Replacement Sire Selection and Genetic Evaluation Strategies for Large Commercial Ranches Robert L....

Replacement Sire Selection and Genetic Evaluation Strategies for Large Commercial Ranches

Robert L. WeaberAssistant Professor,State Extension Specialist-Beef GeneticsUniversity of Missouri-Columbia

National Animal Breeding Seminar SeriesDecember 13, 2004

December 13, 2004 National Animal Breeding Seminar Series 2

Beef Cattle Selection in US

Driven by seedstock segment– Loose stratification of nucleus and multipliers– Supply yearling bulls to commercial producers– “Perceived” needs of commercial segment

Commercial segment– Buys bulls– Minimal selection applied following purchase– Almost no data collection, NO genetic evaluation


Serial Selection at Commercial Level

Can it work??– What type of operation?– What marketing structure? – What traits?– Performance and progeny test strategies? – Genetic evaluation strategies?

Pilot project– Bell Ranch, New Mexico– 4,500 cows, ~300,000 acres– Integrated seedstock unit, but very traditionally

managed


Bell Ranch Pilot Project


Bell Ranch Pilot Project Goals

Create additional selection opportunities– Serial selection (performance and progeny

testing)– Overcome obstacles

• Animal identification• Data collection• Pedigree construction• Genetic evaluation

Demonstrate approach Investigate efficacy of selection strategy


Research Objectives Simulate the serial selection strategy used in

Bell Ranch Pilot Project to investigate the economic returns.

Evaluate two genetic evaluation systems that incorporate information from DNA genotype derived pedigrees

Assess the value of sorting commercial bulls into breeding groups that optimize the probability of single sire paternity assignments

Serial Selection Spring and Fall HerdsSensitivity Analysis Results


Research Question:

Should large commercial ranches consider progeny testing of herd sire replacements as an alternative to performance testing?


Proposed ProgenyTest Protocol

Partition herd– Commercial & progeny test cows– Progeny test cows all same age

Large multi-sire breeding pasture Assign paternity via DNA genotype analysis Progeny test must minimize costs and

operate with minimal management intrusion.


Materials and Methods Simulation in Matlab

– Large commercial ranch with(out) fall herd Optimization of bulls progeny tested,

selected and calves tested per sire Selection differentials for performance tested,

progeny tested and selected bulls– Monte Carlo simulation (200 replicates/scenario)

• True ERT breeding values for bulls• Bull and progeny group phenotypes

– All records evaluated in RAM


Materials and Methods

Economics– Lifetime production of selected progeny

test bull vs. performance tested bull– Costs of progeny test

• DNA genotyping, calf ID, data processing, etc.• Interest charges

– Expected returns of progeny test system vs. performance test system

– Risk analysis of selected optimization


totalcows = 2000 markexclr = 0.97er = 0.04 revlag = 1.00

r = 0.90 intrate = 0.07ptcows = 300 rounds = 200

origsires = 60 fallcowind = 1fracorig = 0.50 totalsprcows = 2000

pt2sel = 1 sprhfrrr = 0.9ptmatings = 1 sprfirstcalvrr = 0.7

unselctdmtgs = 1 sprsecondcalvrr = 0.8selctdmtgs = 6 sprcowrr = 0.9heritability = 0.30 sprcullage = 10

vara = 900 first2fall = 2calfprice = 1 last2fall = 5

cost = 20 fallrr = 0.9prodratio = 0.90 fallcullage = 10

minptacc = 0.10 selctb4fall = 1

Progeny Test Evaluator Parameter Import FileCornell Commercial Ranch Program


Optimization Results

Spring Herd

Base Group: 60

Perf. Test Bulls: 15

Prog. Test Bulls: 16

Selected Bulls: 12

Calves per Sire: 8

System Comp: $ -728.80

Profit Frequency: 45.8%

Spring & Fall Herd

Base Group: 60

Perf. Test Bulls: 15

Prog. Test Bulls: 16

Selected Bulls: 12

Calves per Sire: 16

System Comp: $ 4,474.15

Profit Frequency: 62.3%


Sensitivity Analysis

Test optimization for effect of changing one parameter.

Tested:– Heritability– Additive Genetic Variance– Exposure Rate– Progeny Test Costs per Calf– Value of Unit of Production– Exclusion Rate


0 0.5 18

10

12

14

16

18Progeny Test Optimization Animal Counts

Heritability

Nu

mb

er o

f A

nim

als

in G

rou

p PTPFSPTCalves

0 0.2 0.4 0.6 0.80

1

2

3

4

5

6

7Response to Selection

Heritability

Pro

gen

y P

erfo

rman

ce I

ncr

ease

(kg

.)

PTPFSPT

0 0.1 0.2 0.3 0.4 0.5-200

0

200

400

600

800Net Advantage of Progeny Tested Sire

Heritability

Do

llar

s ($

)

0 0.1 0.2 0.3 0.4 0.5-1000

0

1000

2000

3000

4000

5000Progeny vs. Performance Testing System Comparison

Heritability

Do

llar

s ($

)


0.04 0.06 0.085

10

15

20

25


Exposure Rate

Nu

mb

er o

f A

nim

als

in G

rou

p PTPFSPTCalves

0.04 0.05 0.06 0.07 0.081.5

2

2.5

3

3.5

4

4.5Response to Selection

Exposure Rate

Pro

gen

y P

erfo

rman

ce I

ncr

ease

(kg

.)

PTPFSPT

0.04 0.05 0.06 0.07 0.080

100

200

300

400

500

600


Exposure Rate

Do

llar

s ($

)

0.04 0.05 0.06 0.07 0.08-2000

-1000

0

1000

2000

3000

4000


Exposure Rate

Do

llar

s ($

)


0 10 20 3012

13

14

15

16

17


Progeny Test Cost per Calf

Nu

mb

er o

f A

nim

als

in G

rou

p PTPFSPTCalves

0 10 20 302

2.5

3

3.5

4



Pro

gen

y P

erfo

rman

ce I

ncr

ease

(kg

.)

PTPFSPT

5 10 15 20 25 30300

400

500

600

700

800

900



Do

llar

s ($

)

5 10 15 20 25 300

2000

4000

6000

8000



Do

llar

s ($

)


0.9 0.95 110

12

14

16

18


Exclusion Rate

Nu

mb

er o

f A

nim

als

in G

rou

p PTPFSPTCalves

0.9 0.95 12

2.5

3

3.5

4

4.5

5


Exclusion Rate

Pro

gen

y P

erfo

rman

ce I

ncr

ease

(kg

.)

PTPFSPT

0.9 0.92 0.94 0.96 0.98 10

200

400

600

800

1000


Exclusion Rate

Do

llar

s ($

)

0.9 0.92 0.94 0.96 0.98 1-2000

0

2000

4000

6000

8000


Exclusion Rate

Do

llar

s ($

)


Serial Selection Simulation Results and Conclusions

Simulation of serial selection is a useful managerial decision aid given wide range inputs and variation of economic returns.

Serial selection valuable tool for identifying superior genetics to be returned to seedstock unit to change genetic trend.


Effect of performance and progeny testing on mean genetic merit of selected

replacement bulls.

Time

Mean Genetic Merit

Genetic Trend

Performance Tested

Progeny Tested

All Bulls


Serial Selection Simulation Results and Conclusions

Herds likely to benefit from serial selection systems:– Use bulls for two breeding seasons/year– Select for moderately heritable trait of economic

importance • Simulated weaning wt. observed on animal and progeny• Follow up study: carcass traits

– Can reduce test costs Genetic evaluation system that includes data

on all progeny via paternity prob. from genotype analysis would be useful.

Genetic Evaluation Strategies to Use Paternity Probabilities and Effects of Sorting Sires to Breeding Groups by Genotype


Objectives

Investigate the effect of the incorporation of paternity probabilities on the genetic evaluation of sires– Metrics:

• Correlation of true and predicted progeny differences

• Selection differentials of selected sires

Does sorting bulls to breeding groups improve evaluation?


Use of Paternity Probabilities

Dr. Quaas methodology for computation– Likelihood based approach

Incorporation into Genetic Evaluation– Adaptation of Henderson’s Average

Numerator relationship method– Monte Carlo method


Materials and Methods Simulation of Sire Genotypes and true

progeny differences– 200 replicates of 20 sires– Divided into 2 breeding groups

• Assigned at random or sorted by genotype– 6-15 progeny per sire – 12 poly-allelic markers 2-6 alleles per marker– Bi-allelic panel with 4-28 loci– Simulate co-dominant inheritance of calf genotype

and calf phenotypes (WW) Records evaluated in Sire Model

– Sires considered unrelated (A=I)


Data Structure Focused on structure of Z (incidence matrix

relating calf records to sires) True Pedigree from Simulation

– Typical Z incidence matrix Average Z method

– Substitute matrix of probabilities for typical Z Monte Carlo Z method

– Generate typical incidence matrix Z in the proportions suggested by paternity probabilities

EPD Sets: True pedigree, Sorted, Random


Correlation of true progeny difference and estimated progeny difference (EPD)

Average Z methodology Monte Carlo Z methodology

Progeny per sire Progeny per sire

Correlation 6 15 6 15

2 alleles r(u,u^true) 0.4869 0.6803 0.4869 0.6803

r(u,u^sorted) 0.3902 0.5664 0.3036 0.4583

r(u,u^random) 0.3743 0.5615 0.2891 0.4438

4 alleles r(u,u^true) 0.4959 0.6858 0.4959 0.6858

r(u,u^sorted) 0.4926 0.6832 0.4895 0.6801

r(u,u^random) 0.4901 0.6803 0.4856 0.6758

6 alleles r(u,u^true) 0.5053 0.6684 0.5053 0.6684

r(u,u^sorted) 0.5052 0.6682 0.5051 0.6681

r(u,u^random) 0.5047 0.6693 0.5045 0.6690


Progeny difference selection differentials (kg.) for the best 5 of 20 sires selected

Average Z methodology Monte Carlo Z methodology

Progeny per sire Progeny per sire

Rank Criterion 6 15 6 15

2 alleles True PD 16.45 16.45 16.45 16.45

EPD true 8.10 11.73 8.10 11.73

EPD sorted 6.36 9.59 4.98 7.77

EPD rand 6.24 9.45 4.78 7.47

4 alleles True PD 16.45 16.45 16.45 16.45

EPD true 8.46 11.45 8.46 11.45

EPD sorted 8.40 11.36 8.41 11.37

EPD rand 8.39 11.61 8.44 11.40

6 alleles True PD 16.45 16.45 16.45 16.45

EPD true 8.32 11.05 8.33 11.05

EPD sorted 8.35 11.03 8.32 11.05

EPD rand 8.45 11.26 8.44 11.22


SNP Marker Panel Exclusion Rates for Various Numbers of Loci

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

4 8 12 16 20 24 28 32 36 40 44 48 52 56 60

Number of Marked Loci

Pane

l Exc

lusi

on R

ate

0.50

0.40

0.30

0.20

0.10

0.05


Relationship Between Panel Exclusion Rate and Correlation Between True BV and Estimates of Breeding Value

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

4 8 12 16 20 24 28

Number of SNP Loci

Corr

elat

ion

0.41 0.66 0.80 0.88 0.93 0.96 0.98

Exclusion Rate

r(u,u^t)

Avg Z r(u,u^s)

Avg Z r(u,u^r)

MC Z r(u,u^s)

MC Z r(u,u^r)


Results and Conclusions Inclusion of information from DNA genotype derived

pedigrees can produce useful genetic evaluations and reliable selection decisions.

Sorting sires to breeding groups only marginally useful– Better at low exclusion rates– Maybe important when related sires considered (topic for

follow-up research) Average Z methodology generally produced higher

accuracies of prediction and larger selection differentials than did the Monte Carlo Z method of incorporating paternity probabilities.


Results and Conclusions

Pedigrees do not need to be fully resolved to be useful in genetic evaluation.

Marker panels with exclusion rates of ~0.90 and greater produced adequate pedigree resolution for useful genetic evaluations.

Small panels of SNP markers may provide a low-cost genotyping method that will make serial selection more profitable and available to more commercial producers.


Acknowledgements

Dr. John Pollak Dr. Dick Quaas (SireProb) J. P. Pollak (SireSort) Keith and Bonnie Long,

– Bell Ranch, NM

National Beef Cattle Evaluation Consortium

Replacement Sire Selection and Genetic Evaluation Strategies for Large Commercial Ranches Robert L....

Documents

Transcript of Replacement Sire Selection and Genetic Evaluation Strategies for Large Commercial Ranches Robert L....