Phenotyping vs. ideotyping: Opportunities and  Limitations of model‐assisted crop design 

Phenotyping vs. ideotyping: Opportunities and  Limitations of model‐assisted crop design 

drawing from genetic diversity   drawing from genetic diversity

Delphine Luquet Michael Dingkuhn

Delphine Luquet, Michael Dingkuhn

CIRAD, AGAP research unit

Montpellier France

Montpellier, France

9th of February 2011

CC translates into increasing staple food commodity prices

!

!

!

!

!

400 450

T on

2000 2050 No climate change 2050 CSIRO NoCF 2050 NCAR NoCF

CC will contribute to higher Food prices

2050:

According to models, a less 

favorable climate for agriculture  ₁₀₀ ¹⁵⁰

200 250 300 350

ars Per Metric T

g (Tropics & subtropics) Andrew Jarvis, CIAT/CCAFS

- 50 100

Rice Wheat Maize Soybeans

Doll a

!

!

!

!

!

400 450

T on

2000 2050 No climate change 2050 CSIRO NoCF 2050 NCAR NoCF

CC will contribute to higher Food prices

2050:

According to models, a less 

favorable climate for agriculture  ₁₀₀ ¹⁵⁰

200 250 300 350

ars Per Metric T

g (Tropics & subtropics) Andrew Jarvis, CIAT/CCAFS

- 50 100

Rice Wheat Maize Soybeans

Doll a

Talk Structure 

Crop improvement & CCV 

Place of phenotypic plasticity

Pl t d li t t h t i d id t i

Plant modeling to support phenotyping and ideotyping Plant modeling  vs. molecular breeding: 

Ongoing research

Understanding genetic & physiological architecture of  complex traits

Outlook

Place for phenotypic plasticity p yp p y

• What is PP?

Adapti e changes in plant organi ation d ring ontogenesis – Adaptive changes in plant organization during ontogenesis – Broad adaptation through adjustment to variable conditions

• Why needed under CC?

• Why needed under CC?

– CC will increase variability

– Water will be scarcer (water is a great stabilizator!) Water will be scarcer (water is a great stabilizator!)

• Heat, cold, drought, salinity, soil fertility, weed competition

• Problem of trade‐offs with yield pot.

– Plastic plants = variable plant types;  at what cost?

• Problem of trait complexity

– How to measure?

– Complex genetics?

Inherent capacity to dynamically regulate

Phenotypic plasticity

Inherent capacity to dynamically regulate  +N

morphogenesis    (Nicotra et al. 2010)

• Based on compensatory source‐sink processes 

• Maintains functioning reproduction & production

+N

Maintains functioning, reproduction & production  when conditions fluctuate

• Includes more than morphology: Phenology,  physiological defenses…

‐P

p y g

Examples of traits needed under greater climatic variability Phenology

‐ Adaptive phase duration (temporal compensation and stress escape)

‐ Rapid development for vigour and high yield potential under short duration Morphology

‐ Architecture limiting stress exposure and maximizing resource effiency

‐ Environment responsive morphogenesis Physiology

‐ Effective and rapidly inducible tolerance;  Hardening?

‐ Protection of reproductive processes (e.g., cooling of spikelets) 

Implications of phenotypic plasticity

• Increase of G x E Increase of G x E

• Intelligent use of G x E through management & forecasts

• Avoid or overcome counter‐productive plasticity 

trade‐off on yield (Nicotra et al. 2010)

• Trade offs among multiple yield objectives

• Trade‐offs among multiple yield objectives 

e.g. sweet sorghum for ‘FFF’ (Gutjahr et al. 2010)

Challenge:

Challenge:

Conceive plants having ‘productive’ plasticity Conceive plants having productive plasticity

Lesson from the past: start from available genetic diversity,  t i hf l h i l i l thi ki

not wishful physiological thinking

Understand physiological and genetic architecture of  complex traits 

Reduce complex traits to component traits recombine Reduce complex traits to component traits, recombine 

intelligently

Give room to discovery, and build it in 

Plant modeling to support

Plant modeling to support 

phenotyping & ideotyping

Adapting the Ideotype concept

Morphology & Phenology

• Green revolution for favorable conditions

D fi > Hi h till i & HI l l d i > N i – Dwarfing => High tillering & HI, less lodging => N responsive

• Make traditional systems more productive

– Combine PP‐sensitivity with green revolution traits (African sorghums)

Morphology, phenology and biochemistry

• Multi‐purpose, new purposes

– Grain/forage cowpea, peanut… (FF)

/b /f h ( )

– Sweet grain/bioEtOH/forage sorghum (FFF) – Biomass 2nd generation energy Annuals/Trees – C‐sequestering food/forage crops

Most difficult: Change ecophysiological adaptation (T, drought, CO ₂ )

– Combine multiple adaptations with desired plant type – Transform metabolic type (C4 rice)

T f h bl d (‘ i ’ h ?)

– Transform harvestable product (‘rice ’‐sorghum?)

3 steps in Ideotype development where crop 

d l h l

models  can help

• Characterization of Target Populations of Environments (TPE),  incl. CC scenarios

=> Use simple agronomic crop model as “lens”

=> Use simple agronomic crop model as  lens

• Identification of target trait combinations & plant types for  TPE

TPE

=> Use crop model with GxExM skills to simulate trait expression & 

adaptive value

• Phenotyping process => Association studies => Markers

Heuristics: Extraction of trait parameters from observed variables 

=> Specialized process models with small parameter nb

=> Specialized process models with small parameter nb. 

Role of plant physiology and modelling in  plant breeding: phenotyping

plant breeding: phenotyping

• Novel tools (imagery, remote sensing): maximize data acquisition on  large number of plants

large number of plants

Rapid fluorescence OJIP Handy PEA

High resolution Thermography OJIP Handy PEA

Ö Still need to decorrelate G and E effects (modelling)

g g p y

• Models needed that analyze & predict G response curve to E through

genotypic parameters

Example of role of modelling in phenotyping

Raymond et al. (2003, 2004)

Leaf expansion rate response to drought variables (maize)

Welcker et al. 2007

• LER model QTLs colocate with that of direct measurements  (leaf width)

• More stable across E : QTLxE overcome  (modeling the cause of QTL instability)

13

• Validated in contrasting genetic background (temperate to tropical)

• LER model QTLs colocate with silk expansion QTLs (ASI): 2  crucial traits in 1! 13

From relevant QTLs to ideotype: 

integrative process

• QTL validation = evaluation at plant/pop. scale (crop  performance): When expressed? When relevant?

performance): When expressed? When relevant?

• Modelling must predict accurate G x E x M interactions 

d d ff

and trade‐offs  (Hammer et al. 2010) 

• Even more challenging when addressing CCV  g g g

(extrapolation)

phenotyping (cont )

Simulated effect of  LER QTL i ld Incorporation of LER model 

in APSIM (maize)

phenotyping (cont.)

LER‐QTLs on yield  in APSIM (maize)

From Chenu et al 2009 ; Genetics

Ö First real proof of concept for ideotype simulation using crop  models driven by genetic parameters 

From Chenu et al. 2009 ; Genetics

Ö Doing this for traits for phenotypic plasticity requires models with 

greater detail of trait interactions (morpho/pheno/physio)

Can plant modeling assist molecular Can plant modeling assist molecular 

breeding by analyzing genetic & 

physiological architecture physiological architecture 

of complex traits? 

Ongoing work

Massive use of molecular markers for Massive use of molecular markers for 

agronomic traits and agroecological adaptation

Rice & sorghum are sequenced

Mass sequencing of rice genomes planned Mass sequencing of rice genomes planned

Sorghum is a major source of genes in C4 rice project Mass application of MAS in private seed sector

GRiSP plan for global phenotyping & gene discovery & 

molecular breeding networks

CC&FS plan for ideotype strategies for 2030 CC horizon

Senegal WARDA

Cold    Hot

Montpellier Montpellier

WARDA

C ld H t

Sowing dates

ll

Base Temperature Root vigour

& architecture phenotyping

Cold Heat

Montpellier

Phenotyping

N k

Drought

Network

Cold Heat

Madagascar

Philippines

Heat

Philippines

IRRI Colombia

CIAT

Drought

h l l t h i whole plant morphogenesis

• Body plan construction

• +/‐ / plastic in response to E depending on G

• Many processes related to meristem activity y p y

¾ tillering, leaf initiation, size, expansion…

Important: Simulation of Outcomes vs.  Forcing!  

Example: Partitioning handled differently ifor agronomic or genetic objectives

Genesis of an organ in the plant system:

Plastochron

Meristem Initiation

Productive period

Cell division Senescence

X

Expansion

Si k Source phase

Sink  commitment

Sink

implementation 

Source phase (case of leaf)

Consequences:

• Demand determined before growth

• Demand regulated to match supply Body plan ‐ phenotype 

• Supply feedbacks on meristem behaviour

Ö Physiological linkage (trade‐off) among traits  (Rebolledo et al 2010; Granier & Tardieu 2009) Resource acquisition

C & water status

(Rebolledo et al 2010; Granier & Tardieu 2009) Ö Also genetic linkage  (organ cell vs. organ n°) (W ter Steege et al. 2005; Tisné et al. 2008)

Ö Rebolledo : 

phenotyping for absence of linkage?

impact on rice early vigour p y g

CIRAD, greenhouse (2009), 203 japonica cvs.; pot, well watered and stressed

⇒ DR main trait explaining vigor (RGR) 

d ll t d d d ht diti

under well watered and drought conditions

⇒ Holds up at constant tillering & leaf size 

⇔ Direct effect of DR

⇔ Direct effect of DR

⇔ trait genetic independence?

Issue of signaling

Si k dj t t b th/d l t f db k

Sink adjustment by growth/development process feedbacks

Meristem Development response:

‐Cell division 

‐Development rate

‐Organ size?

Morphogenesis

l

‐Organ size?

Glu + Fru Sucrose OsCINx CIN

Assimilation, Mobilization

Transport &

discharge 

Hormonal Sugar

to apoplast

Stress

signals signal

Stress

EcoMeristem, model of phenotypic plasticity

Ic = index of internal competition = proxy for sugar availability = internal signal

10 20 30 40

total CHO co oncentration of SDW ) (g per g TDW (g) dead leaf number

0.3 0.2 0.1 0.0 10

4

4 8 6

2 0

6 5

3 2

0 1

DR = 0.025 DR= 0.02 DR = 0.017 DR= 0.014 DR = 0.013 DR = 0.011

(a)

(c)

(e)

10 20 30 40

Crop growth r (g of shoot per

rate visual phyllochron (° C.d ) day) tiller number

100 0.6 0.5 0.4 0.3 0.2 0.1 0.0

40 80 60

20 15 12 9 6 3 0

(b)

(d)

(f)

Simulation experiment with Ecomeristem Source ‐ sink processes vs. DR

6 DR values, else parameters constant 40 day simulations

Rapid DR increases...

=>

growth rate

=>

transitory reserve depletion

=>

tillering

But can also cause « trophic crisis »

=>

delayed leaf appearance

=>

smaller leaves

=>

accelerated leaf senescence

During drought:

‐ Stress more severe

( because of greater water use ) ... Followed by faster recovery

Days after germination Days after germination

300 250 200 150 100 50 0

-50

0.010 0.012 0.014 0.016 0.018 0.020 0.022 0.024 0.008 0.012 0.016 0.020 0.024 0.028

d

') RGR (g.g'. °C

1.022 1.020 1.018 1 .016 1.014 1.012 1.010 1.008 1.006

y = 1 + 2.6x-54.8x ² ; r ² =0.67 y = 1+1.55x-34x ² ; r ² =0.74

urce leaves dw)

in main stem so (g

per g of leaf starch

y = 401.4-3.3.10 ⁴ x + 6.7.1 0 ⁷ r ² = 0.2

International Conference on Crop Improvement, Ideotyping and Modeling for African Cropping Systems under Climate Change (CIMAC), 2011/02/07

09, Stuttgart

Hohenheim, Germany

Natural genetic diversity cs. potential (in‐silico) diversity

Natural vs. in ‐ silico population (performance of model

parameter combinations)

Natural relationship of Reserves vs. DR

Developmental rate (°C .d

)

Contribution of parameters ^{MGR Ict Epsib plasto Adjusted R}

1 variable X 0,17

to DR in natural p o p ulation 2variables X X 0 ,28

3 variables X X X 0,52

4 variables X X X X 0,55

lc (Supply/demand)

=>

Fast development provides for earlier tillering

=>

But causes tiller abortion due to competition

D ry matter (kg/ha)

n

s

Phyllo Phyllo Phyllo

55 ^°C.d

LAI green and dead

40 °C.d 70 ^°C.d

6 5 Al L 4

3 2 1

0

0 20 40 60 80 100 120 140 160 180

8 4

Phyllo Phyllo Phyllo

55 °C.d 40 ^°C.d 70 ^°C.d

Tiller number & Ic (Supply /demand)

6 3

mi

-s/1 ^{4 2} lei Til

2 1

0 0

0 20 40 60 80 100 120 140 160 180 14000

Phyllo 55 °C.d

Ag biomass & grain yield

Phyllo 40 °C.d

° Phyllo 70 C.d

12000 10000 8000 6000

2000

0

0 20 40 60 80 100 120 140 160 180 Days after sowing

SAMARA (Risocas product):

Predicting GxExM of process traits in an agronomic context

=>

Fast development increases LAI

=>

But leads to early leaf senescence

=>

Fast development affects biomass yield little

=>

But reduces grain yield (small panicles, poor sink)

Outlook

Analytical modeling:

Reductionist vs integrative (complex) process models Reduce error and calibration effort for complex models Phenotyping:

Methodologies to capture regulation of key processes Phenotype specifically, think systemically

E ager l y awa iti ng th e momen t o f t ru th :

Phenotyping done (w/ & w/o models): internatl. Network Genotyping awaits 600

K SNPs chip

Major loci/alleles? Co

location for different stresses/traits?

Physiologists, be ready for unexpected eye openers!

Merci