1/25/2019

Mission of citrus breeding in NIFTS  Development of diverse varieties to satisfy commercial needs  Improvement of overall fruit quality for premium fruit Citrus breeding 2.0  Minimize the period to release a new citrus variety A novel approach integrating deciphered parentage and Kiyomi Shiranuhi Amakusa Harumi Setoka Seinannohikari Harehime genomics-assisted selection

1979 1995 1999 1999 2002 2002

Tokurou Shimizu Reiko Tamami Benibae Haruhi Asumi Mihaya Rinoka Asuki Mai Minamikawa, Keisuke Nonaka, Terutaka Yoshioka, Eli Kaminuma, Hiroyoshi Iwata

2004 2005 2006 2009 2012 2012 2014 2017

Citrus breeding – three conventional approaches Workflow of citrus breeding in NIFTS  Mutation breeding  Cross breeding is the best approach for developing premium new cultivars • Mutant selection of a sport or from nucellar seedlings  However, it takes about 20 years to select one promising scion from ~ 5,000 • Mutation induction by irradiation or chemicals seedlings • Less efforts to discover a mutant  Genomics assisted selection is highly anticipated to reduce the cost of breeding, and improve the efficiency to select candidate scions • Limited change of trait Workflow of conventional breeding 2nd trial  Cross breeding Sawing Grafting 1st flower 1st trial Clone test • Hybrid selection from a single cross of diploid 22.8 yrs in average • Valid to obtain unique variety • Longer breeding period than mutation breeding ~5,000/yr ~1,000/yr ~10yrs

 Polyploid breeding • Triploid or tetraploid selection • Benefit of seedless variety development • Large efforts to obtain parental lines First priority - improve the breeding efficiency on citrus

Overview of citrus breeding in NIFTS Trilemma in citrus breeding Hybrid selection from various diploid crosses Fast breeding Three conflicting objectives: Released citrus varieties：54 (49 crossbreds) - Fast breeding Germplasm collections: >1,200 - Increase diversity to create unique variety - Improve overall fruit quality Seeds obtained every year: >12,000

Seedlings under evaluation: ~10,000 Enlarge Improve diversity fruit quality

It is difficult to accomplish these three objectives together by conventional breeding

1 1/25/2019

Diversity observed among wide citrus varieties Variation in fruit size and shape in Kiyomi tangor families Parental varieties used for the cross • Size 1. Dancy (125) 1● 2● 3● • Shape 2. Hansen (202) • Rind thickness 3. Rishon (146) 4. Roma (116) • Rind color 5. Shani (208) • Flesh color 6. Sunburst (234) • Seed number • Acidity Distribution curves for fruit height 1● 2● 3●  1st year • Brix  2nd year • Color break • Aroma Fruit height of Kiyomi • Taste Fruit height of another parent • Peelability 4● 5● 6● • Peel puffing  Median in each population is • Disease resistance between the parent heights  Some offspring produce taller or • Fertility shorter fruit • Parthenocarpy  Evaluation for two years showed • ... similar results 4● 5● 6●

Most varieties are not suitable for the breeding. Why?

Combrink, NK et al. (2013) Sci. Hort. 162. 357-364. Fruit height

Top 8 reasons to discard candidate seedlings A few elite varieties are commonly used in breeding programs

Used as a parent of Offspring Used as a parent of Selected 1. Small fruit (less than 100 g) crossbred varieties varieties selected lines lines Selected lines 12 'Kiyomi' 6 'Kiyomi' 10 Satsuma 4 2. Seedy 1 cm Satsuma 7 Ponkan 6 Sweet orange 4 'Encore' 6 'Encore' 3 3. Less Brix content Hassaku 4 Hirado pummelo 4 'Sweet spring' 3 Sweet orange 3 Hassaku 1 4. Acidic or less acidic Clementine 2 Iyo 1 Mato pummelo 2 'Osceola' 1 'Harrumi' 2 5. Hard to peel 'Murcott' 2 Trifoliate orange 1 'Seminole' 1 Natsudaidai 1 Red: Cross-bred varieties 6. Peculiar aroma 'Kara mandarin' 1 'Tanigawa pummelo' 1 Kishu mandarin 1 7. Appearance Lemon 1 Hyuganatsu 1 Banpeiyu 1 8. Nonuniform coloring 'Yoko' 1 'Tsunonozomi' 1 Q: Are small non-elite 'Awa orange' 1 'Minneola' 1 varieties useless for breeding? 'Page' 1 Many non-elite varieties have been considered useless as breeding parents 'Wilking' 1 Compiled from http://www.naro.affrc.go.jp/nifts/kih/files/nifts_hinshu_ichiran.pdf

Small fruit varieties yield small fruit offspring Pedigrees of 67 citrus varieties were determined

A × C × A B A bell-shaped curve observed for polygenic traits C B A

- Most fruit characteristics are polygenic traits - Mean fruit size is a distribution between the parents The pedigrees revealed demonstrated the key roles of non-elite small fruit varieties in expanding citrus diversity

Frequency - Major drawback of small fruit varieties for breeding

Fruit weight (g) Shimizu, T. et al. (2016) PLoS ONE 11(11): e0166969.

2 1/25/2019

Introgression of the genome of Kishu to wide varieties through Kunenbo Diversity of fruit trait observed in half sibs of Kunenbo

Kishu produced Kunenbo by the first cross. 1

Kunenbo backcrossed with Kishu and then produced Satsuma and Yatsushiro. 2

Shimizu, T. et al. (2016) PLoS ONE 11(11): e0166969. Shimizu, T. et al. (2016) PLoS ONE 11(11): e0166969. Shimizu, T. et al. (2017) Front. Genet. 8:180.

Introgression of the genome of Kishu to wide varieties through Kunenbo Diversity of fruit trait on acid citrus varieties observed in half sibs of Yuzu Kunenbo increased citrus diversity as the parent of at least 12 varieties.

Yatsushiro is a sib variety of Satsuma which has the opposite combination of the parents.

Shimizu, T. et al. (2016) PLoS ONE 11(11): e0166969. Shimizu, T. et al. (2016) PLoS ONE 11(11): e0166969. Shimizu, T. et al. (2017) Front. Genet. 8:180.

Diversity of fruit trait observed in half sibs of Kunenbo Diversity of fruit trait on acid citrus varieties observed in half sibs of Yuzu

Shimizu, T. et al. (2016) PLoS ONE 11(11): e0166969. Shimizu, T. et al. (2016) PLoS ONE 11(11): e0166969.

3 1/25/2019

Wide diversity observed among the half sibs of Kaikoukan Wide diversity observed among the sibs of small fruit variety Tachibana

Shimizu, T. et al. (2016) PLoS ONE 11(11): e0166969. Shimizu, T. et al. (2016) PLoS ONE 11(11): e0166969.

Wide diversity observed among the half sibs of Kaikoukan Traits changed during single generation • Size • Shape • Rind thickness • Rind color • Flesh color • Seed number • Acidity • Brix • Color break • Aroma • Taste • Peelability • Peel puffing • Fertility • Parthenocarpy • Flesh texture • ...  Single cross is enough to increase diversity and select promising scions  Multiple generation is not mandatory to exclude undesirable traits

Shimizu, T. et al. (2016) PLoS ONE 11(11): e0166969.  Non-elite varieties show potential as breeding parents

Wide diversity observed among the sibs of small fruit variety Tachibana Using non-elite varieties for breeding will • Contribute to increasing diversity of fruit traits • Decrease the overall fruit quality by introgression of bad trait • Extend total breeding period for continuous cross to segregate out unwanted traits Conventional citrus breeding evaluates more than 20 traits: • Selecting a promising candidate exceeding the thresholds of most of the traits is mandatory • Many fruit traits are complex polygenic quantitative traits - Simple MAS for a specific trait is insufficient • Most fruit characteristics are combination-dependent

Two genomics-assisted selection methods will increase both diversity and fast-breeding

Shimizu, T. et al. (2016) PLoS ONE 11(11): e0166969.

4 1/25/2019

Genomic selection (GS) Prediction accuracy for 17 fruit traits showed variation Prediction accuracy was evaluated in total families. Training population: Parental and1841SNPs breeding citrus cultiv populationarsF1 Pd to F1 totalPedigree

Phenotype 0 10-fold cross validation, 5 repeats

(y) . Reference 1 Associations between rrBLUP(RR) BGLR (BRR) rrBLUP(GAUSS) BGLR (BL) marker genotypes randomForest BGLR (BayesA) glmnet ridge (alpha=0) BGLR (BayesB)

8 cultivars . glmnet elasticnet (alpha=0.5) BGLR (BayesC)

0 and phenotypes glmnet lasso (alpha=1) Mean prediction of all the methods and lines Marker (x)

genotype 6

.

0

Genetic Prediction model r

4

. relationships y = f (x) 0

2

.

0

0 Prediction of . 0 FruW Appear Shape FruH FruC Pericarp Peeling Aroma FleC FleH Juicy SegM Seed Bitter Taste Brix Aci Fruit Rind Flesh Acidity Marker breeding value hardiness color color genotype Seedlings No large differences were observed among the 11 prediction methods for 17 traits

Minamikawa, M.F. et al., 2017. Scientific Reports, 7(1), p.4721.

Genome-wide association study Integrating population data improved the prediction accuracy of GS

(GWAS) Prediction accuracies were improved by integrating additional population data Statistically assess 1841SNPs meanPrediction totalPedigrees

Red 0

. 1 Parental population (n = 111) Phenotype the associations Parental population Breeding populationBreeding population (n = 676) between marker Parental andParental breeding and breeding population population (n = 787)

8

.

Intermed 0 genotype and Marker

6

.

phenotype 0 Yellow genotype Detect QTL r

4

.

0

2

.

0

0 log(P)

. -

GWAS enables the detection 0 FruW Appear Shape FruH FruC Pericarp Peeling Aroma FleC FleH Juicy SegM Seed Bitter Taste Brix Aci Fruit Rind Flesh Acidity and identification of hardiness color color quantitative trait loci (QTL) Minamikawa, M.F. et al., 2017. Scientific Reports, 7(1), p.4721. Chromosome

GS and GWAS in citrus GS is valid for predicting a part of fruit traits

FruitFruW weight AppearanceAppear FruitShape shape Fruit FruHhardiness RindFruC color Plant samples for modeling: 600 3 14 6 9 8 500 12 5 7 2.5 10 • 111 parental varieties 400 4 6 8 5 300 2 3 6 4 200 2 3 4 • 1.5 35 breeding populations, total 676 individuals 2 100 2 1 r = 0.83 r = 0.57 r = 0.22 r = 0.15 r = 0.71 1

0 1 0 0 0 0 100 200 300 400 500 600 1 1.5 2 2.5 3 0 2 4 6 8 10 12 14 0 1 2 3 4 5 6 0 1 2 3 4 5 6 7 8 9

Genotype data: RindPericarp smoothness PeelabilityPeeling AromaAroma intensity FleshFleC color FleshFleH texture

5 5 2.5 7 5 • Verified 1,841 SNPs (Illumina GoldenGate Assays) 6 4 4 4 5 2 3 3 3 • 4 Designed from resequencing data of 15 citrus varieties 3 2 2 2 1.5 2 1 1 1 r = 0.43 r = 0.67 r = 0.13 1 r = 0.86 r = 0.48

Part of pedigrees of plant samples used for modeling 0 0 1 0 0 0 1 2 3 4 5 0 1 2 3 4 5 1 1.5 2 2.5 0 1 2 3 4 5 6 7 0 1 2 3 4 5

JuicyJuicy SegmentSegM hardiness SeedsSeed BitternessBitter Taste

3 5 5 4

2.5 1.6 4 4

2 3 3

1.5 3 Predicted values Predicted 2 2 1.1 1

1 0.5 r = 0.31 r = 0.50 1 r = 0.54 r = 0.32 r = 0.16

0 0 0 0.6 2 0 0.5 1 1.5 2 2.5 3 0 1 2 3 4 5 0 1 2 3 4 5 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.5 3 3.5 4 BrixBrix AcidityAcid Observed values

17 5

16 r = 0.48 4 r = 0.70 15 Heritability and mode of inheritance of trait, 14 3 13

12 2 accuracy of phenotyping, and populations used 11 10 1 for modeling affect prediction accuracy 9

8 0 8 9 10 11 12 13 14 15 16 17 0 1 2 3 4 5 Founder varieties Natural cross Hybrid cultivars Selection Minamikawa, M.F. et al., 2017. Scientific Reports, 7(1), p.4721.

5 1/25/2019

GWAS is valid for identifying QTL of fruit traits

Fruit weight Appearance Fruit shape Fruit hardiness Rind color

Rind smoothness Peelability Aroma intensity Flesh color Flesh texture

Juicy Segment hardiness Seeds Bitterness Taste

Brix Acidity rrBLUP n.PC=4 The use of both GS and GWAS complement the prediction performance of each other

Minamikawa, M.F. et al., 2017. Scientific Reports, 7(1), p.4721.

Citrus breeding 2.0

Citrus breeding 2.0 will • Contribute to increase trait diversity by referring to known pedigrees • Improve overall fruit quality by selecting high-quality seedlings by using GS and GWAS • Achieve fast-breeding by selecting a promising scion from single cross and avoid the trilemma in citrus breeding

In progress: • Increase the number of plant samples and DNA marker genotypes for remodeling • Integration of this strategy for the breeding system to increase diversity • Application in the rebreeding of Satsuma and others

Shimizu, T. in press.

Acknowledgements

Mai Minamikawa Keisuke Nonaka Hiromi Kajiya-Kanegae Shingo Goto Akio Onogi Satoshi Ohta Hiroyoshi Iwata Hiroko Hamada Takeshi Hayashi Eli Kaminuma Terutaka Yoshioka Atsushi Toyoda Asao Fujiyama Yasukazu Nakamura

Funding from  Grants‐in‐aid for Scientific Research (KAKENHI)

 Next Generation Breeding Project of Ministry of Agriculture, Forestry and Fisheries Japan

 A grant from Cabinet Office, Government of Japan

6