Measurements and Models of Primary Productivity Supported by NSERC 1 including OTN John J. Cullen! Department of Oceanography, Dalhousie University Halifax, Nova Scotia, Canada B3H 4R2! 2014 C-MORE Summer Training Course Microbial Oceanography: Genomes to Biomes Provided by the SeaWiFS Project, NASA/Goddard Space Flight Center and ORBIMAGE
Source of some reading http://www.ceotr.ca/publications/ MacIntyre and Cullen 2005 Parkhill et al. 2001 for reading lists, email! John.Cullen@Dal.CA 2 John Cullen: C-MORE 2013
What is marine primary productivity? Net Primary Productivity (Production)!! Net rate of synthesis of organic material from inorganic compounds such as CO 2 and water!! Chemosynthesis: chemical reducing power comes from reduced inorganic compounds such as H 2 S and NH 3!! Photosynthesis: reducing power comes from light energy!! Photosynthetic primary production is usually measured and considered to dominate.!! 3 g C m -3 h -1!!!g C m -2 d -1!
More to it than that, of course 4 Doney, S. C.,et al. (2004), Frontiers in Ecology and the Environment, 2(9), 457-466.
Biological oceanography and phytoplankton ecology Describe the causes and consequences of variations in primary productivity (and food web structure) John Cullen: C-MORE 2013
Why measure and model primary productivity? Ecological prediction! Biogeochemical models! Climate change scenarios 6 Behrenfeld et al. Nature 2006
Oxygenic Photosynthesis This process can be quantified directly by measuring the increase of oxygen, the decrease of CO 2, or the increase of organic carbon. Incubations can be conducted with 14 C, 13 C or H2 18 O added as tracers.!! The process has many chemical consequences that can be traced with measurements of the concentrations of oxygen and carbon, and their 7 isotopic signatures!
There are measurements and there are measurements Annual Review of Marine Science 2013
in vitro vs in situ
In situ estimates need calibration
Cullen, J.J., 2001. Plankton: Primary production methods. In J. Steele, S. Thorpe, K. Turekian (Eds.), Encyclopedia of Ocean Sciences (pp. 2277-2284): Academic Press.
Effects of Light on Photosynthesis Net Photosynthesis = Gross Photosynthesis - Respiration Respiration is measured directly only in the dark. The light-dependent component is much harder to measure ( 18 O-tracer can be used for that). 12 Biological Oceanography (4): John Cullen
Effects of Light on Photosynthesis Net Photosynthesis = Gross Photosynthesis - Respiration Net photosynthesis is negative when irradiance is below the compensation irradiance. 13 Biological Oceanography (4): John Cullen
Relating oxygen evolution to carbon assimilation with the photosynthetic quotient Generalized reactions for growth on nitrate and ammonium 1.0 NO 3 - + 5.7CO 2 + 5.4H 2 O! (C 5.7 H 9.8 O 2.3 N) + 8.5 O 2 +1.0 OH - P.Q. = 1.49 (O 2 evolved / CO 2 consumed) 1.0 NH + 4 + 5.7CO 2 + 3.4H 2 O! (C 5.7 H 9.8 O 2.3 N) + 6.25 O 2 +1.0 H + P.Q. = 1.10 Note that more photosynthesis is required for growth on nitrate because the nitrate must be reduced. 14 Biological Oceanography (4): John Cullen
Inevitable comparisons with satellite-based estimates (models) 15
What is behind this? 16
Still a benchmark: The 14 C method for measuring primary productivity 17 HOT website
The 14 C method for measuring primary productivity 0 Productivity (mgc m -3 d -1 ) 0 1 2 3 4 5 10 Depth (m) 20 30 18 40 50 ± s.e.
An incredibly useful tool for time series and process studies HOT website 19
Ideally, the 14 C method measures net primary productivity 0 Productivity (mgc m -3 d -1 ) 0 1 2 3 4 5 10 Depth (m) 20 30 40 50 20 ± s.e. 12-24 h incubations Sometimes, it does
The measurement is subject to artifacts and biases Depth (m) 0 10 20 30 Productivity (mgc m -3 d -1 ) 0 1 2 3 4 5 Hypothesis: uncertainties inherent in the 14 C method are being forgotten with important consequences for the estimation of primary productivity using other methods. 40 50 21 ± s.e. 12-24 h incubations Cullen, J.J., 2001. Plankton: Primary production methods. In J. Steele, S. Thorpe, K. Turekian (Eds.), Encyclopedia of Ocean Sciences (pp. 2277-2284): Academic Press.
The measurement is subject to artifacts and biases 0 10 Productivity (mgc m -3 d -1 ) 0 1 2 3 4 5 Overestimation due to exclusion of UV-B Depth (m) 20 30 40 50 22 ± s.e. 12-24 h incubations
The measurement is subject to artifacts and biases Productivity (mgc m -3 d -1 ) Depth (m) 0 10 20 30 0 1 2 3 4 5 Underestimation due to static incubation at excessive irradiance 40 ± s.e. 50 23 12-24 h incubations
The measurement is subject to artifacts and biases Productivity (mgc m -3 d -1 ) Depth (m) 0 10 20 30 0 1 2 3 4 5 Underestimation due to dilution of intracellular DIC with respired cellular C 40 ± s.e. 50 24 12-24 h incubations
The measurement is subject to artifacts and biases 0 Productivity (mgc m -3 d -1 ) 0 1 2 3 4 5 Depth (m) 10 20 30 40 50 25 ± s.e. 12-24 h incubations Overestimation due to unnatural accumulation of biomass (disruption or exclusion of grazers)
The measurement is subject to artifacts and biases 0 Productivity (mgc m -3 d -1 ) 0 1 2 3 4 5 Depth (m) 10 20 30 40 50 26 ± s.e. 12-24 h incubations Underestimation due to food-web cycling! (microbial respiration and excretion)
The measurement is subject to artifacts and biases 0 Productivity (mgc m -3 d -1 ) 0 1 2 3 4 5 10 Depth (m) 20 30 40 50 27 12-24 h incubations Overestimation due to inadequate time for respiration to be measured
and that s not all: 0 10 Productivity (mgc m -3 d -1 ) 0 1 2 3 4 5 Toxicity! Relief of iron limitation! Exposure to bright light! Disruption of fragile cells! Depth (m) 20 30 for Simulated in situ:! Poor match of irradiance! 40 Inappropriate temperature! 50 28 12-24 h incubations Possible diel bias
Productivity normalized to Chl is biased by: 0 10 Productivity (mgc m -3 d -1 ) 0 1 2 3 4 5 Changes in Chl during incubations! Depth (m) 20 30 40 Inadequate extraction of Chl by 90% acetone! Interference from Chl b (fluorometric acid-ratio method) 50 29 12-24 h incubations
and in general 0 Productivity (mgc m -3 d -1 ) 0 1 2 3 4 5 Depth (m) 10 20 30 Irradiance is not controlled! Respiration is not accurately measured 40 50 30 12-24 h incubations
The measurement has its limitations Depth (m) 0 10 20 30 40 50 Productivity (mgc m -3 d -1 ) 0 1 2 3 4 5 ± s.e. Net production?! Gross production?! Overestimate?! Underestimate? 31
A less skeptical view
But we use it routinely 0 Productivity (mgc m -3 d -1 ) 0 1 2 3 4 5 Depth (m) 10 20 30 40 50 ± s.e. Water column integral is something like! net primary production! (photosynthesis less losses to respiration) 33 (extent depends on temperature, light and maybe nutrients)
Regardless, it has served us very well Time Series 34 Karl et al. in Williams et al. 2002
Process studies 35
Solving mysteries 36
10 Equatorial Pacific 140 W, 0 ± 2 P B opt Measurements vs Behrenfeld et al. Models "Mystery Solved" 8 Real measurements from JGOFS EqPac P B opt (gc gchl-1 h -1 ) 6 4 VGPM Exponential Model 2 Revised according to insights from Behrenfeld et al. Nature 2006 0 10 15 20 25 30 Temperature ( C)
Direct Measurements will Never Provide Synoptic Estimates of Productivity 38 marine.rutgers.edu/opp/
Continuous measures of optics reveal rates see 39
Ocean chemistry tells the tale even in the most oligotrophic waters 40 3 years of O 2 at POT: Ken Johnson
Models are required for many applications Productivity from ocean color! Ecological prediction! Biogeochemical models! Climate change scenarios 41
Photosynthesis normalized to chlorophyll is the foundation of productivity modeling chl $ P max (1 exp (α chl E) ' chl % & P max ( ) )!##### "##### $ P chl vs. E relationship 42 Behrenfeld and Falkowski L&O 1997
One approach: model the measurements! Optical Depth 43 Behrenfeld and Falkowski 1997a L&O
Simplified functions describe major patterns 44 Behrenfeld and Falkowski 1997a L&O
Is the measured/modeled pattern real? P B (mg C mg Chl -1 d -1 ) 0 20 0 10 20 30 40 Measured nearsurface inhibition Depth (m) 40 60 80 100 45
Is the measured/modeled pattern real? P B (mg C mg Chl -1 d -1 ) 0 20 0 10 20 30 40 or an artifact of fixed-depth incubation? Depth (m) 40 60 80 100 46
Measurements must be made on a shorter time scale Photosynthetron: Controlled laboratory incubation 47 (Lewis and Smith 1983)
A comprehensive approach Cullen, J. J., M. R. Lewis, C. O. Davis, and R. T. Barber. 1992. Photosynthetic characteristics and estimated growth rates indicate grazing is the proximate control of primary production in the equatorial Pacific. Journal of Geophysical Research 97: 639-654. Approach introduced 48by Jitts, H. R., A. Morel, and Y. Saijo. 1976. The relation of oceanic primary production to available photosynthetic irradiance. Aust. J. Mar. Freshwater Res. 27: 441-454.
Short-term measurements can detect near surface inhibition 0 Chlorophyll a (mg m -3 ) 0 0.08 0.2 P B m. Chl (mg C m -3 h -1 ) 0 0 0.4 0.8 10 10 Depth (m) 20 30 Depth (m) 20 30 40 40 50 50 MORNING 49 Principles described by Marra, Harris, Yentsch -- all prior to 1980 0630 h: Chlorophyll and P max are nearly uniform in the 50-m mixed layer
Is measured near-surface inhibition real? Assessment with short-term P vs E 0 Chlorophyll a (mg m -3 ) 0 0.08 0.2 P B m. Chl (mg C m -3 h -1 ) 0 0 0.4 0.8 10 10 Depth (m) 20 30 Depth (m) 20 30 40 40 50 50 MIDDAY 50 1215 h: Chlorophyll and Potential Productivity are still nearly uniform in the 50-m mixed layer
Surface incubation led to artifact 0 Chlorophyll a (mg m -3 ) 0 0.08 0.2 P B m. Chl (mg C m -3 h -1 ) 0 0 0.4 0.8 10 10 Depth (m) 20 30 Depth (m) 20 30 40 40 50 50 INCUBATED 51 1530 h: Fixed-depth incubations at the surface are fried an artifact of sustained exposure 35% underestimation of ML productivity
Conventional 14 C: Near-surface inhibition is overestimated P B (mg C mg Chl -1 d -1 ) 0 20 0 10 20 30 40 Artifact Depth (m) 40 60 80 100 Error at the surface: big Error for integral: 8% 52
Consequences for a model: minor except for irradiance dependence of ΣP at higher daily irradiance Parameterization of an Artifact? 53 Behrenfeld and Falkowski Consumer s Guide: L&O 1997
Many models calculate P from measured relationships Depth (m) P (mg C m -3 d -1 ) 0 0.2 0.4 0.6 0.8 1 0 20 40 60 80 100 From!! P B vs E! and! E vs depth! and!! Chl! to! daily water column net primary productivity 54
Results can be generalized P B / P B opt 0 0.2 0.4 0.6 0.8 1 0 Z / Z 1%(PAR) 0.2 0.4 0.6 0.8 1 P Z = P B opt B f ( E (0) PAR K E ) PAR k 55
Maximum P B is still a key parameter for most models 0 P B (g C g Chl -1 h -1 ) 0 2 4 6 8 10 20 Z (m) 40 60 80 56 100 Even the spectral ones even the quantum-yield models
Compared to other measures, P B opt is relatively insensitive to artifact P B (mg C mg Chl -1 d -1 ) 0 0 10 20 30 40 20 Depth (m) 40 60 80 100 57
Global assessment of primary productivity requires global assessment of maximum P B 58 Behrenfeld and Falkowski 1997b L&O
P B opt modelled as a function of temperature 59 Behrenfeld and Falkowski 1997b L&O
Two commonly used functions 10 Maximum P B (g C g Chl -1 h -1 ) 8 6 4 2 Eppley - type (e.g., Antoine et al 1996) B & F VGPM 60 0 0 5 10 15 20 25 30 Temperature ( C) Behrenfeld and Falkowski 1997b L&O
You may have seen the figure 61 Behrenfeld and Falkowski 1997b L&O
Here are the measurements 40 Maximum P B (g C g Chl -1 h -1 ) 35 30 25 20 15 10 5 62 0 0 5 10 15 20 25 30 Temperature ( C) Rutgers OPP Study Data Base
Compared with more measurements 40 Maximum P B (g C g Chl -1 h -1 ) 35 30 25 20 15 10 5 63 0 0 5 10 15 20 25 30 Temperature ( C) Texas Shelf
and more measurements! Maximum P B (g C g Chl -1 h -1 ) 40 35 30 25 20 15 10 5 Eppley Type Model VGPM Rutgers OPP Data Texas Shelf & Slope HOT BATS EqPac Arabian Sea JGOFS Ross Sea JGOFS Antarctic PFZ JGOFS PBmax Texas (SL) 0 64 0 5 10 15 20 25 30 Temperature ( C)
The foundations of VGPM Published Statistical Fit Measurements (note scale) Behrenfeld and Falkowski 1997 L&O Underlying data + a few more data sets
General functions of T cannot capture major causes of variability in water column productivity 3 2 ln(measured/modeled) 1 0-1 -2 ± 2x error -3 66-5 0 5 10 15 20 25 30 35 Temperature ( C)
Conclusion: General functions of T cannot capture major causes of variability in water column productivity 67 Relative Error
We must appreciate the implications of this unexplained variability 68
and implications of a carbon-based model that isn t! Times Cited: 353 (from Google Scholar, May 2014) Carbon cancels out!
How much confidence should we have? Juranek and Quay 2013 Ann Rev Mar Sci
Estimates like these (NPP) do not yet account for variable physiology beyond central tendencies 71 Behrenfeld et al. Nature 2006
The need to compare models with measurements 72
Validation is important 73
Model 74
Headline 75
Published model result Measurements! show the! opposite trend 76
Time Frame & Statistics Matter Measurements 77
Measurements! (no headlines) 78
It is also useful to remember the distinction between models and observations VGPM model Follows et al. Model 79
No one is immune! Many models cannot be directly verified 0 80 Mixing Depth (m) 20 40 60 80 100 20, 40, 60 min No mixing 120 0.4 0.5 0.6 0.7 0.8 0.9 1.0 P T / P Tpot 8 h b 4 h 2 h
Conclusions Models of primary productivity are a fundamental requirement for describing and explaining ecosystem dynamics and biogeochemical cycling in the sea The models are based on measurements: The measurements are not perfect The models are not perfect Capabilities and limitations of models should always be considered when they are applied Effects of underlying assumptions Comparison with real measurements when available Know 81 your model!