optimality-based modelling: from theory to …...2015/03/04 · 3. current distribution of trait...
TRANSCRIPT
Marine Ecosystem Modelling Symposium, AORI Mar. 4, 2015S. Lan Smith
S. Lan SmithMarine Ecosystem Dynamics Research Group, RCGC, JAMSTEC, Yokohama, Japan
CREST, Japan Science and Technology Agency, Tokyo, Japan
Optimality-based Modelling: from theory to implementation
It’s all about combining Traits & Trade-offs (Smith et al. L&O Review 2011, Smith et al. JPR Horizons 2014)
Sunlight
Nutrient Supply
αVmax
μmax
chl
α
Vmax
μmax
chl
Marine Ecosystem Modelling Symposium, AORI Mar. 4, 2015S. Lan Smith p. 2
マラソン選手は体重が少ないので力は強くないが速く走る事ができる
力士は体重が重いので力は強いが速く走ることはできない
体重
走る速さ
マラソン選手力士一般的な体型の人
Trade-offs: apply to humans as well as plankton
Different body types perform Optimally for different goals.
Marine Ecosystem Modelling Symposium, AORI Mar. 4, 2015S. Lan Smith p. 3
Theory: Optimality-based models of plankton
Assumptions
Natural Selection produces optimally adapted organisms
Fitness = Growth Rate Goal: Maximize Growth Rate
Considerable success over the past decade
Recent Advances in: ‘optimal foraging’, Photoacclimation & Primary Production
Major Challenge
Dynamic models consistent with Evolutionarily Stable Strategy (ESS) ESS formulated in terms of steady state (John Maynard Smith)
i.e., how to describe short-term dynamics, e.g., acclimation, consistent with evolutionary adaptation of the very ability to acclimate.
from the review by Smith et al. (Limnology & Oceanography 56, 2011)
Marine Ecosystem Modelling Symposium, AORI Mar. 4, 2015S. Lan Smith p. 4
Trait x should change in proportion to its effect on fitness, F:
dx = dx∂F(x,E)
dt ∂x dx: flexibility ~ diversity (trait distribution) E: Environment (nutrients, light, temperature, etc.)
For plankton F = Growth; dx/dt depends on E (Smith et al. L&O, 2011)
Acclimation Rates depend on: 1. possible range of trait values (adaptive capacity), 2. environmental variability 3. current distribution of trait values
Remaining Challenge: modelled diversity tends to collapse immigration required to maintain diversity (Bruggeman & Kooijman L&O 2007)
‘Adaptive Dynamics’: evolutionary changes McGill and Brown (An. Rev. Ecol. Evol. Syst. 2007), Litchman et al. (PNAS 2009)
‘adaptive dynamics’: species succession, communities Wirtz & Eckhardt (Ecol. Modell. 1996), Wirtz (J. Biotech. 2002), Abrams (J. Evol. Biol. 2005)
‘adaptive dynamics’: modelling changing trait values
Different trait values are optimal under different environmental conditions.
Marine Ecosystem Modelling Symposium, AORI Mar. 4, 2015S. Lan Smith p. 5
What can optimality be used to model?
Acclimation short-term changes e.g., seasonal change of a dog’s coat of hair changes in chl content of phytoplankton
Ecological Dynamics
species succession (changes in community composition)
Adaptation
long(er)-term changes evolution (genetic changes in a species)
The same approaches can be used to model all three. e.g., Wirtz and Eckhardt (Ecol. Mod., 1996), Merico et al. (Ecol. Mod., 2009)
What I do
Others also model
Marine Ecosystem Modelling Symposium, AORI Mar. 4, 2015S. Lan Smith p. 6
What to Optimize?
Nutrient uptake, e.g., Optimal Uptake
as in Smith et al. (2009)
N used for other enzymes
(max. rate)
N used for uptake
sites(affinity)
nutrient uptake
Autotrophic growth, e.g., iron-light-nitrogen
colimitation as in Armstrong (1999)
phytoplankton
iron used for light
harvesting
iron used for N
assimilation
N uptakelight
phytoplankton
nutrients zooplankton
defenseuptake
Community dynamics, e.g., grazers and phytoplankton
as in Merico et al. (2009)
Maximal net growth rate
Maximal specific growth rate
Maximal uptake rate
Quantity being
optimized
trade-off
(Smith et al. L&O 2011, fig. 2)
Marine Ecosystem Modelling Symposium, AORI Mar. 4, 2015S. Lan Smith p. 7
Trade-offs as ‘Hyper-Parameterizations’
‘Hyper-Parameters’ in hierarchical Bayesian modeling specify prior distributions of model parameters (e.g., Gelman et al. Bayesian Data Analysis, 2nd edition, 2004)
Trade-offs specify how the shapes of functional relationships may change, rather than fixing their shapes. Optimal Uptake kinetics
log NO3 (in seawater)
log
K NO
3
-2.5 -1.0 0.0
-3
-2
-1
0
n = 61 data pts.
0 200 400 600 800 1000
0.0
0.2
0.4
0.6
0.8
1.0
0 200 400 600 800 1000 0 200 400 600 800 1000
Upt
ake
Rat
e
NO3 in incubation expts.
Adaptive Response
Smith et al. (MEPS 2009)
Trade-offV m
ax
α
Marine Ecosystem Modelling Symposium, AORI Mar. 4, 2015S. Lan Smith p. 8
Shape-shifting of the Growth vs. Irradiance curve
RubiscoCHLa CarotPigments
Q Q0
?
f R
DIN
loss
f θ
1-f θ-f R
lossCO2
C-uptake
N-uptake
PONPOC
Gro
wth
Rat
e
Max
. Gro
wth
Rat
eLight
Flexible Response
Wirtz & Pahlow (MEPS 2010)
Optimal Resource Allocation subject to Trade-offs
Optimality-based Photoacclimation models have advantages compared to empirically-based functions (Smith & Yamanaka. Ecol. Modell. 2007)
Marine Ecosystem Modelling Symposium, AORI Mar. 4, 2015S. Lan Smith p. 9
Selected Examples of Trade-offs for Plankton
Affinity for nutrient vs. Maximum Uptake Rate Pahlow (MEPS 2005), Smith et al. (MEPS 2009, 2014)
Light Harvesting vs. Nutrient Uptake Pahlow (MEPS 2005, 2009, 2013), Wirtz and Pahlow (MEPS 2010) and effectively all “photo-acclimation” models *although many (e.g. Geider type) do not explicity employ trade-offs
Ingestion of Prey vs. Cost of Foraging Pahlow and Prowe (MEPS 2010)
a few examples from Smith et al. (L&O, 2011)
Trade-offs allow simple models to Represent Flexible Response
Costs vs. benefits in allocating limited resources (energy, nutrients)
Marine Ecosystem Modelling Symposium, AORI Mar. 4, 2015S. Lan Smith p. 10
�B�a�l�a�n�c�e�d� �G�r�o�w�t�h� �A�s�s�u�m�p�t�i�o�n� �(�B�u�r�m�a�s�t�e�r� �A�m�.� �N�a�t�.� �1�9�7�9�)
V� �=� μQ
�A� �s�i�n�g�l�e�,� �e�x�p�l�i�c�i�t� �e�q�u�a�t�i�o�n�:� μ� �=� �f�(�N�,� �I�)
�i�n�c�l�u�d�i�n�g� �a�d�a�p�t�i�v�e� �r�e�s�p�o�n�s�e�* �*�a�s�s�u�m�i�n�g� �i�n�s�t�a�n�t�a�n�e�o�u�s� �o�p�t�i�m�a�l� � � �r�e�s�o�u�r�c�e� �a�l�l�o�c�a�t�i�o�n
Vmax
α
�O�p�t�i�m�a�l� �U�p�t�a�k�e� �k�i�n�e�t�i�c�s� �(�P�a�h�l�o�w�.� �M�E�P�S�,� �2�0�0�5�;� �S�m�i�t�h� �e�t� �a�l�.� �M�E�P�S�,� �2�0�0�9�)�a�f�f�i�n�i�t�y� �(α�)� � �v�s�.� �m�a�x�.� �u�p�t�a�k�e� �r�a�t�e� �(Vmax�)
� � ∝ fA� � � � � �v�s�.� � � � � � � � � � � � � � � � � � � � ∝ �(1 − fA�)� � � � � � � � � � �.
1 fA 0
�O�p�t�i�m�a�l� �G�r�o�w�t�h� �m�o�d�e�l� �(�P�a�h�l�o�w� �a�n�d� �O�s�c�h�l�i�e�s�.� �M�E�P�S�,� �2�0�1�3�)
� � � � �N� �u�p�t�a�k�e� �(V�^ �)� � �v�s�.� � � � � �C� �f�i�x�a�t�i�o�n� �(μ�^ �I�)� � � � � � � � � � � � � ∝� fV ∝� �(�1� −� �
Qs� − fV�)� � QN
0.5 fV 0
New Simple, Flexible Phytoplankton model: FlexPFT
Just 1 diff. eqn. for dynamics of C biomass.
+ simple calc`ns for chl:C & N:C
submitted with revisions to JPR (March, 2015)
Marine Ecosystem Modelling Symposium, AORI Mar. 4, 2015S. Lan Smith p. 11
0-D (box) model of the mixed layer at stns. K2 & S1
−200
−150
−100
−50
0
MLD
(m)
Time
39.0 39.5 40.0 40.5 41.0 41.5 42.00.00
0.02
0.04
0.06
0.08
0.10
0.12
Day of Year
PAR
(μE/
sq. c
m/s
)
�P�A�R� �F�o�r�c�i�n�g�:� � �C�r�u�i�s�e� �O�b�s�.� �w�h�e�n� �a�v�a�i�l�a�b�l�e� �+� �i�n�t�e�r�p�o�l�a�t�i�o�n� �b�y� �I�d�e�a�l� �A�s�t�r�o�n�o�m�i�c�a�l� �E�q�n�.�
Surface
MLD
Light...
attenuation with depth
Variable Mixed Layer Depth (Forcing)
entrainment ofNO3 below ML
mixing acrossbase of ML
Inflexible PFT vs. FlexPFTa single PFT for each (NPD model), respectively
embedded in the same physical modelFitted to Obs. Data for NO3, chl and PP
Marine Ecosystem Modelling Symposium, AORI Mar. 4, 2015S. Lan Smith p. 12
Tim
e (d
ays)
Tim
e (d
ays)
fA fV
0
200
400
600
800
1000
1.0 0.8 0.6 0.4 0.2 0
0
200
400
600
800
1000
0.5 0.4 0.3 0.2 0.1 0
Optimal Growth model (Pahlow and Oschlies 2013)
N uptake (V^ ) vs. C fixation (μ̂I) ∝ fV ∝ (1 − Qs − fV) QN
0.5 fV 0
Vmax
α
Optimal Uptake kinetics (Pahlow 2005; Smith et al. 2009)
affinity (α) vs. max. uptake rate (Vmax) ∝ fA vs. ∝ (1 − fA)
.
1 fA 0
2010
20
11
2012
2010
20
11
2012
Intracellular Resource Allocation
Inflexible Control:Turn off Optimization:
constant fA constant fV
FlexPFT model Optimal Allocation
optimize fA | N
optimize fV | N, I
stn. K2
stn. S1
more allocation to light gathering at stn. K2
=> lower fA & fV (lower in winter)
more allocation to nutrient uptake at stn. K2 higher fA & fV <=(higher in summer)
Applied in a 0-D (box) model of the mixed layer for two time-series stns.
Marine Ecosystem Modelling Symposium, AORI Mar. 4, 2015S. Lan Smith p. 13
Inflexible PFT vs. FlexPFT applied to stns. K2 & S1
Less Difference at N-poor stn. S1 only
At N-rich stn. K2 Seasonal cycle of chl differs greatly
Each model was fitted to data (red dots) using the Adaptive Metropolis algorithm (Smith JGR 2011)
05
101520253035
0
5
10
15
20
25
30
0.5
1.0
1.5
0.30.40.50.60.70.80.9
0.5
1.0
1.5
2.0
0.12
0.14
0.16
0.18
0.20
Time (days) Time (days)
N (mmol L-1)
chl (mg m-3)
chl:C (g mol-1)
Phy Biomass (mmol C m-3) Q (mol N (mol C)-1)
0 200 400 600 800 10000 200 400 600 800 1000
P. Production(mg C m-3 d-1)
FlexPFT in colorsInflexible Control (grey)
2010 2011 2012 2010 2011 2012
02468
1012
0.0
0.5
1.0
1.5
2.0
0.0
0.2
0.4
0.6
0.8
1.0
0.2
0.4
0.6
0.8
0.20.40.60.81.01.2
0.040.060.080.100.120.140.16
Time (days) Time (days)0 200 400 600 800 10000 200 400 600 800 1000
2010 2011 2012 2010 2011 2012
N (mmol L-1)
chl (mg m-3) chl:C (g mol-1)
Phy Biomass (mmol C m-3) Q (mol N (mol C)-1)
P. Production(mg C m-3 d-1)
Marine Ecosystem Modelling Symposium, AORI Mar. 4, 2015S. Lan Smith p. 14
FlexPFT gives different vertical dist. for chl vs. N biomass, PP 1-D GOTM model, 3 year simulation of subtropical stn. S1.
stn. S1
Importance of ‘Photo-acclimation’ is well known for subtropics (Ayata et al. JMS 2013)
most recent results 2015.01
0.0 0.2 0.4 0.6 0.8 1.0−200
−150
−100
−50
0
0.0 0.2 0.4 0.6 0.8 1.0 0 5 10 15 20 25 30
obs. avg.FlexPFT
obs. avg.FlexPFTchl:N = 1.59
obs. avg.FlexPFT
Dep
th (m
)
(mg m-3) (mmol N m-3) (mg C m-3 d-1)
mean vertical chl mean vertical pon mean vertical P. Prod.
An Inflexible Control Model, chl:N:C ratios
Here the two models perform similarly.
In both, chl decreases too steeply with depth.
Need to add DON to the model to stop ‘nutrient trapping’.
0.0 0.2 0.4 0.6 0.8 1.0−200
−150
−100
−50
0
0.0 0.2 0.4 0.6 0.8 1.0 0 10 20 30
obs. avg.Inflexible PFT
obs. avg.Inflexible PFT
obs. avg.Inflexible PFT
Dep
th (m
)
mean vertical chl mean vertical pon mean vertical P. Prod.
(mg m-3) (mg C m-3 d-1) (mmol N m-3)
Marine Ecosystem Modelling Symposium, AORI Mar. 4, 2015S. Lan Smith p. 15
FlexPFT gives different vertical dist. for chl vs. N biomass, PP 1-D GOTM model, 3 year simulation of subarctic stn. K2
stn. K2
most recent results 2015.01
An Inflexible Control Model, chl:N:C ratios
The inflexible model cannot consistently reproduce both chl & PON.
mean vertical pon
(mmol N m-3)0.0 0.5 1.0 1.5 2.0
mean vertical P. Prod.
(mg C m-3 d-1)0 10 20 30
mean vertical chl
0.0 0.5 1.0 1.5−100
−80
−60
−40
−20
0
(mg m-3)
obs. avg.FlexPFT
obs. avg.FlexPFTchl:N = 1.59
obs. avg.FlexPFT
Dep
th (m
)
(mg m-3) (mmol N m-3) (mg C m-3 d-1)
mean vertical chl mean vertical pon mean vertical P. Prod.
0.0 0.5 1.0 1.5−100
−80
−60
−40
−20
0
0.0 0.5 1.0 1.5 2.0 0 10 20 30
obs. avg.Inflexible PFT
obs. avg.Inflexible PFT
obs. avg.Inflexible PFT
Dep
th (m
)
mean vertical chl mean vertical pon mean vertical P. Prod.
(mg m-3) (mg C m-3 d-1) (mmol N m-3)
Flexible chl:N:C is important at subarctic stn. K2 & likely over much of the ocean -- NOT only at low latitudes.
Marine Ecosystem Modelling Symposium, AORI Mar. 4, 2015S. Lan Smith p. 16
FlexPFT gives dynamic vertical distr. for chl:N and Q1-D GOTM model, 3 year simulations of stns. S1 & K2.
most recent results 2015.01
The Importance of ‘Photo-acclimation’ is well known for sub-tropics (Ayata et al. JMS 2013)
stn. S1
stn. S1
Marine Ecosystem Modelling Symposium, AORI Mar. 4, 2015S. Lan Smith p. 17
New Flexible ZooEFT model developed in FY2014
0 90 180 270 3600
0.1
0.2
0.3
0.4
DIN (nutrient) (mm
ol m
-3)
0 90 180 270 360Grazing Rate
KTWNo KTW
0
0.2
0.4
0.6
0.8
1.0
(d-1
)
Time (d)
Emergent Trade-off KTW: High Phy Size Diversity, more even Growth No KTW: Faster Peak Growth, but more variable, Low Phy size diversity
ZooEFT
Grazing PreferenceRel.
Abu
ndan
ce
0 90 180 270 3600
0.2
0.4
0.6
0.8 Phy size
log
ESD
(μm
)
Time (d)
0 90 180 270 3600
0.2
0.4
0.6
0.8
1.0
Phy Growth Rate
(d-1
)
0 90 180 270 3600
0.2
0.4
0.6
0.8
1.0Phy Size Diversity
log σ
(μm
)
Size as ESD (μm) 0.2 0.5 2.0 5.0 20.0 100.0
0
1
2
3
4
5 N = 10 N = 1 N = 0.1
Gro
wth
Rat
e (d
-1)Simple Trade-off
assumed for Phy:Smaller is better at low NutrientBigger is better at high Nutrient
Periodic Nutrient supply was imposed.
Based on the ‘Kill-the-Winner’ model (Vallina et al. Nat. Comm. 2014; Vallina et al. Prog. Oceanogr. 2014)
Results here are from a simple size-based model using Monod Growth kinetics. We plan to publish this soon.
The ZooEFT model has also been coupled with our PhyEFT model in the 0-D setup for stns. K2 & S1. Next step: try it in the 1-D setup.
Marine Ecosystem Modelling Symposium, AORI Mar. 4, 2015S. Lan Smith p. 18
Non-Adaptive vs. Flexible models
How do changing env. conditions impact:
Overal ecosystem response to, e.g., climate change or human nutrient inputs?
Biodiversity?
B o t h t h e b i o m a s s & r e s p o n s e o f e a c h
F l e x P F T c a n c h a n g e .
Bio
mas
s EnvironmentalChange
・Each FlexPFT adapts to changing environment (e.g., light, nutrients).
Different AdaPFTs, each having different flexible abilities
(b) New Flexible FlexPFT model
O n l y t h e d i s t r i b u t i o n o f b i o m a s s c a n c h a n g e .
Plankton Functional Types (PFT) (a) Existing PFT models
Bio
mas
s
EnvironmentalChange
C o m p e t i t i v e a b i l i t i e s r e m a i n
f i x e d f o re a c h P F T
N uptakeability
C fix’nability
Different PFTs, each having different abilities
Marine Ecosystem Modelling Symposium, AORI Mar. 4, 2015S. Lan Smith p. 19
New size-based PhyEFT model Size scaling of Traits (input parameters)
as in Wirtz (Mar. Biol. 2013)Dark Light
Adaptive Response to env. (light & N)
Big is better at high N
Small is better at low N
(esp. at high light)
0.2 1.0 5.0 20.0 100.01
2
3
4
5
6
7
8
0.2 1.0 5.0 20.0 100.00.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
0.2 1.0 5.0 20.0 100.0
1.0
1.5
2.0
2.5
0.2 1.0 5.0 20.0 100.0
0.02
0.03
0.04
0.05
Pot. max. growth rate
μm
ax (d
-1)
Pot. max. affinity Subsistence cell quota
ESD (micrometers) ESD (micrometers)
Pot. max. uptake rate
A0
(m3
(mm
ol C
)-1 d
-1)
V0
(mol
N (m
ol C
)-1 d
-1)
Q0
(mol
N (m
ol C
)-1)
0
1
2
3
4
0.2 0.5 2 5 20 50
N = 10 mmol m-3
N = 1 mmol m-3
N = 0.1 mmol m-3
Effe
ctiv
e m
ax. (
Mon
od) g
row
th ra
te
μm
ax (d
-1)
Strong Light Limitation, S(I) = 0.2
0
1
2
3
4
0.2 0.5 2 5 20 50
Effe
ctiv
e m
ax. (
Mon
od) g
row
th ra
te μ
max
(d-1
)
Light Replete, S(I) = 1.0
empirical allometriesLitchman et al. (Ecol. Lett. 2007) Edwards et al. (L&O 2007) Marañon et al. (Ecol. Lett. 2013)
�B�a�l�a�n�c�e�d� �G�r�o�w�t�h� �A�s�s�u�m�p�t�i�o�n� �(�B�u�r�m�a�s�t�e�r� �A�m�.� �N�a�t�.� �1�9�7�9�)
V� �=� μQ
�A� �s�i�n�g�l�e�,� �e�x�p�l�i�c�i�t� �e�q�u�a�t�i�o�n�:� μ� �=� �f�(�N�,� �I�)
�i�n�c�l�u�d�i�n�g� �a�d�a�p�t�i�v�e� �r�e�s�p�o�n�s�e�* �*�a�s�s�u�m�i�n�g� �i�n�s�t�a�n�t�a�n�e�o�u�s� �o�p�t�i�m�a�l� � � �r�e�s�o�u�r�c�e� �a�l�l�o�c�a�t�i�o�n
Vmax
α�O�p�t�i�m�a�l� �U�p�t�a�k�e� �k�i�n�e�t�i�c�s�
�(�P�a�h�l�o�w�.� �M�E�P�S�,� �2�0�0�5�;� �S�m�i�t�h� �e�t� �a�l�.� �M�E�P�S�,� �2�0�0�9�)�a�f�f�i�n�i�t�y� �(α�)� � �v�s�.� �m�a�x�.� �u�p�t�a�k�e� �r�a�t�e� �(Vmax�)
� � ∝ fA� � � � � �v�s�.� � � � � � � � � � � � � � � � � � � � ∝ �(1 − fA�)� � � � � � � � � � �.
1 fA 0
�O�p�t�i�m�a�l� �G�r�o�w�t�h� �m�o�d�e�l� �(�P�a�h�l�o�w� �a�n�d� �O�s�c�h�l�i�e�s�.� �M�E�P�S�,� �2�0�1�3�)
� � � � �N� �u�p�t�a�k�e� �(V�^ �)� � �v�s�.� � � � � �C� �f�i�x�a�t�i�o�n� �(μ�^ �I�)� � � � � � � � � � � � � ∝� fV ∝� �(�1� −� �
Qs� − fV�)� � QN
0.5 fV 0
Cell Size (ESD, mm) Cell Size (ESD, mm)
Marine Ecosystem Modelling Symposium, AORI Mar. 4, 2015S. Lan Smith p. 20
Old NPZD approach vs. New Ecologically Flexible Types
‘adaptive dynamics’ can represent the dynamics of the trait distribution (α,σ) using only two differential equations per EFT.
Environmental Change
D i s c r e t e t r a i t v a l u e s
Because traits are fixed, model predictions may become invalid
Flexible trait distributions change to suit environ-mental conditions.
C o n t i n u o u s d i s t r i b u t i o n o f t r a i t v a l u e s
Plankton Functional Types (PFT) Ecologically Flexible Types
Biom
ass
Biom
ass
Biom
ass
Biom
ass
Trait value Trait value
Trait value (here, affinity)Environmental Change
(a) Existing NPZD models (b) EFT-model to be developed
・separate differential eqs. for each PFT・traits are fixed (constant) for each PFT
・ ・ ・
・ ・ ・
Ks Vmax
Gro
wth
Rate
Upta
ke R
ate
nutrient conc. (S)
Affinity-based kinetics
affinity
Trade off*e.g., affinity, Vmax, size
mean trait value:
std. dev. of trait value:
PhyPhyZoo
Zoo
Marine Ecosystem Modelling Symposium, AORI Mar. 4, 2015S. Lan Smith p. 21
Size-Scaling of Traits => Size-Scaling of Growth
0.2 1.0 5.0 20.0 100.00.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
0.2 1.0 5.0 20.0 100.0
1.0
1.5
2.0
2.5
0.2 1.0 5.0 20.0 100.0
0.02
0.03
0.04
0.05
Pot. max. affinity
Subsistence cell quota
ESD (μm)
Pot. max. uptake rate
A^ 0 (
m3
(mm
ol C
)-1 d
-1)
V^ 0 (
mol
N (m
ol C
)-1 d
-1)
Q0
(mol
N (m
ol C
)-1)
Empirical Size-scalings
Response depends on both cell size and EnvironmentNew modelling
framework to relate lab. measurements to the dynamic response of phytoplankton communities.
0
5
10
15
0.0
0.2
0.4
0.6
0.8 αuptake (m
3 (mm
ol C) -1 d -1)
αgr
owth
(m3
(mm
ol N
)-1 d
-1)
0.2 0.5 2.0 5.0 20 50
Light Replete
Effective values of Affinityfor Growth & Uptake, respectively
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5Effective max. (Monod) growth rate
0.2 0.5 2.0 5.0 20 50
μm
ax (d
-1)
eff
0
2
4
6
8
10
w
k
0.2 0.5 2.0 5.0 20 50
KV (dashed)
ESD (μm)
(μm
ol L
-1)
Effective half-saturation ‘constants’for Growth & Uptake, respectively
Kμ (solid)
Very Dark
0.0
0.1
0.2
0.3
0.4
0.5
Effective max. (Monod) growth rate
0.2 0.5 2.0 5.0 20 50
μm
ax (d
-1)
eff
0
2
4
6
8
10
0.0
0.1
0.2
0.3
0.4
0.5
αuptake (m
3 (mm
ol N) -1 d -1)
αgr
owth
(m3
(mm
ol C
)-1 d
-1)
0.2 0.5 2.0 5.0 20 50
Effective values of Affinityfor Growth & Uptake, respectively
ESD (μm)
0
2
4
6
8
10
w
k
0.2 0.5 2.0 5.0 20 50
(μm
ol L
-1)
Effective half-saturation ‘constants’for Growth & Uptake, respectively
KV (dashed)
Kμ (solid)
N = 0.1 μMN = 1.0 μMN = 10 μM
Data for KV (open circles)
(Edwards et al. 2012)
Km < KV (growth) (uptake)
Morel (J. Phycol. 1987)
Here size-scaling for Km depends on light and nutrient environment.
�B�a�l�a�n�c�e�d� �G�r�o�w�t�h� �A�s�s�u�m�p�t�i�o�n� �(�B�u�r�m�a�s�t�e�r� �A�m�.� �N�a�t�.� �1�9�7�9�)
V� �=� μQ
�A� �s�i�n�g�l�e�,� �e�x�p�l�i�c�i�t� �e�q�u�a�t�i�o�n�:� μ� �=� �f�(�N�,� �I�)
�i�n�c�l�u�d�i�n�g� �a�d�a�p�t�i�v�e� �r�e�s�p�o�n�s�e�* �*�a�s�s�u�m�i�n�g� �i�n�s�t�a�n�t�a�n�e�o�u�s� �o�p�t�i�m�a�l� � � �r�e�s�o�u�r�c�e� �a�l�l�o�c�a�t�i�o�n
Vmax
α�O�p�t�i�m�a�l� �U�p�t�a�k�e� �k�i�n�e�t�i�c�s�
�(�P�a�h�l�o�w�.� �M�E�P�S�,� �2�0�0�5�;� �S�m�i�t�h� �e�t� �a�l�.� �M�E�P�S�,� �2�0�0�9�)�a�f�f�i�n�i�t�y� �(α�)� � �v�s�.� �m�a�x�.� �u�p�t�a�k�e� �r�a�t�e� �(Vmax�)
� � ∝ fA� � � � � �v�s�.� � � � � � � � � � � � � � � � � � � � ∝ �(1 − fA�)� � � � � � � � � � �.
1 fA 0
�O�p�t�i�m�a�l� �G�r�o�w�t�h� �m�o�d�e�l� �(�P�a�h�l�o�w� �a�n�d� �O�s�c�h�l�i�e�s�.� �M�E�P�S�,� �2�0�1�3�)
� � � � �N� �u�p�t�a�k�e� �(V�^ �)� � �v�s�.� � � � � �C� �f�i�x�a�t�i�o�n� �(μ�^ �I�)� � � � � � � � � � � � � ∝� fV ∝� �(�1� −� �
Qs� − fV�)� � QN
0.5 fV 0
Marine Ecosystem Modelling Symposium, AORI Mar. 4, 2015S. Lan Smith p. 22
Phy+ZooEFT in 0-D model of stns. K2 & S1 (9 yr. sim.)
stn. S1, subtropicalstn. K2, subarctic
Decreasing diversity
Decreasing diversity
KTW sustains Phy diversity, & its seasonality with modest effects on chl, PP
Next: Examine how diversity relates to function Compare to Obs. size distr.
Kill-the-Winner Passive Prey Switching Kill-the-Winner Passive Prey Switching
Marine Ecosystem Modelling Symposium, AORI Mar. 4, 2015S. Lan Smith p. 23
Summary & Conclusions
Optimality-based modelling has well established theory.and is recently being applied to model plankton in the ocean.
Flexible Phytoplankton Functional Type: FlexPFT developed Optimizes specific growth rate w.r.t. to 2 trade-offs: 1. Optimal Uptake (OU) kinetics 2. C (energy) vs. N aquisition Only 1 diff. eq. per PFT Trait-based => can Directly use reported size-scalings Framework for modelling Size Diversity of phy.
Flexible models respond quite differently vs. Inflexible PFTs e.g., Variable chl & N content => more dynamic Some models of optimal Grazing developed, Much more work needed ...
Marine Ecosystem Modelling Symposium, AORI Mar. 4, 2015S. Lan Smith p. 24
May impact dynamic functioning & resilience of ecosystems e.g., predator-prey feedbacks (Tirok et al. PLoS One 6, 2011)
Major Challenge: How to model realistic biodiversity in dynamic, heterogenous environments?
1. Add more species competitive displacement of many species e.g., Moisan et al. (Ecol. Modell., 2002)
2. Resolve traits better differentiate strategies more precisely e.g., Litchman and Klausmeier (Am. Naturalist 157, 2001)
3. ‘adaptive dynamics’ acclimation within species competitive displacement between species & trait evolution (Wirtz. J. Biotechnol. 97, 2002; Smith et al. L&O. 56, 2011; Tirok et al. PLoS One 6, 2011)
Biodiversity: beyond coarse large-scale patterns
Easiest
MoreDifficult
Simplistic
MoreRealistic
Marine Ecosystem Modelling Symposium, AORI Mar. 4, 2015S. Lan Smith p. 25
Effective Monod (growth) params. vs. MM params.Burmaster (Am. Nat. 1979) MM kinetics (uptake) + Droop model (Growth) => Monod kinetics (growth)Similarly, we combine OU kinetics (uptake) + Optimal Growth (OG) model => Effective Monod Params.
where
or, equivalently
Now, observed size-scalings for MM & Droop parameters can be used to predict size-scalings for Monod parameters.
Marine Ecosystem Modelling Symposium, AORI Mar. 4, 2015S. Lan Smith p. 26
Shannon Diversity Index Differential entropy for N discrete Obs. or PFTs for log-normal distribution (model)
H = -∑i
N pi ln(pi ) h = 12 + ln(s√2p) + m
where pi is the probability of where m is mean (in log space) belonging to the i-th class s is the std. deviation depends on ‘binning’ choice depends on the assumed distr’n. and on N ,which is arbitrary log-normal is reasonable for plankton
Instead, for obs. & PFT models (Quintana et al. 2008, Schartau et al. 2010)
Diversity Index: Discrete vs. Continuous
Qintana et al. (L&O. Methods 2008) who denote diff. entropy as m
applied to obs. by Schartau et al. (JPR 2010)
Non-parametric estimate of h from discrete data (yi)
h
Advantages of h Consistently quantifies diversity despite different # of obs. or different # of PFTs (in models)
Dis-advantages of h not bounded: can be < 0 not as intuitive for intrepretation
Marine Ecosystem Modelling Symposium, AORI Mar. 4, 2015S. Lan Smith p. 27
Comparing h vs. H
0 200 400 600 800 1000012345
initial N = 200
initial N = 50
initial N = 20
0 200 400 600 800 1000−1.0
−0.5
0.0
0.5
1.0
overlayed forN = 20, 50, 200
H
h
Time (days)
FlexPFT (red)Non-adaptive PFT (grey)
2010 2011 2012
Model Set-up for different number, N of either PFTs, or AdaPFTs
in a simple NO3 - PxN - D model (no Zoo)
mortality for mi = m0PavgPi which maintains biodiversity (Record et al. ICES J. Mar. Sci. 2013)
0.5 50cell size (ESD, mm)
PTot
PTot/20PTot/50
PTot/20
= initial total biomass, divided evenly among N
N = 20
0.5 50cell size (ESD, mm)
N = 50
0.5 50cell size (ESD, mm)
N = 200
h is independent of N (# of PFTs), whereas H increases with NOnly slight difference between FlexPFT vs. control
h is better for obs.-model & model-model comparsions
with Size scaling of
Traits
0.2 1.0 5.0 20.0 100.01
2
3
4
5
6
7
8
0.2 1.0 5.0 20.0 100.00.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
0.2 1.0 5.0 20.0 100.0
1.0
1.5
2.0
2.5
0.2 1.0 5.0 20.0 100.0
0.02
0.03
0.04
0.05
Pot. max. growth rate
μm
ax (d
-1)
Pot. max. affinity Subsistence cell quota
ESD (micrometers) ESD (micrometers)
Pot. max. uptake rate
A0
(m3
(mm
ol C
)-1 d
-1)
V0
(mol
N (m
ol C
)-1 d
-1)
Q0
(mol
N (m
ol C
)-1)
3 yr. sim-ulation for
stn. K2
Marine Ecosystem Modelling Symposium, AORI Mar. 4, 2015S. Lan Smith p. 28
Comparing h: FlexPFT vs. Inflexible Control200 PFTs in a simple 0-D model: NO3 - Px - D model (no Zoo)
simplified Kill-the-Winner mortality, mPi = m0PavgPi , maintains biodiversity
(Record et al. ICES J. Mar. Sci. 2013)
0.2
0.51.02.0
5.010.0
0 200 400 600 800 10001.77
1.78
1.79
1.80
1.81
1.82
0.2
0.51.02.0
5.010.0
0 200 400 600 800 10001.0
1.2
1.4
1.6
med
ian
size
(μm
)si
ze d
iver
sity
, h
Time (days) Time (days)
Diversity maintained by inter-dependent mortality ratesstn. K2 stn. S1
FlexPFTcontrol
2010 2011 2012 2010 2011 2012
Marine Ecosystem Modelling Symposium, AORI Mar. 4, 2015S. Lan Smith p. 29
Comparing h: FlexPFT vs. Inflexible Control200 PFTs in a simple 0-D model: NO3 - Px - D model (no Zoo)
independent quadratic mortality, mPi = m0Pi
2, gives Competitive Exclusion (Record et al. ICES J. Mar. Sci. 2013)
0.2
0.51.02.0
5.01020
0 200 400 600 800 1000
0.0
0.5
1.0
1.5
0.2
0.51.02.0
5.01020
0 200 400 600 800 1000
−0.5
0.0
0.5
1.0
1.5
Time (days) Time (days)
med
ian
size
(μm
)si
ze d
iver
sity
, h
stn. K2 stn. S1Competitive exclusion (independent mortality rates)
2010 2011 2012 2010 2011 2012