and Cronin (1982) and
Lee et al. (2000). Thawed filters (usually half of each filter)
were sealed in glass tubes under N2 with 6 mol L21 HCl
and 0.25 wt% phenol added and hydrolyzed at 110uC for
20 h. Hydrolyzates were filtered through combusted glass
wool to remove particles. The supernatant was transferred
to a combusted glass vial, evaporated, and dissolved in
MilliQ water. Amino acids were analyzed by HPLC using
a modification of Lindroth and Mopper (1979). An Alltima
C-18 250-mm 5-mm column (Alltech) equipped with a guard
column was eluted at a flow rate of 0.95 mL min21. A
binary gradient of 0.05 mol L21 sodium acetate (pH 5.7)
and 5% tetrahydrofuran (THF) (eluant A) and methanol
(eluant B) was used, ramping from 22% B to 50% B in
40 min, then to 100% B in 20 min. OPA-derivatized amino
acids were detected by fluorescence and identified by
comparison with retention times of authentic standards.
An amino acid mixture (Pierce, Standard H) was used as
the standard. The nonprotein amino acids, b-alanine and caminobutyric acid (BALA and GABA), were added
individually to the standard mixture. Aspartic acid (ASP)
and glutamic acid (GLU) measurements include the
hydrolysis products of asparagine and glutamine. Analytical errors determined from duplicate analysis were ,10%.
Lipid class analysis—For lipid class analysis, filters were
extracted according to Bligh and Dyer (1959). Each filter
was ground in a monophasic solvent mixture
(CH2Cl2 : CH3OH : H2O, 1 : 2 : 0.8, v : v : v) with 100 mL
internal standard (hexadecanone, Sigma Chemical Ltd,
GC grade), sonicated, and left overnight at 4uC under N2
for extraction. The mixture was then filtered and 1 : 1
CH2Cl2 : H2O (v : v) was added to produce a biphasic
mixture. The aqueous phase was rinsed twice with
dichloromethane; the organic phases that contain the lipids
were combined and evaporated to dryness under N2. Lipid
1648
Table 1.
Goutx et al.
Abbreviation table.
Biogenic silica
Organic carbon
Particulate organic carbon
BSiO2
OC
POC
Total lipids*
hydrocarbons
ketone
wax esters
triacylglycerols
free fatty acids
alcohols
sterols
1,3-diglycerides
1,2-diglycerides
monoglycerides
chloroplast lipids
pigments
monogalactosyl-diglycerides
digalactosyl-diglycerides
phosphatidylglycerides
phosphatidylethanolamines
phosphatidylcholines
metabolites
Total pigments
chlorophyll a
chlorophyll b
phaeophorbide a
pyrophaeophorbide a
phaeophytin a
fucoxanthin
TLip
HC
KET
WE
TG
FFA
ALC
ST
1,3 DG
1,2 DG
MG
CL
PIG
MGDG
DGDG
PG
PE
PC
METAB
TPig
Chl a
Chl b
phide
pyrophide
phytin
fuco
Total hydrolyzed amino acids
aspartic acid
glutamic acid
histidine
serine
arginine
glycine
threonine
b-alanine
alanine
tyrosine
c-aminobutyric acid
methionine
valine
phenylalanine
isoleucine
leucine
lysine
THAA
ASP
GLU
HIS
SER
ARG
GLY
THR
BALA
ALA
TYR
GABA
MET
VAL
PHE
ILE
LEU
LYS
Total sugars
Fucose
Rhamnose
Arabinose
Galactosamine
Glucosamine
Galactose
Glucose
Mannose
Xylose
Fructose
Ribose
TCHO
fuc
rha
ara
gal-am
glucosa
galact
glucose
man
xyl
fru
rib
Lipid biomarkers
tetradecanoic acid
iso-pentadecanoic acid
anteiso-pentadecanoic acid
Pentadecanoic acid
hexadecenoic acid
hexadecanoic acid
octadecatetraenoic acid
octadecadienoic acid
oleic acid
cis-vaccenic acid
octadecanoic acid
eicosapentaenoic acid
eicosenoic acid
eicosanoic acid
docosahexaenoic acid
docosanoic acid
Hexadecanol
Octadecanol
Phytol
cholesta-5,22-dien-3b-ol
cholest-22-en-3b-ol
cholest-5-en-3b-ol
cholestan-3b-ol
24-methylcholesta-5,22-dien-3b-ol
24-methylcholesta-5,24(28)-dien-3b-ol
24-ethylcholesta-5,22-dien-3b-ol
24-ethylcholesta-5-en-3b-ol
4,23,24-trimethylcholest-22-en-3b-ol
C37-C39alkenones
14 : 0
i-15 : 0
a-15 : 0
15 : 0
16 : 1
16 : 0
18 : 4
18 : 2
18 : 1v9
18 : 1v7
18 : 0
20 : 5
20 : 1
20 : 0
22 : 6
22 : 0
16ROH
18ROH
phytol
27(5,22)
27(22)
27(5)
27(0)
28(5,22)
28(5,24/28)
29(5,22)
29(5)
30(22)
alken
* TLip is the sum of all lipid classes except HC, which can be contaminated by anthropogenic compounds.
extracts were stored in dichloromethane under N2 at
220uC until analysis. Lipid extracts were separated into
classes of compounds on chromarods and quantified using
an Iatroscan model MK-6s (Iatron, Tokyo; H2 flow
160 mL min21; air flow 2 L min21) coupled to a PC
equipped with a Chromstar 6.1 integration system (Bionis,
Paris). The Iatroscan thin-layer chromatographic–flame
ionization analyzer (TLC-FID) combines the performance
of TLC for complex mixture resolution with the capacity of
FID quantification. The chromatographic support, a quartz
rod coated with silicic acid, performs like a TLC plate and
passes through the FID burner in an Iatroscan chamber
for detection of separated compounds. Lipid extracts
were thus separated into classes of compounds from
neutral to polar without preliminary fractionation of the
lipid extract. We improved the separation of phosphoglycerides from phosphatidylethanolamine (Gérin and Goutx
1993), and the isolation of degradation metabolites (free
fatty acids, 1,2- and 1,3-diglycerides, and monoglycerides)
from other lipids (Striby et al. 1999). Sixteen classes of
lipids can be separated using the elution sequence presented
in Table 2.
Degradation of marine particles
1649
Table 2. Separation scheme of lipid classes on chromarods performed in six successive developments in elution baths of different
composition and polarity. Wax esters coelute with steryl esters, and ketone (hexadecanone) is used as internal standard. Abbreviations as
in Table 1.
Elution time (min)
Bath composition
Sorted lipid class
28
30
20
7
35
40
Hexane—diethylether—formic acid (97 : 3 : 0.2; v : v : v)
Hexane—diethylether—formic acid (80 : 20 : 0.2; v : v : v)
Hexane—diethylether—formic acid (80 : 20 : 0.2; v : v : v)
Acetone 100%
Chloroform—acetone—formic acid (99 : 1 : 0.2; v : v : v)
Chloroform—methanol—ammonium (50 : 50 : 5; v : v : v)
HC, WE, KET
TG, FFA
ALC, 1,3DG, ST, 1,2DG
no scan
PIG, MG, MGDG, DGDG
PG, PE, PC
The relative standard deviation is usually #612% for
replicate analysis (n 5 3) of natural samples. In the present
work, the Iatroscan analysis of each lipid extract was
performed in duplicate. Variability within duplicates was
on average 64%, except for two samples (613%, NT1-D
12 h, and 616%, NT2-A 120 h).
Lipid biomarker analysis—Analysis of individual lipid
compounds followed the method of Wakeham et al. (1997a).
Samples for analysis of neutral lipids and fatty acids were
filtered onto muffled glass fiber filters (GFFs) and extracted
by sonication with dichloromethane : methanol (2 : 1). Extracts were partitioned into dichloromethane after adding
a 5% NaCl solution and then dried over anhydrous Na2SO4.
An aliquot of the lipid extract was saponified under N2 with
aqueous 0.5 mol L21 KOH in methanol for 2 h at 100uC.
Neutral (nonsaponifiable) lipids were extracted from the
basic solution (pH . 13) using hexane, after which acidic
lipids were extracted with hexane following acidification to
pH , 2. The neutral lipid fraction was derivatized with bis(trimethysilyl)-trifluoroacetamide (BSTFA)/pyridine to
form trimethylsilyl (TMS) ethers of free hydroxyl groups,
while the acids were treated with diazomethane to produce
fatty acid methyl esters (FAMEs). Both fractions were
analyzed on a Carlo Erba 4160 gas chromatograph fitted
with a 60 m 3 0.32 mm inner diameter column coated with
0.25 mm of DB-5 (J&W Scientific), an on-column injector,
and a flame ionization detector. Separations were achieved
using a temperature program of 3uC min21 from 100–320uC
and H2 carrier gas at a head pressure of 1 kg cm22. Data
were acquired and processed with ChromPerfect software
(Justice Laboratories). Prior to analyses, cholestane (Aldrich) and methyleicosanoic acid (Sigma) were added as
internal standards to neutral and acid fractions, respectively.
Reproducibility is about 615%.
For NT2, fatty acids were analyzed in the remaining
lipid extract after Iatroscan lipid class analysis. Free fatty
acids and esterified compounds were derivatized to fatty
acid methyl esters (FAMEs) in the presence of BF3/
methanol/toluene for 1 h at 70uC under N2. Pure water
was added to stop the reaction, and derivatized lipids were
extracted from the aqueous phase with hexane : ether (9 : 1,
v : v). Compounds were purified from the mixture on silica
microcolumns, i.e., hydrocarbons eluted first with hexane,
FAMEs were then eluted with hexane : ethyl acetate
(100 : 1), and polar lipids remained on the silica. FAMEs
were analyzed on a Perkin Elmer Autosystem XL gas
chromatograph fitted with a 30 m 3 0. 25 mm inner
diameter column coated with 0.25 mm of BPX70 (SGE),
a split/splitless injector and a flame ionization detector.
Separations were performed using the following oven
temperature program: 3uC min21 from 50 to 144uC, 1uC
min21 from 144 to 182uC, 7uC min21 from 182 to 250uC,
11 min at 250uC. H2 was used as carrier gas with an initial
pressure of 2.8 kg cm22 (splitless injection method); after
0.95 min the pressure was decreased to 0.84 kg cm22. Data
were acquired and processed using TurboChrom Lite
software (Perkin Elmer). Quantification of individual
components was based on an external calibration using
the response factor of tricosanoic methyl ester (Sigma).
Carbohydrates—Carbohydrates were analyzed as described in Panagiotopoulos and Sempéré (2005). Briefly,
100-mL samples from initial and final incubation times
were filtered onto precombusted GFFs and stored in the
dark (220uC). Dried filters were hydrolyzed under N2 with
0.1 mol L21 HCl at 100uC for 20 h (Burney and Sieburth
1977). After evaporating the acid, the residue was dissolved
in water and filtered and aldose concentrations measured
by high-performance anion-exchange chromatography
with pulse amperometric detection (HPAEC-PAD) according to Mopper et al. (1992) as modified by Panagiotopoulos
et al. (2001). Monosaccharides were separated on a Carbopac PA-1 anion-exchange column (Dionex) by isocratic
elution with 19 mmol L21 NaOH at a flow rate of 0.7 mL
min21 at 17uC and were detected by a Decade electrochemical detector (Antec Leyden BV) using a gold working
electrode and a Pd reference electrode. In procedural
blanks using combusted GFFs hydrolyzed under the same
conditions, the only detectable sugar was glucose at
negligible concentrations (7–10 nmol L21). Analytical
errors determined from duplicate analysis were ,8% for
all sugars except ribose (15%).
POC analysis—POC in incubation samples was measured using a Carlo Erba model 1602 CNS analyzer (filters
were acidified under acid fumes to distinguish total and
organic carbon). For this analyzer, precision for C is 62%.
POC in NT3 samples was determined as described in
Peterson et al. (2005).
Biogenic silica (BSiO2) and silicic acid (DSi)—Sixty to
eighty milliliters of sample was filtered onto 0.4-mm
polycarbonate filters. BSiO2 was determined on filters,
1650
Goutx et al.
and DSi in the corresponding filtrates. BSiO2 analyses were
conducted using a variation of the method of Ragueneau
and Tréguer (1994), where the second digestion step with
HF was not used since lithogenic silica was only one
quarter of the total material (Lee et al. unpubl. data).
Samples were digested in 20 mL of 0.2 mol L21 NaOH for
4 h at 95uC. After cooling, 5 mL of 1 mol L21 HCl were
added before centrifugation and analysis of the silicic acid
concentration as described below.
DSi concentrations were determined according to the
molybdate blue spectrophotometric method (Tréguer and
Le Corre 1975 as modified by Gordon et al. 1993 to
segmented flow colorimetry). A Technicon autoanalyser
(Bran + Luebbe Inc.) was used. Analytical errors determined from duplicate analysis were ,10%.
Kinetic parameters—To compare the loss of organic
matter in the different settling velocity fractions, we
calculated a degradation rate constant k, which is the slope
of the regression line of the log-transformed concentrations
of the various organic compounds over the 5-d incubation
time. As concentrations decreased over time during decomposition and dissolution of particulate compounds, the
slopes were negative. The higher the absolute value of k, the
faster the turnover of the material. The determination
coefficient r2 shows the fit of the data to the regression
slopes, and the p value provides information about the
significance of the slope. For BSiO2, dissolution rates were
calculated for each SV fraction during the incubations. The
initial BSiO2 was calculated from the difference between
initial and final DSi concentrations and the final BSiO2
concentrations, i.e., (BSiO2)t0 5 [(DSi)tf 2 (DSi)t0] +
(BSiO2)tf. BSiO2 dissolution rate constants (k; d21) were
calculated from the slopes of regression lines of DSi
concentrations normalized to initial BSiO2 versus time,
which takes into account the experimental variability.
Thus, the k is based on the initial dissolution rate as
discussed in Greenwood et al. (2001).
Statistical treatment of the data—For the degradation
rate constants, the significance of the regression correlation
coefficient (r) against H0, r 5 0, was tested according to
Sokal and Rohlf (1981). H0 is rejected for r2 $ 0.56 (n 5 7);
p , 0.05. In addition, the robustness of the slope was tested
using the Jackknife test (Miller 1974). For each degradation
experiment, a range of slope values was calculated using all
combination of n-1 observations. The smaller the discrepancy found between minimum and maximum slope values
is, the higher the robustness of the regression slope
calculations, which suggests minimum experimental and
analytical biases.
Particle compositions were compared by means of the
analysis of variance (ANOVA) test. For most bulk
parameters, the significance of differences between deployments (NT1, NT2, NT3) was examined with a degree of
freedom n-2 5 5 to 9. For BSiO2 : POC ratios, ANOVA
tests were done with a higher degree of freedom n-2 5 60.
Because biogenic silica is conservative (BSiO2 + DSi 5
constant), eight values of initial BSiO2 were calculated from
the seven incubation samples and the control.
For individual biomarkers, statistical data treatment
other than principal components analysis (PCA) for each
NetTrap fraction was not possible because different
organic compound classes were analyzed in different splits
of the trap samples, and the low amount of material
available for the multitracer approach did not enable
duplicate sample analysis. PCA is a multivariate ordination
technique that reduces the number of variables in a data set
by constructing ‘‘latent variables,’’ or axes, through which
maximum variability in a data set is explained (Meglen
1992; ter Braak 1994). Our PCA combines multiple organic
matter biomarkers (70 compounds) in the NT samples and
groups compounds together based on compositional
similarities or separates them based on differences.
Abundance data as mol% (amino acids) or normalized to
100% (sugars, lipid classes, pigments, and lipid compounds) were normalized by subtracting the mean of the
observations and dividing by its standard deviation.
Results
The data sets for the three NetTraps (NT) are not
perfectly comparable because we did not measure the same
compounds in all samples. The sampling strategy for the
cruise was to collect two NetTrap samples (NT1 and NT2)
early during the cruise for conducting the two degradation
experiments (5 d each) while at sea. Biogenic silica, organic
carbon, amino acids, lipid classes, lipid fatty acids and
alcohols, and carbohydrates were analyzed on these
samples. A third deployment (NT3) at the end of the
cruise provided a sample for comprehensive identification
of specific lipid biomarkers at the molecular level, which
requires a larger POM sample. POC, amino acids,
pigments, and lipid biomarkers were analyzed in this last
NT sample. The complete analytical data set is available in
Web Appendix 1: http://www.aslo.org/lo/toc/vol_52/issue_4/
1645a1.pdf.
Elutriation of particles—Total organic carbon exhibited
a similar distribution among SV fractions for all NetTrap
samples (Fig. 1), showing the highest amount (48–67% of
total OC) in the fastest settling fraction A (.230 m d21), an
intermediate amount (20–30%) in B (115–230 m d21)
except in NT2 where it was low (,7%), and lower amounts
in the slowest settling fractions C and D (58–115 and 29–
58 m d21, respectively). For NT2, we elutriated the largest
amount of material of the three NetTrap samples. We
believe that the elutriator may have been overloaded with
this large, mucus-rich sample. Hence the different distribution of particles and OC from NT2 might reflect an
incomplete separation of material between the fastest
settling fractions NT2-A and NT2-B.
Composition of particles—The initial composition of
particles having different settling velocities was broadly
similar (Table 3). The mean BSiO2 : OC molar ratio of NT1
and NT2 SV fractions ranged from 0.10 6 0.07 to 0.20 6
0.07 and were not significantly different from each other
(ANOVA, n 5 62). Amino acids made up about 20–40% of
the total OC in all samples, total lipids made up
Degradation of marine particles
Fig. 1. Percentage of total organic carbon in each elutriator
fraction for NT1, NT2, and NT3 samples. Settling velocity ranges
are A, .230 m d21; B, 115–230 m d21; C, 58–115 m d21; D, 29–
58 m d21.
significantly more of the OC in NT1 (29–33%) than in NT2
(9–25%) (n 5 7, p , 0.05), and carbohydrates were a small
portion of the total carbon in both sample sets, significantly
higher in NT2 (9–16%) than NT1 (3–6%) (n 5 7, p , 0.05).
Pigments made up a small percentage of OC (0.1–1.5). In
NT1-C particles, the ratios of compounds to OC (data not
shown) was probably overestimated due to a low OC
concentration measured at T0 caused by a bias during
splitting of the bulk sample into aliquots for compound
analysis. Thus, for NT1-C, we only report particle
compositions based on the qualitative distribution of
individual compounds (see below). These compound classes
together made up 67–82% of the OC in NT1 and 50–67% in
NT2. Carbohydrates and lipid classes were not measured in
NT3.
Differences in biomarker composition were used to
differentiate source and diagenetic status of particles.
Seventy individual compounds were identified and quantified in most cases (cf. Table 1). On the basis of previous
studies, we selected biomarkers among these compounds
that together indicate sources and diagenetic status of
organic matter (Table 4). All fractions have all three
sources of OC (phytoplankton, zooplankton, and bacteria)
and, in particular, all have phytoplankton markers.
However, phytoplankton markers predominated more in
NT1 and NT3 compared with NT2, which had no
phytoplankton biomarker index values above 0.8 (Fig. 2;
abbreviations are given in Table 1). In NT1, the distribution of biomarkers shows clear differences as a function of
the settling velocity of the particles. The fastest settling
particle fractions A and B contained polyunsaturated 20:5
and 22:6 fatty acids and were enriched in several indicators
of fresh undegraded phytoplankton: Chl a, TG, galact,
phytosterols [28(5,22)] and [28(5,24/28)], and long-chain
C37 and C38 alken. Fraction C was dominated by the
pigments fuco and pyrophide, the zooplankton reserve WE
and zooplankton-derived sterols [30(22)], the lipid degradation metabolites METAB, the aminosugars glucosa and
gal-am, and fecal pellets–derived 16ROH and 18ROH,
which taken together indicate phytoplankton, especially
diatomaceous material, that has been ingested by zooplankton and released as fecal pellets. The composition of
D was more heavily influenced by higher values of bacterial
1651
cell or diagenetic process indices: BALA/ASP and PE/PG
ratios, 15:0 and i-15:0, and the ratio of refractory cell-wall
deoxysugars fuc + rha to easily assimilated sugars ara + xyl
(see also Web Appendix 1; Table A1.1 and Table A1.2).
For NT2, we analyzed the same compound classes as in
NT1, except the sterols and alcohols phytol, 16ROH and
18ROH. In general, the biomarker composition for NT2
was less diverse than for NT1. There were no reserve lipids
(TG, WE), whereas CL and METAB tracers of degradative
hydrolysis dominated all other lipid classes (see Web
Appendix 1; Table A1.1). The phytoplankton (Chl a) and
bacteria-fecal pellet (glucosa) biomarkers were largely
undifferentiated between SV fractions NT2-A, NT2-C,
and NT2-D (Fig. 2), although SV fraction A had more
polyunsaturated fatty acids C20:5 and C22:6 in it. Bacterial
lipid biomarkers, PE/PG and the 15:0 and i-15:0 fatty acids,
dominated in NT2-C. NT2-B SV fraction was notably
different from the others, and dominance of fuco and galam suggests diatom material reworked by zooplankton.
For NT3, phytoplankton indices (Chl a, fuco) were
similarly distributed in all SV classes. Among all NetTrap
samples, NT3-A, NT3-B, and NT3-C exhibited the highest
indices of fresh plankton material (20:5 and 22:6) and
chlorophyll pigment degraded by zooplankton grazing
(pyrophide), whereas NT3-D had very low polyunsaturated
fatty acids and high bacteria-reworked fecal matter indices
(18ROH).
Principal components analysis—PCA for NT 1 used
amino acids, lipid biomarkers, lipid classes, sugars, and
pigments (Fig. 3A). Compounds from the three sources
cluster in different regions of the plot, as shown by the
shading. Loadings for the phytoplankton biomarkers, for
example, Chl a, phytosterols [e.g., 28(5,22)], and TG, tend
to group toward the positive direction along PC1, whereas
loadings for zooplankton/grazing biomarkers (METAB,
glucosa, gal-am, pyrophide) cluster toward the negative
direction on PC1, suggesting that source of the organic
matter explains more of the variability along the x-axis
(PC1). These same zooplankton biomarkers along with the
bacterial biomarkers (phytin, GABA, branched-chain i15:0 and a-15:0, 15:0), degradation resistant lipids (CL),
and sugars (rha) group toward the positive side of PC2,
compared with the fresh phytoplankton markers in the
negative region of PC2, suggesting that the y-axis (PC2)
reflects variability due to degradation. The two axes
accounted for 78% of the total variability. In general, the
grouping of compounds in the PCA of NT1 indicates that
settling velocity fraction A is predominately phytoplankton
aggregates, while B and especially C contain diatom-rich
processed zooplankton fecal matter, and D is a mixture of
plankton detritus and bacterial biomass. These conclusions
from PCA agree well with individual biomarker analyses.
PCA for NT2 samples shows a clear picture of
compositional distinctions between NT2-B fraction and
the other fractions NT2-A, NT2-C, and NT2-D (Fig. 3B),
which explains a total of 83% of the total variability. NT2B fraction is most influenced by saturated fatty acids (18:0,
20:0), in addition to zooplankton-reworked material
containing gal-am and fuco biomarkers. Settling velocity
1652
Goutx et al.
Table 3. Biochemical characteristics of particles with different sinking velocities at initial and final times of incubation. A single
measurement 6 analytical error is presented for all parameters except BSiO2 at initial time; for this parameter, the average value 6
standard deviation was calculated from eight replicate analyses. Abbreviation as in Table 1; nd means not determined.
OC-normalized concentrations (%)
Setting velocity (m
Initial time
NT1-A
NT1-B
NT1-D
NT2-A
NT2-B
NT2-C
NT2-D
NT3-A
NT3-B
NT3-C
NT3-D
Final time
NT1-A
NT1-B
NT1-D
NT2-A
NT2-B
NT2-C
NT2-D
d21)
Deployed 15.7 h, 07–08 May 2003
.230
115–230
29–58
Deployed 64.3 h, 08–10 May 2003
.230
115–230
58–115
29–58
Deployed 74.8 h, 10–14 May 2003
.230
115–230
58–115
29–58
.230
115–230
29–58
.230
115–230
58–115
29–58
BSiO2 : OC (%)
THAA
TLip
TCHO
TPig
1267
1367
2067
3263
4364
4364
3260.7
2960.6
3360.7
460.2
360.2
660.4
0.260.02
0.360.03
0.360.03
1466
16610
1166
1067
2663
2462
3263
3564
960.3
1660.4
2561.7
1560.4
1661.0
960.6
1060.6
1160.7
0.560.06
1.560.15
0.160.01
0.460.04
nd
nd
nd
nd
3564
4866
24–36
3364
nd
nd
nd
nd
nd
nd
nd
nd
0.260.02
0.160.01
0.260.02
0.160.01
1161
2462
2863
9.761
9.561
9.061
9.261
2062
3163
3263
2963
2162
2563
2563
1260.5
1760.7
1861.2
961.4
3060.6
1760.6
1460.3
260.1
360.2
660.4
1260.7
nd
960.6
1060.6
0.260.02
0.360.03
0.360.03
0.460.04
1.260.12
0.460.04
0.460.05
fraction A, and to a lesser extent C and D, are all associated
with phytoplankton and zooplankton markers. Fractions C
and D are more influenced by bacterial biomarkers phytin
and branched-chain fatty acids i-15:0.
PCA for NT3 included amino acids, lipid biomarkers,
and pigment compositions (Fig. 3C). As for NT2, the two
axes accounted for 83% of the total variability. Along the
PC1 axis, negative loadings of biomarkers associated with
undegraded OM, for example polyunsaturated 20:5 and
20:6 fatty acids, versus positive loadings of compounds
associated with degraded or reworked OM, GABA, phide,
and zooplankton-derived 16ROH and 18ROH, suggest
that variation along this axis is primarily due to bacterial or
zooplankton reworking. The most rapidly settling particles
(A) were dominated by plankton sterols [e.g., 27(5),
28(5,22), 28(5,24/28), and 30(22)] and alken (see Web
Appendix 1; Table A1.1). Particles having intermediate
settling velocities (B and C) were associated with phide and
pyrophide, polyunsaturated 20:5 and 22:6 fatty acids,
18:1v9 and 16ROH, branched-chain fatty acids (a-15:0
and i-15:0) consistent with zooplankton fecal matter. For
NT3, and in contrast with NT1, the slowest settling
material (fraction D) plotted with phytoplankton pigments
Chl a and fuco and bacterial GABA, suggesting an
enrichment of both small phytoplankton cells (assumed to
be disaggregated) and associated bacteria in this settling
class.
Identical amino acid and pigment measurements in all
three NT sample sets allowed a comparison of NT1, NT2,
and NT3 by PCA. This PCA shows that NT1 and NT3
were more similar in composition to each other than NT2
(Fig. 4). Although phytoplankton, zooplankton, and bacterial organic matter were present in each trap (Fig. 3),
NT1 and NT3 were generally more enriched in fresh
phytoplankton (Chl a) and zooplankton fecal matter
(pyrophide) than NT2, which was relatively more influenced by the bacterial degradation marker GABA and the
diatom markers fuco and SER (Fig. 4). In all three
NetTraps, the slowest settling fractions (D) were separated
from faster settling particles in composition, which suggests
that slower settling particles were the most influenced by
the bacterial biomarkers phytin and BALA. These findings
are in agreement with PCA of individual NT samples that
also included lipid and carbohydrate data (Fig. 3); the
slowest settling fractions (D in NT1, C and D in NT2 and
NT3) were clearly associated with the bacterial markers
phytin, 15:0, and branched-chain fatty acids (a-15:0 and i15:0). Data in Web Appendix 1; Table A1.1 partly support
this interpretation.
Particle degradation and kinetic parameters—Particle
incubations started with initial POC concentrations in the
range 58–225 mmol L21 and BSiO2 initial concentrations in
the range 10.5 to 26.9 mmol L21, except for NT2-A, which
was more concentrated (740 mmol L21 for POC and
102 mmol L21 for BSiO2). For each SV fraction of NT1
and NT2, the decrease of POC, THAA, TLip, and all
individual compound concentrations over time was fit to
a first-order decay equation G 5 G0e2kt where G is the
concentration of a component at time t and G0 the initial
Degradation of marine particles
1653
Fig. 2. Distribution of selected phytoplankton, zooplankton, and bacterial biomarkers in NT1, NT2, and NT3 samples. See
abbreviations and interpretation in Table 1 and 4 and in the text. The biomarker ‘‘index’’ was derived by normalizing the original data to
1 by dividing each value by its maximal value across the NT1, NT2, or NT3 data set. The original data that were normalized were either
the mol% of an individual compound within its compound class for total pigments, amino acids, lipid classes, fatty acids, and alcohols, or
the ratio of those mol% values for selected compound pairs. Blank spots in the graphs were not determined (nd).
concentration. An apparent degradation rate constant (k,
d21) was calculated from the slope of log-transformed
concentrations versus time for total compound classes and
for each individual compound.
Because we collected a limited amount of particles,
replicate incubations for each incubation time were not
possible given our multiparametric approach. However, we
replicated the incubation samples at 48 h that we used to
evaluate the variability within batches. Variability within
the duplicates was ,620% for POC, THAA, and TLip (on
average 7.6 6 6.6 for all classes of compounds) except in
NT2-B, where it was higher for THAA (640%) and TLip
(640%). The variability in BSiO2 concentration between
duplicates was similar to that of carbon with an average
variability of 22% excluding the NT2-B experiment (49%).
The variability of initial BSiO2 measurements between all
batches of each degradation experiment was 40% excluding
the NT2-B experiment (which was 70%).
For the faster settling particles NT1-A, NT1-B, NT2-A,
and NT2-B, most organic compounds did not significantly
degrade during the 5-d incubation (n 5 9, p , 0.05), and
kinetic parameters are not presented. Only a few individual
biomarkers had significant apparent degradation rate
constants. For example, in NT1-A, which was rich in
phytoplankton, PG and phytin had significant degradation
rate constants during incubation (0.16 6 0.04 d21, r2 5
0.75 and 0.12 6 0.04 d21, r2 5 0.62, p , 0.05, n 5 7).
Hydrolyzable lipid classes, TG (7–8% of TLip in NT1-A )
and WE (6% of TLip in NT1-A), degraded within 48 h of
incubation.
For the slower settling particles C and D from NT1 and
NT2, most bulk compound classes had a significant
degradation pattern (k 5 0.03–0.12 d21, r2 . 0.56, n 5 7,
p , 0.05), with exception of NT1-C (Table 5). We
confirmed the robustness of degradation rate constants
that fell at the limit of significance, as for kPOC in NT2-C,
by using the Jackknife test; a small range for k was
calculated (kPOC 5 0.04 6 0.02 d21, r2 5 0.56, range 0.03–
0.05 d21). For NT1-C, decomposition of organic compounds was noticeable although statistically nonsignificant,
e.g., THAA. Several individual amino acid (MET, LEU,
ARG) and lipid class compounds (FFA and MGDG)
exhibited an almost constant degradation pattern in
the four incubation experiments, with higher degradation constants than bulk organic matter (0.10–0.38 d21)
(Table 5).
1654
Table 4.
Goutx et al.
Selected biomarkers and indices.
Biomarkers and indices
chlorophyll a
Triacylglycerols
Galactose
eicosapentaenoic acid
docosahexaenoic acid
24-methylcholesta-5,22-dien-3b-ol
24-methylcholesta-5,24(28)-dien-3b-ol
Alkenones
Fucoxanthin
4,23,24-trimethylcholest-22-en-3b-ol
wax esters
Metabolites
Galactosamine
Glucosamine
Phaeophorbide
Pyrophaeophorbide
c-aminobutyric acid
b-alanine/aspartic acid
phosphatidylethanolamines/phosphatidylglycerides
iso-pentadecanoic acid
pentadecanoic acid
Fucose + rhamnose/arabinose + xylose
Hexadecanol
Octadecanol
Biogenic opal generally dissolved faster in NT2 than in
NT1; BSiO2 dissolution was not statistically significant in
NT1 (Table 6). In NT2, BSiO2 dissolved faster in the
slowest settling particles NT2-D than in NT2-C (Table 6).
Particle composition changes during incubation—During
the incubation period, the mass of organic carbon did not
decrease in the faster settling particles of NT1 and NT2.
However, a significant loss in organic carbon was measured
in slower settling particles of NT2-C and NT2-D (on
average 15.5 6 0.5% of the initial mass, n 5 2); a loss of OC
was also noticeable in NT1-C and NT1-D (on average 5.0
6 2.0% of the initial mass, n 5 2), although not significant.
DSi significantly increased between initial and final incubation times in NT2 samples (ANOVA, p , 0.05, n 5 7),
but not in NT1 (see Web Appendix 1; Table A1.2).
The organic composition of the various settling velocity
fractions from NT1 and NT2 changed between the
beginning and end of the incubation period (Table 3). In
general, the sum of the contribution of amino acids, lipids,
carbohydrates, and pigments to total organic carbon
decreased from 75.2 6 7.0 to 47.2 6 11.6% (n 5 3) in
NT1 and from 57.6 6 8.0 to 50.7 6 1.3% (n 5 4) in NT2
between initial and final incubation times. The percentage
of THAA and TLip, the major contributors to organic
carbon, decreased between initial and final times in all
incubations except in NT2-A and NT2-B. In parallel,
changes in the percentage of pigments in OC and
BSiO2 : OC ratios were not significant. Biomarker distributions at the final incubation time are given in Fig. 5.
Because lipid individual biomarkers were not analyzed at
Sources
phytoplankton
phytoplankton reserves
fresh plankton
diatoms, dinoflagellates, and zooplankton feeding on it
phytosterols
phytosterols
prymnesiophyceae
diatoms
zooplankton-derived lipid
zooplankton reserves
1,3 DG; 1,2 DG; MG, ALC, FFA
bacterial cell wall and peritrophic membrane of fecal pellets
bacterial cell wall and peritrophic membrane of fecal pellets
chlorophyll degradation produced by zooplankton
chlorophyll degradation produced by zooplankton
OM bacterial alteration
microbial
index or bacterial biomass
microbial
microbial
bacterial cell wall
bacteria-reworked fecal matter
bacteria-reworked fecal matter
the end of the incubation, there are fewer indicators than at
the initial incubation time. Thus we present phytoplankton,
processed zooplankton fecal matter, and bacterial indicators in each NetTrap material (NT1 and NT2) in one panel
only. In NT1, phytoplankton biomarker indices for Chl
a and galact increased up to ,0.8–1 in the slower settling
particles NT1-C and NT1-D, suggesting either a low
biodegradability for these compounds or more likely
a contribution of individual (nonaggregated) cells. Index
values of compounds indicative of bacteria-zooplankton
fecal matter aggregates (METAB, gal-am, phide) and
bacterially reworked organic matter (GABA, PE/PG, and
fuc + rha/xyl + ara) were comparable or higher at the end of
the incubation period compared with initial values. The
index value for glucosa peaked in NT1-C, whereas it
decreased in all other SV fractions. In NT2, bacterial
biomarker indices (GABA and fuc + rha/xyl + ara)
increased in particles that exhibited significant degradation
by the end of the incubation.
PCA of the amino acid, pigment, and lipid class
composition of particles with different settling velocity at
different incubation stages showed that their organic
chemical compositions changed over time (Fig. 6). There
was a general trend from left to right along the first PCA
axis. After 120 h of incubation, the C and D particles had
similar end compositions, and the A and B samples were
closer to this end composition after 120 h than they were
initially. These end samples were enriched with the bacterial
biomarkers GABA, BALA and phytin, the phide found in
fecal matter, and had more acyl-lipid degradation metabolites (METAB) in the lipid pool than initially.
Degradation of marine particles
Discussion
Effect of environmental conditions on NetTrap samples—
The well-studied DYFAMED site (Marty 2002) in the
northwestern Mediterranean Sea is protected from major
terrestrial inputs by the presence of the coastal Ligurian
current, conferring a unique advantage to this site for
studying biogeochemical process along a vertical (depth)
dimension. The seasonal hydrological regime varies from
winter mixing (January–February) to strong thermal
stratification in summer and fall. Typically, a diatom
bloom peaks in March and ends in May, followed by
nanoflagellates 1 or 2 months later; cyanobacteria and
prochlorophyte communities develop when stratification
begins. Diatoms are very sensitive to episodic wind events
that bring nutrients to the surface layer. At the end of the
stratification period before winter mixing, wind events
induce a fall diatom bloom. Phytoplankton production is
thus characterized by a succession of predictable mineralsecreting and mineral-free phytoplankton that are grazed
by a small number of fecal pellet forming zooplankton
species (Nival et al. 1975; Carroll et al. 1998).
We began sampling in early May when phytoplankton
biomass was still high, and Chl a concentrations reached
1.7 mg L21 at 50 m (Liu et al. 2005). NetTrap particles were
quite rich in chromatographically resolvable amino acids,
sugars, and lipids indicating freshly biosynthesized organic
matter. The particles were slightly more enriched in
proteins and lipids than during an earlier May 1995 cruise
(10–27% and 10–28%, respectively; Goutx et al. 2000),
whereas carbohydrates were similar (Panagiotopoulos and
Sempéré 2005). Values fell within the range reported for
time-series sediment trap data collected in May 2003 at the
same site (Wakeham et al. unpubl. data). Even though
particles collected by the NetTrap at 200 m were relatively
‘‘fresh,’’ organic compositions indicated that sources of
material between the traps and in the different settling
velocity fractions clearly differed over the 1-week sampling
period. As seen in Peterson et al. (2005), the sampling
period was during a shift from higher spring productivity to
a lower flux period. Further, as we show below, material
collected over the week had different susceptibility to
degradation. The larger contribution in NT1 material of
phytoplankton indicators exported via aggregates and/or
fecal pellets (Fig. 2) was consistent with the annual
phytoplankton pattern of a diatom-based linear trophic
chain at the site (Andersen and Prieur 2000). Material
collected in NT2 was similar in composition, although it
appeared that this NetTrap sampled the end of the diatom
bloom. Indeed, we observed visually during preparation of
the particles for incubation that they were embedded in
mucus that is released as diatoms begin to senesce (this was
also reflected by the high sugar content, 16% of total
carbon, Table 3; Passow et al. 1994). The end of the bloom
is often linked to senescence of diatoms and was illustrated
here by the higher dissolution rate constant of BSiO2 in
NT2 (.0.024 d21), whereas BSiO2 barely dissolved at all in
NT1 (,0.005 d21, except NT1-C, k 5 0.022 d21). This
higher dissolution rate constant in NT2 could be due to
a lower viability of cells or to greater degradation of the
1655
external cellular membrane by bacteria. Zooplankton and
fecal pellet indicators more heavily dominated NT3,
suggesting that the natural progression from phytoplankton to zooplankton was occurring over our sampling
period. Several salps were visually observed in the water
when NetTrap 3 was recovered.
Different deployment times for the NetTraps (see
Table 3) might be another possible cause of differences in
initial NetTrap sample compositions: NT1 (16 h) was
deployed for a shorter time than NT2 (64 h) and NT3
(72 h). Comparison of NT1 and NT2 in which amino acids,
lipid classes, pigments, and carbohydrates were analyzed
suggests that organic matter degradation during the longer
deployment time could explain why NT2 triglycerides,
chlorophyll a and glucose, which have short residence
times, were lower compared with NT1 (Fig. 2 and Web
Appendix 1; Table A1.1).
Elutriation and composition of SV particle classes—
Elutriation requires care in selecting the amounts elutriated, flow rates, and particle introduction (Peterson et al.
2005), and the samples reported here represent the first field
experience with natural settling marine particles. Elutriated
fractions of NT1 and NT3, but not NT2, showed similar
compositional patterns for the four different SV classes.
The relatively higher OC content in the fastest settling
velocity fraction of NT2 (NT2-A) was accompanied by
a reduced yield in fraction B (Fig. 1). Such a pattern likely
resulted from lower elutriator separation efficiency of NT2
particles for two reasons. NT2 particle load in the elutriator
was the highest of the three NetTrap samples (Web
Appendix 1; Table A1.1), and these particles were richer
in mucus than NT1 or NT3. We have observed that shear
in the countercurrent tubes during elutriation can induce
aggregation; this would especially occur if too much
mucus-rich material is loaded into the first fraction A.
Shear may also lead to disaggregation. Thus even though
elutriated fractions may accurately reflect settling velocity
distribution of particles in the elutriator, they may not
reflect in situ distributions. Additional experimental work
is needed to verify how well elutriated samples retain the in
situ character of particles. Nonetheless, particles in nature
also undergo continual exchange as they disaggregate and
reaggregate (e.g., Bacon and Anderson 1982; Wakeham
and Canuel 1988; Sheridan et al. 2002).
In spite of these caveats, we believe that elutriation of
NetTrap material at least to a first approximation
represents the character of particles in nature. The particle
‘‘flux’’ collected by the NetTrap, a device that was not
explicitly designed to accurately measure particle flux, at
200 m (,200 mg m22 d21) was similar to fluxes measured
by moored 200-m time-series arrays just before (153 mg
m22 d21) and after (165 mg m22 d21) the NetTrap deployment (Peterson et al. 2005). Likewise, most of the
organic carbon in elutriated material was in the fraction
with the highest settling velocity (.230 m d21), with
a decreasing carbon contribution from slower settling
fractions. This result is consistent with observations made
using an independent settling-velocity sediment trap
deployed at 200 m at the same time, for which the bulk
1656
Goutx et al.
Degradation of marine particles
1657
Fig. 4. PCA comparing NetTrap 1, 2, and 3 samples. The data set used included mol%
individual amino acids and pigments in fractions after elutriation (but before incubation);
samples and loadings are shown on the same graph. See abbreviations in Table 4.
of mass and organic carbon was associated with particles
settling at rates of .200 m d21 (Peterson et al. 2005). High
particle settling velocities ($600 m d21) were also estimated from the time lag between lipid signatures in surface
particles and 200 m drifting traps at the DYFAMED site
during the DYNAPROC (Dynamique des Processes
Rapides dans la colonne d’eau) cruise (Goutx et al. 2000).
Closer inspection of initial distributions of selected
biomarkers shows clear differences between the samples
separated by SV. Biomarkers were selected on the basis of
previous studies on organic matter composition in sinking
particles and sediments (Wakeham et al. 1997b; Dauwe and
Middelburg 1998; Dauwe et al. 1999; Goutx et al. 2000; Lee
et al. 2000; Panagiotopoulos and Sempéré 2005, and
references therein). Specific indicators for bacterial cells
or diagenetic process were selected from a large variety of
compound classes, sugars (fuc + rha/ara + xyl; Biersmith
and Benner 1998; Opsahl and Benner 1999), amino acids
(BALA/ASP; Lee and Cronin 1984), lipid classes (PE/PG;
Goutx et al. 2003), molecular fatty acids (15:0 and i-15:0;
Kaneda 1991), and alcohols (16ROH and 18ROH; Wakeham 1982). Because individual biomarker distributions
(like mol% GABA) were measured in a single split of the
sample, they are inherently more accurate (relative abun-
r
Fig. 3. Principal components analysis (PCA) of individual compound relative concentrations
(percentage of total compounds class) in NetTrap fractions after elutriation (but before
incubation), where ‘‘samples’’ are the elutriator settling velocity fractions and ‘‘loadings’’ are
organic components; samples and loadings are shown on the same graph. The first two principal
components (PC1 and PC2) explain the greatest variance in the data. (A) NT1 fractions A, B, C,
and D at time zero using the data set of lipid classes and individual amino acid, lipid,
carbohydrate, and pigment compounds. Shading shows areas dominated by phytoplankton
(lower area), zooplankton (upper left), and bacterial (upper right) biomarkers and is included
only to aid in viewing the PCA. (B) NT2 fractions A, B, C, and D using the data set of lipid
classes and individual amino acid, fatty acid (except hydroxy fatty acids), carbohydrate, and
pigment compounds. Shading shows areas dominated by phytoplankton (upper left),
zooplankton, and bacterial biomarkers in fractions C and D (lower area). (C) NT3 fractions
A, B, C, D using the data set of individual amino acid, lipid, carbohydrate and pigment
compounds. Shading shows areas dominated either by phytoplankton and zooplankton or by
phytoplankton and bacterial biomarkers. See abbreviations in Table 1.
1658
Goutx et al.
Table 5. Apparent first-order degradation rate constants, k (d21), obtained from the slope of log-transformed concentrations versus
time for total compound classes and individual compounds; determination coefficient r2; range of k calculated using all combination of n1 observations in NT1-D, NT2-C, and NT2-D. Only variables, with statistically significant results (p , 0.05) in at least two of the slower
settling fractions, are presented, except POC; s, significant; ns, nonsignificant.
Compound classes
POC
THAA
TLip
Individual compounds
MET
LEU
ARG
FFA
MGDG
Sample
k (d21)
r2
p,0.05 n57
NT1-C
NT1-D
NT2-C
NT2-D
NT1-C
NT1-D
NT2-C
NT2-D
NT1-C
NT1-D
NT2-C
NT2-D
0.0460.06
0.0560.04
0.0460.02
0.0460.02
0.1160.08
0.0860.02
0.0960.02
0.0760.02
0.0460.11
0.1260.03
0.1260.04
0.0360.02
0.09
0.20
0.56
0.39
0.27
0.76
0.78
0.72
0.03
0.77
0.70
0.29
ns
ns
s
ns
ns
s
s
s
ns
s
s
ns
NT1-C
NT1-D
NT2-C
NT2-D
NT1-C
NT1-D
NT2-C
NT2-D
NT1-C
NT1-D
NT2-C
NT2-D
NT1-C
NT1-D
NT2-C
NT2-D
NT1-C
NT1-D
NT2-C
NT2-D
0.1460.11
0.1860.06
0.1760.02
0.1360.03
0.1560.18
0.1360.04
0.1060.03
0.1160.02
0.1160.10
0.1260.04
0.1060.03
0.1060.03
0.0460.19
0.0560.07
0.1360.05
0.3860.04
no degradation
0.2760.08
0.1860.06
no degradation
0.24
0.64
0.90
0.79
0.40
0.68
0.75
0.88
0.15
0.67
0.75
0.71
0.01
0.08
0.59
0.96
ns
s
s
s
ns
s
s
s
ns
s
s
s
ns
ns
s
s
0.71
0.63
s
s
dances are internally consistent) than compound class
ratios that were measured in different splits of the sample.
Compound classes may also be less sensitive than individual biomarker analyses because classes add together
compounds that may have different sources or behave in
different ways. These compositional differences in SV
classes (elutriator fractions) are distinct even though the
k range
0.03–0.05
0.07–0.13
0.08–0.10
0.04–0.08
0.12–0.14
0.11–0.24
0.15–0.20
0.15–0.18
0.11–0.14
0.12–0.27
0.07–0.11
0.08–0.12
0.10–0.25
0.07–0.11
0.08–0.11
0.10–0.14
0.33–0.40
0.13–0.29
0.15–0.39
range of BSiO2 : OC ratios (0.10–0.20, Table 2) showed that
all samples were diatom dominated (Ragueneau et al.
2002). These BSiO2 : OC ratios were close to the ratio of
0.13 measured by Brzezinski (1985) on diatoms cultured in
silica-replete conditions. The presence of either degraded
material, fecal pellets (Tande and Slagstad 1985; Cowie and
Hedges 1996), or more silicified species (in response to
Table 6. BSiO2 dissolution rate constants, k (d21), calculated from the increase of DSi normalized to initial BSiO2 over time in each
settling velocity fraction; s, significant; ns, nonsignificant.
NT1-A
NT1-B
NT1-C
NT1-D
NT2-A
NT2-B
NT2-C
NT2-D
Settling velocity (m d21)
k (d21)
r2
p,0.05 n57
.230
115–230
58–115
29–58
.230
115–230
58–115
29–58
0.00260.003
20.00260.007
0.02260.049
0.00560.008
0.02460.002
0.02660.021
0.02660.013
0.06060.010
0.099
0.013
0.007
0.081
0.977
0.856
0.457
0.893
ns
ns
ns
ns
s
s
s
s
Degradation of marine particles
1659
nutrient limitation, Claquin et al. 2002) is known to
increase this ratio.
All samples contained biomarkers indicating the presence of phytoplankton, zooplankton, and bacterial organic
matter, but faster settling particles were dominated by
phytoplankton aggregate indicators, particles of intermediate settling velocity had more zooplankton and fecal
pellet indicators, and slower settling particles were more
associated with bacterial indicators (Figs. 2, 3, and 4).
Accordingly, we use the terms ‘‘fresh’’ and ‘‘reworked’’
material for characterizing the two extreme qualitative
features of organic matter in the particle spectra. These
compositions fit the conventional paradigm that faster
settling particles in the deep ocean contain freshly
biosynthesized material that is not highly decomposed
while slower settling material has been actively altered by
heterotrophy (e.g., Honjo et al. 1982; Wakeham and Lee
1993).
Fig. 5. Biomarker indices using normalized data as in Fig. 2
but at the final incubation stage of samples from NetTraps 1 and 2.
Biodegradability of the different settling velocity classes—
The relationship between the quality of SV particle classes
and their bioavailability (measured as an apparent degradation rate constant) suggests that the faster settling
particles (with indicators of ‘‘fresh material’’) contain less
bioavailable organic matter, whereas slower settling particles (with ‘‘reworked material’’ indicators) are more
bioavailable. For NT1 and NT2 incubation experiments
(NT3 was not incubated), faster settling A and B particles
fell within the first group (fresh matter/nonbioavailable),
whereas slower settling C and D particles fell within the
second group (reworked matter/bioavailable). This conclusion at first may seem counterintuitive.
Fig. 6. PCA using the NT1 data set of individual amino acid, lipid, and pigment compounds
in incubated particles from the different settling velocity fractions. See abbreviations in Table 4.
The gray arrows give a general sense of the trend in composition with time.
1660
Goutx et al.
Table 7. Mean degradation rate constants, k (d21), and
turnover time, 1/k (d), of total amino acids (THAA), total lipids
(TLip), individual amino acids, and lipid classes in slow sinking
particles (data were computed using significant k from Table 5).
K
1/k
Compound classes
THAA
TLip
0.0860.01 (n53)
0.1260.00 (n52)
12.5
8.3
Individual compounds
Amino acids
Lipid classes
0.1360.03 (n59)
0.2460.11 (n54)
7.9
4.2
The presence of attached bacteria could be one reason
for the greater bioavailability of the slowest settling
particles. If slower settling particles from the upper ocean
have a longer residence time in the water column, they
would be relatively ‘‘older’’ in age at the time (or depth) at
which we sampled compared with the faster settling
particles. Therefore, free bacteria from the surrounding
water column could more efficiently colonize them.
Accordingly, fraction D would already be initially enriched
in bacterial tracers (Figs. 2, 3, and 4) relative to the other
fractions, and it is the D fractions that were most affected
by biodegradation during our incubations. Furthermore, in
those incubation samples for which we determined bacterial
carbon (NT2 samples), the largest increase in the proportion of POC contributed by bacteria was observed in C
and less so in D (data not presented). Thus, at the time of
sampling, the slower settling particles having relatively
longer transit times to 200 m contain more (or metabolically more active) bacteria than faster settling particles.
Fresh bacterial biomass/necromass could also be fueling
degradation of these particles, changing their composition
during settling (see discussion below).
Another possible explanation for the lower short-term
bioavailability of faster settling freshly produced particles is
that faster settling material was made up of larger
aggregates, while slower settling particles were much
smaller, based on visual observations of the elutriator
fractions. Phytoplankton cells inside aggregates stay alive
longer than those that are free in the medium (Moriceau et
al. 2007), and a living phytoplankton cell is less subject to
bacterial degradation. Moreover, in large aggregates,
minerals may limit decomposition by physically protecting
organic matter with which they are associated (Arnarson
and Keil 2000). The DYFAMED site receives Saharan dust
input in the spring, and there was a dust event early in May
2003 just before our sampling period (Bartoli et al. 2005).
Thus, clay minerals may have contributed to increase the
resistance of aggregates to microbial decomposition.
In addition, the process of BSiO2 dissolution in
aggregates and/or free cells could affect the susceptibility
of organic compounds to degradation. Since they are
‘‘older’’ at the time (or depth) of sampling compared with
faster settling particles, the slower settling particle fractions
NT1-C, NT1-D, and NT2-D certainly experienced degradation, triggering BSiO2 dissolution. BSiO2 dissolved 10
times faster in NT1-C and two times faster in NT1-D than
in NT1-A and NT1-B, and two times faster in NT2-D than
in NT2-A, NT2-B, and NT2-C (Table 6). While the
embedding of diatom cells inside aggregates or fresh fecal
pellets slows down BSiO2 dissolution (Passow et al. 2003;
Schultes 2004; Moriceau et al. 2007), the breakdown of
external cell membranes by enzymes and/or the breakage of
diatom frustules in fecal pellet by coprophagy (Schultes
2004; Ragueneau et al. unpubl. data) would enhance BSiO2
dissolution rates. Coprophagy would be more likely to
occur during slower settling.
The fact that BSiO2 dissolved faster in the more
degraded particles of NT2 than in the fresher particles of
NT1 (Fig. 2 and Table 6) was consistent with the idea that
BSiO2 dissolution only started after degradation of the
organic matter surrounding the frustule (Bidle and Azam
1999, 2001). The significant degradation rate constants of
several compounds that we observed in faster settling
particles from NT1 (which exhibited lower BSiO2 dissolution) may also indicate decomposition of compounds that
protect BSiO2. On the other hand, the higher BSiO2
dissolution rate constants in faster settling particles in
NT2 compared with faster settling particles in NT1 (one
order of magnitude) were not accompanied by significantly
higher organic compound degradation rate constants,
suggesting that most of the bioavailable matter had been
already decomposed in these particles. These observations
show that in both sets of particles, silica dissolution was far
from being complete during the incubation time. An
alternative explanation is that biogenic silica does not
protect much of the OC in these samples.
Finally, we examined whether the variability within
aliquots might be responsible for the absence of significant
degradation patterns of organic compounds in NT2
fractions A and B. A maximum of about 20% variation
in bulk organic compound concentrations between duplicate samples at 48 h incubation time was observed in NT2A and NT2-B (for TLip-C and THAA-C, respectively),
suggesting that variability could have dampened degradation patterns but was not responsible for the absence of
significant degradation of all compounds in NT2-A and
NT2-B.
Nevertheless, the experiment did not show clear relationships between particle settling velocity, silica content,
dissolution rate constants and organic matter decomposition rate constants, probably because of the heterogeneity
of particulate matter in addition to experimental variability
and short incubation time. As mentioned earlier, another
explanation is that biogenic silica may not protect the OC
in our particle samples.
Evolution of composition during settling—Although previous studies report on the role of bacterial decomposition
of particles by bacteria attached to settling or suspended
particles at DYFAMED (Sempéré et al. 2000; Van
Wambeke et al. 2001; Ghiglione et al. in press), the
MedFlux experiment is the first decomposition study using
natural settling marine particles separated by settling
velocity. As the particles with different settling velocity
decomposed, their compositions changed with time, yet
those compositional changes were related to initial com-
Degradation of marine particles
position. Among the tracers used for characterizing OM
composition, changes in individual and total amino acids,
pigments, and carbohydrates showed the combined results
of assimilation and bacterial biomass production, whereas
changes in total lipid classes better traced early stages of
degradation (hydrolysis), as reported in Goutx et al. (2003).
We observed that the same lipid class in two different
particle fractions had different degradation rate constants
(Table 5). It is probable that the heterogeneity of particles,
the localization of the organic matrix within diatom
frustules (when present), the BSiO2 dissolution rate
constants, the resistance of peritrophic fecal pellet membranes depending on their origin, and/or the presence of
mucus all led to differences in susceptibility of the various
carbon compounds to bacterial attack. This was most
noticeable in NT1, which had more diatom material in it
and exhibited less BSiO2 dissolution over time than NT2.
Although organic compounds were initially present in
different proportions in the four settling velocity particle
classes, degradation led to an overall accumulation of lipid
metabolites, specific amino acids (GABA and BALA),
pigment (phytin), and carbohydrates (deoxysugars). As
illustrated by PCA in Fig. 6, the composition of all four
classes moved in the same direction, with C and D ending
up after 120 h with almost identical compositions. Thus,
regardless of initial particle composition, degradation
would lead to a homogenization in chemical composition
of resolvable organic compounds, which reflects the processes of loss through enzymatic hydrolysis of source
compounds and input of bacterial biomass.
Implications for pelagic organisms—The particle spectra
collected using the NetTrap at 200-m ranges from very
rapidly settling particles (.250 m d21) to slower settling
particles (,50 m d21). Thus, chemical changes and
degradation rate constants observed during the 5-d incubation period reflected the evolution and ‘‘degradation’’
of particles produced in the upper layer as they sink 250–
1,250 m through the water column, below the twilight zone.
All particle classes contained recently biosynthesized
organic compounds.
Organic compounds associated with particles settling at
velocities above 100 m d21 had slower turnover times,
which resulted in little degradation during the incubation
period. These faster settling particles thus are excellent
vehicles for transferring phytoplankton production out of
the mesopelagic layer. Coupling settling velocity and results
of apparent degradation rate constants shows that in low
hydrodynamic systems, disruption/degradation of these
faster settling particles is likely to occur in the bathypelagic
layer and surface sediment where they become bioavailable.
Bacterial communities from deep layers are more likely to
colonize these particles than communities from upper
layers. Surface bacteria attached to particles are grazed
by heterotrophic flagellates during sinking throughout the
mesopelagic layer (Tanaka and Rassoulzadegan 2004),
whereas bathypelagic bacteria are well adapted to pressure
conditions at deep depth (Tholosan et al. 1999; Tamburini
et al. 2002). In these particles however, a few individual
biomarkers exhibited significant apparent degradation rate
1661
constants (data not shown), suggesting that some molecules
carried by the faster settling particles were accessible inside
or outside of aggregates to bacterial attack during transport from surface to depth.
On the other hand, slower settling particles (,100 m
d21) exhibited significant degradation during incubation.
We calculated average degradation rate constants for the
amino acid and lipid pools as well as for the individual
compounds that degraded rapidly (Table 7). These degradation rate constants (0.08 6 0.01 to 0.24 6 0.11 d21) fall
within the range (at the lower limit) of values reported by
Pantoja et al. (2004) and Van Mooy et al. (2002) for settling
matter in oxic conditions. These degradation rate constants
(k) can be used to estimate consumption depth of sediment
trap material collected during the MedFlux cruise. The
turnover time (1/k) of organic compounds in these particles
(4.2–12.5 d), which is the time required for the initial pool
to decrease by a factor of 2.7, indicates that they are mainly
consumed in the mesopelagic layer. Bacterial biomarkers
were accumulated in these particles, suggesting that they
are colonized by bacterial communities, either those from
the upper water column developing in the vicinity of the
phytoplankton or those colonizing particles during settling.
Filter feeding and/or migratory organisms eating smaller
particles would benefit nutritionally from attached bacteria
on these slower settling particles in the mesopelagic layer.
These observations stress the role of attached bacteria in
the consumption of particulate organic carbon in the
northwest Mediterranean Sea. Differences in the dynamics
of faster settling particles and slower settling particles
degradation may have important consequences for evaluating carbon flows through the mesopelagic food chain. A
future goal will be to obtain insight into the species
composition and metabolic characteristics of bacterial
communities attached to these particles. It will also be
important to ascertain whether degradation rate constants
depend on the relative proportions of slower settling and
faster settling particles within a sample.
References
ANDERSEN, V., AND L. PRIEUR. 2000. One-month study in the open
NW Mediterranean Sea (DYNAPROC experiment, May
1995): Overview of hydrobiogeochemical structures and
effects of wind events. Deep-Sea Res. I 47: 397–422.
ANTIA, A. N., AND oTHERS. 2001. Basin-wide particulate carbon
flux in the Atlantic Ocean: Regional export patterns and
potential for atmospheric CO2 sequestration. Glob. Biogeochem. Cycles 15: 845–862.
ARMSTRONG, R. A., C. LEE, J. I. HEDGES, S. HONJO, AND S. G.
WAKEHAM. 2002. A new, mechanistic model for organic
carbon fluxes in the ocean based on the quantitative
association of POC with ballast minerals. Deep-Sea Res. II
49: 219–236.
ARNARSON, T. S., AND R. G. KEIL. 2000. Influence of organicmineral aggregates on microbial degradation of the dinoflagellate Scrippsiella trochoidea. Geochim. Cosmochim. Acta
69: 2111–2117.
BACON, M. P., AND R. F. ANDERSON. 1982. Distribution of
thorium isotopes between dissolved and particulate forms in
the deep sea. J. Geophys. Res. 87: 2045–2056.
1662
Goutx et al.
BARTOLI, G., C. MIGNON, AND R. LOSONO. 2005. Atmospheric
input of dissolved inorganic phosphorus and silicon to the
coastal northwestern Mediterranean Sea: Fluxes, variability
and possible impact on phytoplankton dynamics. Deep-Sea
Res. I 52: 2005–2016.
BIDIGARE, R. R., M. C. KENNICUTT, AND J. M. BROOKES. 1985.
Rapid determination of chlorophylls and their degradation
products by high performance liquid chromatography.
Limnol. Oceanogr. 30: 432–435.
———, AND M. E. ONDRUSEK. 1996. Spatial and temporal
variability of phytoplankton pigment distributions in the
central equatorial Pacific Ocean. Deep-Sea Res. II 43:
809–834.
BIDLE, K. D., AND F. AZAM. 1999. Accelerated dissolution of
diatom silica by marine bacterial assemblages. Nature 397:
508–512.
———, AND ———. 2001. Bacterial control of silica regeneration from diatom detritus: Significance of bacterial ectohydrolases and species identity. Limnol. Oceanogr. 46:
1606–1623.
BIERSMITH, R., AND R. BENNER. 1998. Carbohydrates in phytoplankton and freshly produced dissolved organic matter. Mar.
Chem. 63: 131–144.
BLIGH, E. G., AND W. J. DYER. 1959. A rapid method of total lipid
extraction and purification. Can. J. Biochem. Physiol. 37:
911–917.
BOYD, P. P. W., AND T. W. TRULL. 2006. Understanding the export
of biogenic particles in oceanic waters: Is there consensus?
Prog. Oceanogr. doi:10.1016/J.pocean.2006.10.007
BRZEZINSKI, M. A. 1985. The Si : C : N ratio of marine diatoms:
Interspecific variability and the effect of some environmental
variables. J. Phycol. 21: 347–357.
BURNEY, C. M., AND J. M. c. N. SIEBURTH. 1977. Dissolved carbohydrates in seawater. II. A spectrometric procedure for total
carbohydrate analysis and polysaccharide estimation. Mar.
Chem. 5: 15–28.
CARROLL, M. L., J. C. MIQUEL, AND S. W. FOWLER. 1998. Seasonal
patterns and depth-specific trends of zooplankton fecal pellets
fluxes in the Northwest Mediterranean Sea. Deep-Sea Res. I
45: 1303–1318.
CLAQUIN, P., V. MARTIN-JÉZÉQUEL, J. C. KROMKAMP, M. J. W.
VELDHUIS, AND G. W. KRAAY. 2002. Uncoupling of silicon
compared to carbon and nitrogen metabolism, and role of the
cell cycle, in continuous cultures of Thalassiosira pseudonana
(bacillariophyceae) under light, nitrogen and phosphorus
control. J. Phycol. 38: 922–930.
COPIN-MONTÉGUT, C., AND M. BÉGOVIC. 2002. Distributions of
carbonate properties and oxygen along the water column (0–
2000 m) in the central part of the NW Mediterranean Sea
(Dyfamed site): Influence of winter vertical mixing on air-sea
CO2 and O2 exchanges. Deep-Sea Res. II 49: 2049–2066.
COWIE, G. L., AND J. I. HEDGES. 1996. Digestion and alteration of
the biochemical constituents of a diatom (Thalassiosira
weissflogii) ingested by an herbivorous copepod (Calanus
pacificus). Limnol. Oceanogr. 41: 581–594.
DAUWE, B., AND J. J. MIDDELBURG. 1998. Amino acids and
hexosamines as indicators of organic matter degradation
state in North Sea sediments. Limnol. Oceanogr. 43: 782–
798.
———, ———, P. M. J. HERMAN, AND C. H. R. HEIP. 1999.
Linking diagenetic alteration of amino acids and bulk organic
matter reactivity. Limnol. Oceanogr. 44: 1809–1814.
GÉRIN, C., AND M. GOUTX. 1993. Separation and quantitation of
phospholipids in marine bacteria by Iatroscan TLC/FID
analysis. J. Planar Chromatogr. 6: 307–312.
GHIGLIONE, J. F., G. MEVEL, M. PUJO-PAY, P. LEBARON, AND M.
GOUTX. In press. Diel and seasonal variations in abundance,
production and community structure of particle-attached and
free-living bacteria in offshore NW Mediterranean water
column. Microb. Ecol. doi:10.1007/s00248-006-9189-7.
GORDON, L. I., J. C. JENNINGS, A. A. ROSS, AND J. M. KREST. 1993.
A suggested protocol for continuous flow automated analysis
of seawater nutrients in the WOCE hydrography program
and the Joint Global Ocean Fluxes Study. College of Oceanic
and Atmospheric Sciences, Corvalis, Oregon. Technical Report 93-1.
GOUTX, M., C. GUIGUE, AND L. STRIBY. 2003. Triacylglycerol
biodegradation experiment in marine environmental conditions: Definition of a new lipolysis index. Org. Geochem. 34:
1465–1473.
———, A. MOMZIKOFF, L. STRIBY, V. ANDERSEN, J. C. MARTY,
AND I. VESCOVALI. 2000. High-frequency fluxes of labile
compounds in the central Ligurian Sea, northwestern
Mediterranean. Deep-Sea Res. I 47: 533–556.
GREENWOOD, J., V. W. TRUESDALE, AND A. R. RENDELL. 2001.
Biogenic silica dissolution in seawater—in vitro chemical
kinetics. Prog. Oceanogr. 48: 1–23.
HARVEY, H. R., AND S. A. MACkO. 1997. Kinetics of phytoplankton decay during simulated sedimentation: Changes in lipids
under oxic and anoxic conditions. Org. Geochem. 27:
129–140.
———, J. H. TUTTLE, AND J. T. BELL. 1995. Kinetics of
phytoplankton decay during simulated sedimentation:
Changes in biochemical composition and microbial activity
under oxic and anoxic conditions. Geochim. Cosmochim.
Acta 59: 3367–3377.
HEDGES, J. I., J. A. BALDOCK, Y. GÉLINAS, C. LEE, M. PETERSON,
AND S. G. WAKEHAM. 2001. Evidence for non-selective
preservation of organic matter in settling marine particles.
Nature 409: 801–804.
HONJO, S., S. J. MANGANINI, AND J. J. COLE. 1982. Sedimentation
of biogenic matter in the deep ocean. Deep-Sea Res. 29:
609–625.
INGALLS, A., C. LEE, S. G. WAKEHAM, AND J. I. HEDGES. 2003. The
role of biominerals in the settling flux and preservation of
amino acids in the Southern Ocean along 170uW. Deep-Sea
Res. II 50: 713–738.
ITTEKKOT, V., AND B. HAAKE. 1990. The terrestrial link in the
removal of organic carbon, 319–325. In V. Ittekkot, S.
Kempe, W. Michaelis and A. Spitzy [eds.], Facets of modern
biogeochemistry. Springer.
JACOBSON, T. R., AND F. AZAM. 1984. Role of bacteria in fecal
pellet decomposition: Colonization, growth rates and remineralization. Bull. Mar. Biol. 35: 495–502.
KANEDA, T. 1991. Iso and anteiso-fatty acids in bacteria:
Biosynthesis, function, and taxonomic significance. Microbiol. Rev. 55: 288–302.
KARL, D. M., G. A. KNAUER, AND J. H. MARTIN. 1988. Downward
flux of particulate organic matter in the ocean: A particle
decomposition paradox. Nature 332: 438–441.
KING, L. 1993. Chlorophyll diagenesis in the water column and
sediments of the Black Sea. Ph.D. thesis, W.H.O.I./M.I.T.
Joint Program.
KLAAS, C., AND D. E. ARCHER. 2002. Association of sinking
organic matter with various types of mineral ballast in the
deep sea: Implications for the rain ratio. Glob. Biogeochem.
Cycl. 16: 1116, doi: 10.1029/2001GB001765.
LEE, C., AND C. CRONIN. 1982. The vertical flux of particulate
organic nitrogen in the sea: Decomposition of amino acids in
the Peru upwelling area and the equatorial Atlantic. J. Mar.
Res. 41: 227–251.
Degradation of marine particles
———, AND ———. 1984. Particulate amino acids in the sea:
Effects of primary productivity and biological decomposition.
J. Mar. Res. 42: 1075–1097.
———, S. G. WAKEHAM, AND C. ARNOSTI. 2004. Particulate
organic matter in the sea: The composition conundrum.
Ambio 33: 565–575.
———, ———, AND J. I. HEDGES. 2000. Composition and flux of
particulate amino acids and chloropigments in Equatorial
Pacific seawater and sediments. Deep-Sea Res. I 47:
1535–1568.
LINDROTH, P., AND K. MOPPER. 1979. High performance liquid
chromatographic determination of subpicomolar amounts of
amino acids by precolumn fluorescence derivatization with
ophthaldialdehyde. Anal. Chem. 51: 1667–1674.
LIU, Z., AND oTHERS. 2005. Why do POC concentrations measured
using Niskin bottle collections sometimes differ from those
using in-situ pumps? Deep-Sea Res. I 52: 1324–1344.
MANTOURA, R. F. C., AND C. A. LLEWELLYN. 1983. The rapid
determination of algal chlorophyll and carotenoid pigments
and their breakdown products in natural waters by reverse
phase high performance liquid chromatography Anal. Chim.
Acta 151: 297–314.
MARTY, J.-C. 2002. The DYFAMED time-series program (French
JGOFS). Deep-Sea Res. II 49: 1963–1964.
———, J. CHIAVERINI, AND M. D. PIZAY. 2002. Seasonal and
interannual dynamics of nutrients and phytoplankton pigments in the western Mediterranean sea at the DYFAMED
time-series station (1991–1999). Deep-Sea Res. 49: 1965–1985.
MEGLEN, R. R. 1992. Examining large databases: A chemometric
approach using principle components analysis Mar. Chem.
39: 217–237.
MILLER, R. G. 1974. The jackknife—a review. Biometrika 61:
1–15.
MOPPER, K., C. SCHULTZ, L. CHEVOLOT, C. GERMAIN, R.
REVUELTA, AND R. DAWSON. 1992. Determination of sugars
in unconcentrated seawater and other natural waters by liquid
chromatography. Environ. Sci. Technol. 26: 133–137.
MORICEAU, B., M. GARVEY, O. RAGUENEAU, AND U. PASSOW. 2007.
Evidence for reduced biogenic silica dissolution rates in
diatom aggregates. Mar. Ecol. Prog. Ser. 333: 129–142.
NIVAL, P., S. NIVAL, AND A. THIRIOT. 1975. Influences des
conditions hivernales sur les productions phyto- et zooplanctoniques en Méditerranée nord-occidentale. V. Biomasse et
production zooplanctonique—Relations phyto-zooplancton.
Mar. Biol. 31: 249–270.
OPSAHL, S., AND R. BENNER. 1999. Characterization of carbohydrates during early diagenesis of five vascular plant tissues.
Org. Geochem. 30: 83–94.
PANAGIOTOPOULOS, C., AND R. SEMPÉRÉ. 2005. Molecular distribution of carbohydrates in large marine particles. Mar. Chem.
95: 31–49.
———, ———, R. LAFONT, AND P. KERHERVÉ. 2001. Effects of
temperature in the determination of monosaccharides by
using an anion-exchange column. J. Chromatogr. 920: 13–
22.
———, AND oTHERS. 2002. Bacterial degradation of large organic
particles in the Southern Indian Ocean using in vitro
degradation experiments. Org. Geochem. 33: 985–1000.
PANTOJA, S., J. SEPULVEDA, AND H. E. GONZALES. 2004. Decomposition of sinking proteinaceous material during fall in
the oxygen minimum zone off northern Chile. Deep-Sea Res. I
51: 51–70.
PASSOW, U., A. ALLDREDGE, AND B. E. LOGAN. 1994. The role of
particulate carbohydrate exudates in the flocculation of
diatom blooms. Deep-Sea Res. I 41: 335–357.
1663
———, A. ENGEL, AND H. PLOUGH. 2003. The role of aggregation
for the dissolution of diatom frustules. FEMS Microbiol.
Ecol. 46: 247–255.
PETERSON, M. L., S. G. WAKEHAM, C. LEE, M. A. ASKEA, AND J. C.
MIQUEL. 2005. Novel techniques for collection of sinking
particles in the ocean and determining their settling rates.
Limnol. Oceanogr. Methods 3: 520–532.
RAGUENEAU, O., N. DITTERT, P. PONDAVEN, P. TRÉGUER, AND L.
CORRIN. 2002. Si/C decoupling in the world ocean: Is the
Southern Ocean different? Deep-Sea Res. II 49: 3127–3154.
———, AND P. TRÉGUER. 1994. Determination of biogenic silica in
coastal waters: Applicability and limits of the alkaline
digestion method. Mar. Chem. 45: 43–51.
SCHULTES, S. 2004. The role of mesozooplankton grazing in the
biogeochemical cycle of silicon in the Southern Ocean.
Dissertation, Universität Bremen.
SEMPÉRÉ, R., S. C. YORO, F. VAN WAMBEKE, AND B. CHARRIÉRE.
2000. Microbial decomposition of large particles in northwestern Mediterranean Sea. Mar. Ecol. Prog. Ser. 198: 61–72.
SHERIDAN, C. C., C. LEE, S. G. WAKEHAM, AND J. K. B. BISHOP.
2002. Suspended particle organic composition and cycling in
surface and midwaters of the equatorial Pacific Ocean. DeepSea Res. I 49: 1983–2008.
SOKAL, R. R., AND F. J. ROHLF. 1981. Biometry. Freeman.
STRIBY, L., R. LAFONT, AND M. GOUTX. 1999. Improvement in the
Iatroscan thin-layer chromatography-flame ionisation detection analysis of marine lipids. Separation and quantification of mono- and diacylglycerols in standards and natural
samples. J. Chromatogr. 849: 371–380.
TAMBURINI, C., J. GARCIN, M. RAGOT, AND A. BIANCHI. 2002.
Biopolymer hydrolysis and bacterial production under
ambient hydrostatic pressure through a 2000 m water
column in the NW Mediterranean. Deep-Sea Res. II 49:
2109–2123.
———, ———, G. GRÉGORI, K. LEBLANC, P. RIMMELIN, AND D.
L. KIRCHMAN. 2006. Pressure effects on surface Mediterranean prokaryotes and biogenic silica dissolution during
a diatom sinking experiment. Aquat. Microb. Ecol. 43:
267–276.
TANAKA, T., AND RASSOULZADEGAN, F. 2004. Vertical and seasonal
variations of bacterial abundance and production in the
mesopelagic layer of the NW Mediterranean Sea: Bottom-up
and top-down controls. Deep-Sea Res. I 51: 531–544.
TANDE, K. S., AND D. SLAGSTAD. 1985. Assimilation efficiency in
herbivorous aquatic organisms: The potential of the ratio
methods using 14C and biogenic silica as markers. Limnol.
Oceanogr. 30: 1093–1099.
TER BRAAK, C. J. F. 1994. Canonical community ordination. Part
I: Basic theory and linear methods. Ecoscience 1: 127–140.
THOLOSAN, O., J. GARCIN, AND A. BIANCHI. 1999. Effects of
hydrostatic pressure on microbial activity through a 2000 m
deep water column in the NW Mediterranean. Mar. Ecol.
Prog. Ser. 168: 273–283.
TRÉGUER, P., AND P. LE CORRE. 1975. Manuel d’analyse des sels
nutritifs dans l’eau de mer: utilisation de l’auto-analyseur
Technicon II., Université de Bretagne Occidentale, Brest.
110 p.
VAN MOOY, B. A. S., R. G. KEIL, AND A. H. DEVOL. 2002. Impact
of suboxia on sinking particulate organic carbon: Enhanced
carbon flux and preferential degradation of amino acids via
denitrification. Geochim. Cosmochim. Acta 66: 457–465.
VAN WAMBEKE, F., M. GOUTX, L. STRIBY, R. SEMPÉRÉ, AND F.
VIDUSSI. 2001. Bacterial dynamics during the transition from
spring bloom to oligotrophy in the Northwestern Mediterranean Sea. Relationships with particulate detritus and dissolved organic matter. Mar. Ecol. Prog. Ser. 212: 89–105.
1664
Goutx et al.
WAKEHAM, S. G. 1982. Organic matter from a sediment trap
experiment in the equatorial North Atlantic: Wax esters,
steryl esters, triacylglycerols and alkyldiacylglycerols. Geochim. Cosmochim. Acta 46: 2239–2257.
———, AND E. A. CANUEL. 1988. Organic geochemistry of
particulate matter in the eastern tropical North Pacific Ocean:
Implications for particle dynamics. J. Mar. Res. 46: 183–213.
———, J. I. HEDGES, C. LEE, M. L. PETERSON, AND P. J. HERNES.
1997a. Compositions and transport of lipid biomarkers
through the water column and surficial sediments of the
equatorial Pacific Ocean. Deep-Sea Res. II 44: 2131–2162.
———, AND C. LEE. 1993. Production, transport, and alteration
of particulate organic matter in the marine water column, p.
145–169. In M. Engel and S. Macko [eds.], Organic geochemistry. Plenum Press.
———, ———, J. I. HEDGES, P. J. HERNES, AND M. L. PETERSON.
1997b. Molecular indicators of diagenetic status in marine
organic matter. Geochim. Cosmochim. Acta 61: 5363–5369.
Received: 30 December 2005
Amended: 26 January 2007
Accepted: 12 February 2007