1. Ocean color remote sensing of phytoplankton physiology & primary production
Toby K. Westberry1, Mike J. Behrenfeld1  Emmanuel Boss2, David A. Siegel3
1Department of Botany, Oregon State University
2School of Marine Sciences, University of Maine
3Institute for Computational Earth System Science, UCSB

2. Outline 1. Introduction to problem - Phytoplankton Chl v. Carbon
- NPP modeling
2. Model
- bio-optics
- physiology
- photoacc./light limitation/nutrient stress
3. Results
- surface & depth patterns
- global patterns
4. Validation
5. Future directions

3. Carbon v. Chlophyll Chl is NOT biomass How to quantify phytoplankton
Historically, net primary production (NPP) has been modeled as a function of chlorophyll concentration
BUT, cellular chlorophyll content is highly variable and is affected by acclimation to light & nutrient stress and species composition
Chlorophyll a
Pros: uniquely algal trait; easily measured; detectable from space; large historical database; NPP ~ 0(chl)
Cons: chl synthesis depends on physiological state!
Carbon
Pros: -represents actual biomass; can be directly related to export flux
Cons: -difficult to measure
Chl is NOT biomass

4. Modeling NPP NPP ~ [biomass] x physiologic rate NPP ~ [Chl] x Pbopt
General
NPP ~ [Chl] x Pbopt
Chl-based
NPP ~ [C] x m
C-based
However, chl contains a mixture of biomass AND physiology, so that getting Pbopt is complex
Chl does NOT represent biomass, C does !
Scattering
(cp or bbp)
Ratio of Chl to
scattering (Chl:C)

5. Phytoplankton C Scattering covaries with particle abundance
(Stramski & Kiefer, 1991; Bishop, 1999; Babin et al., 2003)
Scattering also covaries with phytoplankton carbon
(Behrenfeld & Boss, 2003; Behrenfeld et al., 2005)
Chlorophyll variations independent of carbon (C) are an
index of changing cellular pigmentation

6. Scattering:Chl
From Behrenfeld & Boss (2003)

7. 1. Chl:C is consistent with lab dataNP SP SA NA CP SI NI CA SO L0 L1 L2 L3 L4
SO-all
Variance Level
Chlorophyll
excluded
28 Regional
Bins based on
seasonal Chl
variance
‘cell size domain?’
C = (bbp – intercept) x scalar
= (bbp – ) x 13,000
‘biomass domain’
bbp (m-1)
Intercept = m-1 based on Stramski & Kiefer (1991) & Cho & Azam (1990)
Phytoplankton Carbon = 25 – 35% POC
Eppley et al. (1992), DuRand et al. (2001), Gundersen et al. (2001), Obuelkheir et al. (2005)
1. Chl:C is consistent with lab data
Mean Chl:C=0.010, range=
(see synthesis in Behrenfeld et al. (2002))
2. C ~ 25-40% of POC
(Eppley et al. (1992); DuRand et al. (2001);
Gundersen et al. (2001), Obuelkheir et al. (2005),
Loisel et al., (2001), Stramski et al., (1999))
‘physiology domain’
Chlorophyll (mg m-3)

8. Chl:C registers physiology
Chl:C (mg mg-1)
Chl:C (mg mg-1)
Laboratory
Space
Light (moles photons m-2 h-1)
Chl:C
Chl:C
Low Nutrient stress High
Low Nutrient stress High
Growth rate (div. d-1)
Temperature (oC)
after Behrenfeld et al. (2005)

9. Model

10. CbPM overview Invert ocean color data to estimate [Chl a] & bbp(443)
(Garver & Siegel, 1997; Maritorena et al., 2001)
Relate bbp(443) to carbon biomass (mg C m-3)
(Behrenfeld et al., 2005)
Use Chl:C to infer physiology (photoacclimation & nutrient stress)
Propagate information through water column
Estimate phytoplankton growth rate (m) and NPP
Results
1. large spatial & temporal differences in carbon-based NPP from chl-based results (e.g., > ±50%)
2. seasonal cycles dampened in tropics, but strengthened and delayed in “spring bloom” areas
3. differences due to photoacclimation and nutrient-stress related changes in Chl : C
Carbon-based Production Model (CbPM)

11. CbPM details (1) 1. Let surface values of Chl:C
indicate level of nutrient-stress
-nutrient stress falls off
as e-Dz (Dz=distance from nitracline)
2. Let cells photoacclimate through
the water column
Chl : C
m (divisions d-1)
Ig (Ein m-2 h-1)

12. CbPM details (2) 3. Spectral accounting for underwater light field
-both irradiance & attenuation
4. Phytoplankton growth rate, m
5. Net primary production, NPP(z) = m(z) x C(z)
Chl : C
m (divisions d-1)
Ig (Ein m-2 h-1)
Max. growth rate
Nutrient limitation
(& temperature)
Light limitation

13. * red arrows indicate relationships exist ONLY when z>MLD
SeaWiFS
FNMOC
WOA01
INPUTS
nLw
Kd(490)
PAR(0+)
MLD
NO3
Maritorena et al. (2001)
Austin & Petzold (1986)
DNO3 > 0.5 mM
bbp
chl
Kd(l)
Ed(l)
zno3, Dzno3
Morel (1988) C
Chl:C
Photoacclimation
PAR(z)
DChl:Cnut
Light limitation
NPP m
OUTPUTS
* if z<MLD,
* red arrows indicate relationships exist ONLY when z>MLD
* Run with 1° x1° monthly mean climatologies ( )

14. Results

15. Example profiles (1) Stratified, shallow mixed layer, oligo- trophic
Sargasso Sea (35°N, 65°W, Aug)
Stratified, shallow
mixed layer, oligo-
trophic
MLD =25m
zNO3 =110m
zeu =105m
Uniform mixed layer
Subsurface Chl a max.
Realistic K(l), PAR, and euphotic depth

16. Example profiles (2) Deep mixed layer, nutrient replete MLD =95m
North Atlantic (50°N, 30°W, Apr)
Deep mixed layer,
nutrient replete
MLD =95m
zNO3 =0m
zeu =40m
Uniform mixed layer
Subsurface Chl a max.
Realistic K(l), PAR, and euphotic depth

17. Example profiles (mean)
Annual mean northern hemisphere
Chl m
NPP
Depth (m)
mg Chl m-3
d-1
mg C m-3 d-1
- c.f. Morel & Berthon (1989)

18. Surface patterns South Pacific (L0) (central gyre) Equatorial (L3)
Chl (mg Chl m-3)
C (mg C m-3)
Chl:C (mg mg-1)
South Pacific (L2)
(non-gyre)
North Atlantic (L3)
Month # since 1997

19. Growth rate, m Summer (Jun-Aug) Persistently elevated in upwelling
regions
Chronically depressed in open ocean
Can see effects of mixing depth &
micro-nutrient limitation
Winter (Dec-Feb)
Annual mean
Annual mean (L0 only)
m (d-1)
m (d-1)
m (d-1)

20. NPP patterns O(1) looks like Chl - gyres, upwelling, seasonal blooms
Summer (Jun-Aug)
O(1) looks like Chl
- gyres, upwelling,
seasonal blooms
Large seasonal cycle at
high latitudes (ex., N. Atl.)
Winter (Dec-Feb)
∫NPP (mg C m-2 d-1)

21. NPP patterns (2) large spatial (& temporal)
differences in carbon-based
NPP from chl-based results
(e.g., > ±50%)
differences due to photo-
acclimation and nutrient-stress
related changes in Chl : C
mg C m-2 d-1
differences due to photoacclimation and nutrient-stress manifest in the Chl : C

22. Seasonal NPP patterns (N. Atl.)
Western N. Atl
CBPM
VGPM
Eastern N. Atl

23. Seasonal NPP patterns seasonal cycles dampened in tropics,
CbPM
VGPM
seasonal cycles
dampened in tropics,
but strengthened and
delayed in “spring
bloom” areas

24. Annual NPP Although total NPP doesn’t change much (~15%),
∫NPP (Pg C)
VGPM
This model
Annual 45 52
Gyres
5 (11%)
13 (26%)
High latitudes
19 (42%)
12 (23%)
Subtropics?
18 (39%)
25 (48%)
Southern Ocean
(q<-50°S)
2 (4%)
3 (5%)
Add in comments about growth rate and ability to calc standing stocks,
Uncertainties, etc.
Although total NPP doesn’t change much (~15%),
where and when it occurs does

25. Validation

26. Surface Chl:C at HOT Prochlorococcus cellular HOT fluorescence at HOT
~(in situ Chl : C)
(Winn et al., 1995)
Satellite Chl :C
1998
1999
2000
2001
2002

27. Chl(z) & Kd(z) at BATS Model compared to Bermuda Atlantic Time-
series Study/Bermuda Bio- Optics Project (BATS/BBOP)
HPLC Chl & CTD fluorometer

28. ∫NPP at HOT & BATS
∫NPP (mg C m-2 d-1)

29. NPP(z) at HOT
NPP (mg C m-3 d-1)
Serial day since 09/1997

30. NPP(z) at HOT
- Uniform mixed layer (step function) v. in situ incubations
- Discrepancies due to satellite estimates, NOT concept

31. Future directions

32. Next steps (model) Sensitivity to inputs (e.g., MLD, MODIS)
Error budget
Inclusion of CDOM(z)
Change photoacclimation with depth
change bbp to C relationship
-diatoms, coccolithophorids, coastal
Further validation

33. Next steps (applications)
Look at finer spatial/temporal scales
Knowledge of m & dC/dt allow statements about loss
processes
Recycling efficiency (wrt nutrients)
Characterization of ocean in terms of nutrient and light
limitation patterns
Inclusion of concepts/data into coupled models

34. Thanks Princeton Jorge Sarmiento Patrick Shultz Mike Hiscock OSU UCSB
Norm Nelson
Stephane Maritorena
Manuela Lorenzi-Kayser
OSU
Robert O’Malley
Julie Arrington
Allen Milligen
Giorgio Dall’Olmo

35. Extra

36. Laboratory Chl:C physiology 3 primary factors Light Temperature
Nutrients
Chl:Cmax
Dunaliella tertiolecta
20 oC
Replete nutrients
Exponential growth phase
Geider (1987) New Phytol. 106: 1-34
16 species
= Diatoms
= all other species
Laws & Bannister (1980)
Limnol. Oceanogr. 25:
Thalassiosira fluviatilis
= NO3 limited cultures
= NH4 limited cultures
= PO4 limited cultures
Chl:C (mg mg-1)
Chl:Cmin
Light (moles m-2 h-1)
Laboratory
Chl:Cmax
Temperature (oC)
Chl:Cmin
Low Nutrient stress High
Growth rate (div. d-1)

37. Depth-resolved CBPM Uniform Nutrient-limited &/or light-limited
z=0
Uniform
z=MLD
Nutrient-limited &/or light-limited
+ photoacclimation
z=zNO3
Light-limited + photoacclimation
z=∞
Relative PAR
Relative NO3
* Iterative such that values at z=zi+1 depend on values at z=zi *

38. GSM01 (Maritorena et al., 2002)
Non-linear least squares problem with 3 unknowns and 5 equations
Solved by minimization of of squared sum of residuals (between obs & estimate)
Result is Chl, acdm(443), bbp(443)

39. The Model (con’t)

40. CBPM data sources INPUT (surface) OUTPUT ((z))
- SeaWiFS: nLw(l), PAR, Kd(490)
- GSM01: Chl a, bbp(443)
- FNMOC: MLD
- WOA 2001: ZNO3
- Chl, C, &
Chl:C
- m
- NPP
Run with 1° x1° monthly mean climatologies ( )

41. Example profiles (3) Deep winter mixing, Very low light,
Southern Ocean (50°S, 130°W, Aug)
Deep winter mixing,
Very low light,
Nutrient replete
MLD =>300m
zNO3 =0m
zeu =
Uniform mixed layer
Subsurface Chl a max.
Realistic K(l), PAR, and euphotic depth

42. Growth rate, m (2)
Annual mean
Annual mean (L0 only)
m (d-1)
m (d-1)

43. NPP patterns (Jun-Aug)
This work
∫NPP (mg C m-2 d-1)
large spatial & temporal
differences in carbon-based
NPP from Chl-based results
(e.g., > ±50%)
Chl-based model interprets high
Chl areas as high NPP
differences due to photo-
acclimation and nutrient-stress
related changes in Chl : C
VGPM (Chl-based model)
∫NPP (mg C m-2 d-1)

44. NPP patterns (2) large spatial & temporal differences in carbon-based
NPP from chl-based results
(e.g., > ±50%)
seasonal cycles dampened in
tropics, but strengthened and
delayed in “spring bloom”
areas
differences due to photo-
acclimation and nutrient-stress
related changes in Chl : C
mg C m-2 d-1
C-based
Chl-based
differences due to photoacclimation and nutrient-stress manifest in the Chl : C

45. Annual NPP Models are very sensitive to input sources VGPM CBPM
This model
Annual ∫NPP
(Pg C)
45 (61) 75 52
DMLD -- 18 8
DChl
8-10 ?? 4
DKd 26 37 29
OR SHOW BY OCEAN BASIN AND/OR SEASON TO SHOW REDISTRIBUTION??
D∫NPP for change
In input
Models are very sensitive to input sources

46. Conclusions Spectral, depth-resolved NPP model that includes
photoacclimation, light & nutrient limitation
- based on phytoplankton scattering-carbon relationship
Consistencies with field data  ongoing validation
Spatial patterns in ∫PP markedly different than Chl-based models
- also different seasonal cycles (timing/magnitude)