Genomics-assisted Hybrid Breeding in Cassava
-- Inbred Progenitor
Xiaofei Zhang, Randall Holley, Giovanny Eduardo Covarrubias Pazaran, 
Marlee Rose Labroo, Hernan Ceballos
May, 2022
Pain Point #1:  LOW efficiency in trait introgression
Inbred progenitor is essential 
P1 P2
3 6 to make change quickly.
7 2
4 4
6 7
Heterozygous 
2 5
parents P1 P2
1 1
1 3
2 1
5 1
1 2
2 1
Homozygous 
1 1
6 2 7 1 7 4 6 parents
1 1
5 1 2 3 2 2 3
1 1
7 5 6 4 3 6 2
2 4 4 7 6 7 4
1 7 1 5 4 3 5
1 1 1 1 1 1 1
4 6 3 6 1 5 1
2 2 2 1 1 2 1
3 3 5 2 5 1 7
2 2 1 1 2 2 1
1 2 2 2 2 1 1
1 1 1 1 1 1 1
1 1 1 1 1 1 1
1 1 1 1 1 1 1
Mix up everything
Targeted improvement
“ In a changing world, we 
must change quickly.
”
Upgraded TME419 with CBSD resistance
Upgraded KU50 with CMD resistance
New varieties are similar with the one farmers have been planting.
Pain Point #2:  SMALL between-family variance
Euphytica (2016) 210:79-92
Feasibility: Biology
Diploid
Cross-pollinated
Self-compatible
Inbreeding depression
Clonal propagation 
– no seed production system
Feasibility: Technology #1
Flower Inducing Technology
Photoperiod Extension induced 
early flowering by 2-3 months 
for progenitors with erect plant 
architecture.
Flower Inducing Technology
Pruning and Growth 
Regulators increased 
the seed number per 
plant up to >20.
Flowering at Similar Time
2022 April 15
Towards Speed Breeding
-- Implement Flower Inducing Technology
Two lots, each 0.5 Ha  
Feasibility: Technology #2
Double Haploid
Semi-inbred progenitors with purged genetic load make DH feasible.
Genomics-assisted Hybrid Breeding in Cassava
-- Inbred Progenitor
Understand inbreeding depression
Develop semi-inbred progenitors
Improve population using rapid cycling
Create heterotic groups
Hybrid Cassava Breeding – Inbreeding Depression
Selfing clones Variety Testcross Evaluate testcross Evaluate selfing 
Stake increase Pop. Improvement
(Inbred clones) Development (heterotic pattern and predict (inbreeding depression for 
(variety development) GCA) MAS and GS) (reduce inbreeding depression)
Year 1 Elite progenitors
30
S1 GH propagation*
30-50
Year 2
Selection S1 at TPE
30-50
Year 3
ID& S1 at TPE
Prediction 30-50
Year 4
Year 5
* Genomewide marker for prediction or association; solid arrow, germplasm delivery; broken arrow, information sharing; ID, inbreeding depression.
Hybrid Cassava Breeding – Semi-inbred Progenitors
Selfing clones Variety Testcross Evaluate testcross Evaluate selfing 
(Inbred clones) Stake increase Pop. Improvement
Development (heterotic pattern and predict (inbreeding depression for 
(variety development) (reduce inbreeding depression)
GCA) MAS and GS)
Year 1 Elite progenitors
30
S1 GH propagation*
30-50
S1
Year 2 30-50
Selection S1 at TPE
30-50
S1
30-5
Year 3 S2 GH propagation*
150-5
S2
30-5
Year 4 S3 GH propagation*
150-5
S3 paired cross
200-5
Year 5
Evaluation at TPE*
* Genomewide marker for prediction or association; solid arrow, germplasm delivery; broken arrow, information sharing; ID, inbreeding depression.
Hybrid Cassava Breeding – Population Improvement 
Selfing clones Variety Testcross Evaluate testcross Evaluate selfing 
Stake increase Pop. Improvement
(Inbred clones) Development (heterotic pattern and predict (inbreeding depression for 
(variety development) GCA) MAS and GS) (reduce inbreeding depression)
Year 1 Elite progenitors
30
S1 GH propagation*
30-50
Year 2
S1 at TPE Selection
30-50
Cross S1 within a 
family
150-20
Year 3
Increase & evaluate F1
150-20
Year 4
Year 5
Selection
Evaluation at TPE*
* Genomewide marker for prediction or association; solid arrow, germplasm delivery; broken arrow, information sharing; ID, inbreeding depression.
Hybrid Cassava Breeding – Heterotic Groups
Selfing clones Variety Testcross Evaluate testcross Evaluate selfing 
(Inbred clones) Stake increase Pop. Improvement
Development (heterotic pattern and predict (inbreeding depression for 
(variety development) (reduce inbreeding depression)
GCA) MAS and GS)
Year 1 Elite progenitors
30
S1 GH propagation*
30-50
S1
Year 2 30-50
Selection S1 at TPE
30-50
Testcross
150 x 6 testers
Year 3
Testcross clones*
900-10
Year 4
Heterotic Testcross clones at TPE Heterotic 
pattern& pattern& 
900-10
prediction prediction
Year 5
Selection
Evaluation at TPE*
* Genomewide marker for prediction or association; solid arrow, germplasm delivery; broken arrow, information sharing; ID, inbreeding depression.
Hybrid Cassava Breeding – Variety Development and Population Improvement
Selfing clones Variety Testcross Evaluate testcross Evaluate selfing 
(Inbred clones) Stake increase Pop. Improvement
Development (heterotic pattern and predict (inbreeding depression for 
(variety development) (reduce inbreeding depression)
GCA) MAS and GS)
Year 1 Elite progenitors
30
S1 GH propagation*
30-50
S1
Year 2 30-50
Selection
Selection S1 at TPE
30-50
S1
30-5 Testcross Cross S1 within a 
150 x 6 testers family
150-20
Year 3 S2 GH propagation*
150-5
S1 at TPE
Testcross clones* ID& 
Prediction 30-50 Increase & evaluate F1
900-10 150-20
S2
30-5
Year 4 S3 GH propagation*
150-5
Heterotic Testcross clones at TPE Heterotic 
pattern& pattern& 
900-10
prediction prediction
S3 paired cross
200-5
Year 5
Selection
Evaluation at TPE*
* Genomewide marker for prediction or association; solid arrow, germplasm delivery; broken arrow, information sharing; ID, inbreeding depression.
Cost Estimation of Cassava Breeding
Xiaofei Zhang & Randall Holley
April-22-2022
Cost components (years 1-10):
1) Breeding populations for major markets
2) Start of inbred and hybrid development
3) Double haploid development
4) Genetic studies
5) Building and transferring enhanced breeding populations
6) Breeding population maintenance
● 1.5 FTE for each market (=1.5 x 450,000)
● 20% for operation cost and 80% for personnel
● genomics-assisted recurrent selection
Annual Annual 
Item New model activities years 1-10 Cost Current / ongoing Cassava model years 1-10 Cost Comments / clarifications
● 20% for opera on cost and 80% for personnel
1.1 ~5 Breeding programs covering the 5 major markets at ~5 Breeding programs covering the 5 major markets at ● 1.5 FTE for each market (=1.5 x 450,000)
$675,000 each $  3,375,000 $675,000 each $    3,375,000 ● genomics-assisted recurrent selection
Costs for start up are much less than a full program because 
the pipeline materials coming out of it will be transferred to 
1.2
the local conventional program for advanced testing & product 
Start of inbred & hybrid development $      250,000 development
1.3
Double haploid tech development $      100,000 start later when we have improved selfing clones
Genetic studies to better understand genetic correlations Genetic studies to better understand genetic correlations Genetic studies without inbreds are more costly due to the 
1.4
among key traits and the number of genes / QTL among key traits and the number of genes / QTL controlling need for higher entry # to generate the same quality of 
controlling them (integrated in #1.1 and #1.2) them (structured populations for genetic mapping) $        100,000 information
Cost of building and transferring enhanced breeding Extra cost as stacking genes in heterozygous populations 
1.5
populations to others $        50,000 $        100,000 requires higher entry # and more cycles of selection
●  $75,000 for maintain the historical popula ons;
●  the other cost for cleaning and tes ng virus of new breeding 
1.6 materials
Maintaining cell cultures and grow outs to support ● maintain S3 or inbred seeds in storage room in the new 
Breeding population management costs $100,000 distributions $150,000 model, rather than tissue culture
Total annual costs $  3,875,000 $    3,725,000
Cost components:
1) Breeding populations for major markets
2) Start of inbred and hybrid development
3) Double haploid development
4) Genetic studies
5) Building and transferring enhanced breeding populations
6) Breeding population maintenance
Hybrid Cassava Breeding – Variety Development and Population Improvement
Selfing clones Variety Testcross Evaluate testcross Evaluate selfing 
(Inbred clones) Stake increase Pop. Improvement
Development (heterotic pattern and predict (inbreeding depression for 
(variety development) (reduce inbreeding depression)
GCA) MAS and GS)
Year 1 Elite progenitors
30
S1 GH propagation*
30-50
S1
Year 2 30-50
Selection
Selection S1 at TPE
30-50
S1
30-5 Testcross Cross S1 within a 
150 x 6 testers family
150-20
Year 3 S2 GH propagation*
150-5
S1 at TPE
Testcross clones* ID& 
Prediction 30-50 Increase & evaluate F1
900-10 150-20
S2
30-5
Year 4 S3 GH propagation*
150-5
Heterotic Testcross clones at TPE Heterotic 
pattern& pattern& 
900-10
prediction prediction
S3 paired cross
200-5
Year 5
Selection
Evaluation at TPE*
* Genomewide marker for prediction or association; solid arrow, germplasm delivery; broken arrow, information sharing; ID, inbreeding depression.
Towards Rapid Breeding
-- Double Haploid
Semi-inbred progenitors with purged genetic load make DH feasible.
Annual Annual 
Item New model activities years 1-10 Cost Current / ongoing Cassava model years 1-10 Cost Comments / clarifications
● 20% for opera on cost and 80% for personnel
1.1 ~5 Breeding programs covering the 5 major markets at ~5 Breeding programs covering the 5 major markets at ● 1.5 FTE for each market (=1.5 x 450,000)
$675,000 each $  3,375,000 $675,000 each $    3,375,000 ● genomics-assisted recurrent selection
Costs for start up are much less than a full program because 
the pipeline materials coming out of it will be transferred to 
1.2
the local conventional program for advanced testing & product 
Start of inbred & hybrid development $      250,000 development
1.3
Double haploid tech development $      100,000 start later when we have improved selfing clones
Genetic studies to better understand genetic correlations Genetic studies to better understand genetic correlations Genetic studies without inbreds are more costly due to the 
1.4
among key traits and the number of genes / QTL among key traits and the number of genes / QTL controlling need for higher entry # to generate the same quality of 
controlling them (integrated in #1.1 and #1.2) them (structured populations for genetic mapping) $        100,000 information
Cost of building and transferring enhanced breeding Extra cost as stacking genes in heterozygous populations 
1.5
populations to others $        50,000 $        100,000 requires higher entry # and more cycles of selection
●  $75,000 for maintain the historical popula ons;
●  the other cost for cleaning and tes ng virus of new breeding 
1.6 materials
Maintaining cell cultures and grow outs to support ● maintain S3 or inbred seeds in storage room in the new 
Breeding population management costs $100,000 distributions $150,000 model, rather than tissue culture
Total annual costs $  3,875,000 $    3,725,000
Cost components:
1) Breeding populations for major markets
2) Start of inbred and hybrid development
3) Double haploid development
4) Genetic studies
5) Building and transferring enhanced breeding populations
6) Breeding population maintenance
Annual Annual 
Item New model activities years 1-10 Cost Current / ongoing Cassava model years 1-10 Cost Comments / clarifications
● 20% for opera on cost and 80% for personnel
1.1 ~5 Breeding programs covering the 5 major markets at ~5 Breeding programs covering the 5 major markets at ● 1.5 FTE for each market (=1.5 x 450,000)
$675,000 each $  3,375,000 $675,000 each $    3,375,000 ● genomics-assisted recurrent selection
Costs for start up are much less than a full program because 
the pipeline materials coming out of it will be transferred to 
1.2
the local conventional program for advanced testing & product 
Start of inbred & hybrid development $      250,000 development
1.3
Double haploid tech development $      100,000 start later when we have improved selfing clones
Genetic studies to better understand genetic correlations Genetic studies to better understand genetic correlations Genetic studies without inbreds are more costly due to the 
1.4
among key traits and the number of genes / QTL among key traits and the number of genes / QTL controlling need for higher entry # to generate the same quality of 
controlling them (integrated in #1.1 and #1.2) them (structured populations for genetic mapping) $        100,000 information
Cost of building and transferring enhanced breeding Extra cost as stacking genes in heterozygous populations 
1.5
populations to others $        50,000 $        100,000 requires higher entry # and more cycles of selection
●  $75,000 for maintain the historical popula ons;
●  the other cost for cleaning and tes ng virus of new breeding 
1.6 materials
Maintaining cell cultures and grow outs to support ● maintain S3 or inbred seeds in storage room in the new 
Breeding population management costs $100,000 distributions $150,000 model, rather than tissue culture
Total annual costs $  3,875,000 $    3,725,000
Cost components:
1) Breeding populations for major markets
2) Start of inbred and hybrid development
3) Double haploid development
4) Genetic studies
5) Building and transferring enhanced breeding populations
6) Breeding population maintenance
Annual Annual 
Item New model activities years 1-10 Cost Current / ongoing Cassava model years 1-10 Cost Comments / clarifications
● 20% for opera on cost and 80% for personnel
1.1 ~5 Breeding programs covering the 5 major markets at ~5 Breeding programs covering the 5 major markets at ● 1.5 FTE for each market (=1.5 x 450,000)
$675,000 each $  3,375,000 $675,000 each $    3,375,000 ● genomics-assisted recurrent selection
Costs for start up are much less than a full program because 
the pipeline materials coming out of it will be transferred to 
1.2
the local conventional program for advanced testing & product 
Start of inbred & hybrid development $      250,000 development
1.3
Double haploid tech development $      100,000 start later when we have improved selfing clones
Genetic studies to better understand genetic correlations Genetic studies to better understand genetic correlations Genetic studies without inbreds are more costly due to the 
1.4
among key traits and the number of genes / QTL among key traits and the number of genes / QTL controlling need for higher entry # to generate the same quality of 
controlling them (integrated in #1.1 and #1.2) them (structured populations for genetic mapping) $        100,000 information
Cost of building and transferring enhanced breeding Extra cost as stacking genes in heterozygous populations 
1.5
populations to others $        50,000 $        100,000 requires higher entry # and more cycles of selection
●  $75,000 for maintain the historical popula ons;
●  the other cost for cleaning and tes ng virus of new breeding 
1.6 materials
Maintaining cell cultures and grow outs to support ● maintain S3 or inbred seeds in storage room in the new 
Breeding population management costs $100,000 distributions $150,000 model, rather than tissue culture
Total annual costs $  3,875,000 $    3,725,000
Cost components:
1) Breeding populations for major markets
2) Start of inbred and hybrid development
3) Double haploid development
4) Genetic studies
5) Building and transferring enhanced breeding populations
6) Breeding population maintenance
Ability to store seed of inbreds with traits of interest over long time 
periods in a cold room in contrast to maintaining cell cultures or 
growing out the clones each year.
Annual Annual 
Item New model activities years 1-10 Cost Current / ongoing Cassava model years 1-10 Cost Comments / clarifications
● 20% for opera on cost and 80% for personnel
1.1 ~5 Breeding programs covering the 5 major markets at ~5 Breeding programs covering the 5 major markets at ● 1.5 FTE for each market (=1.5 x 450,000)
$675,000 each $  3,375,000 $675,000 each $    3,375,000 ● genomics-assisted recurrent selection
Costs for start up are much less than a full program because 
the pipeline materials coming out of it will be transferred to 
1.2
the local conventional program for advanced testing & product 
Start of inbred & hybrid development $      250,000 development
1.3
Double haploid tech development $      100,000 start later when we have improved selfing clones
Genetic studies to better understand genetic correlations Genetic studies to better understand genetic correlations Genetic studies without inbreds are more costly due to the 
1.4
among key traits and the number of genes / QTL among key traits and the number of genes / QTL controlling need for higher entry # to generate the same quality of 
controlling them (integrated in #1.1 and #1.2) them (structured populations for genetic mapping) $        100,000 information
Cost of building and transferring enhanced breeding Extra cost as stacking genes in heterozygous populations 
1.5
populations to others $        50,000 $        100,000 requires higher entry # and more cycles of selection
●  $75,000 for maintain the historical popula ons;
●  the other cost for cleaning and tes ng virus of new breeding 
1.6 materials
Maintaining cell cultures and grow outs to support ● maintain S3 or inbred seeds in storage room in the new 
Breeding population management costs $100,000 distributions $150,000 model, rather than tissue culture
Total annual costs $  3,875,000 $    3,725,000
Cost components (years 11-20):
1) Breeding populations for major markets
2) Inbred and hybrid development for trait donors
3) Double haploid development
4) Genetic studies
5) Building and transferring enhanced breeding populations
6) Breeding population maintenance
New Model Current Model
ONE foundation population and four conversion population vs FIVE separate populations
● 2 FTE  for the base population for two pools  and DH development.
● 0.8 FTE for each line conversion pipeline
Annual Annual 
Item New model activities years 11-20 Cost Current / ongoing Cassava model years 11-20 Cost Comments / clarifications
In the new model general agronomic improvements continue 
only in the base population for the largest TPP while feeding 
2.1
~5 Breeding programs covering the 5 major markets at new lines for the secondary programs to be used in trait 
$900,000 for the base program & $360,000 for each of the 4 ~5 Breeding programs covering the 5 major markets at conversion programs to generate products that better fit their 
line conversion / testing programs $  2,340,000 $675,000 each $    3,375,000 TPPs.
There are additional costs for developing inbreds partly 
2.2 Inbred & hybrid development continues with most costs because the development of inbreds will be expanding into 
covered within the larger base program (trait donors) $      150,000 donor parents to facilitate utilization of gene bank materials
2.3
Double haploid tech development (integrated in #2.1)
Genetic studies without inbreds are more costly due to the 
Genetic studies to better understand genetic correlations Genetic studies to better understand genetic correlations need for higher entry # to generate the same quality of 
2.4
among key traits and the number of genes / QTL among key traits and the number of genes / QTL controlling information & the poorly quality of results requires ongoing 
controlling them (integrated in #1.1 and #1.2) them (structured populations for genetic mapping) $        100,000 efforts over a longer time period
Cost of building and transferring enhanced breeding Extra cost as stacking genes in heterozygous populations 
2.5
populations to other $        25,000 $       1 00,000 requires higher entry # and more cycles of selection
● Widespread use of inbreds as storage & transfer unit of 
germplasm
2.6 ●  $75,000 for maintain the historical popula ons;
Maintaining cell cultures and grow outs to support ● in the new model, the historical popula on size will be 
Breeding population management costs $50,000 distributions $150,000 reduced.
Start of dihaploid development with associated reduction Cost reduction based on lower cost of genetic gains per unit 
in breeding cycle time & improved selection accuracy ($100,000) time
Total annual costs $  2,465,000 $    3,725,000
Annual Annual 
Item New model activities years 11-20 Cost Current / ongoing Cassava model years 11-20 Cost Comments / clarifications
In the new model general agronomic improvements continue 
only in the base population for the largest TPP while feeding 
2.1
~5 Breeding programs covering the 5 major markets at new lines for the secondary programs to be used in trait 
$900,000 for the base program & $360,000 for each of the 4 ~5 Breeding programs covering the 5 major markets at conversion programs to generate products that better fit their 
line conversion / testing programs $  2,340,000 $675,000 each $    3,375,000 TPPs.
There are additional costs for developing inbreds partly 
2.2 Inbred & hybrid development continues with most costs because the development of inbreds will be expanding into 
covered within the larger base program (trait donors) $      150,000 donor parents to facilitate utilization of gene bank materials
2.3
Double haploid tech development (integrated in #2.1)
Genetic studies without inbreds are more costly due to the 
Genetic studies to better understand genetic correlations Genetic studies to better understand genetic correlations need for higher entry # to generate the same quality of 
2.4
among key traits and the number of genes / QTL among key traits and the number of genes / QTL controlling information & the poorly quality of results requires ongoing 
controlling them (integrated in #1.1 and #1.2) them (structured populations for genetic mapping) $        100,000 efforts over a longer time period
Cost of building and transferring enhanced breeding Extra cost as stacking genes in heterozygous populations 
2.5
populations to other $        25,000 $       1 00,000 requires higher entry # and more cycles of selection
● Widespread use of inbreds as storage & transfer unit of 
germplasm
2.6 ●  $75,000 for maintain the historical popula ons;
Maintaining cell cultures and grow outs to support ● in the new model, the historical popula on size will be 
Breeding population management costs $50,000 distributions $150,000 reduced.
Start of dihaploid development with associated reduction Cost reduction based on lower cost of genetic gains per unit 
in breeding cycle time & improved selection accuracy ($100,000) time
Total annual costs $  2,465,000 $    3,725,000