Element Fluxes and Cycling in Marine Particles During VERTIGO 2004: Preliminary Results

Element Fluxes and Cycling in Marine Particles During VERTIGO 2004: Preliminary Results

near Station ALOHA

Thanks to: Captain and crew of R/V Kilo Moana, Science Party of KM0414, Lary Ball, Suilou Huang, Mak Saito, Matt Charrette, Meagan Gonneea, Brian Guest, NSF-OCE, WHOI Post-doctoral Scholar Program (esp. John Farrington, Janet Fields and Christine Charrette), Penzance Endowed Discretionary Fund.

Carl Lamborg, Ken Buesseler, Jim Valdes, Tom Trull, Jim Bishop, Karen Casciotti, Steve Pike, John Andrews, Steve Manganini, Chanda Bertrand and Dave Schneider WHOI ICPMS Facility

•Introduction to VERTIGO•Our Interests in Element Fluxes and Cycling•VERTIGO 2004 (North Central Pacific Ocean)

•Methods•Fluxes

•How Accurate? Solubilization of Poisoned Trap Materials

•Remineralization. Unpoisoned Incubations

•Remineralization Time and Space Length Scales…Factors Affecting Flux Attenuation

•Summary/Implications

At the heart of VERTIGO is the question:What controls the efficiency of particle transport between the

surface and deep ocean?•There is an obvious mismatch between spatial patterns in primary production and the export of carbon to the deep ocean that indicates a complex suite of transformations must occur in the “twilight zone”, the region below the surface euphotic zone and the deep ocean.

C flux to seafloor -benthic O2 demand. Jahnke, 1996

C uptake in surface ocean-SeaWiFS global primary production. Behrenfeld & Falkowski, 1997

Sinking Marine Particles –detritus from the euphotic zone and lithogenic material

Swimmers

Conventionally studied using shallow “sediment” traps, at the base of euphotic zone (100-150 m)

The VERTIGO P.I.’s have identified two general mechanistic possibilities:

1.particle source characteristics are the dominant control on the efficiency of particle transport.

2.mid-water processing, either by zooplankton or bacteria, controls transport efficiency.

and/or

The Twilight Zone is Undersampled!

A compilation of all particle flux vs. depth data available from the last decade of JGOFS studies (redrawn from Berelson, 2001; not all shallow data plotted). Solid line is Martin curve for POC with b=-0.858.

This is likely to be where a lot of the action is…but few data.

Our Interests in Element Fluxes and Cycling•Fluxes for many elements other than C, N and P are not well known…especially in the top 500m.

e.g., Fe flux data lacking…fundamental information for understanding micronutrient cycling, and efficacy of Fe fertilization…

•Multi-element determinations may provide multiple tracers to constraining rates of particle remineralization, variable sinking rates (particle dispersion), and sources.

biomineral and crustal elements in particular, in conjunction with other fluxes…

•Additional tests on Th-derived approach to calculating POC fluxes.

•e.g., will (Fe/234Th)part when combined with 234Th deficits match collections in sediment traps?

Site Lat/Lon Chl avg (mg m-3)

Shallow POC Flux (g m-2 y-1)

f-ratio Deep POC flux (g m-2 y-1)

bSi/Cinorg deep trap (mole)

Dust Flux(g m-2 y-1)

ALOHA 22.75º N158º W 0.1 17-22 0.15 0.51 <1 (est.) ~0.5

K2 47º N160º E 0.5 86 0.42 0.82 4.2 ~10

Time Series sites provide context, data.

Gadgets!CLAP Trap-Buesseler

NBST-ValdesMULVFS-Bishop

Splitter

MOCNESS-UNOLS/SteinbergIRS Trap-Trull

2 tubes poisoned w/ HgCl2

2 tubes preserved w/formalin

1 tube covered

blank

1 tube preserved w/formalin

biological i.d. work

“Clap” Trapssurface tethered, drogued arrays

screen, combine

, split, filter

filters dried, stored frozen

1 Ag (d#1) or 1 nucl. (d#2)

filtrate

Consists of 10 tubes on each drifting array.Deployed at 3 depths: 150, 300 and 500m

2 tube Trull’s Gels

screen, combine

, split, filter


1 Ag (d#1) or 1 nucl. (d#2)

filtrate

screen, combine

, split, filter


1 Ag (d#1) or 1 nucl. (d#2)

filtrate

2 tubes aging expts. 1st HgCl2;

2nd formalin

NBSTs3 deployed @ 150 m, 2 @ 300 m and 2 @ 500m.Consists of 5 tubes on each array.

Time0.0 0.5 1.0 1.5 2.0

Mas

s in

Sed

. Tra

p

0.0

0.2

0.4

0.6

0.8

1.0

)1(' ktekt

FF

From incubation experiments, one acquires the value of k, the remineralization rate constant

Accuracy Issue: Solubilization of material in the sediment traps lowers estimate of flux.•some also believe “interstitial fluids” are an important component of flux…becoming enriched during sinking. squeezed out during processing (e.g., Avan Antia).

slope=F

slope=F’

VERTIGO Poisoned Particulate Solubilization Experiments1. brine/particle mixture collected from 2 combined sediment tubes designated for incubation experiments. experiments conducted with samples from 150, 300 and 500m. poisons change with deployment.

2. mixture split 8 ways on splitter, into 500 mL bottles.

4. at pre-determined times, incubation bottle filtered on 0.2 µm PC, with liquid subsamples for ICPMS and DOC. half PC filter washed onto Ag filter for PC/PN/PP/bSi. filters dried and frozen.

t = 3d 5d 30d1d

3. splits combined. samples incubated at in-situ temperature

Ca/Al

1 day3 days5 days1 day3 days5 days1 day3 days5 days1 day3 days5 days1 day3 days5 days1 day3 days5 days

mas

s ra

tio

0

10

20

30

40

50

150m

300m

500m

1st Dep. - Hg 2nd Dep. - Formalin

Fe/Al


mas

s ra

tio

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

150m

300m 500m


P/Al


mas

s ra

tio

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

150m

300m

500m


DOC in overlying brine


DO

C (µ

M)

0

50

100

150

200

150m

300m500m


Incubations of Poisoned Trap Material (and DOC from brine)

Remineralization k (d-1)

-0.2 0.0 0.2 0.4 0.6

Cor

rect

ion

Fact

or

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

3 days (150m)4 days (300m)5 days (500)

k = 0.1 d-1 suggests correction of 15-27%.In general, modest corrections needed from in-trap remineralization.

Poisoned Solubilization Rates

PC 1st Deployment

Flux (mg m-2 d-1)

0 10 20 30

Dep

th (m

)0

100

200

300

400

500

600

PC 2nd Deployment

Flux (mg m-2 d-1)

0 5 10 15 20 25 30 35D

epth

(m)

0

100

200

300

400

500

600

PN 1st Deployment

Flux (mg m-2 d-1)

0 1 2 3 4 5

Dep

th (m

)

0

100

200

300

400

500

600

PN 2nd Deployment

Flux (mg m-2 d-1)

0 1 2 3 4 5

Dep

th (m

)

0

100

200

300

400

500

600

HOTS Primary ProductionJune ’04: 531 mg C m-2 d-1

August ’04: 539 mg C m-2 d-1

f-ratio: 0.04

Deep Trap Limit(Karl, unpublished)

Though there’s variability in the absolute magnitude in the traps…the particles do not appear to be sorting very much…they index pretty well. The organic matter is decaying as one would expect. Makes up about 60% of attenuating mass.

The zero OM endmember is 4 g m-2 y-1…average dust flux is 0.5 g m-2 y-1.

red=150mgreen=300mblue=500mopen=NBSTclosed=CLAP

average dust Al fluxred=CLAPblue=NBST

Vertigo '04 - All Traps, All Depths

Ca/Al (mass/mass)0 20 40 60 80 100 120

Ba/

Al (

mas

s/m

ass)

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35Filled - STDAOpen - NBSTRed - 150 mGreen - 300 mBlue - 500 m

Vertigo '04 - All Traps, All Depths

Ca/Al (mass/mass)0 20 40 60 80 100 120

Sr/A

l (m

ass/

mas

s)

0

1

2

3

4

5

6

red=150mgreen=300mblue=500mopen=NBSTclosed=CLAP

Vertical Flux Behavior Elements

Decreasing with Depth Sr, P, Ca, Mg, Total Mass, Ni, Th, PC, PN

Constant with Depth Al, Sc, Fe, Cu

Increasing with Depth Ba, Mn

Hard to Tell Ti, V, Co, Zn, Cd, Pb

Summary of Flux Measurements

No statistically significant difference between poisons.Little difference between CLAP and NBST for most elements...suggests

little hydrodynamic particle sorting and/or systematic under/oversampling.

The Martin CurveParaphrasing Tom Trull from the VERTIGO website…it’s the Law, but is it a good Idea?

Does it fit the data any better than other functional forms, for instance that

proposed by Lutz et al. (2002):

If flux attenuation is completely due to in-situ remineralization, then degradation rates of unpoisoned material can be used to estimate sinking rates through comparison of remineralization length (time/space) scales…

define remineralization length scale as e-folding scale (time or depth over which flux changes by 1/e):

refrac

zWD

labilez FeFF

b

zzFF

100100

1

1

1

100

1001

1

0

112

1

2

1

2

1100

2100

bMartin

time

ktt

bdepth

b

b

b

b

ekzSk

L

eFF

ezzzL

ezz

zz

zF

zF

e

Martin

Martin

DkW

tzS

zWDkt

FeFFeF

Lutz

refrac

zWD

labilerefrackt

labile

The Martin Attenuation Curve is matched with exponential decay to zero, while Lutz Attenuation Curve is matched with exponential decay to a constant.

Lutz Curve matches the data better (esp. if you include deep trap data)…but Martin Curve appears to require increasing bulk sinking rates, which is consistent with increasing importance of ballasting biomineral phases.

Is Reality somewhere in between?

sinking rate accelerates with depth!

sinking rate is constant in relation to remineralization

t = 0 1d 2d

VERTIGO Unpoisoned Particulate Degradation Experiments

1. quarter filter washed into incubation bottle using 0.2 µm filtered 500m water

2. water/particle mixture occasionally agitated, incubated at depth appropriate temperature in dark.150m: 25 ºC; 300m: 15 ºC; 500m: 4 ºC

3. at particular times, incubation bottle agitated and aliquot removed by clean syringe. mixture is filtered on 0.2 µm PC, with liquid subsamples going for nitrification experiments (15NH4

+), ICPMS and DOC. filters rinsed and dried for ICPMS. half PC filter washed onto Ag filter for POC.

3d 5d

Ca/Al

T0 1

50T1

150

T2 1

50T3

150

T5 1

50T0

300

T1 3

00T2

300

T3 3

00T5

300

T0 5

00T1

500

T2 5

00T3

500

T5 5

00

ratio

0

5

10

15

20

25

Ba/AlT0

150

T1 1

50T2

150

T3 1

50T5

150

T0 3

00T1

300

T2 3

00T3

300

T5 3

00T0

500

T1 5

00T2

500

T3 5

00T5

500

ratio

0.0

0.1

0.2

0.3

0.4

P/AlT0

150

T1 1

50T2

150

T3 1

50T5

150

T0 3

00T1

300

T2 3

00T3

300

T5 3

00T0

500

T1 5

00T2

500

T3 5

00T5

500

ratio

0.0

0.2

0.4

0.6

0.8

1.0

Unpoisoned Remineralization Rates

More elements showed remineralization in these experiments than in the poisoned experiment, but not all…

Fe/Al

T0 1

50T1

150

T2 1

50T3

150

T5 1

50T0

300

T1 3

00T2

300

T3 3

00T5

300

T0 5

00T1

500

T2 5

00T3

500

T5 5

00

ratio

0.0

0.1

0.2

0.3

0.4

0.5

0.6

Remineralization of Unpoisoned MULVFS material

Ba

150

Ba

300

Ba

500

Mg

150

Mg

300

Mg

500

P150

P300

P500

Ca

150

Ca

300

Ca

500

Sc 1

50Sc

300

Sc 5

00

V150

V300

V500

Mn

150

Mn

300

Mn

500

Fe 1

50Fe

300

Fe 5

00

Co

150

Co

300

Co

500

Sr 1

50Sr

300

Sr 5

00

Rem

iner

aliz

atio

n k

(d-1

)

0.0

0.5

1.0

1.5

2.0

2.5

3.0

Element Apparent sinking Rate (m d-1)

Ba 114 - 233

Mg slow - 294

Al undefined

P 14 - 43

Ca 51- fast

Sc fast

Mn slow

Fe fast

Co Slow - 107

Ni undefined - slowSr 77 - 219

Sinking Rates for Comparison of Remineralization Rates and Attenuation Lengths (deployment 1)

For elements where attenuation and remineralization both seen, the analysis suggests sinking rates comparable to those observed directly…remineralization can explain the vertical profiles!

For the others, analysis suggests constrained fast or slow rates…either approach is breaking down or perhaps other factor important (particle type, grazing)

slow=attenuation in flux profile seen, but little remineralization on 5 day timescale.

fast=vertical profile doesn’t show much attenuation, but some remineralization seen.

Martin, Lutz give comparable results.

Summary/ImplicationsSediment traps were successfully used for trace level work for many elements.In the quiet conditions of the N. Pacific Gyre, NBSTs and tethered traps mostly compared well. Some differences, with tethered often higher.In-trap solubilization of most elements, as indicated by poisoned incubations, did not significantly alter flux estimates.

shallow traps, deployed for short times, generally not as affected as deep traps as suggested by Antia and others.

Though there appears to be something of a biological imprint on Fe cycling, Fe “remineralization length scales” are longer for Fe than C.

no catalytic effect can be expected for Fe additions in removing CO2.Sinking rates suggested by matching unpoisoned remineralization rates and vertical flux attenuation are, for many particle constituents, consistent with direct sinking rate measurements.

this suggests that much of the control on the flux of this material was through mid-water remineralization processes. microbial and physical/chemical processes likely dominant. particle source/type and grazing less important.

Documents

Element Fluxes and Cycling in Marine Particles During VERTIGO 2004: Preliminary Results