Dr. Scott Sebastian, Research Fellow, Pioneer Hi-Bred International

11-2-09 Plant Breeding Seminar at University of California Davis

Accelerated Yield TechnologyTM

Context-Specific MAS for Grain Yield

Pioneer Soybean Breeding

3

Yield: Genetic Gain vs. Precision

Mean yield gain per year: ~ 1%

Precision in our best trials: +/- 5%

*courtesy of James Specht:

Crop Science 39:1560-1570

USA Soybean Yield Trends (1972-2003)

USA Trend: y = +0.412x - 785 R2 = 0.678

15

20

25

30

35

40

45

50

55

1970 1975 1980 1985 1990 1995 2000 2005Production Year

See

d Y

ield

(b

u/a

c)

4

Soybean Yield Map (one inbred) typical yield range: 30 to 70 bu/a

depending on position in the field

5

Corn Yield Map (one hybrid) yield range: 109 to 243 bu/a

depending on position in the field

6

The paradigm for mapping additive traits

Mapping yield QTL as an additive trait

Do we need a new paradigm for yield?

Context-Specific Mapping

Breeding Bias and genomic hotspots

AYT: a combination of many tools

Outline

7

Simple Trait Mappinge.g. SCN Resistance in Soybean

Resistant Parent x Susceptible Parent

R R R R S S S S

good correlation phenotype: genotype

Phenotype

Genotype

poor correlation phenotype: genotype

putative QTL hit

segregating progeny

8

0.03.5

14.723.027.728.028.129.030.931.132.746.564.771.474.993.294.295.295.597.8

101.6102.3

0.02.15.39.1

28.435.0

51.5

100.1105.2108.8109.8110.9115.9116.6116.7119.6125.4128.4128.9129.9145.6154.1162.0165.7

0.0

22.028.332.533.036.546.457.969.873.878.180.981.982.984.285.989.795.196.4

102.6

125.7132.2

0.06.0

11.917.8

34.951.555.257.065.667.771.772.172.572.973.278.887.691.197.9

121.0

0.0

9.0

65.173.374.274.475.576.280.684.885.490.1

120.1123.8

135.6

0.0

26.630.538.044.7

56.5

82.2

112.2113.4115.5117.8121.3122.0126.2128.2

151.9157.9

0.0

11.212.0

50.255.056.458.358.461.963.564.365.265.769.870.771.873.882.5

120.9

0.06.7

26.637.240.043.946.6

59.672.674.874.975.776.187.2

100.9

116.4

140.0

0.03.2

16.8

39.3

53.9

79.280.284.685.787.988.089.289.8

105.5113.6115.0124.3129.0133.9

0.03.7

12.918.219.330.332.132.334.235.841.743.143.644.945.145.447.556.356.764.271.3

1.93.03.43.64.05.4

15.320.6

50.2

70.671.472.573.074.377.778.185.391.9

102.1117.6119.2124.6130.6135.1

151.0

5.0

6.612.212.723.123.927.543.848.949.950.552.953.456.056.562.268.869.980.487.194.496.6

100.0102.8107.1116.8

0.00.68.5

27.6

38.9

46.9

58.968.569.172.285.886.591.193.7

124.0

0.0

20.328.031.531.934.035.3

50.1

65.6

77.882.8

99.8

112.7113.4

125.2

0.012.315.724.125.526.127.829.732.136.737.838.239.841.242.543.152.771.978.889.891.0

0.014.421.730.341.542.743.344.0

46.2

46.449.549.650.952.978.678.7

104.8

117.0

0.08.0

11.1

27.930.630.933.736.138.256.159.564.766.570.2

106.4107.2112.3115.1

0.05.07.8

18.6

33.535.9

56.359.962.167.073.975.676.477.287.195.4

107.7111.1112.8

133.8140.7142.2

0.0

26.127.129.431.834.534.636.937.438.038.140.853.270.672.675.976.584.692.6

116.7

0.05.49.5

17.320.4

39.842.343.649.752.153.754.255.155.856.356.957.068.471.182.193.495.4

100.4106.0118.1119.5135.1146.4

QTL detected in Population 1

PR

P1

P1

P2

9

Population 1

Parent1 (Resistant) x Parent2 (susceptible)

‘Major QTL’

‘Minor QTL’

Disease QTL detected within a specific population

P1

P1 P2

10

Population 1

RES x SUS

‘Validation’ of QTL Across Populations

Major ‘additive’ gene

These QTL did not ‘validate’ across populations. Does that mean they are not real ?

Population 2

RES x SUS

Population 3

RES x SUS

Chromosome G position 3

11

0

.

20

.

40

.

60

.

80

.

100

.

120

.

Map PositionChromosome G

A validated SCN resistance gene ‘Rhg1’

Rhg1

But what is the effect of Rhg1 on yield?

12

Effect of a Rhg1 on Yield

Global conclusion: Rhg1 does not affect yield.

Reality: the effect of Rhg1 on yield can be positive, neutral, or negative depending on the population.

Trait gene IBD

Effect of Rhg1 on disease

Effect on Yield (bu/a)

Statistical Signif

Rhg1 93B86 YB32K01 R +4.0 **

Rhg1 93B86 EX36Y01 R +1.9 *

Rhg1 93B86 92B52 R +1.2 ns

Rhg1 93B86 XB23Y02 R 0.0 ns

Rhg1 93B15 92B74 R -0.2 ns

Rhg1 93B15 ST2870 R -1.9 *

Rhg1 93B15 ST3630 R -6.3 **

Rhg1 across all across all R -0.2 ns

Population Parent 1 x Parent 2

13

0

.

20

.

40

.

60

.

80

.

100

.

120

.

Chromosome G

Why do yield effects of a QTL differ across populations?

Rhg1

Yield Effect

Yield effects are not distinguishable as single genes.

At best, a yield QTL can be assumed as the net effect of an

entire region within a given population.

Direction and magnitude of effect can change dramatically

with both population and environment (the context)

14

Attempts to Map Yield QTLin the old paradigm

15

Population1 Population2

Population3

Attempts to ‘validate’ Yield QTL

Many QTL found, NONE have validated across all populations.

16

Do we need a different

paradigm for mapping Yield?

17

Population1 Population2

Population3

What if ?

These QTL are valid for

Population 1

These QTL are valid for

Population 2

These QTL are valid for

Population 3

18

Population1

How valid are the Yield QTL within a given context?

QTL are only as valid as the data used to detect them !

More progeny + more environments = more confidence

Context-Specific Mapping

19

Implications for MAS ina breeding program

20

Development of One Product (before AYT)

Hundreds of Crosses (Parent1 x Parent2)

MAS for simple traits

Yield Testing

20,000 lines x 1 rep

5,000 lines x 2 reps

500 lines x 6 reps

20 lines x 25 reps

4 lines x 50 reps

1 product (better than parents?)

Year0

Year1

Year2

Year3 R1

Year4 R2

Year5 R3

Year6 R4

Year7 R5

inbreeding

Many choices but terrible precision

error is ~ +/- 30% (15 bu/a)

Few choices but better precision

error ~ +/- 5% (2 to 3 bu/a)

21

First Yield Screen: Progeny Row Yield Test

~ 85% of plot-to-plot variation is not heritable

22

AA

aaAA

aaAA

aa AA AA aa

AAaa

AAAA

aa

aa aa

aaAA AA

AA

aa

AYT: markers as ‘heritable covariates’

23

bb

BB

bb

BB

bb

BB

More marker coverage = more power to detect yield QTL

Large populations, multiple environments = more power

bb bb BB

bbBB

bb

bbBB

BB BB

BB

bb

bb

bbBB

24

AYT analysis can be simple: AA vs. aa

… or more sophisticated

Yield (predicted) = Mean + 2xAA + 4xbb + 2xDD + …. + epistasis …

QTL Favorable Alleles Magnitude location P1 alleles P2 alleles

Region A: AA > aa 2 bu/a

Region B: BB < bb 4 bu/a

Region C: CC = cc 0

Region D: DD > dd 2 bu/a

Region E: EE = ee 0

25

Select winners by Target Genotype

AA bb DD …

26

Product Development (before AYT)

Hundreds of Crosses F1

F2

F3

Forward selection for simple traits

Yield Testing

20,000 lines x 1 rep

5,000 lines x 2 reps

500 lines x 6 reps

20 lines x 25 reps

4 lines x 50 reps

1 product

Year0

Year1

Year2

Year3

Year4

Year5

Year6

Resources

20,000 micro plots

10,000 small plots

3,000 med plots

500 large plots

200 large plots

34,000 plots + 6 years

27

Product Development with AYT

Only the Best Crosses F1

F2

F3

Forward Selection for (simple traits)

Context-Specific MAS for Yield

Much better selection precision

Advance only the most promising genotypes

Fewer lines = better characterization in fewer years

Better Products, Faster to Market

Year0

Year1

Year2

Year3

Year4

28

What about the cost of genotyping?

29

Genotyping Efficiency

Are some genomic regions yield hotspots?

Can this reduce genotyping costs?

Can this improve QTL detection rate?

30

‘Breeding Bias’aka ‘Genetic Hitchhiking’ aka ‘Selection Sweep’

1995: US Patent 5,437,69. Sebastian, Hanafey, Tingey (soy example)

1998: US Patent 5,746,023. Hanafey, Sebastian, Tingey (corn example)

2004: Crop Science 44:436-442. Smalley, Fehr, Cianzio, Han, Sebastian, Streit

2006: Maydica 51: 293-300 Feng, Sebastian, Smith, Cooper.

Multiple lines of evidence

Very powerful tool

31

Ancestral Population

Elite Population

60+ years of recurrent selection for

Yield

History of Soybean

32

Yield-associated region

Marker: genetic hitchhiker

33

Ancestral Population

Elite Population

60+ years of recurrent selection for

Yield

Loci with evidence of selection

Reliable measure of:

1) which genomic regions were most important over time

2) response to the ‘average environment’

implicitly leverages a century of breeding progress!

change in allele frequency

34

5.1 5.714.617.018.0

19.1

27.128.548.2

69.975.383.286.4

87.3

96.4

A1

0.0 2.0 5.0 8.619.320.023.333.2

50.0

73.578.3

89.993.796.2

108.7

119.6123.4132.4135.1136.0138.2

154.7

161.8

173.5175.2

184.0

A2

22.526.7

34.939.045.056.668.171.673.374.174.876.480.085.091.992.1

117.3120.0

B1

All Markers on First 3 Chromosomes

35

A1

A2

B1

Regions of Breeding Bias

36

Breeding Bias hotspots across the entire genome

0.03.5

14.723.027.728.028.129.030.931.132.746.564.771.474.993.294.295.295.597.8

101.6102.3

A1

0.02.15.39.1

28.435.0

51.5

100.1105.2108.8109.8110.9115.9116.6116.7119.6125.4128.4128.9129.9145.6154.1162.0165.7

A2

0.0

22.028.332.533.036.546.457.969.873.878.180.981.982.984.285.989.795.196.4

102.6

125.7132.2

B1

0.06.0

11.917.8

34.951.555.257.065.667.771.772.172.572.973.278.887.691.197.9

121.0

B2

0.0

9.0

65.173.374.274.475.576.280.684.885.490.1

120.1123.8

135.6

C1

0.0

26.630.538.044.7

56.5

82.2

112.2113.4115.5117.8121.3122.0126.2128.2

151.9157.9

C2

0.0

11.212.0

50.255.056.458.358.461.963.564.365.265.769.870.771.873.882.5

120.9

D1a

0.06.7

26.637.240.043.946.6

59.672.674.874.975.776.187.2

100.9

116.4

140.0

D1b

0.03.2

16.8

39.3

53.9

79.280.284.685.787.988.089.289.8

105.5113.6115.0124.3129.0133.9

D20.03.7

12.918.219.330.332.132.334.235.841.743.143.644.945.145.447.556.356.764.271.3

E

0.01.93.03.43.64.05.4

15.320.6

50.2

70.671.472.573.074.377.778.185.391.9

102.1117.6119.2124.6130.6135.1

151.0

F0.03.3

5.0

6.612.212.723.123.927.543.848.949.950.552.953.456.056.562.268.869.980.487.194.496.6

100.0102.8107.1116.8

G

0.00.68.5

27.6

38.9

46.9

58.968.569.172.285.886.591.193.7

124.0

H

0.0

20.328.031.531.934.035.3

50.1

65.6

77.882.8

99.8

112.7113.4

125.2

I

0.012.315.724.125.526.127.829.732.136.737.838.239.841.242.543.152.771.978.889.891.0

J

0.014.421.730.341.542.743.344.0

46.2

46.449.549.650.952.978.678.7

104.8

117.0

K

0.08.0

11.1

27.930.630.933.736.138.256.159.564.766.570.2

106.4107.2112.3115.1

L

0.05.07.8

18.6

33.535.9

56.359.962.167.073.975.676.477.287.195.4

107.7111.1112.8

133.8140.7142.2

M

0.0

26.127.129.431.834.534.636.937.438.038.140.853.270.672.675.976.584.692.6

116.7

N

0.05.49.5

17.320.4

39.842.343.649.752.153.754.255.155.856.356.957.068.471.182.193.495.4

100.4106.0118.1119.5135.1146.4

O

= Yield Loci= SCN Loci= BSR Loci= Rps Loci

37

Hotspots segregating in a given cross

0.03.5

14.723.027.728.028.129.030.931.132.746.564.771.474.993.294.295.295.597.8

101.6102.3

A1

0.02.15.39.1

28.435.0

51.5

100.1105.2108.8109.8110.9115.9116.6116.7119.6125.4128.4128.9129.9145.6154.1162.0165.7

A2

0.0

22.028.332.533.036.546.457.969.873.878.180.981.982.984.285.989.795.196.4

102.6

125.7132.2

B1

0.06.0

11.917.8

34.951.555.257.065.667.771.772.172.572.973.278.887.691.197.9

121.0

B2

0.0

9.0

65.173.374.274.475.576.280.684.885.490.1

120.1123.8

135.6

C1

0.0

26.630.538.044.7

56.5

82.2

112.2113.4115.5117.8121.3122.0126.2128.2

151.9157.9

C2

0.0

11.212.0

50.255.056.458.358.461.963.564.365.265.769.870.771.873.882.5

120.9

D1a

0.06.7

26.637.240.043.946.6

59.672.674.874.975.776.187.2

100.9

116.4

140.0

D1b

0.03.2

16.8

39.3

53.9

79.280.284.685.787.988.089.289.8

105.5113.6115.0124.3129.0133.9

D20.03.7

12.918.219.330.332.132.334.235.841.743.143.644.945.145.447.556.356.764.271.3

E

0.01.93.03.43.64.05.4

15.320.6

50.2

70.671.472.573.074.377.778.185.391.9

102.1117.6119.2124.6130.6135.1

151.0

F0.03.3

5.0

6.612.212.723.123.927.543.848.949.950.552.953.456.056.562.268.869.980.487.194.496.6

100.0102.8107.1116.8

G

0.00.68.5

27.6

38.9

46.9

58.968.569.172.285.886.591.193.7

124.0

H

0.0

20.328.031.531.934.035.3

50.1

65.6

77.882.8

99.8

112.7113.4

125.2

I

0.012.315.724.125.526.127.829.732.136.737.838.239.841.242.543.152.771.978.889.891.0

J

0.014.421.730.341.542.743.344.0

46.2

46.449.549.650.952.978.678.7

104.8

117.0

K

0.08.0

11.1

27.930.630.933.736.138.256.159.564.766.570.2

106.4107.2112.3115.1

L

0.05.07.8

18.6

33.535.9

56.359.962.167.073.975.676.477.287.195.4

107.7111.1112.8

133.8140.7142.2

M

0.0

26.127.129.431.834.534.636.937.438.038.140.853.270.672.675.976.584.692.6

116.7

N

0.05.49.5

17.320.4

39.842.343.649.752.153.754.255.155.856.356.957.068.471.182.193.495.4

100.4106.0118.1119.5135.1146.4

O

A a

B b

C c

D d

E e

F f

G g

J jH h

I iK k

L l

R r

T t

S sV v

U u W wM m

N nO o

P p

Q q

38

MAS for simple traits across populations

Breeding Bias & other tools to find hotspots

Context-Specific MAS for yield within each pop

Accelerated Yield TechnologyTM

a combination of many tools

39

USA Soybean Yield Trends (1972-2003)

USA Trend: y = +0.412x - 785 R2 = 0.678

15

20

25

30

35

40

45

50

55

1970 1975 1980 1985 1990 1995 2000 2005Production Year

See

d Y

ield

(b

u/a

c)

Our Goal: Double the Rate of Genetic Gain

*courtesy of James Specht:

Crop Science 39:1560-1570

Thank You!