effects of physiological gxe on selection · water productivity ... • timing (of everything !) 15...
TRANSCRIPT
Interpreting effects of physiological GxE on marker and genomic selection
PLANT INDUSTRY
Scott ChapmanSenior Principal Research Scientist, CSIRO Plant IndustryAdjunct Professor, The University of Queensland
7 Sep 2012 EUCARPIA meeting
Incorporating genetics of traits into physiological models – QTL networks in maize
• Chenu et al PC&E 2009• Chenu et al Genetics 2009• van Eeuwijk et al COPB 2010
2 | Interpreting effects of physiological GxE | Scott Chapman
Understanding and breeding on GxE landscapes
Podlich and Cooper (1998) Bioinform.Chapman et al (2002) AJARChapman et al (2003) Agron. J. Wang et al (2004) Crop Sci.Cooper et al (2005) AJARHammer et al (2005) AJARHammer et al (2006) Trends in Plant Sci.Wang et al (2007) Crop Sci.Chapman (2008) EuphyticaWang et al (2009) TAG
Interpreting effects of physiological GxE | Scott Chapman
http://www.uq.edu.au/lcafs/qugene/
3 |
Gene to Phenotype Modelling‐ value of physiological knowledge (but with fake QTL !)
Cycle of selection0 2 4 6 8 10 12
Yiel
d in
TPE
(kg
ha-1
)
3800
4000
4200
4400
4600
4800
5000
5200
Marker selectionWeighted marker selectionPhysiologically weighted marker selection
G PUnexplainedExplained
Fullydescribed
Contextdependent
Chapman et al 2003; Hammer et al 2005 AJAR
4 | Interpreting effects of physiological GxE | Scott Chapman
Sorghum production and breeding in Australian dryland environments
Interpreting effects of physiological GxE | Scott Chapman5 |
Water productivity and G x E x M
• Sorghum ~ 1.5M ha, 3M t per year• 0.5 to 6 t/ha• Minimum tillage, rotation with wheat
• Water Productivity• Stored water + in‐season rainfall• Managing dynamics of– Canopy development– Root exploration
• Timing (of everything !)
Interpreting effects of physiological GxE | Scott Chapman6 |
Breeding challenges‐ complex E x M• Small(ish) market• State pre‐breeding program• 5 companies
• Large geographic region• 1000 km N‐S; 300km E‐W
• Extreme variability• Soils 80 to 300mm water capacity• In‐season rainfall 0 to 700mm
• Management• Sow Sep to Feb• Pop density of 25k to 75k /ha• Row spacing of 0.75 to 3m
Interpreting effects of physiological GxE | Scott Chapman7 |
Chapman et al 2000 AJAR
Breeding challenges‐ large GxE
• ca. 700 final stage trials• Within seasons• Error ca. 0.2 t2/ha• G ~ 0.05 to 0.5• GxL ~ 0.05 to 1.0 • GxL ~ 70% due to lack of genetic correlation
• G/(G+GxL) ~ 0.1 to 1• Across seasons• G ~ GxL ~ GxY ~ 0.2 • GxLxY ~ 10 times G• G/(G+GxE) ~ 0.2
Interpreting effects of physiological GxE | Scott Chapman8 |
Chapman et al 2000 AJAR
Dynamics of water supply/demand‐ how does this generate GxE?Genotype Environment
RootAngle: 30 RootAngle: 35 RootAngle: 40 RootAngle: 45 RootAngle: 50
-2.0-1.5-1.0-0.5+0
+0.5+1.0+1.5+2.0
-2.0-1.5-1.0-0.5+0
+0.5+1.0+1.5+2.0
-2.0-1.5-1.0-0.5+0
+0.5+1.0+1.5+2.0
-2.0-1.5-1.0-0.5+0
+0.5+1.0+1.5+2.0
-2.0-1.5-1.0-0.5+0
+0.5+1.0+1.5+2.0
MaxTR
ate: 0.6M
axTRate: 0.7
MaxTR
ate: 0.8M
axTRate: 0.9
MaxTR
ate: 1.0
130
137.
514
515
2.5
160
167.
517
518
2.5
190
130
137.
514
515
2.5
160
167.
517
518
2.5
190
130
137.
514
515
2.5
160
167.
517
518
2.5
190
130
137.
514
515
2.5
160
167.
517
518
2.5
190
130
137.
514
515
2.5
160
167.
517
518
2.5
190
TTEJ_INIT
Tille
rMod
Yield
2000
2500
3000
3500
Year=2001,Site=EMERALD_BE120,Pop=5,Skip=solid,Sowing=medium,SowWater=medium
DALBY_Vert180 EMERALD_BE120 GOONDIWINDI_GC120
2000
4000
6000
8000
10000
2000
4000
6000
8000
10000
2000
4000
6000
8000
10000
solidsingle
double
1995 2000 2005 2010 1995 2000 2005 2010 1995 2000 2005 2010Year
Yie
ld (k
g/ha
)
Population
5
7.5
9 | Interpreting effects of physiological GxE | Scott Chapman
Interpreting effects of physiological GxE on marker and genomic selection
1.A generic modelling framework• complex phenotypes as emergent properties of biological models
2.Characterising GEM landscapes• Simulating effects of traits on water productivity in sorghum
3.Exploring GEM landscapes• Breeding for adaptation, selecting for different environments or traits
Interpreting effects of physiological GxE | Scott Chapman10 |
1. APSIM ‐ A generic modelling framework
2. Characterising GEM landscapes
3. Exploring GEM landscapes with breeding
Interpreting effects of physiological GxE | Scott Chapman11 |
Drought patternCrop
System control
Soil
SWIM
ManagerReportClock
SoilWat
SoilNSoilPHSoilP
ResidueEconomicsFertiliz
Irrigate
Canopy Met
ErosionOther Crops
Maize
Sorghum
Legume
Wheat
New Module
Manure
Management
ENGINE
Weather
Transpiration
Evaporation
Uptake
Rainfall
Runoff
Infiltration
Drainage
Radiation
APSIM – integrate interaction of plant & environment
Interpreting effects of physiological GxE | Scott Chapman12 |
Predicts crop yield, given a physiological model driven by weather, soil inputs and parameters that drive traits and are related to gene network controls
Development Growth
Physiol Maturity
Initiation
Anthesis
Emergence
T, W&N
T, W&N
T, PP
Grain Yield
Grain Number Grain Size & N
BiomassRADN
TE T RUE Rint
vpd
kl LAISLNRoots k
TN
LNo
LNo
A A >A
APSIM ‐ A Deterministic Physiological Framework‐ daily timestep, 300+ state variables
Hammer et al JExpBot 2010
13 | Interpreting effects of physiological GxE | Scott Chapman
Biomass at Maturity
y = 0.8816x + 116.35R2 = 0.8649
0
500
1000
1500
2000
2500
0 500 1000 1500 2000 2500Observed
Pred
icte
d
Grain Yield at Maturity
y = 0.9081x + 78.134R2 = 0.7957
0
200
400
600
800
1000
0 200 400 600 800 1000Observed
Pred
icte
d
Phenology
40
60
80
100
120
140
160
40 60 80 100 120 140 160Observed
Pred
icte
d
Days to Flower
Days to Marturity
Biomass at Flowering
y = 0.7605x + 102.55R2 = 0.7274
0
200
400
600
800
1000
0 200 400 600 800 1000Observed
Pred
icte
d
Crop model validation
• Iterations of experiments & model design
• Aim to capture functional controls
14 | Interpreting effects of physiological GxE | Scott Chapman
Water productivity – supply and demand
• Water Demand• Maturity• Tillering• Leaf transpiration rate• Management
• Water Supply• Rainfall• Soil storage• Root exploration• Management
• Timing (of everything !)
Interpreting effects of physiological GxE | Scott Chapman15 |
Genotypic traits affecting water productivity
• Input traits• TTEJ_INIT – thermal time from end juvenile to FI at least 40 QTL• TillerMod – ‘tillering propensity’ 10 to 20 QTL• RootAngle – angle of nodal roots, affects root depth 5 to 10 QTL• MaxTRate – reduced transpiration with high VPD QTL unknown
• Output traits• Yield• Flowering date, leaf area, biomass, water use
16 | Interpreting effects of physiological GxE | Scott Chapman
Development Growth
Physiol Maturity
Initiation
Anthesis
Emergence
T, W&N
T, W&N
T, PP
Grain Yield
Grain Number Grain Size & N
BiomassRADN
TE T RUE Rint
vpd
kl LAISLNRoots k
TN
LNo
LNo
A A >A
Genotypic traits affecting water productivity
Hammer et al JExpBot 2010
TTEJ_INIT
TillerMod
MaxTRate
RootAngle
17 | Interpreting effects of physiological GxE | Scott Chapman
1. APSIM ‐ A generic modelling framework
2. Characterising GEM landscapes
3. Exploring GEM landscapes with breeding
Interpreting effects of physiological GxE | Scott Chapman18 |
GEM landscape for sorghum
• Genotype or Trait bins (9 x 9 x 5 x 5) = 3645• TTEJ_INIT (130 to 190 Cd – ca. 2‐3 weeks)• TillerMod (9 values – 0 to 4 tillers)• MaxTRate (0.6 to 1.0 mm/h)• RootAngle (30 to 50 deg from vertical – deep vs flat)
• Environment (3 x 110) = 330• Site/soil depth (Emerald 120 mm, Dalby 180, Goondiwindi 120)• Year (1890 to 2010)
• Management (3 x 3 x 3 x 3) = 81• Sowing time (early, medium, late)• Starting soil water (low, medium, high)• Density (2.5, 5.0, 7.5 plants/m2)• Row spacing (solid, skip, double skip)
• ~ 100 million phenotypes in 27k environments• ~ 42 days on one PC; 2 days on 10k processors
Interpreting effects of physiological GxE | Scott Chapman19 |
RootAngle: 30 RootAngle: 35 RootAngle: 40 RootAngle: 45 RootAngle: 50
-2.0-1.5-1.0-0.5+0
+0.5+1.0+1.5+2.0
-2.0-1.5-1.0-0.5+0
+0.5+1.0+1.5+2.0
-2.0-1.5-1.0-0.5+0
+0.5+1.0+1.5+2.0
-2.0-1.5-1.0-0.5+0
+0.5+1.0+1.5+2.0
-2.0-1.5-1.0-0.5+0
+0.5+1.0+1.5+2.0
MaxTR
ate: 0.6M
axTRate: 0.7
MaxTR
ate: 0.8M
axTRate: 0.9
MaxTR
ate: 1.0
130
137.
514
515
2.5
160
167.
517
518
2.5
190
130
137.
514
515
2.5
160
167.
517
518
2.5
190
130
137.
514
515
2.5
160
167.
517
518
2.5
190
130
137.
514
515
2.5
160
167.
517
518
2.5
190
130
137.
514
515
2.5
160
167.
517
518
2.5
190
TTEJ_INIT
Tille
rMod
Yield
2000
2500
3000
3500
Year=2001,Site=EMERALD_BE120,Pop=5,Skip=solid,Sowing=medium,SowWater=medium
DALBY_Vert180 EMERALD_BE120 GOONDIWINDI_GC120
2000
4000
6000
8000
10000
2000
4000
6000
8000
10000
2000
4000
6000
8000
10000
solidsingle
double
1995 2000 2005 2010 1995 2000 2005 2010 1995 2000 2005 2010Year
Yie
ld (k
g/ha
)
Population
5
7.5
Environment x Management effects ‐– site, year, soil water, population, row spacing
Interpreting effects of physiological GxE | Scott Chapman20 |
DALBY_Vert180 EMERALD_BE120 GOONDIWINDI_GC120
2000
4000
6000
8000
10000
2000
4000
6000
8000
10000
2000
4000
6000
8000
10000
solidsingle
double
1995 2000 2005 2010 1995 2000 2005 2010 1995 2000 2005 2010Year
Yie
ld (k
g/ha
)
Population
5
7.5
Genotype effects, e.g. early maturity– 1990 – 2010, Dalby
Interpreting effects of physiological GxE | Scott Chapman21 |
DALBY_Vert180
4000
5000
6000
7000
8000
9000
10000
1995 2000 2005 2010Year
Yie
ld (k
g/ha
) TTEJ_INIT130
160
190
Interpreting effects of physiological GxE | Scott Chapman22 |
DALBY_Vert180
4000
5000
6000
7000
8000
9000
10000
1995 2000 2005 2010Year
Yie
ld (k
g/ha
) TTEJ_INIT130
160
190
Chapman et al 2000 AJAR
Genotype effects, e.g. early maturity– 1990 – 2010, Dalby
Growing the whole breeding program(in one environment)
• Displays two traits• Yield as heatmap• Expand to four trait dimensions
Interpreting effects of physiological GxE | Scott Chapman23 |
Thermal time to floral initiation(increasing time to maturity)
Tilleringpropensity
Maxyield
Minyield
DALBY_Vert180
EMERALD_BE120
GOONDIWINDI_GC120
2000
4000
6000
8000
10000
2000
4000
6000
8000
10000
2000
4000
6000
8000
10000
1900 1920 1940 1960 1980 2000Year
Yie
ld (k
g/ha
)
TTEJ_INIT130
160
190
Growing the whole breeding programOne of 27 000 environments ‐ Emerald 2004
Interpreting effects of physiological GxE | Scott Chapman24 |
Growing the whole breeding programOne of 27 000 environments ‐ Emerald 2004
Interpreting effects of physiological GxE | Scott Chapman25 |
Statistical characteristics of landscapes
Interpreting effects of physiological GxE | Scott Chapman26 |
Emerald 2001‐2010
Interpreting effects of physiological GxE | Scott Chapman27 |
• Flips 2001, 2004, 2005, 2010
Scatter Plot Matrix
200130003500
3000
200025002000
200255006000
5500
450050004500
2003600070006000
45005500500
200460006500
6000
500055005000
200530003500
3000
200025002000
200670008000
7000
500060005000
20076000700080006000
400050006000000
20086000700080006000
400050006000000
200970008000
7000
500060005000
201055006000
5500
450050004500
Constrained landscape has similar Vg to ‘real world’ breeding population
• Variance for• all traits• maturity in central 30%• tillering in central 30%• Maturity and tillering constrained
Interpreting effects of physiological GxE | Scott Chapman28 |
2002 2004 2006 2008 2010
0.0
0.5
1.0
1.5
2.0
Year
Gen
otyp
ic v
aria
nce
Chapman et al 2000 AJAR
Scatter Plot Matrix
200130003500
3000
20002500000
2002550060005500
50005500
5000
2003600065006000
50005500000
2004600065006000
55006000
5500
2005250030002500
20002500000
200660006500
6000
500055005000
200760006500
6000
500055005000
200860006500
6000
500055005000
20096000650070006000
500055006000
000
2010550060005500
50005500
5000
Correlations between years‐ changes with trait physiology
Interpreting effects of physiological GxE | Scott Chapman29 |
0.0 0.2 0.4 0.6 0.8 1.0
0.0
0.2
0.4
0.6
0.8
1.0
0.0 0.2 0.4 0.6 0.8 1.0
0.0
0.2
0.4
0.6
0.8
1.0
TillerMod = 0.0 (midpoint) TillerMod = +2.0
‘yo‐yo’ effects in GxE
• System describes G effects on water use patterns• Water demand – maturity, tillering, transpiration rate• Water supply ‐ root angle/depth
• Has broad characteristics of ‘real’ landscape
• How can we build breeding strategies to explore landscapes• e.g. early‐generation screening of traits, later generation screening of yield; imputing responses in different environments
Interpreting effects of physiological GxE | Scott Chapman30 |
1. APSIM ‐ A generic modelling framework
2. Characterising GEM landscapes
3. Exploring GEM landscapes with breeding
Interpreting effects of physiological GxE | Scott Chapman31 |
Genotype
Trait genetics
APSIM
Manager
BiologicalModules
Surface Residue
EnvironmentalModules
ErosionB
ErosionA
Other N moduleorSoilN
CropC
CropB
CropA
PastureC
PastureB
PastureA
SwimorSoilwat
Economics Climate
APSIM
Simulate Crop Improvement
Strategies
Experiments –physiology and
genetics
Trait dissection and functional physiology
Cooper et al. 2002, In Silico Biol.
Phenotype
Software and Database Tools
A Research framework to capture physiological responses as part of breeding simulations
32 | Interpreting effects of physiological GxE | Scott Chapman
ENGINE
QUG
GESG-E System
POPGermplasm
POPGermplasm
QMPParameters
QU-LINE
QU-MASS
QU-PEDRRS
G-P Plugin
Output1
Output2
Output...
Outputn
QU-RRS
QU-MAS
QU-LINEQU-LINE
QU-RRS
QU-MAS
QU‐Gene
• Models populations of genotypes undergoing recombination, crossing and selection
• Engine to generate starting population and modules to simulate alternative breeding methods
• Podlich and Cooper 1998 Bioinformatics
Contact for licence:[email protected]
http://www.uq.edu.au/lcafs/qugene/
33 | Interpreting effects of physiological GxE | Scott Chapman
A simple experiment‐ GxE interactions under perfect marker selection
• Environments – 10 years, 1 location (Emerald)• Genetic trait architecture• Two chromosomes, eight QTL, four traits• Starting population of 100 lines with average favourable allele freq of 20%• H2 = 1.0 (perfect marker selection)
• Recurrent selection for 5 cycles (5 runs)• Selection among 50 F3s, 10% selection pressure
• Selection experiments• Yield in one environment (ET1 = Emerald 2001)• Yield in 10 environments (ET2 = Emerald 2001‐2010)• Selection by input or output traits
Interpreting effects of physiological GxE | Scott Chapman34 |
Development Growth
Physiol Maturity
Initiation
Anthesis
Emergence
T, W&N
T, W&N
T, PP
Grain Yield
Grain Number Grain Size & N
BiomassRADN
TE T RUE Rint
vpd
kl LAISLNRoots k
TN
LNo
LNo
A A >A
Generating GxE with four traits
Hammer et al JExpBot 2010
TTEJ_INIT
TillerMod
MaxTRate
RootAngle
35 | Interpreting effects of physiological GxE | Scott Chapman
2 chromosomes, 8 QTL, 4 interacting traits
Interpreting effects of physiological GxE | Scott Chapman36 |
• Input traits• TTEJ_INIT – thermal time from
end juvenile to FI 8 QTL• TillerMod – ‘tillering propensity’ 3
QTL• RootAngle – angle of nodal roots,
affects root depth 2 QTL• MaxTRate – reduced transpiration
with high VPD 4 QTL
• Output traits• Yield• Flowering date, leaf area,
biomass, water use
Change in trait fitness
• ET1 is negatively correlated with mean of 2001‐2010..
So…• Selection for yield in ET1 decreases yield in ETALL
Interpreting effects of physiological GxE | Scott Chapman37 |
Yield.ET1 Yield.ETALL LAIA.ET1
LAIA.ETALL TTEJ_INIT.ET1 TillerMod.ET1
MaxTrate.ET1 RootAngle.ET1
2000
3000
4000
5000
6000
2000
3000
4000
5000
6000
2000
3000
4000
5000
6000
1 2 3 4 5 6 1 2 3 4 5 6as.numeric(as.character(cycle))
Yie
ld
environmentET1
ET2
ETALL
Selection for yield‐ 5 cycles in one environment (2001)
• Severe drought environment
• Yield selection• Lowered MaxTRate(saved water)
• Increased RootAngle(explored more soil)
Interpreting effects of physiological GxE | Scott Chapman38 |
Selection for yield in 2001 ‘environment’ (ET1)Poor progress on 2001 vs 2001‐2010 landscapes‐ need more tillers !
Interpreting effects of physiological GxE | Scott Chapman39 |
Selection for yield 2001 (ET1) or 2001‐2010 (ET2)ET2 2001‐2010 selection increased tillers and root angle
Interpreting effects of physiological GxE | Scott Chapman40 |
Change in allele frequency
• 3 tillering QTL favoured at different rates depending on linkage with maturity and nearby traits
• Complex responses related to QTL model and relative favourability of traits at specific alleles
Interpreting effects of physiological GxE | Scott Chapman41 |
Yield.ET1 Yield.ETALL LAIA.ET1 LAIA.ETALLTEJ_INIT.ETTillerMod.ET1MaxTrate.ET ootAngle.ET
0.00.20.40.60.81.0
0.00.20.40.60.81.0
0.00.20.40.60.81.0
0.00.20.40.60.81.0
0.00.20.40.60.81.0
0.00.20.40.60.81.0
0.00.20.40.60.81.0
0.00.20.40.60.81.0
Q_N
AM
_1_10Q
_NA
M_1_53
Q_N
AM
_1_66Q_N
AM
_1_112Q_N
AM
_1_182Q_N
AM
_2_23Q_N
AM
_2_123Q_N
AM
_2_147
1 2 3 4 5 6 1 2 3 4 5 6 1 2 3 4 5 6 1 2 3 4 5 6 1 2 3 4 5 6 1 2 3 4 5 6 1 2 3 4 5 6 1 2 3 4 5 6as.numeric(as.character(cycle))
valu
e allele
1
2
Genotype
Trait genetics
APSIM
Manager
BiologicalModules
Surface Residue
EnvironmentalModules
ErosionB
ErosionA
Other N moduleorSoilN
CropC
CropB
CropA
PastureC
PastureB
PastureA
SwimorSoilwat
Economics Climate
APSIM
Simulate Crop Improvement
Strategies
Experiments –physiology and
genetics
Trait dissection and functional physiology
Cooper et al. 2002, In Silico Biol.
Phenotype
Software and Database Tools
A Research framework to capture physiological responses as part of breeding simulations
42 | Interpreting effects of physiological GxE | Scott Chapman
Representing Genotype‐Phenotype landscapes– alternative models
P = G_Fully‐described + G_Context‐dependent + G_Unexplained +
G PUnexplainedExplained
Fully described
ContextDependent
epistasis,pleiotropy,
GxE
(Cooper et al AJAR 2005) Int. Crop Science Congress, Brisbane 200443 | Interpreting effects of physiological GxE | Scott Chapman
Conclusions
• Deterministic model of crop growth• Predicts crop yield, given a physiological model driven by weather, soil inputs and parameters that drive traits and are related to gene network controls
• System generates ‘real‐world’ GxE signals• Similar variances and correlations for yield
• A platform for investigation of questions in• Systems biology of plant‐crop growth and development• Statistical analysis of gene‐to‐phenotype relationships• Role of GxE in marker and genomic selection
Interpreting effects of physiological GxE | Scott Chapman44 |
Acknowledgements
CSIRO• Bangyou Zheng, Adrian Hathorn
University of Queensland/QAAFI• Erik van Oosterom, Graeme Hammer, David Jordan, Karine Chenu, Vijaya Singh, Zonjian Yang
• Queensland Government• Greg McLean, Al Doherty, Emma Mace
• GRDC andGeneration Challenge Program
|Exploring complexity | Scott Chapman45 | Interpreting effects of physiological GxE | Scott Chapman