ESTIMATION OF GENETIC PARAMETERS FOR ECONOMICALLY IMPORTANT TRAITS IN SOUTHERN HIGHBUSH BLUEBERRIES

By

CATHERINE CELLON

A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE

UNIVERSITY OF FLORIDA

2015

1

© 2015 Catherine Cellon

2

To my parents and friends, thanks for always being there for me

3

ACKNOWLEDGMENTS

I would like to sincerely thank my main advisors, Dr. Jim Olmstead and Dr. Patricio

Munoz, for all their help and patience. I know you are both incredibly busy and I really appreciate the time you took to answer all my questions and listen to my concerns.

Thank you Dr. Matias Kirst for your knowledge and input on the project. Thank you

Werner Collante and David Norden for all your help in the field and the lab.

I would like to thank all the undergraduate students who helped with my project, especially Allison Bechtloff, Kaitlin Quincy, Ashley Leonard, Elliot Norden, and Micah

Weiss. I know the work was incredibly tedious, but I appreciate all your positive attitudes and attention to detail. Thank you Rodrigo for always answering my software questions and developing the packages I used throughout my project. Finally, I would like to acknowledge the blueberries. If those tiny blue guys weren’t so popular, I would not have had a fantastic research project.

4

TABLE OF CONTENTS

page

ACKNOWLEDGMENTS ...... 4

LIST OF TABLES ...... 7

LIST OF FIGURES ...... 8

LIST OF ABBREVIATIONS ...... 9

ABSTRACT ...... 10

CHAPTER

1 LITERATURE REVIEW ...... 12

Economic Importance ...... 12 Taxonomy and Description ...... 13 Primary Blueberry Species ...... 15 Blueberry Breeding ...... 16 Early Cultivation ...... 16 Breeding Southern Highbush Blueberries for Florida’s Climate ...... 16 Importance of 2n Gametes in Blueberry Breeding ...... 17 Recurrent Selection ...... 18 Best Linear Unbiased Prediction ...... 20 Implications of Autotetraploids ...... 21 Cytological Behavior ...... 24 Current Blueberry Research ...... 25

2 FOUNDING CLONES, COANCESTRY, AND INBREEDING IN THE UNIVERSITY OF FLORIDA SOUTHERN HIGHBUSH BLUBERRY BREEDING PROGRAM ...... 27

Materials and Methods...... 29 Pedigree Information ...... 29 Coefficient of Coancestry ...... 29 Inbreeding Coefficient ...... 30 Results and Discussion...... 30 Identification of the Founding Clones ...... 30 Genetic Contribution From The Founding Clones ...... 33 Inbreeding Coefficient ...... 37 Conclusion ...... 39

3 ESTIMATION OF DOUBLE REDUCTION LEVELS FOR SOUTHERN HIGHBUSH BLUEBERRY SELECTION TRAITS ...... 55

Materials and Methods...... 58

5

Blueberry Breeding Population and Pedigree ...... 58 Phenotyping ...... 59 Analysis and Model Comparison ...... 60 Matrix Construction with AGHmatrix ...... 61 Results and Discussion...... 63 Double Reduction in Blueberry ...... 63 Effect of Double Reduction on Inheritance ...... 64 Effect of Double Reductions on the A-matrix Construction ...... 65 Changes in Depth of Pedigree on Double Reduction Estimates ...... 66 Conclusions ...... 68

4 ESTIMATION OF GENETIC PARAMETERS IN SOUTHERN HIGHBUSH BLUEBERRIES ...... 73

Materials and Methods...... 76 Blueberry Breeding Population ...... 76 Phenotyping Methods ...... 77 Statistical Analysis ...... 79 Estimation of Expected Genetic Gain ...... 81 Results and Discussion...... 81 Population Distributions ...... 81 Estimation of Genetic Parameters ...... 82 Breeding Values ...... 87 Estimated Genetic Gain ...... 88 Conclusion ...... 89

5 SUMMARY AND CONCLUSIONS ...... 97

LIST OF REFERENCES ...... 100

BIOGRAPHICAL SKETCH ...... 110

6

LIST OF TABLES

Table page

2-1 Founding clones from high chilling requirement germplasm used in the University of Florida southern highbush blueberry breeding program ...... 40

2-2 Founding clones from low chilling requirement germplasm used in the University of Florida southern highbush blueberry breeding program ...... 40

2-3 The coefficient of coancestry between the high chilling requirement founders and the southern highbush blueberry cultivars released by the University of Florida ...... 41

2-4 The coefficient of coancestry between the low chilling requirement founders and southern highbush blueberry cultivars released by the University of Florida ...... 42

2-5 Inbreeding coefficients for the southern highbush blueberry cultivars released from the University of Florida ...... 43

2-6 The coefficient of coancestry between the high chilling requirement founders and the parents of the 2011 southern highbush blueberry breeding population ...... 44

2-7 The coefficient of coancestry between the low chilling requirement founders and the parents of the 2011 southern highbush blueberry breeding population ...... 48

2-8 Inbreeding coefficients for the parents of the 2011 southern highbush blueberry breeding population ...... 51

3-1 The heritability of five breeding traits calculated assuming disomic and tetrasomic inheritance ...... 69

3-2 Statistical summary of the kinship coefficient assuming various amounts of double reduction ...... 69

7

LIST OF FIGURES

Figure page

2-1 Heat map indicating the coefficient of coancestry among the 32 cultivars released by the University of Florida between 1976 and 2015 ...... 54

3-1 The Akaike information criterion (AIC) difference between the diploid model and the tetraploid models assuming double reduction values from 0 to 0.25 ..... 69

3-2 The estimated narrow-sense heritability for firmness assuming disomic inheritance and tetrasomic inheritance with increasing coefficients of double reduction ...... 70

3-3 Correlation of breeding values for firmness estimated assuming disomic and tetrasomic inheritance with w = 0.15...... 71

3-4 Pedigree depths of 5, 10, 20, and 30 years and their influence on the estimated amount of double reduction for each trait ...... 72

4-1 Phenotypic distributions of blueberry fruit quality and plant characteristics ...... 90

4-2 Narrow-sense heritability and standard errors of fruit firmness, fruit diameter, scar diameter, weight, and yield rating according to the year the data were taken ...... 91

4-3 Narrow-sense heritability and relative standard errors for flower bud density, soluble solids content, and fruit pH ...... 92

4-4 Genotype-by-year interaction (G × Y) and standard error for fruit firmness, fruit diameter, scar diameter, weight, and yield rating ...... 93

4-5 Genetic correlations between the primary selection traits of total yield, flower bud density, and fruit weight, firmness, diameter, scar diameter, pH, and soluble solids...... 94

4-6 Boxplots for fruit firmness, fruit diameter, weight, scar diameter, and crop rating showing the variation of the estimated breeding values (EBVs) between 2014, 2015, and combined data from 2014 and 2015 ...... 95

4-7 Potential genetic gain for fruit weight when using parents from different selection stages and intensity ...... 96

8

LIST OF ABBREVIATIONS

AIC Akaike information criteria

BLUP Best linear unbiased prediction

EBVs Estimated breeding values

NHB Northern highbush blueberry

REML Restricted maximum likelihood

SHB Southern highbush blueberry

UF The University of Florida

9

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science

ESTIMATION OF GENETIC PARAMETERS FOR ECONOMICALLY IMPORTANT TRAITS IN SOUTHERN HIGHBUSH BLUEBERRIES

By

Catherine Cellon

December 2015

Chair: James W. Olmstead Cochair: Patricio R. Munoz Major: Horticultural Sciences

Until recently, blueberries were bred using phenotypic recurrent selection. Best linear unbiased prediction (BLUP) and restricted maximum likelihood (REML) are methods that can be used to achieve larger genetic gains per breeding cycle than phenotypic selection. These methods incorporate the pedigree of the tested population, assuming disomic inheritance. Southern highbush blueberries (SHB) are autotetraploids, with polysomic inheritance and can experience double reduction. The goal of this study was to estimate the genetic parameters of the SHB program at the

University of Florida (UF), while accounting for this complex inheritance pattern. To accomplish this, phenotypic data were collected for eight traits (yield, flower bud density, fruit weight, fruit firmness, fruit diameter, fruit soluble solids, fruit pH, and fruit scar diameter) from 1,996 individuals developed in 2011 and a complete pedigree file was constructed. Two founding events were identified comprised of northern highbush blueberries (NHB) and SHB. The largest genetic contributions were from NHB cultivars

‘Brooks’, ‘Sooy’, and ‘Rubel’ and SHB ‘FL 4A’. All SHB cultivars released from UF, except ‘Meadowlark’ were related and had a low degree of inbreeding. Similar levels of

10

inbreeding were present among the parents used in the 2011 crosses, indicating that these levels of inbreeding may not be detrimental to this species and that the diversity has been well managed in the breeding program. Yield, weight, fruit diameter, and scar diameter best fit the genetic model assuming tetrasomic inheritance with no double reduction. Firmness had the best fit when double reduction was from 0.15 to 0.20.

Higher estimations of double reduction decreased the narrow-sense heritability, but there was a high correlation between the breeding values estimated assuming both modes of inheritance. Shallow pedigrees led to models that fit better assuming larger proportions of double reduction because they underestimated the kinship coefficient.

The largest narrow-sense heritabilities were for weight, yield, and scar diameter. Most genetic correlations were under 0.20, with the highest correlation between weight and fruit diameter. The results from this study will allow for parent selection based on BLUP estimated breeding values and REML estimated variance components in order to maximize genetic gains.

11

CHAPTER 1 LITERATURE REVIEW

Economic Importance

Blueberry (Vaccinium spp.) consumption has dramatically increased in the past decade as new research revealed the many nutritional benefits and fresh fruit became available throughout the year due to cultivars adapted to a variety of climates

(Brazelton, 2013; Finn et al., 2014; Isogai et al., 2009) Similarly, the development of new, ready to eat products containing blueberries has further increased consumer availability (Brazelton, 2013). Blueberry production has become global with industries in

North America, South America, Europe, Australia, and New Zealand (Ballington, 1990,

2001; Finn et al., 2014). However, in 2012 over half of the world’s blueberries were produced in the United States (214,708 tonnes out of 399,309 tonnes) (FAO, 2015).

The blueberry industry in the U.S. is valued at $824.9 million (USDA, 2015). This number is expected to grow as consumption and demand increase worldwide

(Brazelton, 2013). The blueberry industry in Florida has capitalized on this growth with production nearly quadrupling over the past decade (USDA, 2015). This surge in production is due to low chill, early ripening cultivars known as southern highbush blueberries (SHB) developed by the University of Florida (UF). The fruit from these cultivars can be harvested at an exclusive time in the northern hemisphere production season (April to mid-May) and thus sold at a premium (Lyrene & Sherman, 1984;

Sharpe, 1953). In 2014, the Florida blueberry industry was valued at $75.6 million, with average fresh market prices at least $1.50 to $2.80 more per pound than other states

(USDA, 2015).

12

Taxonomy and Description

Cultivated blueberries are a member of the Ericaceae family, genus Vaccinium, and subgenus Cyanococcus (Camp, 1942; Lyrene & Ballington, 1986; Vander Kloet,

1983). Members of this family are stress-tolerant, shade adverse, and prefer acidic, infertile soils in disturbed areas (Ballington, 1990; Camp, 1942; Coville, 1937; Lyrene &

Ballington, 1986). The fruit from this family are nontoxic and develop from inferior ovaries (Ballington, 2001; Camp, 1942; Lyrene & Ballington, 1986). The Vaccinium genus is an ancient group, which has gone through considerable evolutionary modification (Camp, 1942). The center of origin is thought to be northern South America

(Ballington, 1990). Today there are about 450 species distributed around the globe, with the most abundant populations found in temperate to tropical regions (Ballington, 2001;

Camp 1942; Lyrene & Ballington, 1986). The genus is divided into 30 subgenera with economically significant sections including: Cyanococcus (cultivated blueberries),

Oxycoccus (large cranberry), and Vitis-idaea (lingonberry) (Ballington, 2001; Camp,

1942).

In the section Cyanococcus, there are about 10 to 26 blueberry species depending on the taxonomic definition (Ballington, 1990). These species are connected by a reduced amount of gene flow, a term known as syngameon (Grant, 1981).

Because these species are not genetically isolated, intermediate phases of speciation have been created and from this the disagreement originated. Interspecific hybridization is facilitated by the outcrossing nature of the species, with the most significant barrier being different chromosome numbers (Camp, 1942; Lyrene & Ballington, 1986).

Morphological differentiation between the homoploid species is likely a result of time and genetic drift due to microhabitats (Camp, 1942; Vander Kloet, 1983).

13

Vaccinium spp. is a highly polymorphic species which can be found in: coastal sand dunes, margins of lakes or rivers, headlands, shallow soils on mountainous slopes, pine flatwoods, barrens, dry ridges, and frequently burned areas (Lyrene et al.,

2003; Vander Kloet, 1983). In North America, the geographical range is widespread from central Florida to Canada and west to the Pacific Northwest. The plants are either deciduous or evergreen with a chilling requirement ranging from 0 to 1000 hours

(Lyrene, 1997).

The long-lived, woody shrubs can be tall and crown-forming, or short and rhizomatous (Camp, 1942; Lyrene et al., 2003; Vander Kloet, 1983). The inflorescences are racemose with urceolate flowers that are both perfect and pendant (Camp, 1942;

Lyrene, 1997; Lyrene et al., 2003). The floral buds are rotund and twice the size of the vegetative buds (Vander Kloet, 1983). The leaf blade can be a combination of glaucous, glabrous, glandular, pale, green, or pubescent; and the leaf margin is either entire or serrate (Vander Kloet, 1983). Berries contain up to 80 small seeds, which facilitates easy dispersion by birds and other mammals (Lyrene et al., 2003).

Self-fertility in Vaccinium spp. depends on the ploidy, genotype, and relatedness of parents (Krebs & Hancock, 1990; MacKenzie, 1997). Research suggests that the reduced fertility seen in self-crosses is not a consequence of self-incompatibility, but instead a type of early acting inbreeding depression resulting in abortion of the fertilized zygote (Harrison et al., 1994; Krebs & Hancock, 1990). Harrison et al. (1994) concluded that post-fertilization abortion affected seed formation, and in turn fruit set. Comparing outcrossing and self-pollinations of V. corymbosum, Krebs and Hancock (1990)

14

concluded that zygotic inbreeding coefficients of 0.30 were the threshold and inbreeding coefficients greater than that would result in infertility.

Primary Blueberry Species

The primary species used by blueberry breeders consist of lowbush (V. angustifolium), highbush (V. corymbosum), rabbiteye (V. virgatum syn. V. ashei), and V. darrowii (Ballington, 1990; Lyrene & Ballington 1986). Lowbush blueberry is a tetraploid

(2n=4x=48) native to the northeastern United States. These species spread through underground rhizomes, grow best when burned very 2 to 3 years, and may be a source of resistance to stem-blight (Ballington, 2001). While improved cultivars are available, they are not widely used because of the high establishment cost (Lyrene & Ballington,

1986).

Highbush blueberry is also a tetraploid (2n = 4x = 48) crown-forming shrub that has benefitted the most from interspecific hybridization resulting in a genetically complex species (Camp, 1942). Crosses between wild selections and cultivated species has led to a widespread production area stretching from central Florida up to southern

Canada, and across to California (Lyrene & Ballington, 1986; Qu et al., 1998).

Rabbiteye blueberries are hexaploid (2n = 6x = 72) and can be found in

Alabama, Florida, and Georgia. Rabbiteyes have lower chilling requirements and produce fruit that ripen later in the season (Lyrene & Ballington, 1986). V. darrowii (2n =

2x = 24) is an evergreen species that ripens later in the summer and retains its leaves into the winter (Lyrene, 1997). V. darrowii can also be found in the Florida peninsula

(Lyrene, 1997).

15

Blueberry Breeding

Early Cultivation

Blueberry breeding began in the early 1900s under the direction of Frederick

Coville (Coville, 1937). Coville began the domestication and improvement process by selecting superior wild bushes from forested areas in the northeastern U.S. ‘Brooks’ (V. corymbosum) was the first wild selection made followed by ‘Russell’ (V. angustifolium).

Other early selections included ‘Chatsworth’, ‘Grover’, ‘Harding’, ‘Rubel’, and ‘Sooy’.

Coville developed a breeding program using these wild accessions that had been selected for their light color, large size, dry picking scar, firmness, crop load, and superior flavor (Coville, 1937). ‘Katharine’ and ‘Pioneer’ were derived from crosses between these wild selections and became the first improved cultivars released from this program. These and subsequent cultivars released from the program are referred to as northern highbush blueberries (NHB).

Breeding Southern Highbush Blueberries for Florida’s Climate

In the 1920s to 1930s there was a small wild rabbiteye industry in northern Florida consisting of 3,000 acres (Sharpe, 1953). Blueberries from these plants were harvested and sold, but the fruit were small, dark, and variable in quality and production.

Ultimately, this industry collapsed due to improper marketing and low quality fruit compared to the improved northern highbush cultivars (Ballington, 2001; Lyrene &

Sherman, 1984;).

In the 1950s a second attempt was initiated to create a blueberry industry for

Florida (Lyrene, 2002). Ralph Sharpe, a fruit breeder at the University of Florida (UF), saw that Florida’s climate offered the opportunity to produce high-quality blueberries that could that could be harvested from April to mid-May and sold before the other

16

states entered the market. In order to meet these goals, traits such as low chilling requirement, precocious flowering, and short ripening period were heavily selected for

(Lyrene & Sherman, 1984; Sharpe & Darrow, 1959). After experimenting with of a variety of crosses, two separate breeding programs arose focused on hexaploid rabbiteye and tetraploid highbush.

The UF blueberry breeding program focused on tetraploid highbush has been the most successful. Germplasm improvement initially focused on crosses between NHB cultivars and advanced selections with the native southeastern species, V. darrowii and rabbiteye (Lyrene & Sherman 1984; Lyrene, 2002; Sharpe & Darrow, 1959). Later introgressions into the germplasm included V. elliottii, V. fuscatum, and V. arboreum

(Brevis et al., 2008; Lyrene, 1997, 2008; Olmstead et al., 2013). The NHB selections used in the first crosses provided fruit quality characteristics such as large size and superior taste, while traits from the southern species such as low chill requirement and heat tolerance allowed for climatic adaptation. The wild germplasm not only enhanced fruit quality in Florida’s warm and humid conditions, but also boosted plant vigor

(Lyrene, 2002; Sharpe & Darrow, 1959). As a result of these efforts the first southern highbush (SHB) cultivars released from the program were ‘Sharpblue’, ‘Avonblue’, and

‘Flordablue’ (Lyrene & Sherman, 1984).

Importance of 2n Gametes in Blueberry Breeding

Vaccinium spp. has a strong triploid block, which has limited direct crosses between species of different ploidy levels (Lyrene et al., 2003). Dweikat and Lyrene

(1988) pollinated 7,000 tetraploid highbush flowers with pollen from the diploid V. elliottii and obtained only three triploid hybrids. The best way to circumvent this triploid block is through unreduced gametes. When surveying diploid species for 2n gamete formation,

17

Ortiz et al., (1992) found unreduced gamete formation varied between species and among individuals within the species. V. darrowii, especially ‘FL 4B’ had the highest frequency of 2n gamete production (Cockerham & Galletta, 1976; Dweikat & Lyrene,

1988; Ortiz et al., 1992). The production of unreduced gametes allowed V. darrowii to generate fertile hybrids with both northern highbush cultivars and rabbiteye selections, making it instrumental in the development of SHB. Furthermore, V. darrowii clones acted as a genetic bridge for gene transfer between the two species (Draper, 1977;

Lyrene, 1997; Lyrene et al., 2003; Qu & Hancock, 1995; Qu et al., 1998; Vorsa & Novy,

1995).

Unreduced gametes can be formed by either first-division restitution (FDR) or second-division restitution (SDR). Unreduced gametes created during FDR are composed of non-sister chromatids, whereas unreduced gametes formed by SDR are made up of sister chromatids (Qu & Hancock, 1995). Therefore, 2n gametes created by

FDR will transmit a larger amount of heterozygosity. Using RAPD markers Qu and

Hancock (1995) determined ‘US 75’, a tetraploid hybrid from ‘FL 4B’ × ‘Bluecrop’, was a product of FDR because it had 70% of ‘FL 4B’s heterozygosity. A similar study analyzing progeny of ‘FL 4B’ and V. corymbosum cultivars with RAPD markers found heterozygosity levels to range from 5.3% to 100%, with a mean of 73% (Vorsa &

Rowland, 1997). While this indicates the majority of 2n gametes are created by FDR, a minor amount of gametes are likely created by SDR. The method of formation will affect the overall heterozygosity transferred to the progeny.

Recurrent Selection

Currently, the breeding program at UF is based on phenotypic recurrent selection. This method focuses on crossing heterozygous parents to produce

18

segregating populations containing individuals with more outstanding phenotypes than their parents (Lyrene, 2002). These exceptional progeny are then used as parents in the next round of crosses, so that each subsequent generation will phenotypically surpass the parents. This generates additive progress by changing the frequency of favorable alleles in the germplasm, resulting in an overall genetic improvement of the population over time (Lyrene, 1981, 2005). To maximize the benefits of recurrent selection the time between selection cycles should be minimized and traits should be simultaneously selected for. Priority should be given to traits with high heritability and economic importance (Lyrene, 2005). For blueberries, selection is based on both fruit quality

(large size, small scar, firmness, color, and flavor) and vegetative (high vigor, disease resistance, upright growth, chilling requirement) characteristics (Lyrene, 2002).

The cultivar development typically takes 12 to 15 years with over 100 parents used each year in over 100 crosses (Lyrene, 2002). Parent selection is critical in order to decrease inbreeding depression, maintain genetic diversity, and minimize the time it takes to release a cultivar. Under the current breeding strategy at UF, nearly half the parents selected for crossing in a given year are advanced selections or released cultivars, one-third are young seedlings that had exceptional phenotypes after the first year of evaluation, and the remaining parents are either superior wild types or highly- vigorous selections from test plots (Lyrene, 2002). The challenges of recurrent selection are the cost of replications across years and environments, difficulty in accurately selecting traits with low heritability, and the fixation of undesirable alleles (Lyrene &

Sherman, 1984; Lyrene, 2005).

19

Best Linear Unbiased Prediction

Best linear unbiased prediction (BLUP), along with restricted maximum likelihood

(REML) has been used with great success in livestock and forest breeding programs to estimate variance components for complex traits and predict breeding values

(Henderson, 1975). The breeding values are used to rank the breeding population in order to more accurately select families, parents, and individuals (Munoz et al., 2014;

Piepho et al., 2008) These methodologies are well suited for breeding programs because they are able to handle highly unbalanced data sets composed of complex genetic relationships (Lynch & Walsh, 1998). Classical models measure genotypic values assuming fixed effects. However, these models result in inaccurate estimations because they violate the assumption of random sampling by analyzing populations that have been selected on (Henderson, 1975). BLUP is able to account for selection by considering genetic effects as random effects and modeling the relationship of individuals across generations in the A-matrix (see below).

BLUP assumes an infinitesimal model, meaning that assumes that multiple genes control the trait and each has a small effect on the overall phenotype. These models produce shrinkage of the effects of genotypes depending on the amount of information is available for them. Furthermore, these models allow for incorporation of the tested population’s pedigree, which increases the accuracy when predicting breeding values, and controls for selection. The pedigrees are incorporated through the numerator relationship matrix (A-matrix). The A-matrix is computed from the coefficient of coancestry, which is the probability that two alleles selected at random will be identical by descent in two given individuals depending on their relationship. The incorporation of the A-matrix allows the genetic merit of the relatives to be predicted,

20

even though phenotypic data was not collected for them. The relatives in the pedigree will share a portion of alleles so their response will be correlated and thus can be predicted.

Implications of Autotetraploids

Polyploids are classified as allopolyploids or autopolyploids based on the origin of their genomes (Kihara & Ono, 1926). Allopolyploids arise when two distinct genomes merge, creating homeologous chromosomes. There is a strong barrier prohibiting pairing between the two different genomes. This barrier allows pairing to follow normal disomic inheritance, with preferential bivalent pairing between homologous chromosomes of the same genome (Gallais, 2003; Mather, 1936). Autopolyploids result from chromosomes of the same genome doubling, producing twice the basic set of chromosomes. Cultivated blueberries are autotetraploids with four homologous chromosomes, each with an equal opportunity to pair. During meiosis, these homologous chromosomes can pair to form bivalents or multivalents. This occurrence is known as polysomic inheritance and results in random chromatid segregation (Fisher,

1947). Polysomic inheritance creates segregation ratios that differ from those assumed in independent assortment (Krebs & Hancock, 1989; Mather, 1936; Wu et al., 2001).

For example, one locus with two alleles will have up to 5 possible genotypes instead of the 3 in diploids. The possibilities are two homozygous genotypes: aaaa (monogenic nulliplex) and AAAA (monogenic quadriplex); and three heterozygous genotypes: Aaaa

(digenic simpex), AAaa (digenic duplex), and AAAa (digenic triplex) (Gallias, 2003).

Segregation ratios are further skewed because autotetraploids can have up to 4 different alleles at one locus. This multiallelism creates up to 70 possible genotypes for one locus compared to the 15 in diploids (Gallias, 2003).

21

Another unique phenomenon that autopolyploids experience is double reduction.

This arises from multivalent formation and leads to identical chromatid segments derived from the same chromosome in one gamete (Mather, 1936). Inbreeding is then generated in the absence of consanguineous mating resulting in an increase in homozygosity (Gallais, 2003). During zygotene, chromosomes start to associate and pair. This can begin at any spot along the chromosome so that at pachytene one chromosome may have multiple partners, resulting in multivalent formation. This is known as a partner switch and occurs when one chromosome end pairs with one chromosome and the other end pairs with a different chromosome. In quadrivalents this type of pairing can take on a ring or chain formation depending on the number and position of the chiasmas. By the end of pachytene, chiasmas are formed and genetic crossing-over occurs between the paired chromosomes. The recombinant sister chromatids then become associated with different spindle-fiber attachments. During anaphase I, adjacent chromosomes in the ring or chain formation can migrate to the same or opposite poles. Multivalent formation is essential for double reduction because bivalent formations will always result in chromosomes migrating to different poles in anaphase I. Equational separation occurs when the sister chromatid segments migrate to the same pole, but have different spindle fiber attachments (Mather, 1935). If the sister chromatids were joined at the same spindle attachment, they would migrate to opposite poles in anaphase II and double reduction would not occur (Mather, 1936). If in anaphase II the sister chromatid segments again migrate to the same pole double reduction has been completed.

22

The proportion of double reduction thus depends on four parameters: probability of multivalent formation, probability of first equational separation, probability of the centromeres of sister chromatids migrating to the same pole in anaphase I, and the probability of two sister chromatids ending up in the same gamete (Doyle, 1973). The theoretical maximum coefficient of double reduction has been speculated to be 1/7,1/6, or 1/25; depending on the model assumptions (Bouffette, 1966; Demarly, 1963; Luo et al., 2006). Experiments analyzing the proportion of double reduction in autotetraploid species have found varying amounts ranging from 0 to 0.30 (Wu et al., 2001). Krebs and Hancock (1989) found no double reduction when modeling tetrasomic inheritance in

V. corymbosum using four isoenzyme markers. Slater et al., (2013) found the proportion of double reduction to vary from 0 to 0.25 depending on which trait was analyzed in potato. Double reduction is position dependent and varies between chromosomes and their tendency to form multivalents, and where the studied loci are located on the chromosome (Wu et al., 2001). Double reduction frequencies will be higher at the distal ends of the chromosome, and reduced near the centromere (Bourke et al., 2015;

Levings & Alexander, 1966; Welch, 1962). The increase in double reduction is due to the higher probability of a recombination event occurring between the centromere and the locus. Luo et al. (2006) and Bourke et al. (2015) observed increased frequencies of double reduction at the tips of linkage groups when analyzing molecular markers in autotetraploid potato. Therefore, because it is a position dependent phenomenon, double reduction will vary from gene to gene (Mather, 1936).

For the most part, the effects of double reduction and polysomic inheritance are often neglected due to the low theoretical maximum, little empirical information and

23

analytical complexities (Gallias, 2003). However, the amount of double reduction will affect the total heterozygosity in the breeding population and thus the breeder’s strategy and mating design.

Cytological Behavior

Inheritance studies have been limited in cultivated blueberries due to the lack of single gene polymorphisms and the small size of the chromosomes (1.5 to 2.5 µm), which makes exact meiotic pairing indistinguishable (Galleta, 1975; Hall & Galletta,

1971; Lyrene, 1997; Qu et al., 1998). Furthermore, complications arise when interpreting marker segregation in autotetraploids, often making it impossible to distinguish between allele dosage (Gallias, 2003; Qu & Hancock, 1995).

The few inheritance studies of V. corymbosum and V. corymbosum × V. darrowii crosses have shown inheritance patterns that follow expected autotetraploid segregation ratios (Draper & Scott, 1971; Krebs & Hancock, 1989; Qu & Hancock,

1995). However, Vorsa and Novy (1995) found evidence of preferential pairing in a tetraploid V. corymbosum × V. darrowii hybrid when using 14 RAPD makers. Draper and Scott (1971) also found evidence of preferential paring in V. corymbosum, but concluded the majority of the seedlings exhibited tetrasomic inheritance. This combined behavior has been seen in other autotetraploids such as Lotus corniculatus L. (a forage legume) and the male rainbow trout (Allendorf & Danzmann, 1997; Fjellstrom et al.,

2001). Species that exhibit this behavior are known as segmental allopolyploids.

Cytogenic studies of meiotic pairing found the majority of pairing to be bivalent with few quadrivalent formations in V. corymbosum (Jelenkovic & Harrington, 1971;

Jelenkovic & Hough, 1970; Qu et al., 1998; Vorsa & Novy, 1995). Qu et al. (1998) summarized that 90% of chromosome pairings in of ‘Bluecrop’ and a wild V.

24

corymbosum selection and over 75% of parings in ‘US 75’ (‘Bluecrop × ‘FL 4B’) were bivalent.

Current Blueberry Research

There are limited results available for heritability of quantitative breeding traits in blueberries. Most studies have focused on the general combining ability and the specific combining ability, which are biased in their conclusions because calculations were made assuming fixed effects (Aalders & Hall, 1975; Edwards et al., 1974; Erb et al.,

1990; Finn & Luby 1986,1992). The few studies that estimated the heritability did so from calculations of the offspring on the mid-parent value (Edwards et al., 1974; Finn &

Luby, 1986). These studies were done in breeding populations created from crosses between 6 (Edwards et al., 1974) or 17 parents (Finn & Luby, 1986) and assumed disomic inheritance. Furthermore, data for yield, firmness, scar, and fruit size was collected subjectively using a limited number of categories (5-10) (Darrow et al., 1939;

Draper, Galletta, & Ballington, 1982; Edwards et al., 1974; Finn & Luby, 1986, 1992;

Finn et al., 2003).

There is also growing concern about the depletion of genetic diversity in the cultivated germplasm and its implications for genetic gain in long-term selection. The original germplasm is very small and the founders ‘Brooks’, ‘Sooy’, and ‘Rubel’ make up a large genetic portion of present day NHB cultivars (Ehlenfeldt, 1994; Hancock &

Siefker, 1982). SHB cultivars were thought to be more heterozygous because of the genetic contribution from multiple species, but Brevis et al. (2008) found that SHB cultivars were genetically similar to the oldest NHB cultivars by using a panel of microsatellite DNA markers. Moreover, several researchers have found a parallel

25

increase between selection cycles and inbreeding coefficients in NHB cultivars (Brevis et al., 2008; Ehlenfeldt, 1994; Hancock & Siefker, 1982)

The objective of this study was to identify the founding clones and their genetic contribution to the UF SHB breeding program, the overall genetic diversity of the program, and impact inbreeding has had on it. A second objective was to use the pedigree information from the program to build A-matrices that account for tetrasomic inheritance and double reduction. From these matrices genetic parameters such as narrow-sense heritability, genetic correlations, and environmental variance can be estimated for primary breeding traits (yield, fruit diameter, fruit firmness, fruit weight, stem scar diameter, flower bud density, fruit soluble solids, and fruit pH). This will lay the foundation for parental selection in the UF breeding program based on BLUP predicted breeding values and REML estimated variance components in order to maximize genetic gains.

26

CHAPTER 2 FOUNDING CLONES, COANCESTRY, AND INBREEDING IN THE UNIVERSITY OF FLORIDA SOUTHERN HIGHBUSH BLUBERRY BREEDING PROGRAM

In the early 1900s, Frederick Coville made the first efforts to domesticate blueberries through his selections of wild V. corymbosum (Coville, 1937). The cultivars released as a result of these efforts are referred to as northern highbush blueberries

(NHB). Development of southern highbush blueberry (SHB) cultivars began approximately 50 years later at the University of Florida (UF). The visionary fruit breeder

Ralph Sharpe saw that Florida climate’s offered the unique opportunity to breed high- quality fruit that could be harvested before the other states entered the market (Lyrene

& Sherman, 1984; Lyrene, 2005; Sharpe, 1953). The initial breeding germplasm for what would become SHB cultivars was developed from crosses between NHB cultivars and native Florida species such as V. darrowii and rabbiteye (V. virgatum, syn. V. ashei) selections followed by several rounds of recurrent selection (Sharpe, 1953; Sharpe &

Darrow, 1959). As a result of this program, the first SHB cultivars (Sharpblue,

Flordablue, and Avonblue) were released in the late 1970s (Lyrene & Sherman, 1984).

These and the multiple cultivars released subsequently have been successfully used by producers in Florida to develop an important high-value specialty crop industry. In 2014 blueberries produced in Florida were priced $1.50 to $2.80 more per pound than any other state (USDA, 2015).

Blueberries are traditionally bred using phenotypic recurrent selection. This process allows for population improvement by increasing the frequency of favorable alleles in the germplasm, but can also deplete genetic diversity and lead to inbreeding

(Lyrene, 1981, 2005). Several researchers have highlighted a parallel increase between

27

selection cycles and inbreeding coefficients in NHB (Brevis et al., 2008; Ehlenfeldt,

1994; Hancock & Siefker, 1982). By analyzing pedigree information, Ehlenfeldt (1994) found that genes from the founding cultivars ‘Brooks’ and ‘Rubel’ account for nearly

25% of the genetic composition for 79 highbush cultivars (Ehlenfeldt, 1994). SHB blueberries are thought to be more heterozygous due to the diversity of the founding species used to create the initial breeding germplasm (Lyrene, 2002). However, NHB cultivars were heavily used to recover standard fruit quality traits during the initial SHB crosses (Sharpe & Darrow, 1959). Brevis et al. (2008), using a panel of microsatellite

DNA markers, found that SHB cultivars were similar to NHB cultivars that had undergone only one to four cycles of selection. This increase in inbreeding and narrow gene pool has led to a growing concern about the potential for genetic gain from long- term selection (Ballington, 1990; Brevis et al., 2008; Ehlenfeldt, 1994; Hancock &

Siefker, 1982).

Cultivated blueberries are autotetraploids with up to four alleles at one locus, creating 70 possible genotypes and a higher level of gene interaction (Gallias, 2003).

This multiallelism generates a more genetically diverse population than in diploids and can act as a buffer to inbreeding depression. This has been documented in Epilobium angustifolium, where diploids exhibited significantly higher inbreeding depression compared to autotetraploids (Husband & Schemske, 1997). Furthermore, Galloway et al. (2003) found inbreeding depression was not significant for most seed and germination characteristics between inbred and outcrossed autotetraploid Campanula americana.

28

This objective of this study was to analyze the genetic diversity and impact of

inbreeding on the UF SHB breeding program. The founding clones in the SHB breeding

program were identified and used to determine the coefficient of coancestry between

the founders, SHB cultivars released by UF, and the parents utilized for one cycle of

crosses made in 2011. The inbreeding coefficient of each released SHB cultivar and the

parents of the 2011 breeding population was calculated and compared to see if

inbreeding is stable or increasing over time.

Materials and Methods

Pedigree Information

A pedigree file was created using 32 SHB cultivars released from the University

of Florida as well as 146 parents used to make the 2011 crosses that were first

evaluated in 2013. The file was assembled from in house pedigree records, patent

filings, available literature (Brooks & Olmo, 1952, 1997), and the National Clonal

Germplasm Repository (NCGR) – Corvallis Vaccinium Catalog. The file was ordered

chronologically, beginning with the founding parents in 1908. Genotypes of unknown

origin, including founders, were assumed to be unrelated and not inbred.

Coefficient of Coancestry

The coefficient of coancestry was obtained from a pedigree-based relationship

matrix (A-matrix). The A-matrix was constructed as described in Slater et al. (2013), for

every individual with parents and :

� If and 푘of are unknown,푠 they푑 are assumed unrelated, and:

� 푠 푑 푘 = = 0; = 1( 1)

푘�� 푘�� 1 + 푗 푖 − = 4 푤 푘��

29

If only is known, is assumed unrelated, and:

푠 푑 = = 0.5 ; = 1( 1)

�� �� �� 푘 5푘+ 7 +�4푘 �(1푗 )푖 − = 24 푤 푘�� − 푤 푘�� If and are known:

푠 푑 = = 0.5 + ; = 1( 1)

�� �� �� �� 1 푘+ 2 푘+ (1 �)푘 +푘(1� 푗 ) 푖+−3 = 6 푤 − 푤 푘�� − 푤 푘�� 푘�� 푘�� Finally, the matrix Aw is given by = 4 . Because the matrix is built assuming uniform

풘 ploidy, all calculations were made푨 assuming퐊 tetrasomic inheritance with no double

reduction, = 0.

Inbreeding푤 Coefficient

The inbreeding coefficient ( ) was obtained from the equation described in

� Gallias (2003): 퐹

[ + ( ) ] = ퟏ ퟐ풗 − ퟏ 퐹� 푘�� where is the individuals relation withퟐ풗 itself found on the diagonal of the matrix

�� and is the gametic푘 ploidy level, 2.

풗 Results and Discussion

Identification of the Founding Clones

The objective of the SHB breeding program at UF was to develop cultivars that

would produce high-quality fruit adapted to Florida’s climate, and that could be

harvested in the early season (April to mid-May). Thus, the breeding program was

founded on the resulting progeny of crosses between V. darrowii × rabbiteye, V.

darrowii × NHB, and NHB × NHB. Composite crosses were made between (NHB × V.

30

darrowii) × NHB, (V. darrowii × rabbiteye) × (V. darrowii × NHB), and (V. darrowii × rabbiteye) × NHB (Sharpe & Darrow, 1959). These complex hybrids were then crossed to advanced northern selections in order to maximize genetic diversity and maintain high fruit quality. After the crosses were made, the progeny underwent several rounds of recurrent selection (Lyrene, 1997). The low chilling requirement obtained from the native species V. darrowii and rabbiteye enabled production in Florida and other areas worldwide with a similar climate. V. darrowii, a diploid in nature, was unique in its ability to generate fertile hybrids with both NHB and rabbiteye due to unreduced gamete formation. This allowed the various clones of V. darrowii to act as a genetic bridge between the two species with differing ploidal levels (Draper, 1977; Lyrene, 1997;

Lyrene et al., 2003; Qu & Hancock, 1995; Qu et al., 1998; Vorsa & Novy, 1995).

The NHB cultivars used in these crosses included ‘Angola’, ‘Berkeley’, ‘Bluecrop’,

‘Earliblue’, and ‘Ivanhoe’ (Sharpe & Darrow, 1959). GA23-46 (‘Dixi’ × ‘Hildenbrant’), F6

(‘Crabbe 4’ × ‘Stanley’) and F72 (‘Wareham’ × ‘Pioneer’) were advanced selections in use by various breeding programs at the time that were also used in the initial SHB crosses. Other selections used for SHB germplasm development included those with an

E prefix (e.g. E66), which were brought to Florida by Arthur Elliott and provided for use as parents in crossing (P.M. Lyrene, pers. comm.). Unique among the parents used in the founding event was one referred to as Clone ‘No.3’, which was an interspecific hybrid between the lowbush (V. angustifolium) cultivar ‘North Sedgwick’ and the NHB cultivar ‘Coville’ (Table 2-1). The lineage of the NHB selections and cultivars can be traced back to the seven original selections by Frederick Coville as well as the later wild selections ‘Crabbe 4’, ‘Hildenbrant’, and ‘North Sedgwick (Table 2-1). In this study a

31

founder was defined as a wild selection that has contributed genetic material to more than half of the cultivars released by UF or parents of the 2011 breeding progeny. The founders ‘Brooks’, ‘Rubel’, and ‘Sooy’ are in over 95% of the released SHB cultivars and the parents of the 2011 breeding progeny (Table 2-1). High percentages (> 85%) can also be seen for ‘Chatsworth’, ‘Grover’, and ‘Russell’. When comparing the SHB selections to the parents of the 2011 breeding population, the percentage of SHB selections that include ‘Crabbe 4’ and ‘North Sedgwick’ as founders increased 18%.

‘Crabbe 4’ is a grandparent to ‘Wolcott’ and ‘Murphy’ and great-grandparent to ‘O’Neal’, while ‘North ‘Sedgwick’ is a grandparent to ‘E22”, all which were frequently used as parents in the establishment of the program.

Introgression of the southeastern Vaccinium species allowed production to geographically expand by contributing genes for drought tolerance, disease resistance, low chilling requirement, and improved fruit quality in warm and humid conditions

(Sharpe, 1953). These southern species were represented predominantly by wild selections of V. darrowii including ‘FL 4B’, ‘FL 4A’, and ‘FL 6A’. ‘FL 4A’ can be found in

97% of the SHB cultivars and 95% of the 2011 breeding population (Table 2-2). The genetic contribution of ‘FL 4B’ increased 27% between the released SHB cultivars and the 2011 parent population. This is likely due to the use of selections from the North

Carolina State University blueberry breeding program (NC1523’, ‘NC1528’, ‘NC1524’) which have been introduced into the UF breeding germplasm over the past 25 years.

These selections were derived from an interspecific cross between ‘Bluecrop’ and ‘FL

4B’.

32

Rabbiteye and V. tenellum genomes were incorporated through initial crosses of the rabbiteye cultivar ‘Woodard’ (‘Ethel’ × ‘Callaway’) and interspecific hybrid ‘US17’

(‘Callaway’ × V. tenellum) (Table 2-2). ‘Callaway’ is the progeny of a cross between wild rabbiteye selections ‘Black Giant’ and ‘Myers’. Their presence has decreased in the

2011 parent population in comparison to the released SHB cultivars because of the extensive time between when they were incorporated into the program and the current parent population. In the 1980s, native V. elliottii and V. fuscatum clones were added to the breeding germplasm and have a greater genetic presence in the 2011 breeding population (Table 2-2). Their genetic contribution is expected to grow as more of their progeny are used in crosses.

Genetic Contribution From The Founding Clones

Northern Highbush Genetic Contributions. Coefficients of coancestry between the 32 released cultivars and the founding clones from the original NHB domestication are reported in Tables 2-3. Similarly, coefficients of coancestry among the parents of the 2011 parent population, used to determine genetic parameters (see Chapters 3 and

4), and the NHB founding clones are given in Table 2-6. The mean coefficient of coancestry between the UF SHB cultivar releases and NHB selections ranged from 0.01 to 0.13. The parent population had very similar mean coefficients, ranging from 0.01 to

0.10. All the cultivars except ‘Meadowlark’ have a relationship with ‘Brooks’, ‘Sooy’ and

‘Rubel’ (Table 2-3). ‘Meadowlark’ results from a recent introgression project using sparkleberry (Vaccinium arboreum), as well as wild highbush blueberry native to Florida and V. darrowii (Olmstead et al., 2013). None of the founders in the pedigree for

‘Meadowlark’ have any similarity to the original NHB founders (Table 2-3).

33

Similar to the findings of Hancock and Siefker (1982), and Ehlenfeldt (1994) for

NHB cultivars, the largest genetic contributions for the SHB released cultivars and the parent population came from ‘Brooks’, ‘Sooy’, and ‘Rubel’. This supports the conclusion of Brevis et al. (2008) that the SHB blueberries are not genetically diverse from the oldest NHB cultivars. The heavy backcrossing to NHB in the beginning of the program may explain their strong genetic presence.

Coefficients of coancestry reported for ‘Sharpblue’, ‘Misty’, and ‘Avonblue’ were analogous to previous studies (Ehlenfeldt, 1994; Hancock & Siefker, 1982). As a whole,

‘Brooks’, ‘Sooy’, and ‘Rubel’ have the closest relationships with ‘Avonblue’, ‘Misty’, and

‘Star’, with coefficients ranging from 0.09 to 0.20. Apart from the previously noted outlier in ‘Meadowlark’, the newer releases ‘Arcadia’, ‘Avanti’, ‘Bobolink’, ‘Chickadee’, ‘Endura’,

‘Flicker’, ‘Indigocrisp’, ‘Kestrel’, ‘Raven’, and ‘Vireo’ (2009-2015) have coefficients comparable to the overall mean (Table 2-3). The stability in the newest releases as well as the similarity between the coefficients obtained for both populations suggests that the alleles from the NHB founders are established in the population and frequencies are likely not decreasing with increasing selection cycles. In autotetraploids alleles from one parent occur together with a one-third probability, creating an inherent linkage drag, which may explain the maintenance of allele frequencies over time (Gallias, 2003).

Southern Species Genetic Contributions. Contributions of the southern selections to the released cultivars and the 2011 parent population are given in Tables

2-5 and 2-8. Overall, the mean coefficients between these two groups were very similar.

The largest genetic contributions for V. darrowii clones used in the first crosses came from ‘FL 4A’, with minor contributions from ‘FL 4B’, and ‘FL 6A’. Cultivars with the

34

largest amount of ‘FL 4A’ were some of the first SHB cultivar releases from UF:

‘Sharpblue’, ‘Sapphire’, ‘Avonblue’, and ‘Sebring’ (Table 2-5). Coefficients for ‘FL 4B’ have increased in recent releases with the highest coefficient 0.19 obtained in parental selection FL05-442 (Table 2-7). This may result from selection towards individuals with a lower chilling requirement. There is currently an interest in developing cultivars more adapted to evergreen management, which may lead to an increase the genetic contribution of V. darrowii in selection of future cultivars. In contrast, the rabbiteye clones used in the development of SHB have contributed very little to the pedigree of released cultivars, with mean contributions of 0.01 in the cultivars and 0 in the 2011 parent population. It is possible that the coefficients may be higher than what are found in this study because parents used in crosses weren’t always distinguished by individual clone, but instead general species. Contributions from V. darrowii and rabbiteye selections were similar to calculations from Brevis et al. (2008) for ‘Emerald’, ’Millenia’,

‘Misty’, and ‘Star’. For rabbiteye selections, the contributions to ‘Sebring’ and

‘Sharpblue’ were found to be slightly smaller in this study 0.05 and 0.07 compared 0.09 and 0.15; and for V. darrowii, the contributions were found to be greater in this study for

‘Avonblue’ (0.13 compared to 0.05) and slightly less for ‘Sharpblue’ (0.25 compared to

0.29).

Vaccinium elliottii and V. fuscatum, two additional native Vaccinium species in

Florida with a similar low-chill phenotype as V. darrowii, were later introgressions to the program. Coefficients of coancestry for the clones V. elliottii 1 (Ve1) and V. fuscatum 2

(Vf2) (several different clones were used in crossing without specific selection numbers given) appear frequently in recently released cultivars. The largest genetic relationships

35

are between ‘Southmoon’ and Vf2 (0.25) and ‘Snowchaser’ and Ve1 (0.19) (Table 2-7).

VF2 is a grandparent for ‘Southmoon’ and Ve1 is a great-grandparent for ‘Snowchaser’.

Genetic Diversity Among the University of Florida Released Cultivars

The genetic relationships among the released cultivars from UF are shown in

Figure 2-1. The strongest relationships are in the bottom right corner and weaken moving left and upward. Coefficients of coancestry range from 0 to 0.68, with an overall mean of 0.19. The largest mean coefficients were 0.31, 0.27, and 0.27 and found for

‘Windsor’, ‘Sebring’ and ‘Springwide’, respectively. Twenty of the reported relationships have coefficients of coancestry larger than the expected parent, progeny and full-sib relationships of 0.50. This implies inbreeding has occurred due to individuals appearing more than once in the same lineage.

The largest coefficients of coancestry were between ‘Farthing’ and ’Avanti’

(0.68), ‘Windsor’ and ‘Farthing’ (0.66), ‘Windsor’ and ‘Indigocrisp’ (0.66), and ‘Farthing’ and ‘Vireo’ (0.65). An explanation of these large relationships follows: ‘Windsor’ is a parent for the full-sibs ‘Farthing’ and ‘Indigocrisp’, and grandparent to ‘Vireo’ and

‘Avanti’ through ‘Farthing’. ‘Sharpblue’ is a parent for ‘Windsor’, grandparent and great- great grandparent for ‘Farthing’ and ‘Indigocrisp’. Furthermore, ‘FL83-132’ is ‘the female parent of ‘Windsor’ and thus, grandparent for ‘Farthing’ and ‘Indigocrisp’, while its full- sib ‘FL83-135’ is a great-grandparent of ‘Farthing’ and ‘Indigocrisp’. ‘O’Neal’,

‘Sharpblue’, 61-3, 61-7, 65-28, ‘E30’, and ‘Earliblue’ appear in the pedigrees of both of the parents of Avanti (‘Farthing’ and FL02-12), as well as the parents of ‘Vireo’

(‘Farthing’ and FL02-13). This example of repetition of individuals in the pedigree accounted for the larger than expected relationships seen here.

36

The strong relationships among the released cultivars suggest a narrowing gene pool. However, despite the narrow base, sizable genetic gains are still being achieved for traits under selection. Rasmusson and Phillips (1997) proposed that genetic variation was not limited to the founding population and could arise through epistasis due to de novo generated diversity. For example, the barley cultivars ‘Robust’ and

‘Excel’ have a coefficient of coancestry estimated at 0.87, yet ‘Excel’ surpassed ‘Robust’ in malt extract and alpha amylase. Furthermore, multiallelism and polysomic inheritance has the potential to create greater genetic diversity by allowing for higher orders of allelic interactions and more recombination events (Gallias, 2003). These factors could explain the genetic gains despite the limited genetic diversity.

Inbreeding Coefficient

Inbreeding coefficients from the SHB cultivars and 2011 parent population are reported in Tables 2-5 and 2-8. The coefficients for the released cultivars range from 0 to 0.055, with a mean coefficient of 0.028. The coefficients for the 2011 parent population have a wider range from 0 to 0.08, with a mean coefficient of 0.023.

‘Meadowlark’ is the only cultivar to not have any inbreeding, due to its unique genetic background using sparkleberry in its pedigree (Table 2-5). ‘Avanti’, ‘Snowchaser’,

‘Arcadia’ and ‘Star’ displayed the highest inbreeding coefficients of 0.055, 0.055, 0.050, and 0.049, respectively. ‘Avanti’ and ‘Arcadia’ are two of the most recent releases from

UF. ‘Star’ and ‘Snowchaser’ were selected in 1987 and 1998. This range in years and the similar mean inbreeding coefficients between the 2011 parent population and released cultivars suggest that inbreeding is not necessarily associated with cycles of selection. Brevis et al. (2008) found no significant correlation between the selection cycle and inbreeding coefficient when examining SHB cultivars released from various

37

breeding programs. However, reports on NHB cultivars have found inbreeding to increase with cycles of selection (Brevis et al., 2008; Ehlenfeldt, 1994; Hancock &

Siefker, 1982) SHB cultivars have been under selection for half as long as NHB, which may explain the lack of correlation between inbreeding and cycles of selection.

Moreover, SHB germplasm began with a wide hybridization between multiple species and the addition of new species has broadened the genetic base (Table 2-2).

Hancock and Siefker (1982) and Ehlenfeldt (1994) reported inbreeding coefficients of 0.0381 and 0.0083 for ‘Avonblue’ and 0.0967 and 0.0052 for ‘Sharpblue’, respectively; this study found the coefficients to be 0.022 for ‘Avonblue’ and 0.031 for

‘Sharpblue’. Hancock and Siefker (1982) assumed disomic inheritance and Ehlenfeldt

(1994) accounted for variation in ploidy depending on the species. This study assumed all genotypes in the pedigree were tetraploids.

One of the consequences of recurrent selection is the inbreeding that naturally ensues. The large inbreeding coefficients found in this study suggest that inbreeding may not be as detrimental in autopolyploids. In fact, Ehlenfeldt (1994) suggested that it may allow for the accumulation of favorable alleles in blueberry. Galloway et al. (2003) found inbreeding depression in autotetraploid Campanula americana did not significantly affect most seed or germination characteristics and only manifested later in life. Husband and Schemske (1997) found inbreeding depression in Epilobium angustifolium was significantly higher in diploids (0.97) compared to than tetraploids

(0.67). The greater heterozygosity in autotetraploid species may act as a buffer to the negative effects of inbreeding depression.

38

Conclusion

The founding clones for the UF SHB breeding program can be split into two groups: the NHB cultivated germplasm dating back to the original domestication of the species, and the southeastern species that contributed alleles for low chilling requirement. The NHB genetic contribution has remained stable between the released

SHB cultivars and the parents used in 2011 crosses, with the only increase having been seen for ‘Crabbe 4’ and ‘North Sedgwick’. The contributions for the southeastern species have been more dynamic, with an increase in ‘FL 4B’ (V. darrowii), V. elliottii, V. fuscatum, and a resulting decline of V. tenellum and rabbiteye in the 2011 parent population. Overall, ‘Brooks, ‘Sooy’, ‘Rubel’ and ‘FL 4A’ have the highest genetic contributions to the released SHB cultivars and the 2011 parent population. There was no correlation seen between the inbreeding coefficient and selection cycle, nor was inbreeding seen to be detrimental. This could be a result of the genetic buffer created by multiallelism in this autotetraploid species. Although these are the best estimates given the available pedigree data, potential problems may still be present because pedigrees can be erroneous, the genetic contribution from each parent is assumed to be equal, and the clones used in crosses were not always specified. Future work should include molecular marker genotyping to verify these pedigrees. This knowledge will allow the breeder to accurately select parents for future crosses in order to maximize genetic gain and minimize potential inbreeding.

39

Table 2-1. Founding clones from high chilling requirement germplasm used in the University of Florida southern highbush blueberry breeding program. Founder Species Founder for UF Founder for UF cultivars (%) germplasm (%) Brooks V. corymbosum 97% 96% Chatsworth V. corymbosum 91% 91% Crabbe 4 V. corymbosum 72% 90% Grover V. corymbosum 88% 92% Harding V. corymbosum 66% 60% Hildenbrant V. corymbosum 66% 62% North Sedgwick V. angustifolium 66% 84% Rubel V. corymbosum 97% 95% Russell V. angustifolium 88% 92% Sooy V. corymbosum 97% 95%

Table 2-2. Founding clones from low chilling requirement germplasm used in the University of Florida southern highbush blueberry breeding program. Clone Species Founder for UF Founder for UF cultivars (%) germplasm (%) 4A V. darrowii 97% 95% 4B V. darrowii 41% 68% 6A V. darrowii 63% 56% Black Giant V. virgatum 69% 36% Ethel V. virgatum 50% 21% Myers V. virgatum 69% 36% V. elliottii 1 V. elliottii 19% 49% V. fuscatum 2 V. fuscatum 34% 60% V. tenellum 1 V. tenellum 78% 58%

40

Table 2-3. The coefficient of coancestry between the high chilling requirement founders and the southern highbush blueberry cultivars released by the University of Florida. Cultivar Brooks Chatsworth Crabbe 4 Grover Harding Hildenbrant North Sedgwick Rubel Russell Sooy Avonblue 0.16 0.02 0 0.03 0.06 0 0 0.20 0.01 0.14 Sharpblue 0.11 0.02 0 0.02 0 0.06 0 0.14 0.01 0.07 Misty 0.15 0.02 0 0.03 0.03 0 0 0.20 0.01 0.12 Sapphire 0.13 0.02 0 0.03 0.03 0.03 0 0.17 0.01 0.11 Sebring 0.15 0.05 0.06 0.01 0 0.03 0.03 0.18 0.03 0.07 Southmoon 0.03 0 0 0.01 0.01 0.01 0 0.04 0 0.03 Windsor 0.14 0.04 0.03 0.02 0 0.03 0.02 0.17 0.02 0.08 Springwide 0.14 0.04 0.03 0.02 0 0.03 0.02 0.17 0.02 0.08 Star 0.17 0.06 0.06 0.02 0.01 0 0.05 0.20 0.03 0.09 SantaFe 0.08 0.01 0 0.02 0.03 0 0 0.10 0 0.07 Millenia 0.10 0.04 0.06 0 0 0 0.03 0.11 0.02 0.04 Primadonna 0.10 0.04 0.06 0 0 0 0.03 0.11 0.02 0.04 San Joaquin 0.12 0.04 0.05 0.01 0.01 0.01 0.02 0.14 0.02 0.06 FLX-2 0.04 0 0 0.01 0.02 0 0 0.05 0 0.04 Abundance 0.11 0.03 0.03 0.01 0.01 0.01 0.02 0.13 0.02 0.06 Springhigh 0.10 0.03 0.03 0.01 0.01 0.01 0.02 0.12 0.01 0.06 Emerald 0.13 0.01 0 0.04 0.04 0 0 0.17 0.02 0.11 Flicker 0.08 0.02 0.02 0.01 0.01 0.02 0.01 0.11 0.01 0.06 Indigocrisp 0.11 0.03 0.02 0.02 0 0.02 0.02 0.14 0.02 0.07 Farthing 0.11 0.03 0.02 0.02 0 0.02 0.02 0.14 0.02 0.07 Snowchaser 0.14 0.03 0.03 0.02 0.02 0 0.03 0.17 0.02 0.09 Scintilla 0.09 0.03 0.02 0.01 0.01 0.02 0.01 0.12 0.01 0.06 Sweetcrisp 0.05 0.01 0.01 0.01 0.01 0.01 0 0.07 0.01 0.04 Meadowlark 0 0 0 0 0 0 0 0 0 0 Kestrel 0.10 0.03 0.05 0.01 0.01 0.01 0.01 0.11 0.02 0.05 Bobolink 0.08 0.02 0.02 0.02 0 0.01 0.01 0.10 0.01 0.04 Chickadee 0.04 0.01 0.03 0 0 0 0 0.05 0.01 0.02 Raven 0.13 0.03 0.04 0.02 0.01 0.02 0.02 0.16 0.02 0.08 Vireo 0.12 0.02 0.02 0.02 0.01 0.02 0.01 0.15 0.02 0.08 Avanti 0.13 0.03 0.03 0.02 0.01 0.01 0.02 0.17 0.02 0.08 Endura 0.10 0.03 0.02 0.02 0.01 0.01 0.01 0.13 0.02 0.06 Arcadia 0.10 0.02 0.02 0.02 0.01 0.02 0.01 0.12 0.02 0.06 Mean Coefficient 0.10 0.03 0.02 0.02 0.01 0.01 0.01 0.13 0.02 0.07

41

Table 2-4. The coefficient of coancestry between the low chilling requirement founders and southern highbush blueberry cultivars released by the University of Florida. Cultivar 4A 4B 6A Black Ethel M y e r s V. elliottii V. fuscatum V. tenellum Giant 1 2 1 Avonblue 0.13 0 0 0.02 0 0.02 0 0 0.03 Sharpblue 0.19 0 0.06 0.02 0.03 0.02 0 0 0 Misty 0.06 0 0 0.02 0 0.02 0 0 0.03 Sapphire 0.16 0 0.03 0.02 0.02 0.02 0 0 0.02 Sebring 0.13 0 0.03 0.01 0.02 0.01 0 0 0.01 Southmoon 0.04 0 0.01 0 0 0 0 0.25 0 Windsor 0.12 0.01 0.03 0.01 0.02 0.01 0 0 0.01 Springwide 0.12 0.01 0.03 0.01 0.02 0.01 0 0 0.01 Star 0.05 0.02 0 0.01 0 0.01 0 0 0.02 SantaFe 0.06 0 0 0.01 0 0.01 0 0 0.02 Millenia 0.03 0 0 0 0 0 0 0 0.01 Primadonna 0.03 0 0 0 0 0 0 0 0.01 San Joaquin 0.05 0 0.01 0.01 0 0.01 0 0 0.01 FLX-2 0.03 0 0 0 0 0 0 0 0.01 Abundance 0.06 0 0.01 0.01 0 0.01 0 0.03 0.01 Springhigh 0.07 0 0.01 0.01 0.01 0.01 0 0.13 0.01 Emerald 0.02 0.13 0 0.01 0 0 0 0 0.01 Flicker 0.06 0 0.01 0.01 0.01 0.01 0 0.06 0 Indigocrisp 0.08 0.07 0.02 0.01 0.01 0.01 0 0 0.01 Farthing 0.08 0.07 0.02 0.01 0.01 0.01 0 0 0.01 Snowchaser 0.05 0.04 0 0.01 0 0.01 0.19 0 0.01 Scintilla 0.06 0 0.01 0.01 0.01 0.01 0 0.05 0.01 Sweetcrisp 0.03 0 0 0 0 0 0 0 0 Meadowlark 0 0 0 0 0 0 0 0 0 Kestrel 0.04 0 0.01 0 0 0 0.03 0.06 0.01 Bobolink 0.04 0.07 0.01 0 0.01 0 0.03 0 0 Chickadee 0.01 0 0 0 0 0 0 0.02 0 Raven 0.07 0.04 0.02 0.01 0.01 0.01 0 0.02 0.01 Vireo 0.07 0.06 0.02 0.01 0.01 0.01 0.03 0.02 0.01 Avanti 0.07 0.06 0.01 0.01 0.01 0.01 0.03 0 0.01 Endura 0.07 0.03 0.01 0.01 0.01 0.01 0 0.02 0.01 Arcadia 0.06 0.07 0.01 0.01 0.01 0.01 0.03 0.02 0.01 Mean 0.07 0.02 0.01 0.01 0.01 0.01 0.01 0.02 0.01 Coefficient

42

Table 2-5. Inbreeding coefficients for the southern highbush blueberry cultivars released from the University of Florida. Cultivar Year Inbreeding Coefficient Avonblue 1964 0.022 Sharpblue 1966 0.031 Misty 1972 0.035 Sapphire 1982 0.037 Sebring 1983 0.040 Southmoon 1985 0.037 Windsor 1985 0.001 Springwide 1986 0.037 Star 1987 0.049 SantaFe 1987 0.004 Millenia 1988 0.007 Primadonna 1990 0.007 San Joaquin 1994 0.022 FLX-2 1994 0.001 Abundance 1995 0.031 Springhigh 1995 0.026 Emerald 1995 0.026 Flicker 1996 0.021 Indigocrisp 1998 0.041 Farthing 1998 0.041 Snowchaser 1998 0.055 Scintilla 1999 0.032 Sweetcrisp 2000 0.003 Meadowlark 2001 0 Kestrel 2002 0.037 Bobolink 2003 0.016 Chickadee 2004 0.002 Raven 2005 0.034 Vireo 2005 0.041 Avanti 2006 0.055 Endura 2006 0.030 Arcadia 2007 0.050 Mean Inbreeding Coefficient 0.028

43

Table 2-6. The coefficient of coancestry between the high chilling requirement founders and the parents of the 2011 southern highbush blueberry breeding population. Parent Brooks Chatsworth Crabbe 4 Grover Harding Hildenbrant North Sedgwick Rubel Russell Sooy 97-136 0.06 0.02 0.03 0.01 0 0.01 0.01 0.07 0.01 0.03 00-180 0.07 0.02 0.03 0.01 0 0.01 0.02 0.08 0.01 0.03 01-25 0.02 0.01 0.01 0 0 0 0 0.02 0 0.01 Corindi 02-073 0.01 0 0 0.02 0.02 0 0 0.09 0.01 0.05 03-52 0.10 0.02 0.05 0.02 0.01 0 0.01 0.13 0.02 0.06 Corindi 04-014 0.01 0.03 0.03 0.01 0 0 0.02 0.10 0.01 0.04 Corindi 04-017 0.01 0.03 0.03 0.01 0 0 0.02 0.10 0.01 0.04 Corindi 04-22 0 0 0 0 0 0 0 0 0 0 04-17 0.09 0.02 0.02 0.02 0.02 0 0.01 0.12 0.01 0.06 04-173 0.06 0.02 0.03 0.01 0 0.01 0.01 0.07 0.01 0.03 05-442 0.17 0.03 0 0.04 0.02 0 0 0.23 0.03 0.11 06-12 0.06 0.01 0.02 0.01 0.01 0 0.01 0.08 0.01 0.04 06-161 0.15 0.05 0.05 0.01 0 0 0.03 0.18 0.03 0.07 06-340 0.10 0.02 0.02 0.02 0.01 0.02 0.01 0.12 0.02 0.06 06-362 0.11 0.03 0.03 0.01 0 0.02 0.02 0.13 0.02 0.06 06-379 0.16 0.04 0.06 0.02 0.02 0 0.02 0.19 0.03 0.09 06-410 0.12 0.04 0.05 0.01 0 0 0.02 0.14 0.02 0.06 06-411 0.12 0.04 0.05 0.01 0 0 0.02 0.14 0.02 0.06 06-435 0.13 0.05 0.06 0.02 0 0 0.04 0.16 0.03 0.06 06-436 0.10 0.05 0.06 0.02 0 0 0.04 0.16 0.03 0.06 06-503 0.06 0.02 0.02 0.01 0 0.01 0.01 0.07 0.01 0.03 06-556 0.08 0.02 0.02 0.02 0.01 0.01 0.01 0.11 0.01 0.05 07-23A 0.09 0.03 0.03 0.01 0 0.01 0.01 0.10 0.01 0.05 07-26 0.06 0.02 0.02 0.01 0 0.01 0.01 0.07 0.01 0.03 07-158 0.09 0.03 0.03 0.01 0 0.01 0.01 0.10 0.01 0.05 07-166 0.06 0.02 0.02 0.01 0 0.01 0.01 0.07 0.01 0.03 07-224 0.09 0.02 0.02 0.02 0.02 0.01 0.01 0.11 0.01 0.06 07-246 0.13 0.02 0.02 0.03 0.02 0.01 0.01 0.16 0.02 0.08 07-338 0.10 0.02 0.02 0.02 0.01 0.01 0.01 0.12 0.01 0.06 07-339 0.10 0.02 0.02 0.02 0.01 0.01 0.01 0.12 0.01 0.06 08-39 0.10 0.03 0.04 0.01 0.01 0.01 0.02 0.12 0.02 0.05 08-61 0.07 0.02 0.02 0.01 0.01 0.01 0.01 0.08 0.01 0.04 09-01 0.04 0 0 0.01 0.01 0 0 0.06 0.01 0.03 09-06 0.08 0.03 0.03 0.01 0 0.01 0.02 0.09 0.01 0.04 09-10 0.08 0.02 0.02 0.01 0.01 0.01 0.02 0.10 0.01 0.05 09-12 0.10 0.03 0.04 0.01 0 0 0.02 0.11 0.02 0.05 09-14 0.11 0.02 0.02 0.02 0.01 0.01 0.01 0.13 0.01 0.07

44

Table 2-6. Continued. Parent Brooks Chatsworth Crabbe 4 Grover Harding Hildenbrant North Sedgwick Rubel Russell Sooy 09-29 0.08 0.02 0.02 0.01 0.01 0.01 0.01 0.09 0.01 0.04 09-32 0.08 0.02 0.02 0.01 0.01 0.01 0.01 0.09 0.01 0.04 09-37 0.10 0.03 0.03 0.02 0 0.01 0.01 0.12 0.02 0.06 09-45 0.02 0.01 0.02 0 0 0 0 0.02 0 0.01 09-68 0.07 0.03 0.04 0.01 0 0 0.01 0.08 0.01 0.03 09-82 0.08 0.01 0.01 0.02 0.02 0 0 0.10 0.01 0.06 09-216 0.04 0 0 0.01 0.01 0 0 0.06 0.01 0.03 09-267 0.12 0.03 0.03 0.02 0.01 0.02 0.02 0.15 0.02 0.07 10-03 0.12 0.03 0.03 0.02 0.01 0.02 0.02 0.15 0.02 0.07 10-04 0.10 0.03 0.02 0.02 0.01 0.01 0.02 0.13 0.01 0.07 10-15 0.05 0.01 0.01 0.01 0 0 0.01 0.06 0.01 0.03 10-18 0.09 0.02 0.02 0.02 0 0 0.01 0.12 0.02 0.05 10-22 0.06 0.01 0.01 0.01 0 0.01 0.01 0.08 0.01 0.04 10-26 0.10 0.03 0.04 0.01 0.01 0 0.01 0.13 0.02 0.06 10-40 0.09 0.02 0.02 0.01 0.01 0.02 0.01 0.11 0.01 0.05 10-54 0.04 0 0 0.01 0.01 0 0 0.05 0 0.03 10-99 0 0 0 0 0 0 0 0 0 0 10-195 0.08 0.02 0.01 0.02 0.01 0.01 0.01 0.11 0.01 0.06 10-532 0 0 0 0 0 0 0 0 0 0 10-617 0 0 0 0 0 0 0 0 0 0 11-1 0.08 0.02 0.03 0.01 0.01 0 0.01 0.09 0.01 0.04 11-2 0.04 0.01 0.01 0.01 0.01 0.01 0 0.05 0 0.03 11-4 0.12 0.03 0.03 0.02 0.01 0.02 0.02 0.15 0.02 0.07 11-5 0.07 0.02 0.01 0.01 0.01 0.01 0.01 0.09 0.01 0.04 11-6 0.06 0 0 0 0 0 0 0 0 0 11-7 0.06 0.01 0.01 0.01 0.01 0.01 0.01 0.07 0.01 0.03 11-8 0.06 0.01 0.01 0.01 0.01 0.01 0.01 0.07 0.01 0.03 11-9 0.06 0.01 0.01 0.01 0.01 0.01 0.01 0.07 0.01 0.03 11-10 0.06 0.01 0.01 0.01 0.01 0.01 0.01 0.07 0.01 0.03 11-11 0.06 0.01 0.01 0.01 0.01 0.01 0.01 0.07 0.01 0.03 11-12 0.06 0.01 0.01 0.01 0.01 0.01 0.01 0.07 0.01 0.03 11-13 0.06 0.01 0.01 0.01 0.01 0.01 0.01 0.07 0.01 0.03 11-14 0.09 0.03 0.03 0.01 0.01 0.01 0.01 0.11 0.01 0.05 11-15 0.09 0.02 0.03 0.01 0.01 0.01 0.01 0.11 0.01 0.05 11-16 0.09 0.02 0.02 0.02 0 0.01 0.01 0.11 0.01 0.05 11-17 0.09 0.02 0.02 0.02 0 0.01 0.01 0.11 0.01 0.05 11-18 0.09 0.02 0.02 0.02 0 0.01 0.01 0.11 0.01 0.05 11-19 0.06 0.01 0.01 0.01 0.01 0.01 0.01 0.07 0.01 0.03

45

Table 2-6. Continued. Parent Brooks Chatsworth Crabbe 4 Grover Harding Hildenbrant North Sedgwick Rubel Russell Sooy 11-20 0.12 0.03 0.02 0.02 0.01 0.02 0.02 0.16 0.02 0.08 11-22 0.06 0.01 0.01 0.02 0.01 0 0.01 0.08 0.01 0.04 11-23 0.11 0.03 0.02 0.02 0.01 0.01 0.01 0.14 0.01 0.07 11-24 0.11 0.03 0.03 0.02 0.02 0.01 0.01 0.14 0.02 0.07 11-25 0.11 0.03 0.03 0.02 0.02 0.01 0.01 0.14 0.02 0.07 11-26 0.09 0.03 0.03 0.01 0 0 0.02 0.11 0.02 0.04 11-27 0.08 0.02 0.02 0.01 0.01 0.01 0.01 0.10 0.01 0.05 11-28 0.07 0.02 0.01 0.01 0.01 0.01 0.01 0.09 0.01 0.05 11-29 0.08 0.02 0.01 0.01 0.01 0.01 0.01 0.09 0.01 0.05 11-30 0.08 0.02 0.01 0.01 0.01 0.01 0.01 0.09 0.01 0.05 11-32 0.10 0.02 0.02 0.02 0.01 0.01 0.01 0.13 0.01 0.06 11-33 0.08 0.02 0.01 0.02 0.01 0.01 0.01 0.11 0.01 0.05 11-34 0.08 0.02 0.01 0.02 0.01 0.01 0.01 0.11 0.01 0.05 11-35 0.08 0.02 0.01 0.02 0.01 0.01 0.01 0.11 0.01 0.05 11-36 0.11 0.03 0.02 0.02 0.01 0.01 0.01 0.14 0.02 0.07 11-37 0.11 0.03 0.02 0.02 0.01 0.01 0.01 0.14 0.02 0.07 11-38 0.12 0.03 0.03 0.02 0.02 0.01 0.02 0.15 0.02 0.08 11-40 0.09 0.03 0.03 0.01 0.01 0.01 0.01 0.10 0.01 0.05 11-41 0.09 0.02 0.02 0.02 0.01 0.01 0.01 0.11 0.01 0.05 11-42 0.09 0.02 0.02 0.02 0.01 0.01 0.01 0.11 0.01 0.05 11-43 0.11 0.03 0.03 0.02 0.01 0.01 0.01 0.13 0.02 0.06 11-45 0.04 0.01 0.01 0.01 0 0 0 0.05 0.01 0.03 11-46 0.09 0.02 0.02 0.02 0.01 0.01 0.01 0.11 0.01 0.06 11-47 0.07 0.02 0.03 0.01 0 0 0.02 0.08 0.01 0.03 11-48 0.10 0.02 0.02 0.02 0.01 0.01 0.01 0.12 0.01 0.06 11-49 0.10 0.02 0.02 0.02 0.01 0.01 0.01 0.12 0.01 0.06 11-50 0.10 0.02 0.02 0.02 0.01 0.01 0.01 0.12 0.01 0.06 11-51 0.06 0.02 0.02 0.01 0 0 0.01 0.07 0.01 0.03 11-52 0.09 0.02 0.02 0.01 0.01 0.01 0.01 0.11 0.01 0.05 11-54 0.10 0.02 0.02 0.01 0.01 0.01 0.01 0.12 0.01 0.06 11-55B 0.06 0.01 0.01 0.02 0 0 0.01 0.08 0.01 0.04 11-56 0.06 0.02 0.01 0.01 0 0.01 0.01 0.08 0.01 0.04 11-57 0.05 0.01 0.01 0.01 0.01 0 0 0.06 0.01 0.03 11-58 0.08 0.02 0.02 0.01 0.01 0 0.01 0.09 0.01 0.05 11-59 0.08 0.02 0.02 0.01 0.01 0 0.01 0.09 0.01 0.05 11-60 0.08 0.02 0.02 0.01 0.01 0 0.01 0.09 0.01 0.05 11-61 0.08 0.02 0.02 0.01 0.01 0 0.01 0.09 0.01 0.05 11-62 0.08 0.02 0.02 0.01 0.01 0 0.01 0.09 0.01 0.05

46

Table 2-6. Continued. Parent Brooks Chatsworth Crabbe 4 Grover Harding Hildenbrant North Sedgwick Rubel Russell Sooy 11-63 0.08 0.02 0.02 0.01 0.01 0 0.01 0.09 0.01 0.05 11-64 0.04 0.01 0.01 0.01 0 0 0.01 0.09 0.01 0.02 11-65 0.04 0.01 0.01 0.01 0 0 0.01 0.09 0.01 0.02 11-66 0.10 0.02 0.02 0.02 0.01 0.01 0.01 0.12 0.01 0.06 11-67 0.10 0.02 0.02 0.02 0.01 0.01 0.01 0.12 0.01 0.06 11-68 0.12 0.04 0.06 0.01 0.01 0.01 0.02 0.14 0.02 0.05 11-69 0.10 0.02 0.01 0.02 0.01 0.02 0.01 0.12 0.01 0.06 11-70 0.07 0.02 0.03 0.01 0 0.01 0.01 0.08 0.01 0.04 11-71 0.07 0.02 0.03 0.01 0 0.01 0.01 0.08 0.01 0.04 11-72 0.09 0.03 0.04 0.01 0 0 0.02 0.10 0.02 0.04 11-73 0.09 0.03 0.04 0.01 0 0 0.02 0.10 0.02 0.04 11-75 0.10 0.03 0.03 0.02 0.01 0.01 0.01 0.13 0.02 0.06 11-76 0.08 0.02 0.03 0.01 0 0.01 0.01 0.10 0.01 0.04 11-77 0.08 0.02 0.03 0.01 0 0.01 0.01 0.10 0.01 0.04 11-79 0.08 0.03 0.04 0.01 0 0 0.01 0.09 0.01 0.04 11-80 0.05 0.01 0.01 0.01 0.01 0.01 0 0.06 0.01 0.03 11-81 0.05 0.01 0.01 0.01 0.01 0.01 0 0.06 0.01 0.03 11-82 0.04 0 0 0.01 0.01 0 0 0.06 0.01 0.03 11-83 0.01 0 0 0 0 0 0 0.01 0 0.01 11-84 0.03 0.01 0.01 0 0 0 0.01 0.04 0.01 0.02 Corindi 0 0 0 0 0 0 0 0 0 0 HBcomp Mean 0.08 0.02 0.02 0.01 0.01 0.01 0.01 0.10 0.01 0.05 Coefficient *Arcadia’, ’Bobolink’, ‘Emerald’, ‘Farthing’, ‘Flicker’, ‘Indigocrisp’, ‘Meadowlark’, ‘Primadonna’, ‘Raven’, ‘Vireo’, and ‘Windsor’ were also parents, their coefficients can be found in Table 2-3, 2-4, 2-5.

47

Table 2-7. The coefficient of coancestry between the low chilling requirement founders and the parents of the 2011 southern highbush blueberry breeding population. Parent 4A 4B 6A Black Giant Ethel M y e r s V. elliottii 1 V. fuscatum 2 V. tenellum 1 97-136 0.04 0 0 0 0 0 0.06 0 0 00-180 0.03 0 0 0 0 0 0 0.03 0 01-25 0.01 0 0 0 0 0 0.03 0.03 0 Corindi 0.01 0.06 0 0 0 0 0 0 0 02-073 03-52 0.03 0 0 0.01 0 0.01 0.03 0.03 0.01 Corindi 0.02 0.01 0 0 0 0 0 0 0.01 04-014 Corindi 0.02 0.01 0 0 0 0 0 0 0.01 04-017 Corindi 0 0 0 0 0 0 0 0 0 04-22 04-17 0.04 0.04 0 0.01 0 0.01 0.06 0 0.01 04-173 0.04 0 0.01 0 0 0 0.06 0.05 0 05-442 0.01 0.19 0 0 0 0 0 0 0 06-12 0.02 0.03 0 0 0 0 0.03 0 0 06-161 0.04 0.04 0 0.01 0 0.01 0 0.02 0.01 06-340 0.06 0.07 0.01 0.01 0.01 0.01 0 0.02 0.01 06-362 0.08 0 0.02 0.01 0.01 0.01 0 0.02 0.01 06-379 0.05 0.02 0 0.01 0 0.01 0 0.03 0.01 06-410 0.04 0.01 0 0.01 0 0.01 0.03 0.02 0.01 06-411 0.04 0.01 0 0.01 0 0.01 0.03 0.02 0.01 06-435 0.04 0.07 0 0.01 0 0.01 0 0 0.01 06-436 0.04 0.07 0 0.01 0 0.01 0 0 0.01 06-503 0.04 0 0.01 0 0 0 0 0 0 06-556 0.05 0.03 0.01 0 0.01 0 0 0 0 07-23A 0.04 0 0.01 0 0 0 0 0 0 07-26 0.03 0 0 0 0 0 0 0.02 0 07-158 0.05 0 0.01 0 0 0 0 0 0 07-166 0.03 0 0 0 0 0 0 0.02 0 07-224 0.04 0.03 0.01 0 0 0 0.03 0.03 0 07-246 0.04 0.06 0.01 0.01 0 0.01 0 0 0.01 07-338 0.04 0.03 0.01 0 0 0 0 0.02 0.01 07-339 0.04 0.03 0.01 0 0 0 0 0.02 0.01 08-39 0.07 0 0.01 0.01 0.01 0.01 0.02 0.02 0.01 08-61 0.04 0 0.01 0 0 0 0 0.03 0 09-01 0.01 0.03 0 0 0 0 0 0 0 09-06 0.04 0 0.01 0 0 0 0.02 0.02 0 09-10 0.05 0 0.01 0 0 0 0.02 0.02 0 09-12 0.03 0.01 0 0 0 0 0 0.03 0 09-14 0.06 0.06 0.01 0 0.01 0 0.02 0.02 0 09-26 0.03 0.03 0 0 0 0 0 0.01 0 09-29 0.04 0 0.01 0 0 0 0.02 0.03 0 09-32 0.04 0 0.01 0 0 0 0.02 0.03 0 09-37 0.05 0.03 0.01 0.01 0.01 0.01 0 0.02 0.01 09-45 0.01 0 0 0 0 0 0 0.02 0 09-68 0.03 0 0 0 0 0 0.02 0.03 0 09-82 0.04 0.03 0 0 0 0 0 0.02 0.01 09-216 0.01 0.03 0 0 0 0 0 0 0 09-216 0.01 0.03 0 0 0 0 0 0 0

48

Table 2-7. Continued. Parent 4A 4B 6A Black Giant Ethel M y e r s V. elliottii 1 V. fuscatum 2 V. tenellum 1 09-267 0.08 0.01 0.02 0.01 0.01 0.01 0 0.02 0.01 10-03 0.09 0.01 0.02 0.01 0.01 0.01 0 0 0.01 10-04 0.07 0.01 0.01 0.01 0.01 0.01 0 0.04 0.01 10-15 0.02 0.01 0 0 0 0 0.01 0.01 0 10-18 0.01 0 0 0 0 0 0 0 0 10-22 0.03 0.03 0 0 0 0 0 0 0 10-26 0.03 0.02 0 0 0 0 0 0.02 0.01 10-40 0.06 0.02 0.01 0.01 0.01 0.01 0 0.04 0.01 10-54 0.02 0.02 0 0 0 0 0 0 0.01 10-99 0 0 0 0 0 0 0 0 0 10-195 0.05 0.04 0.01 0.01 0 0.01 0 0 0.01 10-532 0 0 0 0 0 0 0 0 0 10-617 0 0 0 0 0 0 0 0 0 11-1 0.03 0.02 0 0 0 0 0 0 0.01 11-2 0.03 0 0.01 0 0 0 0 0.03 0 11-4 0.09 0.01 0.02 0.01 0 0.01 0.03 0.01 0.01 11-5 0.05 0.02 0.01 0.01 0.01 0.01 0 0.03 0 11-6 0 0 0 0 0 0 0 0 0 11-7 0.03 0 0.01 0 0 0 0.01 0.02 0 11-8 0.03 0 0.01 0 0 0 0.01 0.02 0 11-9 0.03 0 0.01 0 0 0 0.01 0.02 0 11-10 0.03 0 0.01 0 0 0 0.01 0.02 0 11-11 0.03 0 0.01 0 0 0 0.01 0.02 0 11-12 0.03 0 0.01 0 0 0 0.01 0.02 0 11-13 0.03 0 0.01 0 0 0 0.01 0.02 0 11-14 0.05 0.01 0.01 0.01 0 0.01 0 0.02 0.01 11-15 0.05 0 0.01 0.01 0 0.01 0.01 0.03 0.01 11-16 0.06 0.05 0.01 0.01 0.01 0.01 0.01 0.01 0.01 11-17 0.06 0.05 0.01 0.01 0.01 0.01 0.01 0.01 0.01 11-18 0.06 0.05 0.01 0.01 0.01 0.01 0.01 0.01 0.01 11-19 0.03 0.01 0.01 0 0 0 0.02 0.02 0 11-20 0.08 0.02 0.02 0.01 0.01 0.01 0.02 0 0.01 11-22 0.01 0.08 0 0 0 0 0.01 0 0 11-23 0.07 0.01 0.01 0.01 0.01 0.01 0.02 0.02 0.01 11-24 0.05 0.04 0.01 0.01 0 0.01 0 0.07 0.01 11-25 0.05 0.04 0.01 0.01 0 0.01 0 0.07 0.01 11-26 0.02 0 0 0 0 0 0 0 0 11-27 0.04 0.03 0.01 0 0 0 0 0.01 0.01 11-28 0.05 0.02 0.01 0.01 0 0.01 0 0.01 0.01 11-29 0.05 0.02 0.01 0.01 0 0.01 0.02 0.02 0.01 11-30 0.05 0.02 0.01 0.01 0 0.01 0.02 0.02 0.01 11-32 0.05 0.03 0.01 0.01 0 0.01 0.01 0 0.01 11-33 0.04 0.04 0.01 0 0 0 0.04 0.02 0.01 11-34 0.04 0.04 0.01 0 0 0 0.04 0.02 0.01 11-35 0.04 0.04 0.01 0 0 0 0.04 0.02 0.01 11-36 0.07 0.03 0.01 0.01 0.01 0.01 0.03 0.02 0.01 11-37 0.07 0.03 0.01 0.01 0.01 0.01 0.03 0.02 0.01 11-38 0.06 0.02 0.01 0.01 0 0.01 0.03 0.02 0.01 11-40 0.03 0.02 0 0 0 0 0.02 0.04 0.01 11-41 0.05 0.04 0.01 0.01 0 0.01 0.03 0.02 0.01 11-42 0.05 0.04 0.01 0.01 0 0.01 0.03 0.02 0.01 11-43 0.04 0.03 0.01 0.01 0 0.01 0.02 0 0.01

49

Table 2-7. Continued. Parent 4A 4B 6A Black Ethel M y e r s V. elliottii V. fuscatum V. tenellum Giant 1 2 1 11-45 0.01 0.03 0 0 0 0 0 0 0 11-46 0.02 0.02 0.01 0 0 0 0.03 0.03 0.01 11-47 0.02 0 0 0 0 0 0 0 0.01 11-48 0.06 0.03 0.01 0.01 0.01 0.01 0.01 0.02 0.01 11-49 0.06 0.03 0.01 0.01 0.01 0.01 0.01 0.02 0.01 11-50 0.06 0.03 0.01 0.01 0.01 0.01 0.01 0.02 0.01 11-51 0.02 0 0 0 0 0 0.02 0.02 0 11-52 0.05 0.04 0.01 0.01 0.01 0.01 0.02 0.02 0 11-54 0.04 0.02 0.01 0.01 0 0.01 0 0.02 0.01 11-55B 0.02 0.11 0 0 0 0 0 0 0 11-56 0.04 0.02 0.01 0 0 0 0 0 0.01 11-57 0.02 0.02 0 0 0 0 0 0 0 11-58 0.03 0.02 0 0 0 0 0.03 0 0.01 11-59 0.03 0.02 0 0 0 0 0.03 0 0.01 11-60 0.03 0.02 0 0 0 0 0.03 0 0.01 11-61 0.03 0.02 0 0 0 0 0.03 0 0.01 11-62 0.03 0.02 0 0 0 0 0.03 0 0.01 11-63 0.03 0.02 0 0 0 0 0.03 0 0.01 11-64 0.02 0.01 0 0 0 0 0 0 0 11-65 0.02 0.01 0 0 0 0 0 0 0 11-66 0.04 0.05 0.01 0 0 0 0.02 0.02 0.01 11-67 0.04 0.05 0.01 0 0 0 0.02 0.02 0.01 11-68 0.04 0.02 0.01 0 0 0 0 0.04 0.01 11-69 0.06 0.03 0.01 0.01 0.01 0.01 0.02 0 0.01 11-70 0.04 0 0.01 0 0 0 0.01 0.02 0.01 11-71 0.04 0 0.01 0 0 0 0.01 0.02 0.01 11-72 0.03 0.01 0 0 0 0 0.03 0 0.01 11-73 0.03 0.01 0 0 0 0 0.03 0 0.01 11-75 0.05 0.02 0.01 0.01 0 0.01 0 0.02 0.01 11-76 0.05 0.02 0.01 0 0 0 0 0.02 0.01 11-77 0.05 0.02 0.01 0 0 0 0 0.02 0.01 11-79 0.03 0 0 0 0 0 0.02 0.03 0.01 11-80 0.03 0.02 0 0 0 0 0.03 0.05 0 11-81 0.03 0.02 0 0 0 0 0.03 0.05 0 11-82 0.01 0.03 0 0 0 0 0 0 0 11-83 0.01 0 0 0 0 0 0 0 0 11-84 0.02 0 0 0 0 0 0 0.01 0 Corindi 0 0 0 0 0 0 0 0 0 HBcomp Mean 0.04 0.02 0.01 0 0 0 0.01 0.01 0.01 Coefficient

50

Table 2-8. Inbreeding coefficients for the parents of the 2011 southern highbush blueberry breeding population. Parent Year Selected Inbreeding Coefficient 97-136 1997 0.004 00-180 2000 0.009 01-25 2001 0 Corindi 02-073 2002 0.004 03-52 2003 0.019 Corindi 04-014 2004 0.008 Corindi 04-017 2004 0.008 Corindi 04-22 2004 0 04-17 2004 0.015 04-173 2004 0.017 05-442 2005 0.057 06-12 2006 0.047 06-161 2006 0.055 06-340 2006 0.030 06-362 2006 0.038 06-379 2006 0.056 06-410 2006 0.043 06-411 2006 0.043 06-435 2006 0.069 06-436 2006 0.069 06-503 2006 0.007 06-556 2006 0.019 07-23A 2007 0.019 07-26 2007 0.008 07-158 2007 0.019 07-166 2007 0.008 07-224 2007 0.014 07-246 2007 0.026 07-338 2007 0.020 07-339 2007 0.020 08-39 2008 0.056 08-61 2008 0.010 09-01 2009 0.002 09-06 2009 0.031 09-10 2009 0.029 09-12 2009 0.023 09-14 2009 0.028 09-26 2009 0.010 09-29 2009 0.041 09-32 2009 0.041 09-37 2009 0.031 09-45 2009 0 09-68 2009 0.017 09-82 2009 0.014 09-216 2009 0.002 09-267 2009 0.056 10-03 2010 0.080 10-04 2010 0.038 10-15 2010 0.004 10-18 2010 0.013 10-22 2010 0.007

51

Table 2-8. Continued Parent Year Selected Inbreeding Coefficient 10-26 2010 0.028 10-40 2010 0.029 10-54 2010 0.004 10-99 2010 0 10-195 2010 0.024 10-532 2010 0 10-617 2010 0 11-1 2011 0.016 11-2 2011 0.004 11-4 2011 0.064 11-5 2011 0.017 11-6 2011 0 11-7 2011 0.009 11-8 2011 0.009 11-9 2011 0.009 11-10 2011 0.009 11-11 2011 0.009 11-12 2011 0.009 11-13 2011 0.009 11-14 2011 0.028 11-15 2011 0.040 11-16 2011 0.063 11-17 2011 0.063 11-18 2011 0.063 11-19 2011 0.006 11-20 2011 0.047 11-22 2011 0.009 11-23 2011 0.044 11-24 2011 0.035 11-25 2011 0.035 11-26 2011 0.019 11-27 2011 0.040 11-28 2011 0.016 11-29 2011 0.017 11-30 2011 0.017 11-32 2011 0.031 11-33 2011 0.022 11-34 2011 0.022 11-35 2011 0.022 11-36 2011 0.047 11-37 2011 0.047 11-38 2011 0.043 11-40 2011 0.025 11-41 2011 0.025 11-42 2011 0.025 11-43 2011 0.030 11-45 2011 0.004 11-46 2011 0.021 11-47 2011 0.011 11-48 2011 0.033 11-49 2011 0.033 11-50 2011 0.033 11-51 2011 0.014

52

Table 2-8. Continued. Parent Year Selected Inbreeding Coefficient 11-52 2011 0.033 11-54 2011 0.047 11-55B 2011 0.022 11-56 2011 0.009 11-57 2011 0.004 11-58 2011 0.008 11-59 2011 0.008 11-60 2011 0.008 11-61 2011 0.008 11-62 2011 0.008 11-63 2011 0.008 11-64 2011 0.004 11-65 2011 0.004 11-66 2011 0.021 11-67 2011 0.021 11-68 2011 0.037 11-69 2011 0.045 11-70 2011 0.017 11-71 2011 0.017 11-72 2011 0.021 11-73 2011 0.021 11-75 2011 0.030 11-76 2011 0.022 11-77 2011 0.022 11-79 2011 0.020 11-80 2011 0.006 11-81 2011 0.006 11-82 2011 0.002 11-83 2011 0 11-84 2011 0.002 Corindi HBcomp Unknown 0 Mean Inbreeding Coefficient 0.023

53

Figure 2-1. Heat map indicating the coefficient of coancestry among the 32 cultivars released by the University of Florida between 1976 and 2015.

54

CHAPTER 3 ESTIMATION OF DOUBLE REDUCTION LEVELS FOR SOUTHERN HIGHBUSH BLUEBERRY SELECTION TRAITS

Several important agronomic and horticultural crops such as alfalfa (Medicago sativa), potato (Solanum tuberosum), and highbush blueberry (Vaccinium corymbosum) are autotetraploids, with four homologous chromosomes present in each linkage group.

In autotetraploids random chromosome pairing during meiosis can occur between two homologous chromosomes (bivalents) or between more than two homologous chromosomes (multivalents). This occurrence is known as polysomic inheritance

(Fisher, 1947) and results in different segregation ratios for a given locus compared to diploid species, as up to four copies of an allele can lead to higher orders of allele interaction (Gallais, 2003). Thus, considering a bi-allelic marker, there are five possible states for a given A allele (referred hereafter as dosage): nulliplex, simplex, duplex, triplex and quadruplex, representing aaaa, Aaaa, AAaa, AAAa, and AAAA, respectively.

Polysomic inheritance can lead to the phenomenon known as double reduction, where sister chromatids segregate into the same gamete, creates difficulties when studying quantitative or population genetics in autotetraploid species. Double reduction also has practical implications in plant breeding because it can result in increased inbreeding even when two unrelated individuals are crossed (Gallais, 2003; Kerr et al.,

2012; Mather, 1935, 1936). These challenges have been identified as the main reasons the rate of genetic improvement has been limited and sometimes null in autotetraploid species as compared with their diploid counterparts (Brummer, 1999; Jansky, 2009;

Katepa-Mupondwa et al., 2002). As an example, the genetic gain for yield in alfalfa since 1940 has been between 0-0.30% per year on average, while 2% has been estimated in annual diploid crops (Brummer, 1999).

55

Breeding of autotetraploid species can take between 10-20 years to develop an improved cultivar (Bradshaw & Bonierbale, 2010; Lyrene, 2008). This could be due to the difficulty of eliminating deleterious alleles resulting from the greater number of possible states (i.e. Aaaa, AAaa, AAAa) (Gallais, 2003). Even though progress has been made in these species, current improvement strategies (e.g. phenotypic recurrent selection) and analysis methods are rudimentary and known to be inefficient (Slater et al., 2013).

In breeding, the use of linear mixed models, popularly known as BLUP (best linear unbiased prediction), is a method used to estimate variance components through

REML (restricted maximum likelihood) and predict breeding values. This methodology was developed by Henderson and first applied to livestock breeding (Henderson 1975,

Van Raden, 2008). In the past two decades, BLUP has gained popularity among plant breeders; first with perennial and forestry crops, and more recently with annual crops

(Piepho et al., 2008). In this period, the applicability of BLUP has been tested and adopted in breeding diploid crops (Bernardo, 1994, 1996a; Crossa, 2010; Massman et al., 2013; Resende et al., 2012), although it is still completely novel for many polyploid species like blueberry.

Blueberries (Vaccinium spp.), including northern highbush, southern highbush rabbiteye, and lowbush, are economically important small fruit crops for the U.S., with

543.5 million pounds harvested for fresh production in 2014 and valued at $824.9 million (USDA, 2015). U.S. production has increased an average of 10% every year since 2008, while acreage has increased by nearly 75% between 2005 and 2012.

These figures are expected to be maintained in the future as expansion moves into

56

many non-traditional production areas (Brazelton, 2013). This increase is based in part on the reputation of blueberries as healthy foods due to the high amounts of antioxidants such as anthocyanins that have been correlated with beneficial health effects for immunology, ophthalmology, cardiology, neurology, metabolism, and inflammation (Cassidy et al., 2013; Fan et al., 2012; Heim et al., 2012; Krikorian et al.,

2010; Li et al., 2013; Liu et al., 2012; McAnulty et al., 2011; Shukitt-Hale, 2012; Tipton et al., 2013; Yousef et al., 2013).

The breeding program at the University of Florida develops tetraploid (2n = 4x =

48) southern highbush blueberry (SHB) cultivars. These hybrid cultivars are derived mainly from crosses between tetraploid northern highbush cultivars (V. corymbosum) and the diploid species (V. darrowii), with lesser contributions from various other

Vaccinium species such as V. arboreum, V. elliottii, V. fuscatum, V. tenellum, and V. virgatum (Brevis et al., 2008; Lyrene, 1997, 2008; Olmstead et al., 2013). Many traits under selection in this species are polygenic, (Lyrene, 1993). The few inheritance studies of tetraploid V. corymbosum and V. corymbosum × V. darrowii crosses have shown inheritance patterns that follow expected autotetraploid segregation ratios

(Draper & Scott, 1969; Krebs & Hancock, 1989; Qu & Hancock, 1995). However, Vorsa and Novy (1995) found evidence for preferential pairing between subgenomes when using RAPD markers in a V. corymbosum × V. darrowii cross. When examining segregation ratios, Krebs and Hancock (1989) found random chromosome segregation, indicating no double reduction, when using four isoenzyme markers on the progeny derived from four highbush blueberry cultivars. Because double reduction is position dependent, locus position on the chromosome in relation to the centromere and which

57

chromosome the locus is located on will affect the level of double reduction observed

(Wu et al., 2001). Cytogenetic studies of meiotic pairing in highbush blueberry found the majority of pairing to be bivalent, with few quadrivalent formations (Jelenkovic & Hough,

1970; Jelenkovic & Harrington, 1971; Vorsa & Novy, 1995; Qu et al., 1998). In addition, the small size of highbush blueberry chromosomes (1.5 to 2.5 µm) could limit multivalent formation (Hall & Galletta, 1971; Krebs & Hancock, 1989).

Kerr et al. (2012) developed an extension of the numerator relationship matrix based on pedigree values ( matrix) for autopolyploids. This calculation includes the ploidy level and w coefficient퐀 (proportion of the alleles which are identical by descent).

Using this new approach, Slater et al. (2013) doubled genetic gains in autotetraploid potato by using BLUP estimated breeding values instead of conventioanl schemes such as phenotypic recurrent selecion.

The objective of this study was to estimate the level of double reduction present in the University of Florida SHB breeding germplasm and to study its effects on model fitting and genetic parameter estimations. This information will allow the future implementation of BLUP/REML methods to estimate breeding values in the breeding program, study genetic diversity, inbreeding, and enable genomic prediction.

Materials and Methods

Blueberry Breeding Population and Pedigree

The SHB population used in this study was generated from the breeding program at the University of Florida. This population was developed through 124 controlled crosses of 146 selected parents made in February 2011. Seeds were extracted from the resulting mature fruit, cold-stratified for 5 months, and planted in November 2011 as a family in 2 L pots in a greenhouse. One hundred seedlings from each family were

58

transplanted to trays in January 2012 and planted in May 2012 in a high-density

(~20,000 plants per 0.2 ha) nursery at the Plant Science Research and Education Unit

(PSREU) in Citra, Florida. In May 2013, 5 to 32 plants were selected from each family, and the remaining plants were removed from the planting, leaving 1,996 plants.

The pedigree describing the University of Florida SHB breeding program was assembled using internal pedigree records, the NCGR-Corvallis Vaccinium Catalog

(NCGR-Corvallis, 2012), and various literature (Brooks & Olmo 1952, 1997). The file was ordered chronologically beginning with the founding parents in 1908. The pedigree file contains 7755 lines with annotated pedigree since 1908.

Phenotyping

Phenotypic data collected for 1,996 seedlings in 2014 and 2015 included plant yield, fruit weight, fruit diameter, fruit firmness and fruit stem scar diameter. Yield was measured using a rating scale from 1 to 5, where 1 equaled none to very few berries on the plant and 5 equaled a yield comparable to commercial cultivars of the same age plant. Yield ratings were recorded over a two-week period before the blueberries were fully mature.

Fruit traits were determined using five randomly sampled representative berries from each genotype. Due to the diversity of harvest timing among so many individual selections, the berries were collected over a six-week period from the beginning of April to the middle of May in each year. The total sample of berries was hand-harvested from each genotype when they were fully mature and had no insect or visual damage. After harvest, the berries were kept in a cooler and stored at 4 C overnight prior to analysis.

Weight (g) was measured on each individual berry using an analytical scale (CP2202S,

Sartorius Corp., Bohemia, NY). The same five berries were oriented equatorially for

59

diameter (mm) and firmness (g·mm-1 compression force) measurement with a FirmTech

II (BioWorks Inc., Wamego, KS). The minimum and maximum force thresholds were set at 50 g and 350 g, respectively. Subsequently, each of the berries were placed with the stem scar positioned upward on a tray in a lightbox with a digital SLR (Pentax K-x,

Ricoh Imaging, Denver, CO) camera that was positioned 50 cm above the berries. The camera was set at a shutter speed of 1/50, aperture F 5.6, and ISO 200. A ruler was placed in each image as a size reference. Subsequently, the images were uploaded into FIJI (Schindelin et al., 2012) and the scale was set using the ruler. Scar diameter

(mm) was recorded for each fruit. The average weight (g), diameter (mm), firmness

(g·mm-1), and scar diameter (mm) of the five berries was calculated and used for subsequent analyses.

Analysis and Model Comparison

Variance components were estimated with Restricted maximum likelihood (REML) and breeding values with best-linear unbiased predictions (BLUP) using ASReml version 4.0 (Gilmour et al., 2014) with the following univariate linear mixed models:

y = Xb + Z1c + Z2r + Z3a + Z4b×a + e

where y is a vector of the trait being analyzed (i.e. yield, weight, firmness, diameter, or scar), b is the fixed effect of year, c is the random effect of column nested

2 2 within year ~ N(0, c), r is the random effect of row nested within year ~N(0, r), a is

2 the random effect ofσ genotype ~N(0, A g), b×a is the random effect of the interactionσ

2 2 of year by genotype ~N(0, b×a), e is theσ random residual effect ~N(0, I e), X, and Z1-

Z4 are incidence matrices forσ year, column, row, genotype and year-by-genotypeσ

2 2 2 interaction, respectively. The unbiased heritability was calculated as h = g /( g+

σ σ σ

60

2 2 2 2 c+ r+ bxg+ e). The A matrix was constructed with the AGHmatrix R package as

describedσ σ below.σ

Matrix Construction with AGHmatrix

In order to build the pedigree-based relationship matrices, we utilized AGHmatrix

which is freely-available for download at https://github.com/rramadeu/AGH-matrix. To

build the relationship matrix derived from the pedigree using the AGHmatrix R package,

the pedigree information needs to be formatted in three columns (individual, parent 1,

and parent 2) and read as a table in R (using read.table or read.csv functions). The

pedigree file should be sorted from older to newer generations, thus the oldest

generation (founders) without pedigree information should appear at the top of the file,

while the current generation being phenotyped is at the bottom of the file (see example

1). Before proceeding to build the relationship matrix, the AGHmatrix verifies that the

above order is correct, and if not, it will flag and permutate the order of individuals

misplaced.

The autotetraploid pedigree-based relationship matrix , was calculated as

풘 presented in Slater et al. (2013). The algorithm first computes푨 the matrix considering

the proportion of parental gametes that are IBD due to double reduction 퐊. Thus for

every individual with parents and : 푤

� If and 푘of are unknown,푠 they푑 are assumed unrelated, and:

� 푠 = 푑= 0;푘 for = 1( 1)

푘�� 푘�� 푗 푖 − 1 + = 4 푤 푘�� If only is known, is assumed non-related, and:

= 푠 = 0.5 푑; for = 1( 1)

푘�� 푘�� �푘��� 푗 푖 − 61

5 + 7 + 4 (1 ) = 24 푤 푘�� − 푤 푘�� If and are known:

푠 = 푑= 0.5 + ; for = 1( 1)

�� �� �� �� 푘 푘 �푘 1푘+ �2 + 푗(1 푖)− + (1 ) + 3 = 6 푤 − 푤 푘�� − 푤 푘�� 푘�� 푘�� Finally, the matrix Aw is given by = 4 . Six Aw matrices were constructed

풘 from the pedigree information by varying푨 the levels퐊 of double reduction (w): 0, 0.05,

0.10, 0.15, 0.20 and 0.25. Additionally, we also built the traditional diploid pedigree-

based relationship matrix ( matrix) using the AGHmatrix R package, which was first

described by Henderson (1975)퐀 and calculated according to the recursive method

presented in Mrode et al. (2014). Finally, the AGHmatrix R package can also build the

genomic-base relationship matrix, G matrix, for diploids using either the method

proposed by VanRaden (2008) or from Powell, Visscher, and Goddard (2010). To

calculate the G matrix, molecular data need to be in table format where n is the number

of individuals (in rows) and m is the number of loci (in columns). The loci information is

set to -1, 0, 1, for the genotypes aa, Aa, and AA, respectively. Although the G matrix

capability is present in the package, the genomic matrix was not built or used in this

study.

The autotetraploid matrices with different levels of double reduction were used in

the linear mixed model described above. These autotetraploid models were compared

against the diploid counterpart using the Akaike information criteria (AIC) value to

measure the goodness of fit (Akaike 1974). As the level of double reduction (w) is

62

unknown, the model that maximized the likelihood of the data given the parameters of the linear mixed model was used to estimate the proportion of double reduction.

Results and Discussion

Double Reduction in Blueberry

The mode of inheritance for yield, weight, firmness, diameter, and scar was analyzed using seven different models assuming disomic inheritance or tetrasomic inheritance with a level of double reduction (w) ranging from 0 to 0.25. The fit of each model was assessed using the AIC value obtained from the REML analysis and the difference between the diploid model and the tetraploid model was plotted, with larger differences indicating a better fit (Figure 3-1). Yield, weight, and diameter had similar negative slopes, each decreasing more dramatically with increasing double reduction, whereas the quality of fitness for scar and firmness was similar over the various levels of double reduction. The tetraploid model assuming w = 0 was the best fit for 4 of the 5 traits. Firmness was the only trait that had the best fit when double reduction was larger than zero; the fitness of the model was maximized when double reduction ranged from

0.15 to 0.20. The absence of double reduction for the majority of traits is similar to the findings of Krebs and Hancock (1989) using four isoenzyme makers in V. corymbosum.

A similar study in autotetraploid potato that compared the fit of each model found the proportion of double reduction varied from 0 to 0.25, depending on the trait (Slater et al.,

2013).

Double reduction is expected to be vary from trait to trait because it is a position dependent phenomenon, fluctuating between chromosomes based on the frequency of multivalent formation and within chromosomes depending on where the measured loci reside (Wu et al., 2001). Bourke et al. (2015) found a parallel increase between the rate

63

of double reduction and the distance from the centromere using a high-density set of markers in autotetraploid potato. The rate of double reduction increases towards telomeres because there is a greater probability of a crossover to occur between the centromere and the loci (Bourke et al., 2015; Luo et al., 2006; Welch, 1962; Wu et al.,

2001). The high proportion of double reduction for firmness suggests that genes controlling this trait may reside toward the distal end of a linkage group. The genes for yield, weight, diameter, and scar may be located near the centromere, quantitative in nature and dispersed over several chromosomes, or located on chromosomes that don’t exhibit multivalent formation.

Evidence has been seen for these last two hypotheses. Breeding traits for blueberry are polygenic, with each gene contributing a small effect to the overall phenotype (Lyrene & Ballington, 1986; Lyrene, 1993). The small V. corymbosum chromosome size could limit multivalent formation (Hall & Galletta, 1971; Krebs &

Hancock, 1989), and cytogenetic studies of meiotic pairing in V. corymbosum found the majority of pairings to be bivalent (Jelenkovic & Harrington, 1971; Jelenkovic & Hough,

1970; Qu et al., 1998; Vorsa & Novy, 1995). Furthermore, preferential pairing has been reported in V. corymbosum and V. corymbosum x V. darrowii hybrids (Draper & Scott,

1969; Vorsa & Novy, 1995). Preferential pairing lowers the expected frequency of multivalent formation between homologous chromosomes (Sybenga, 1996).

Effect of Double Reduction on Inheritance

The estimated narrow-sense heritability for each trait was compared assuming disomic inheritance and tetrasomic inheritance with the proportion of double reduction that best fit each trait (Table 3-1). The tetrasomic inheritance for yield, weight, diameter, and scar was calculated assuming w = 0, while firmness was estimated assuming w =

64

0.15. Yield, weight, diameter, and scar exhibited effectively the same heritability under both models. The heritability estimations for firmness had the largest difference between the two models, decreasing from 0.40 to 0.35 due to the larger proportion of double reduction used in the calculation. As seen in Figure 3-2, there is a small difference between the disomic estimated heritability for firmness and the tetrasomic estimated heritability with no double reduction. The difference becomes more pronounced and the narrow-sense heritability for the trait decreases as the proportion of double reduction increases. This trend was seen in all the traits measured (data not shown). Comparing disomic and tetrasomic inheritance with w = 0.10 in autotetraploid potatoes, Slater et al.

(2013) observed no consistent trend between the two modes of inheritance. For example, the estimated heritability for tuber specific gravity was 0.73 assuming disomic inheritance and 0.74 assuming tetrasomic inheritance with w = 0.10, while the estimated heritability for early blight resistance was 0.57 assuming disomic inheritance and 0.44 assuming tetrasomic inheritance with w = 0.10. The difference between these two studies could be due the varying genetic architecture between the traits measured.

The negative slope for heritability in firmness in Figure 3-2 could be a result of the increase in inbreeding and the resulting lack of allelic variation. Because heritability is the measure of the genetic variance that contributes to the overall phenotypic variance, the heritability will decrease as the genetic variance decreases.

Effect of Double Reductions on the A-matrix Construction

The statistical summary of the kinship coefficient for each level of double reduction can be seen in Table 3-2. The maximum coefficient was determined by the amount of double reduction, whereas the minimum kinship coefficient was unaltered by the level of double reduction. A large number of the individuals in the pedigree had

65

unknown parents, which negatively skewed the distribution. The unknown parents were a result of missing information, unspecified clones, and wild selections. These individuals were assumed to be unrelated and thus unaffected by double reduction. As double reduction increased the kinship coefficient increased as well. This can be seen by comparing the kinship coefficients obtained for w = 0 and those obtained for w = 0.25

(Table 3-2). The effects of double reduction are cumulative with larger kinship coefficients more susceptible to its effects. In the first quartile, the difference between the kinship coefficients estimated assuming w = 0 and the coefficient estimated assuming w = 0.25 was 0.021. Using the same scale, the difference between the two models more than doubles (0.046) in the third quartile.

BLUP uses these kinship coefficients to predict breeding values and inaccurate estimations of double reduction could lead to bias predictions. Because double reduction has a larger impact on stronger relationships, full-sib relationships and parent/progeny relationships are more vulnerable. However, there was a strong correlation (r = 0.999) between the diploid and tetraploid (w = 0.15) estimated breeding values for firmness (Figure 3-3). This strong correlation was seen for all traits (data not shown). Similarly, Slater et al. (2013) found a strong correlation (>0.95) for 8 of the 9 traits when comparing the diploid estimated breeding values with the autotetraploid estimated breeding values. The deep pedigree could help balance the erroneous estimations, resulting in small changes to the breeding values over various levels of double reduction.

Changes in Depth of Pedigree on Double Reduction Estimates

Various pedigree depths, going back to 2009, 2004, 1994 and 1984 (5, 10, 20, and 30 years of pedigree, respectively) and their influence on the estimated proportion

66

of double reduction are shown in Figure 3-4. A pedigree depth of five years indicated a double reduction level of 0.25 was the best fit for four of the five traits. Each of the traits in this graph exhibited positive slopes, with fruit diameter plateauing at 0.15 and 0.20.

The fit of the model assuming no double reduction was equivalent to the fit of the diploid model for all of the traits. The slopes for the graphs using 10, 20, and 30 years of pedigree information varied for each trait. The addition of pedigree information changes the majority of slopes from positive to negative. Furthermore, the larger the pedigree the more traits that have a better fit when w = 0 and a worse fit when w = 0.25.

Shallow pedigrees underestimate the actual kinship coefficient between two individuals. Boyce (1983) found that shallow pedigrees miscalculated the level of inbreeding when analyzing 10 Standardbred stallions with 30 generations worth of pedigree information. Inbreeding has been documented in northern highbush blueberries and SHB (see Chapter 2, Brevis et al., 2008; Ehlenfeldt, 1994; Hancock &

Siefker, 1982). This is due to the small genetic pool used in the founding event, the recurrence of the same individual in a single pedigree, and the effects of recurrent selection. The shallow pedigrees cannot accurately account for the inbreeding so the higher proportions of double reduction compensate for the lack of information by increasing the kinship coefficient between the individuals. Shallow pedigrees can lead the erroneous conclusions in the estimation of relationship coefficients and affect the correct level of double reduction for each trait and ultimately impact the estimated heritability, inbreeding, and breeding values.

67

Conclusions

The majority of traits (yield, weight, diameter, and scar) had the best fit assuming tetrasomic inheritance with no double reduction. The only exception was firmness, which had the best fit when double reduction was between 0.15 and 0.20. These estimations could indicate where the genes for each trait reside on the chromosomes.

The estimated narrow-sense heritability was essentially the same when assuming disomic inheritance or tetrasomic inheritance with no double reduction. Changes in the estimation of heritability occurred with larger proportions of double reduction.

Furthermore, there was a high correlation between breeding values predicted assuming disomic inheritance and those predicted assuming tetrasomic inheritance. The effects of double reduction are cumulative and shallow pedigree depths can underestimate the true kinship coefficient and as a result report inflated levels of double reduction. These accurate estimations of double reduction will allow for precise selection based on BLUP predicted breeding values and REML estimated variance components.

68

Table 3-1. The heritability of five breeding traits calculated assuming disomic and tetrasomic inheritance. Tetrasomic inheritance was estimated assuming no double reduction for all traits except firmness. Disomic Inheritance Tetrasomic Inheritance Trait Heritability Standard Error Heritability Standard Error Yield 0.46 0.02 0.45 0.02 Weight 0.58 0.02 0.58 0.02 Firmness* 0.40 0.03 0.35* 0.02* Fruit Diameter 0.26 0.03 0.26 0.03 Scar Diameter 0.54 0.02 0.54 0.02 * Tetrasomic inheritance was calculated assuming double reduction of 0.15

Table 3-2. Statistical summary of the kinship coefficient assuming various amounts of double reduction. Double Reduction Minim um First Quartile Median Mean Third Quartile Maxim um 0 0 0.067 0.106 0.115 0.148 1.000 0.05 0 0.071 0.113 0.123 0.158 1.050 0.10 0 0.076 0.120 0.130 0.168 1.100 0.15 0 0.080 0.126 0.138 0.177 1.150 0.20 0 0.084 0.133 0.145 0.186 1.200 0.25 0 0.088 0.139 0.151 0.194 1.250

Figure 3-1. The Akaike information criterion (AIC) difference between the diploid model and the tetraploid models assuming double reduction values from 0 to 0.25. The AIC values were obtained from ASReml 4.0.

69

Figure 3-2. The estimated narrow-sense heritability for firmness assuming disomic inheritance and tetrasomic inheritance with increasing coefficients of double reduction.

70

Figure 3-3. Correlation of breeding values for firmness estimated assuming disomic and tetrasomic inheritance with w = 0.15.

71

Figure 3-4. Pedigree depths of 5, 10, 20, and 30 years and their influence on the estimated amount of double reduction for each trait.

72

CHAPTER 4 ESTIMATION OF GENETIC PARAMETERS IN SOUTHERN HIGHBUSH BLUEBERRIES

Blueberries (Vaccinium spp.) are among the top five economically important, non-citrus fruits in North America (USDA, 2015). Consumption continues to dramatically increase with new research supporting their superfood status and year round availability due to cultivars adapted to a range of climates (Brazelton, 2013; Isogai et al, 2009). The industry in Florida has capitalized on this growth with total production nearly quadrupling since 2005 (USDA, 2015). This surge in production is a result of low chill, early ripening cultivars known as southern highbush blueberries (SHB) developed by the University of

Florida (UF). SHB cultivars originated through crosses between northern highbush

(Vaccinium corymbosum) cultivars and advanced selections and native southeastern

Vaccinium species (Sharpe & Darrow, 1959). These fruit are harvested at a unique time in the season, resulting in prices at least $1.50 to $2.80 more per pound than any other state (USDA, 2015).

To date, the breeding program at UF has been based on phenotypic recurrent selection. This method focuses on population improvement by crossing highly heterozygous parents to produce segregating populations with more outstanding phenotypes than their parents (Lyrene, 2002, 2005). Many successful cultivars have been developed using this method, but it can take 12 to 15 years to evaluate a selection before releasing it as a cultivar. Parent selection is paramount in order to decrease inbreeding depression, maintain genetic diversity, and maximize genetic gains (Lyrene,

2005). Each year, over 100 parents are crossed creating more than 100 families

(Lyrene, 2002). Approximately one-third of the selected parents are young seedlings that have exhibited exceptional phenotypes after one year of evaluations. However,

73

early evaluations have limited use in phenotypic selection when the heritability for a trait is low (Bernardo, 1991). Primary selection traits for blueberries are assumed to be polygenic, with many genes contributing a small effect to the overall phenotype (Lyrene,

1993; Lyrene & Ballington, 1986). These traits include high yielding plants, excellent fruit quality, dry picking scar, and a balanced sweet, but acidic flavor (Coville, 1937;

Draper et al., 1982; Finn et al. 2014; Lyrene, 1993, 2005; Sharpe, 1953).

Best linear unbiased prediction (BLUP), along with restricted maximum likelihood

(REML) has been integral in livestock and forest breeding programs in order to predict breeding values and estimate variance components for complex traits (Henderson,

1975). This information allows the breeder to rank the breeding population to more accurately select parents, families, and individuals (Munoz et al., 2014; Piepho et al.,

2008). BLUP assumes an infinitesimal model with equal shrinkage towards the mean

(Henderson, 1975). BLUP shrinks the observed regression of the progeny towards the mean allowing for the improved analysis of quantitative traits, especially those with low heritability (Peipho et al., 2008). Slater et al. (2013) found genetic gains could potentially double for potatoes by using breeding values estimated by BLUP because of its greater accuracy compared to phenotypic values or progeny means. For low heritability traits, mid-parent estimated breeding values were considerably more accurate than mid- parent values (0.74 compared to 0.14) (Slater et al., 2013).

In blueberry, most studies on quantitatively inherited traits have been limited to general (GCA) and specific combining ability (SCA) (Aalders & Hall, 1975; Edwards et al., 1974; Erb et al., 1990; Finn & Luby 1986,1992). The few studies on the correlation between traits and their narrow-sense heritability were calculated from regression of the

74

offspring on the mid-parent value (Edwards et al., 1974; Finn & Luby, 1986). These studies used populations that have undergone selection, which violates the assumption of randomness and in turn biases their estimations and predictions (Henderson, 1975;

Lynch & Walsh, 1998). Furthermore, data for yield, firmness, scar, and fruit size was collected subjectively using a limited number of categories (5-10) (Darrow et al., 1939;

Draper et al., 1982; Edwards et al., 1974; Finn & Luby, 1986, 1992; Finn et al., 2003).

BLUP and REML remove this bias by assuming the genetic effects are random and by incorporating the pedigree of the tested population, dating back to the known founders (Henderson, 1975; Kerr et al, 2012; Piepho et al, 2008). The pedigree is integrated through the numerator relationship matrix, or A-matrix. The A-matrix accounts for the genetic covariances among individuals due to the expected shared alleles. This allows predicting the genetic merit of relatives, even if phenotypic data were not collected from them. The relatives in the pedigree will share an expected proportion of alleles depending on their relationship so their response is correlated and thus can be predicted. REML and BLUP are able to handle complex relationships, which make it well suited for the blueberry breeding program at UF. Bernardo (1996b) found that even with pedigree errors, such as missing parents or unexpected parental contributions to the progeny, the correlation between the predicted and observed performance was not affected when using BLUP.

While BLUP is standard practice in livestock and animal breeding it has not been widely adopted in plant breeding (Kerr et al., 2012; Peipho et al., 2008). The presence of polyploid species, particularly autopolyploids with complex patterns of inheritance may be a contributing factor to this low adoption rate. Until recently, there has been no

75

simple method to construct the A-matrix accounting for this complicated inheritance, a necessary step for utilizing REML and BLUP (Kerr et al., 2012; Slater et al., 2013).

Cultivated blueberries are autotetraploids, and have the possibility of four different alleles at one locus, compared to the two for diploids (Gallias, 2003). Furthermore, autotetraploids can experience the phenomenon of double reduction, leading to identical segments of sister chromatids segregating to the same gamete (e.g., AABB progeny from the biallelic simplex × nulliplex parental cross of AAAB/AAAA) (Mather,

1936). Double reduction generates inbreeding in the absence of consanguineous mating, which affects the heritability of a trait and thus the subsequent genetic gain (see

Chapter 3, Slater et al., 2013). The effects of double reduction on the inheritance of traits for blueberries is not accounted for in the analyses to date (Draper et al., 1982;

Edwards et al., 1974; Erb et al., 1990; Finn & Luby, 1986, 1992; Finn et al., 2003).

The aim of this study was to lay the foundation for selection based on estimated breeding values (EBVs) and variance components estimated by BLUP and REML. Our objectives were to: estimate the genetic parameters (heritability and genetic correlations) of primary selection traits in the UF blueberry breeding program while accounting for tetrasomic inheritance and the double reduction level found to best fit each evaluated trait (yield rating, flower bud density, fruit weight, fruit diameter, fruit firmness, fruit stem scar diameter, fruit soluble solids, and fruit pH), determine the genotype-by-environment interaction over two years, and estimate the possible genetic gain that could be obtained by using BLUP EBVs.

Materials and Methods

Blueberry Breeding Population

The blueberry breeding program at UF developed the SHB population used in this

76

study. This population was generated through 124 controlled crosses of 146 selected parents made in February 2011. Seeds were extracted from the resulting mature fruit, cold-stratified for 5 months, and planted as a family in 2 L pots in a greenhouse in

November 2011. One hundred seedlings from each family were transplanted to larger pots in January 2012 and planted in May 2012 in a high-density (~20,000 plants per 0.2 ha) nursery at the Plant Science Research and Education Unit (PSREU) in Citra,

Florida. In May 2013, 5 to 32 plants were selected from each family to advance in the breeding program. The remaining plants were removed, leaving 1,996 individuals.

These plants were tagged and labeled to insure proper identification throughout the course of the experiment

Seedlings were planted by family so that the same family made up one or two rows in the field. Pine bark mulch was incorporated into the soil profile for additional organic matter and used as weed mulch on the surface, similar to standard Florida blueberry production practices. Before the seedlings were planted, 22.5 kg of elemental sulfur was added to the field nursery (to reduce soil pH) and 11.3 kg of Epsom salt

(magnesium source) was added a month after planting. Granular ammonium nitrate fertilizer with a 12-4-8 NPK ratio was applied as needed. Overhead irrigation was used to supplement rainfall and for freeze protection. During the harvest period, the nursery was netted so birds could not affect the experiment results.

Phenotyping Methods

Phenotypic data were collected in 2014 and 2015 when the plants were 3 and 4 years of age. The seedlings were evaluated in both years for yield, fruit weight, fruit diameter, fruit firmness, and scar diameter. The genetic parameters for these traits were estimated based on the 2-year data. Flower bud density was measured in 2015, while

77

soluble solids and pH were measured on fruit harvested in 2014. These genetic parameters were estimated based on data from 1 year.

Yield was recorded on a 1 to 5 scale similar to Finn et al. (2003), with 1 having no to very few berries on the plant and 5 indicating a yield comparable to commercial cultivars. For both years, yield ratings were recorded over a two-week period before the blueberries were fully mature and harvest began. Flower bud density was measured over a three-week period in January 2015 only. The number of flower buds (including opened flowers) was counted on the top 20 cm of one characteristic upright shoot from the main cane and was reported as buds per 20 cm of shoot. Fruit traits were determined using five randomly sampled representative berries from each genotype.

The berries were harvested when the fruit were fully mature and only berries that exhibited peak quality and had no visual or pathogen, insect, or bird damage were chosen. Due to low fruit set, some seedlings were represented by as few as one fruit.

Due to the differing maturity timing for different families, fruit were collected over a six- week period starting at the beginning of April and ending in the middle of May for each year. The harvested fruit were separated by genotype and placed in Ziploc bags (5 x 5 cm). Samples were kept on ice until transported to the lab where they were stored in a

4C cooler overnight. The following day, the berries were analyzed at room temperature

(21C) for weight, fruit diameter, firmness, and scar diameter.

The weight (g) of the fruit was measured using an analytical scale (CP2202S,

Sartorius Corp., Bohemia, NY). Firmness (g·mm-1) and fruit diameter (mm) were measured simultaneously using a FirmTech II firmness tester (BioWorks Inc., Wamego,

KS). The fruit were placed perpendicular to the turntable and the firmness was

78

measured as the amount of force required to compress the fruit 1 mm. The minimum and maximum compression force thresholds were set at 50 g and 350 g, respectively.

The scar diameter (mm) was measured using FIJI software (Schindelin et al., 2012).

Pictures of the stem scar for each berry were taken using a digital SLR (Pentax K-x,

Ricoh Imaging, Denver, CO) camera positioned 50 cm above the berries. The camera was set at a shutter speed of 1/50, aperture F 5.6, and ISO 200. A ruler was included in every picture to act as a size reference. The images were imported into FIJI where the stem scar diameter (mm) was measured and recorded. After analysis, fruit samples were placed in their original collection bags and stored in a -20C freezer for 6 to 9 months prior to chemical analysis. The average weight (g), fruit diameter (mm), firmness

(g·mm-1), and scar diameter (mm) of the five berries was calculated and used for subsequent analyses.

Before chemical analysis, the frozen fruit were thawed 4-5 hours at room temperature. The fruit remained in Ziploc bags and were manually squeezed to extract the juice. Soluble sugar content (°Brix) was measured using a digital pocket refractometer (Atago U.S.A, Inc., Bellevue, WA) and 300 mL blueberry juice. The pH was measured using a glass pH electrode (Mettler-Toldeo, Inc., Schwerzenbach,

Switzerland).

Statistical Analysis

Two Year Data. Variance components were estimated with restricted maximum likelihood (REML) and breeding values predicted with best-linear unbiased predictions

(BLUP) using ASReml version 4.0 (Gilmour et al., 2014). The following univariate linear mixed model was used to analyze yield, weight, firmness, fruit diameter, and scar diameter:

79

y = Xb + Z1c + Z2r + Z3a + Z4b×a + e where y is a vector of the trait being analyzed (i.e. yield, weight, firmness, fruit diameter, or scar diameter), b are the fixed effect of year of overall mean and year, c is the

2 random effect of column nested within year ~N(0, σ c), r is the random effect of row

2 2 nested within year ~N(0, σ r), a is the random effect of genotype ~N(0, Aσ g), b×a is

2 the random effect of the interaction of year by genotype ~N(0, σ b×a), e is the random

2 residual effect ~N(0, Iσ e), X, and Z1-Z4 are incidence matrices for year, column, row, genotype and year-by-genotype interaction, respectively. The A-matrix was constructed using the AGH matrix R package assuming tetrasomic inheritance and the proportion of double reduction that best described each trait from (see Chapter 3). The unbiased

2 2 2 2 2 2 2 heritability was calculated as h = σ g /(σ g+ σ c+ σ r+ σ b x g + σ e). The genotype-by-year

2 2 2 interaction was calculated as G x Y = σ g /(σ g+ σ bxg .) Genetic correlations were estimated by fitting a bivariate model for each pair of traits that included the same factors as shown in equation 1. Because each trait was uniquely affected by double reduction, with the majority of traits experiencing no double reduction, genetic correlations were fit with a relationship matrix run assuming no double reduction.

A model assuming no correlation was fit and used as a base model to test for significant correlations. A log-likelihood ratio test, following a chi-square distribution with one degree of freedom and α = 0.05, was used to determine if the correlations were significant.

One Year Data. Variance components, estimated breeding values, and genetic correlations for flower bud density, soluble solids, and pH were measured using the above equations, without considering the year factor or the effect of the genotype by

80

year interaction (Z4b×a). The genetic correlations for soluble solids and pH were run with the data collected in 2014, while the genetic correlations involving flower bud density were run with the data collected in 2015.

Estimation of Expected Genetic Gain

Genetic gain was calculated based on the following equation from Falconer

2 2 (1960): ∆G =(h ×VP × i)/L (h = narrow-sense heritability, Vp = phenotype variation, i = selection intensity in standard deviation units, L = length of breeding cycle). The length of the breeding cycle was represented by parents chosen for use while in the various stages of selection used in the UF SHB breeding program. Stage I represents 3 years, stage II was 5 years, stage III was 8 years, and cultivars were 12 years.

Results and Discussion

Population Distributions

Phenotypic data were collected from 124 families totaling 1,996 individuals.

Family size varied from 5 to 32 individuals with a mean size of 17. These families resulted from crosses among 146 diverse parents, each with deep and complex pedigree information. The parents were autopolyploid, and descended from the wide hybridization between cultivated highbush blueberry (V. corymbosum) and native southeastern Vaccinium species such as evergreen blueberry (V. darrowii), mayberry

(V. elliottii), rabbiteye blueberry (V. virgatum syn. V. ashei), and sparkleberry (V. arboreum) (Brevis et al., 2008; Lyrene 1997, 2008; Olmstead et al., 2013).

Primary characteristics that are used for stage I selection in the UF blueberry breeding program were analyzed. There was considerable phenotypic variation for all of the traits, which was expected in a large and genetically diverse progeny set (Figure 4-

81

1). Firmness (xbar = 235.14g·mm-1), weight (xbar = 2.56g), and yield (xbar = 3.05) data were collected for two successive fruiting seasons, and the averaged data showed no significant departures from normal distributions. Fruit diameter (xbar = 18.81mm) and scar diameter (xbar = 1.95mm), both also averaged over two seasons, were positively skewed with greater numbers of individuals having smaller fruit and scars. In this study, the larger berries, tended towards more oblate shapes instead of spherical.

Three traits could only be measured in one year. Soluble solids (xbar =

11.73°Brix) were normally distributed, while pH (xbar =3.65) was positively skewed; both of these distributions exhibited larger ranges than those previously reported in

Gilbert et al. (2015) for similar germplasm (8.1-13.9 and 2.8-4.4 compared to 6.8-17.9 and 2.4-5.5). This is likely due to the larger sample size in the present study. Flower bud density (xbar = 4.44buds/20 cm), was less phenotypically diverse than the other measured traits with almost 60% of the observations between 2 to 5 buds per 20 cm of shoot. A small portion of the population (15%) had multiple buds at one node, and the highest amount recorded was seven buds at one node.

Estimation of Genetic Parameters

Narrow-sense Heritability. All of the traits varied in their yearly estimations of narrow-sense heritability between the two years (Figure 4-2). The standard errors were much smaller when using the combined information from 2014 and 2015 compared to using separate years. This suggests the heritability estimated using the collective data is more reliable. Heritabilities were similar for fruit weight in 2014 and 2015, however when the data for both years was merged, the heritability increased by ~ 0.15 due to the increase in genotypic variance. Interestingly, fruit diameter, a common alternative measure of fruit size, displayed the lowest combined heritability in the present study

82

(0.26). Earlier studies have found fruit weight to be highly heritable in V corymbosum and V. angustifolium (Aalders & Hall, 1975; Draper & Scott, 1969). Finn and Luby

(1986) estimated the narrow-sense heritability to be 0.78 with a standard error of 0.29 in a population derived from crosses between V corymbosum, V. angustifolium, and corymbosum × V. angustifolium hybrids. This study found the heritability to be 0.58 with a standard error of 0.02. This higher estimate could be due to the assumption of disomic inheritance, neglecting of genotype by year effect, or calculations made assuming a fixed effects model. In contrast, the GCA and SCA for weight was found to be significant only once in a two-year study (Erb et al., 1990). However, these plants were established in atypical conditions using non-amended mineral soil, and were derived from a complex five parent cross. Studying early UF breeding selections, Edwards et al. (1974) concluded fruit size was highly heritable using a 1-5 categorical method with estimations obtained from mid-parent offspring regression.

The heritability of fruit firmness showed the greatest variation between the two years, varying from 0.43 in 2014 to 0.70 in 2015. This may be due to differences in harvest period, where nearly half of the fruit harvest in 2014 occurred over a five-day period, while in 2015 the fruit were collected more uniformly over the season. Edwards et al. (1974) estimated fruit firmness to be moderately heritable over two years, similar to the narrow-sense heritability estimated from 2014 (0.35).

Scar diameter and yield had the largest heritabilities in 2014, decreasing by 0.1 and 0.18 in 2015, respectively. The combined data gave estimations slightly less than those made in 2014. Edwards et al (1974) and Finn and Luby (1992) found GCA to be significant for fruit scar. However, both studies concluded parental phenotype was not

83

an accurate predictor of progeny performance. The discrepancies in heritability estimations between the previous studies and the current study could be due to the larger parent population 146 in the current study compared to 17 in Edwards et al

(1974) and 6 in Finn and Luby (1992) or the difference in measurement methods. The present study measured scar diameter (mm) on a continuous scale instead of the discrete scale used in the above studies. Aalders and Hall (1975) found the GCA and

SCA to be significant for yield in V. angustifolium, but additive and genetic components were very low, with a few negative estimations.

Soluble solids and pH had moderately low heritabilities of 0.33 and 0.36 (Figure

4-3). Applying BLUP and REML analysis to a large population, de Souza et al. (1998b) determined the narrow-sense heritability to be 0.33 and 0.39 for soluble solids and pH, although the total variability for pH was close to 0. Both soluble solids and pH have shown to be effected by genetic and environmental factors, varying by year, location, harvest date, species, and genotype (Ballinger et al., 1963; Ballington, Ballinger et al.,1984; Gilbert et al., 2015; Saftner et al., 2008). Studying 11 Vaccinium species

Ballington et al. (1984) found means for soluble solids to vary across three years, but population differences remained consistent, demonstrating a strong genetic component.

Flower bud density had the lowest heritability (0.16) among the traits measured

(Figure 4-3). De Souza et al. (1998a) found the heritability of flower bud density to be

0.41 when measuring six replicate peach canes. Measuring one cane in the current study may have limited the observed phenotypic variation (Figure 4-1), leading to a lower heritability estimate.

84

Genotype-by-Year Interaction. Fruit weight had the lowest genotype-by-year (G

× Y) interaction, measured by the higher genotype by year correlation, with the smallest standard error (Figure 4-4). Finn and Luby (1986) drew similar conclusions when they found that the cross-by-year interaction was significant for fruit weight, but there was no significant variation over the two years. Yield and scar diameter had the same interaction estimations of 0.85. This is conflicting with previous studies that found yield to have a large environmental component (Aalders & Hall, 1975; Finn et al., 2003).

These differences could be due to the subjective nature in which the trait was measured and a result of plant maturity differences. Studies on fruit scar found that year effects have no significant influence over the trait (Edwards et al., 1974; Finn et al., 2003).

However, Finn and Luby (1992) found the cross-by-year interaction was not significant, but the year variation was.

Fruit diameter and firmness had the highest interactions, indicating that these traits were more susceptible to environmental variation. Other studies have found firmness to be significantly affected by the harvest date and year (Aalders & Hall, 1975;

Ballinger et al., 1963; Edwards et al., 1974). Furthermore, it has been noted that rainfall, heavy irrigation, and warm temperatures can result in softer fruit (Ehlenfeldt & Martin,

2002; Ehret et al., 2012). Conversely, Finn and Luby (1992) found the cross-by-year interaction to be significant for firmness, but there were no yearly variations. Finally, fruit size has been reported to have very little variation over years (Edwards et al., 1974;

Finn et al., 2003).

Trait Genetic Correlations. Out of 26 estimations for genetic correlation among traits, 19 were significantly correlated (Figure 4-5). Correlations measured over two

85

years were all significantly correlated with much smaller standard errors (0.06 compared to 0.15). Firmness was significantly correlated with every trait measured indicating selection for firmness elicits a response in all traits.

The largest significant correlations were between weight and fruit diameter (0.90) and flower bud density and yield (0.75). Due to the strong correlation, the larger heritability of fruit weight and the stability of the trait seen over the two years, improvement for fruit diameter via direct selection for fruit weight may be more efficient than direct selection on fruit diameter. Fruit diameter and weight had a similar pattern of correlations among the other traits. However, highly correlated flower bud density and yield had opposite patterns, with yield typically positively correlated with the other traits while flower bud density was negatively correlated. Correlations for flower bud density were estimated using data from 2015, and consequently were subject to larger standard errors than those obtained for yield. Using data from only one year may be unreliable for estimating correlations estimated flower bud density and primary breeding traits.

Yield had significant correlations with all the traits except for pH. While these correlations were significant, the majority were small, less than 0.20. The minor correlation is comparable to previous reports, which were unable to find correlations between yield ratings and weight, firmness, size, and pH (Ballinger et al., 1963; Draper

& Scott, 1969; Draper et al., 1982; Finn & Luby, 1986). Ballinger et al. (1963) did find a significant interaction between increasing crop load and decreasing soluble solids and decreasing fruit weight in a single genotype, ‘Wolcott’. Awad, et al. (2001) found soluble solids in apples to be significantly higher in low crop loads.

86

Fruit firmness had small significant correlations with weight (0.06) and fruit diameter (-0.06), two elements of fruit size. Using similar equipment (FirmTech II), Ehret et al. (2012) found no relationship in the regression of firmness on fruit diameter. Draper et al. (1982) also found no relationship between fruit firmness and size, but Edwards et al. (1974) found a significant and positive correlation. Initially, Ballinger et al. (1973) found size to have no effect on firmness, but after increasing the sample size from 4 to

9 genotypes, a tendency towards smaller diameter and firmer fruit was noted. These studies were conducted on small populations containing 1 to 9 selections ad progeny developed from crosses between 6 selections, which may be the reason for the conflicting results. The negative correlation between firmness and scar diameter is desirable when making selections for mechanical harvesting; however, the resulting decrease in soluble solids is not (Ballington, 1990; Edwards et al., 1974; Olmstead &

Finn, 2014).

Overall, the genetic correlations between most traits were small, with 18 relationships having a correlation less than 0.20. The low correlations allows for simultaneous selection for all traits without adversely affecting each other.

Breeding Values

The yearly variation in estimated breeding values was compared for each trait using data from 2014, 2015, and combined data from 2014 and 2015 (Figure 4-6). The mean and distribution of the breeding values obtained in 2014 were very similar to those obtained using information from both 2014 and 2015 for firmness, fruit diameter, scar diameter, and yield. Breeding values for fruit diameter measured in 2014 had a large number of outliers, due to more the oblate shaped berries found only that year. The mean breeding value for weight was similar over all data sets. This supports the findings

87

in (Figure 4-3), that weight has a large genetic variance and breeding values estimated after one year are accurate and can be used for selection. Interestingly, breeding values estimated using the pooled data strongly resemble breeding values estimated in 2014, instead of an average between the two years. This may be a result of the greater number of observations made in 2014 compared to 2015.

Estimated Genetic Gain

Genetic gain is a function of the heritability and phenotypic variance of a trait, selection intensity, and the length of the breeding cycle (Falconer, 1960). In the present study, the breeding cycle was defined as the length of time from seed to seed

(Wellensiek, 1962). An example of the genetic gain that can be obtained for direct selection on fruit weight is seen in (Figure 4-7). Genetic gain increases with increasing selection intensity and with parents selected earlier in the breeding program. The highest genetic gain (0.35 g) was acquired when selections from stage I were used as parents and the top 1% of the population was selected. By waiting longer the breeder can select parents more confidently and apply a more stringent selection intensity.

However, in this study, the stage the parents are selected in has a larger impact on the genetic gain than the applied selection pressure. The breeding program typically selects the top 10% of individuals, reducing the genetic gain achieved from stage I selections to

0.23 g, although this is still larger than the genetic gain obtained when using selections from stage II and the highest selection intensity.

Because genetic gain is the progress of the population per unit of time, minimizing the period between initial planting and parent selection is crucial. The breeding program at UF uses a wide array of parents when making breeding crosses.

Over half of parents chosen in this study were selections from stage I germplasm, 19%

88

were selections from stage II germplasm, 22% were selections from stage III, and 7% were cultivars. BLUP estimated breeding values and knowledge of the genetic parameters will allow for more accurate selection of parents used in future crosses, allowing a larger number of parents to be selected from stage I.

Conclusion

The large phenotypic variation present in the breeding germplasm indicates progress can be made for all economically important traits in the UF SHB breeding program. The largest narrow-sense heritabilities were obtained for scar diameter, weight, and yield, while firmness and fruit diameter were the most effected by the environmental variance from year to year. The genetic correlation between most traits was less than 0.20, suggesting that simultaneous selection can be applied without adversely affecting the genetic gain of other traits. The estimated breeding values obtained in 2014 strongly resembled the breeding values in 2014 and 2015. This resemblance combined with the larger genetic gain that can be achieved from early parental selection validates choosing parents after one year of evaluation. However, selection and crossing designs in the program have resulted in breeding populations with highly unbalanced family sizes that are composed of numerous kinds of relationships. BLUP/REML analyses accurately estimate the variance components and predict breeding values even though the data is highly unbalanced. The results from this study will allow for accurate parent selection based on REML estimated variance components such as narrow-sense heritability and genotype by year correlation as well as prediction of breeding values.

89

Figure 4-1. Phenotypic distributions of blueberry fruit quality and plant characteristics. Fruit firmness, fruit diameter, fruit scar diameter, fruit weight, and yield collected in 2014 and 2015. Flower bud density, pH, and soluble solids collected over one year.

90

Figure 4-2. Narrow-sense heritability and standard errors of fruit firmness, fruit diameter, scar diameter, weight, and yield rating according to the year the data were taken.

91

Figure 4-3. Narrow-sense heritability and relative standard errors for flower bud density, soluble solids content, and fruit pH. Each trait was only measured in a single year.

92

Figure 4-4. Genotype-by-year interaction (G × Y) and standard error for fruit firmness, fruit diameter, scar diameter, weight, and yield rating.

93

Figure 4-5. Genetic correlations between the primary selection traits of total yield, flower bud density, and fruit weight, firmness, diameter, scar diameter, pH, and soluble solids. An asterisk (*) indicates the correlation is significant according to a log-likelihood ratio test. No correlation could be obtained between flower bud density and pH and soluble solids because they were measured indifferent years.

94

Figure 4-6. Boxplots for fruit firmness, fruit diameter, weight, scar diameter, and crop rating showing the variation of the estimated breeding values (EBVs) between 2014, 2015, and combined data from 2014 and 2015.

95

Figure 4-7. Potential genetic gain for fruit weight when using parents from different selection stages and intensity.

96

CHAPTER 5 SUMMARY AND CONCLUSIONS

In the past, blueberries have been bred using phenotypic recurrent selection.

While this has been very successful it is difficult to accurately select parents for low heritability traits and the development time to release a cultivar typically takes 12 to 15 years. To maximize genetic gains, accurate selection of parents with limited information is critical in the UF SHB breeding program where up to half of the parents selected for future crosses have been evaluated for only one year.

BLUP and REML have been used with great success in the livestock and forest breeding programs to estimate variance components for complex traits and predict breeding values. The accuracy of these predictions is improved by incorporating the pedigree of the tested population through the numerator relationship matrix (A-matrix).

These methods are well suited for the SHB breeding program because they are able to handle intricate genetic relationships and highly unbalanced data sets. However, these models are constructed assuming disomic inheritance. Cultivated blueberries are autotetraploids and can experience double reduction. Therefore, the present study was developed to estimate the genetic parameters (founders, inbreeding, genetic diversity, narrow-sense heritability of primary breeding traits, genetic correlations, and environmental variance) of the UF SHB breeding program, while accounting for the complex inheritance that autotetraploids have.

The founding clones for the UF SHB breeding program consist of cultivated NHB germplasm and southeastern species that contributed alleles for climatic adaptation.

The largest genetic contributions were from ‘Brooks’, ‘Sooy’, ‘Rubel’, and ‘FL 4A’. The genetic contribution of the NHB founders has remained stable over the program history,

97

while genetic contributions from the southern species have been more dynamic. Almost all the cultivars exhibited some level of inbreeding, however the largest inbreeding coefficient of the parents of the 2011 SHB breeding population was only 0.08.

Furthermore, there was no correlation between the inbreeding coefficient and selection cycle. This suggests that inbreeding has been well controlled in the UF blueberry breeding program.

The proportion of double reduction each trait experiences can effect the estimated heritability and predicted breeding values. The majority of primary breeding traits (fruit yield, fruit weight, fruit diameter, and fruit stem scar diameter) had the best fit when assuming tetrasomic inheritance with no double reduction. Fruit firmness had the best fit when double reduction was 0.15 to 0.20, indicating that the genes involved in firmness may reside towards the end of a linkage group. As double reduction increased, the narrow-sense heritability of each trait decreased. However, there was a high correlation between the breeding values estimated assuming disomic inheritance and those estimated assuming tetrasomic inheritance. The depth of the pedigree had a large impact on the estimation of double reduction that affected each trait. Shallow pedigrees underestimated the true kinship coefficient and as a result estimated large portions of double reduction.

The estimated narrow-sense heritability varied between the two years for fruit firmness, fruit diameter, fruit stem scar diameter, fruit weight, and yield rating. Fruit weight, fruit stem scar diameter, and yield rating had the highest heritability when combining the information from both years and were the least effected by the environmental variance from year to year. The largest genetic correlations were seen

98

between fruit diameter and fruit weight (0.90) and flower bud density and yield rating

(0.75). Overall, the genetic correlation between most traits was less than 0.20, suggesting that selection can be made simultaneously without affecting genetic gain of other traits. The estimated breeding values varied over the two years for all the traits except weight. However, there was a strong resemblance between the breeding values estimated in 2014 and the values estimated from the combined information of both years. This strong resemblance, combined with the larger genetic gain that can be achieved from early parental selection validates the current method of choosing parents after one year of evaluations.

The results from this study will allow for parental selection of SHB based on

BLUP predicted breeding values and REML estimated variance components. These methodologies will maximize the genetic gain that can be achieved, while accounting for the complex inheritance that characterizes autotetraploid species.

.

99

LIST OF REFERENCES

Aalders, L. E., & Hall, I. V. (1975). A study of variation in fruit yield and related characters in two diallels of the lowbush blueberry, Vaccinium angustifolium Ait. Canadian Journal of Genetics and Cytology, 17, 401-404.

Akaike, H. (1974). A new look at the statistical model identification. Automatic Control, IEEE Transactions on, 19, 716-723.

Allendorf, F. W., & Danzmann, R. G. (1997). Secondary tetrasomic segregation of MDH-B and preferential pairing of homeologues in rainbow trout. Genetics, 145, 1083-1092.

Awad, M. A., De Jager, A., Dekker, M., & Jongen, W. M. (2001). Formation of flavonoids and chlorogenic acid in apples as affected by crop load. Scientia Horticulturae, 91, 227-237.

Ballinger, W. E., Kushman, L. J., & Brooks, J. F. (1963). Influence of crop load and nitrogen applications upon yield and fruit qualities of Wolcott blueberries. Proceedings of the American Society for Horticultural Sciences 82, 264-276.

Ballington, J. R. (1990). Germplasm resources available to meet future needs for blueberry cultivar improvement. Fruit Varieties Journal, 44(2), 54-62.

Ballington, J. R. (2001). Collection, utilization, and preservation of genetic resources in Vaccinium. HortScience, 36, 213-220.

Ballington, J. R., Ballinger, W. E., Swallow, W. H., Galletta, G. J., & Kushman, L. J. (1984). Fruit quality characterization of 11 Vaccinium species. Journal of the American Society for Horticultural Science, 109, 684-689.

Bernardo, R. (1991). Correlation between testcross performance of lines at early and late selfing generations. Theoretical and Applied Genetics, 82, 17-21.

Bernardo, R. (1994). Prediction of maize single-cross performance using RFLPs and information from related hybrids. Crop Science, 34, 20-25.

Bernardo, R. (1996a). Best linear unbiased prediction of maize single-cross performance. Crop Science, 36, 50-56.

Bernardo, R. (1996b). Best linear unbiased prediction of maize single-cross performance given erroneous inbred relationships. Crop Science, 36, 862-866.

Bouffette, A. R. (1966). Contribution mathématique à la génétique des tétraploïdes (Doctoral dissertation).

100

Bourke, P. M., Voorrips, R. E., Visser, R. G., & Maliepaard, C. (2015). The Double Reduction Landscape in Tetraploid Potato as Revealed by a High-Density Linkage Map. Genetics, 3, 853-863.

Boyce, A. J. (1983). Computation of inbreeding and kinship coefficients on extended pedigrees. Journal of Heredity, 74, 400-404.

Bradshaw, J. E., & Bonierbale, M. (2010). Root and Tuber Crops. Potatoes (pp. 1-52). New York, NY: Springer.

Brazelton, C. (2013). 2012 World blueberry acreage and production report. Folsom, CA: North American Highbush Blueberry Council.

Brevis, P. A., Bassil, N. V., Ballington, J. R., & Hancock, J. F. (2008). Impact of wide hybridization on highbush blueberry breeding. Journal of the American Society for Horticultural Science, 133, 427-437.

Broman, K. (2015). R package primer. http://kbroman.org/pkg_primer/.

Brooks, R. M., & Olmo, H. P. (1952). Register of new fruit and nut varieties, 1920-1950. Berkeley and Los Angeles, CA: University of California Press.

Brooks, R. M., & Olmo, H. P. (1997). Brooks and Olmo register of fruit & nut varieties (3rd ed.). Alexandria, VA: Ashs Press.

Brummer, E. C. (1999). Capturing heterosis in forage crop cultivar development. Crop Science, 39, 943-954.

Camp, W. H. (1942). On the structure of populations in the genus Vaccinium. Brittonia, 4, 189-204.

Cassidy, A., Mukamal, K. J., Liu, L., Franz, M., Eliassen, A. H., & Rimm, E. B. (2013). High anthocyanin intake is associated with a reduced risk of myocardial infarction in young and middle-aged women. Circulation, 127, 188-196.

Cockerham, L. E., & Galletta, G. J. (1976). A survey of pollen characteristics in certain Vaccinium species. Journal of the American Society for Horticultural Sciences 101, 671-675.

Coville, F. V. (1937). Improving the wild blueberry. Yearbook of Agriculture (pp. 559- 574). Washington, DC: United States Department of Agriculture.

Crossa, J., de los Campos, G., Pérez, P., Gianola, D., Burgueno, J., Araus, J. L., & Braun, H. J. (2010). Prediction of genetic values of quantitative traits in plant breeding using pedigree and molecular markers. Genetics, 186, 713-724.

101

Darrow, G. M., Clark, J. H., & Morrow, E. B. (1939). The inheritance of certain characters in the cultivated blueberry. Proceedings of the American Society for Horticultural Sciences, 37, 611-616.

Demarly, Y. (1963). The genetics of tetraploids and plant breeding. Annales de l’amelioration des Plantes, 13, 307-400.

Doyle, G. G. (1973). Autotetraploid gene segregation. Theoretical and Applied Genetics, 43, 139-146.

Draper, A. D. (1977). Tetraploid hybrids from crosses of diploid, tetraploid, and hexaploid Vaccinium species. Acta Horticulturae, 61, 33-38.

Draper, A. D., Galletta, G. J., & Ballington, J. R. (1982). Breeding methods for improving southern tetraploid blueberries. Journal American Society for Horticultural Science, 107, 106-109.

Draper, A. D., & Scott, D. H. (1969). Fruit size inheritance in highbush blueberries, Vaccinium australe Amall. Proceedings of the American Society for Horticultural Sciences, 94, 417-418.

Draper, A. D., & Scott, D. H. (1971). Inheritance of albino seedling in tetraploid highbush blueberry. Journal of the American Society for Horticultural Sciences, 96, 791- 792.

Dweikat, I. M., & Lyrene, P. M. (1988). Production and viability of unreduced gametes in triploid interspecific blueberry hybrids. Theoretical and Applied Genetics, 76, 555-559.

Edwards Jr, T. W., Sherman, W. B., & Sharpe, R. H. (1974). Evaluation and inheritance of fruit color, size, scar, firmness and plant vigor in blueberry. HortScience 9, 20- 22.

Ehlenfeldt, M. K. (1994). The genetic composition and tetrasomic inbreeding coefficients of highbush blueberry cultivars. HortScience, 29, 1342-1345.

Ehlenfeldt, M. K., & Martin, R. B. (2002). A survey of fruit firmness in highbush blueberry and species-introgressed blueberry cultivars. HortScience, 37, 386-389.

Ehret, D. L., Frey, B., Forge, T., Helmer, T., & Bryla, D. R. (2012). Effects of drip irrigation configuration and rate on yield and fruit quality of young highbush blueberry plants. HortScience, 47, 414-421.

Erb, W. A., Draper, A. D., Galletta, G. J., & Swartz, H. J. (1990). Combining ability for plant and fruit traits of interspecific blueberry progenies on mineral soil. Journal of the American Society for Horticultural Science, 115, 1025-1028.

102

Falconer, D. S. (1960). Introduction to quantitative genetics. New York, NY: Ronald Press.

Fan, Z. L., Wang, Z. Y., Zuo, L. L., & Tian, S. Q. (2012). Protective effect of anthocyanins from lingonberry on radiation-induced damages. International Journal of Environmental Research and Public Health, 9, 4732-4743.

FAO. (2015). FAOSTAT: Total World Blueberry Production in 2013. Retrieved from http://faostat.fao.org/site/567/default.aspx#ancor

Finn, C. E., Hancock, J. F., Mackey, T. & Serce, S. (2003). Genotype× environment interactions in highbush blueberry (Vaccinium sp. L.) families grown in Michigan and Oregon. Journal of the American Society for Horticultural Science, 128, 196-200.

Finn, C. E., & Luby, J. J. (1986). Inheritance of fruit development interval and fruit size in blueberry progenies. Journal of the American Society for Horticultural Science, 111, 784-788.

Finn, C. E., & Luby, J. J. (1992). Inheritance of fruit quality traits in blueberry. Journal of the American Society for Horticultural Science, 117, 617-621.

Finn, C. E., Olmstead, J. W., Hancock, J. F., & Brazelton, D. M. (2014). Welcome to the party! Blueberry breeding mixes private and public with traditional and molecular to create a vibrant new cocktail. Acta Horticulturae, 1017, 51-62.

Fisher, R. A. (1947). The theory of linkage in polysomic inheritance. Philosophical Transactions of the Royal Society B: Biological Sciences, 233, 55-87.

Fjellstrom, R. G., Beuselinck, P. R., & Steiner, J. J. (2001). RFLP marker analysis supports tetrasonic inheritance in Lotus corniculatus L. Theoretical and Applied Genetics, 102, 718-725.

Gallais A. (2003). Quantitiative genetics and breeding methods in autopolyploid plants. Paris: Institut National de la Recherche Agronomique.

Galloway, L. F., Etterson, J. R., & Hamrick, J. L. (2003). Outcrossing rate and inbreeding depression in the herbaceous autotetraploid, Campanula americana. Heredity, 90, 308-315.

Gilbert, J. L., Guthart, M. J., Gezan, S. A., de Carvalho, M. P., Schwieterman, M. L., Colquhoun, T. A., & Olmstead, J. W. (2015). Identifying Breeding Priorities for Blueberry Flavor Using Biochemical, Sensory, and Genotype by Environment Analyses. PloS One, 10(9), e0138494

103

Gilmour, A. R., Gogel, B. J., Cullis, B. R., & Thompson, R. (2014). ASReml user guide Release 4.0. Hemel Hempstead, UK: VSN International Ltd.

Grant, V. (1981). Plant speciation (2nd ed.). New York, NY: Columbia University Press.

Hall, S. H., & Galletta, G. J. (1971). Comparative chromosome morphology of diploid Vaccinium species. Journal of the American Society for Horticultural Sciences, 96, 289-292.

Hancock, J. F., & Siefker, J. H. (1982). Levels of inbreeding in highbush blueberry cultivars. HortScience, 17, 363-366.

Harrison, R. E., Luby, J. J., & Ascher, P. D. (1994). Poolen Source Affects Yield Components and Reproductive Fertility of Four Half-high Blueberry Cultivars. Journal of the American Society for Horticultural Science, 119, 84-89.

Heim, K. C., Angers, P., Léonhart, S., & Ritz, B. W. (2012). Anti-inflammatory and neuroactive properties of selected fruit extracts. Journal of Medicinal Food,15, 851-854.

Henderson, C. R. (1975). Best linear unbiased estimation and prediction under a selection model. Biometrics, 423-447.

Husband, B. C., & Schemske, D. W. (1997). The effect of inbreeding in diploid and tetraploid populations of Epilobium angustifolium (Onagraceae): implications for the genetic basis of inbreeding depression. Evolution, 3, 737-746.

Isogai, M., Ishii, K., Umemoto, S., Watanabe, M., & Yoshikawa, N. (2009). First report of blueberry red ringspot disease caused by Blueberry red ringspot virus in Japan. Journal of General Plant Pathology, 75, 140-143.

Jansky, S. (2009). Breeding, genetics and cultivar development. Advances in potato chemistry and technology (pp. 27-62). Burlington, VT: Academic Press.

Jelenkovic, G., & Harrington, E. (1971). Non-random chromosome associations at diplotene and diakinesis in a tetraploid clone of Vaccinium australe Small. Canadian Journal of Genetics and Cytology, 13, 270-276.

Jelenkovic, G., & Hough, L. F. (1970). Chromosome associations in the first meiotic division in three tetraploid clones of Vaccinium corymbosum L. Canadian Journal of Genetics and Cytology, 12, 316-324.

Katepa-Mupondwa, F. M., Christie, B. R., & Michaels, T. E. (2002). An improved breeding strategy for autotetraploid alfalfa (Medicago sativa L.). Euphytica, 123, 139-146.

104

Kerr, R. J., Li, L., Tier, B., Dutkowski, G. W., & McRae, T. A. (2012). Use of the numerator relationship matrix in genetic analysis of autopolyploid species. Theoretical and Applied Genetics, 124, 1271-1282.

Kihara, H., & Ono, T. (1926). Chromosomenzahlen und systematische Gruppierung der Rumex-Arten. Cell and Tissue Research, 4, 475-481.

Krebs, S. L., & Hancock, J. F. (1989). Tetrasomic inheritance of isoenzyme markers in the highbush blueberry, Vaccinium corymbosum. Heredity, 63, 11-18.

Krebs, S. L., & Hancock, J. F. (1990). Early-acting inbreeding depression and reproductive success in the highbush blueberry, Vaccinium corymbosum L. Theoretical and Applied Genetics, 79, 825-832.

Krikorian, R., Shidler, M. D., Nash, T. A., Kalt, W., Vinqvist-Tymchuk, M. R., Shukitt- Hale, B., & Joseph, J. A. (2010). Blueberry Supplementation Improves Memory in Older Adults†. Journal of Agricultural and Food Chemistry, 58, 3996-4000.

Levings III, C. S., & Alexander, D. E. (1966). Double reduction in autotetraploid maize. Genetics, 54, 1297.

Li, C., Feng, J., Huang, W. Y., & An, X. T. (2013). Composition of polyphenols and antioxidant activity of rabbiteye blueberry (Vaccinium ashei) in Nanjing. Journal of Agricultural and Food Chemistry, 61, 523-531.

Liu, Y., Song, X., Zhang, D., Zhou, F., Wang, D., Wei, Y., & Ji, B. (2012). Blueberry anthocyanins: protection against ageing and light-induced damage in retinal pigment epithelial cells. British Journal of Nutrition, 108, 16-27.

Luo, Z. W., Zhang, Z. E., Leach, L., Zhang, R. M., Bradshaw, J. E., & Kearsey, M. J. (2006). Constructing genetic linkage maps under a tetrasomic model. Genetics, 172, 2635-2645.

Lynch, M., & Walsh, B. (1998). Genetics and analysis of quantitative traits. Sunderland, MA: Sinauer Associates Inc. Publishers.

Lyrene, P. M. (1981). Recurrent selection in breeding rabbiteye blueberries (Vaccinium asheiReade). Euphytica, 30, 505-511.

Lyrene, P. M. (1993) Some problems and opportunities in blueberry breeding. V International Symposium on Vaccinium Culture, 346, 63-71.

Lyrene, P. M. (1997). Value of various taxa in breeding tetraploid blueberries in Florida. Euphytica, 94, 15-22.

Lyrene, P. M. (2002). Development of highbush blueberry cultivars adapted to Florida. Journal-American Pomological Society, 56, 79-85.

105

Lyrene, P. M. (2005). Breeding low-chill blueberries and peaches for subtropical areas. HortScience, 40, 1947-1949.

Lyrene, P.M. (2008). Breeding southern highbush blueberries. Plant Breeding Reviews, 30, 353-414.

Lyrene, P. M., & Ballington Jr, J. R. (1986). Wide hybridization in Vaccinium. HortScience, 21, 52-57.

Lyrene, P. M., & Sherman, W. B. (1984). Breeding early-ripening blueberries for Florida. Proceedings of the Florida State Horticultural Society, 97, 322-325.

Lyrene, P. M., Vorsa, N., & Ballington, J. R. (2003). Polyploidy and sexual polyploidization in the genus Vaccinium. Euphytica, 133, 27-36.

MacKenzie, K. E. (1997). Pollination requirements of three highbush blueberry (Vaccinium corymbosum L.) cultivars. Journal of the American Society for Horticultural Science, 122, 891-896.

Massman, J. M., Jung, H. J. G., & Bernardo, R. (2013). Genomewide selection versus marker-assisted recurrent selection to improve grain yield and stover-quality traits for cellulosic ethanol in maize. Crop Science, 53, 58-66.

Mather, K. (1935). Reductional and equational separation of the chromosomes in bivalents and multivalents. Journal of Genetics, 30, 53-78.

Mather, K. (1936). Segregation and linkage in autotetraploids. Journal of Genetics, 32, 287-314.

McAnulty, L. S., Nieman, D. C., Dumke, C. L., Shooter, L. A., Henson, D. A., Utter, A. C., & McAnulty, S. R. (2011). Effect of blueberry ingestion on natural killer cell counts, oxidative stress, and inflammation prior to and after 2.5 h of running. Applied Physiology, Nutrition, and Metabolism, 36, 976-984.

Mrode, R. (2014). Linear models for the prediction of animal breeding values (3rd ed.). Oxfordshire, UK: CABI.

Munoz, P. R., Resende, M. F., Huber, D. A., Quesada, T., Resende, M. D., Neale, D. B., & Peter, G. F. (2014). Genomic relationship matrix for correcting pedigree errors in breeding populations: impact on genetic parameters and genomic selection accuracy. Crop Science, 54(3), 1115-1123.

Olmstead, J. W., Armenta, H. P. R., & Lyrene, P. M. (2013). Using sparkleberry as a genetic source for machine harvest traits for southern highbush blueberry. HortTechnology, 23(4), 419-424.

106

Olmstead, J. W., & Finn, C. E. (2014). Breeding highbush blueberry cultivars adapted to machine harvest for the fresh market. HortTechnology, 24(3), 290-294.

Ortiz, R., Vorsa, N., Bruederle, L. P., & Laverty, T. (1992). Occurrence of unreduced pollen in diploid blueberry species, Vaccinium sect. Cyanococcus. Theoretical and Applied Genetics, 85(1), 55-60.

Piepho, H. P., Möhring, J., Melchinger, A. E., & Büchse, A. (2008). BLUP for phenotypic selection in plant breeding and cultivars testing. Euphytica, 161(1-2), 209-228.

Powell, J. E., Visscher, P. M., & Goddard, M. E. (2010). Reconciling the analysis of IBD and IBS in complex trait studies. Nature Reviews Genetics,11(11), 800-805.

Qu, L., & Hancock, J. F. (1995). Nature of 2n gamete formation and mode of inheritance in interspecific hybrids of diploid Vaccinium darrowii and tetraploid V. corymbosum. Theoretical and Applied Genetics, 91(8), 1309-1315.

Qu, L., Hancock, J., & Whallon, J. (1998). Evolution in an autopolyploid group displaying predominantly bivalent pairing at meiosis: genomic similarity of diploid Vaccinium darrowii and autotetraploid V. corymbosum (Ericaceae). American Journal of Botany, 85(5), 698-698.

Rasmusson, D. C., & Phillips, R. L. (1997). Plant breeding progress and genetic diversity from de novo variation and elevated epistasis. Crop Science, 37(2), 303-310.

Resende, M. F. R., Munoz, P., Acosta, J. J., Peter, G. F., Davis, J. M., Grattapaglia, D., ... & Kirst, M. (2012). Accelerating the domestication of trees using genomic selection: accuracy of prediction models across ages and environments. New Phytologist, 193(3), 617-624.

Saftner, R., Polashock, J., Ehlenfeldt, M., & Vinyard, B. (2008). Instrumental and sensory quality characteristics of blueberry fruit from twelve cultivars. Postharvest Biology and Technology, 49(1), 19-26.

Schindelin, J., Arganda-Carreras, I., Frise, E., Kaynig, V., Longair, M., Pietzsch, T. & Cardona, A. (2012). Fiji: an open-source platform for biological-image analysis. Nature Methods, 9(7), 676-682.

Sharpe, R. H. (1953). Horticultural development of Florida blueberries. Proceedings of the Florida State Horticultural Society, 66,188-190.

Sharpe, R. H., & Darrow, G. M. (1959). Breeding blueberries for the Florida climate. Proceedings of the Florida State Horticultural Society, 72, 308-311.

Shukitt-Hale, B. (2012). Blueberries and neuronal aging. Gerontology, 58, 518-523.

107

Slater, A. T., Wilson, G. M., Cogan, N. O., Forster, J. W., & Hayes, B. J. (2013). Improving the analysis of low heritability complex traits for enhanced genetic gain in potato. Theoretical and Applied Genetics, 127, 809-820. de Souza, V. A., Byrne, D. H., & Taylor, J. F. (1998a). Heritability, genetic and phenotypic correlations, and predicted selection response of quantitative traits in peach: I. An analysis of several reproductive traits. Journal of the American Society for Horticultural Science, 123, 598-603. de Souza, V. A., Byrne, D. H., & Taylor, J. F. (1998b). Heritability, genetic and phenotypic correlations, and predicted selection response of quantitative traits in peach: II. An analysis of several fruit traits. Journal of the American Society for Horticultural Science, 12, 604-611.

Sybenga, J. (1996). Recombination and chiasmata: few but intriguing discrepancies. Genome, 39, 473-484.

Tipton, D. A., Babu, J. P., & Dabbous, M. K. (2013). Effects of cranberry components on human aggressive periodontitis gingival fibroblasts. Journal of Periodontal Research, 48, 433-442.

USDA. (2015) Noncitrus fruits and nuts 2014 summary. Retrieved from http://usda.mannlib.cornell.edu/usda/current/NoncFruiNu/NoncFruiNu-07-17- 2015

Vander Kloet, S. P. (1983). The taxonomy of Vaccinium § Cyanococcus: a summation. Canadian Journal of Botany, 61, 256-266.

VanRaden, P. M. (2008). Efficient methods to compute genomic predictions. Journal of Dairy Science, 91, 4414-4423.

Vorsa, N., & Novy, R. (1995). Meiotic behavior and RAPD segregation in a tetraploid hybrid (Vaccinium darrowii× V. corymbosum) suggests genomic divergence in blueberry. HortScience, 30, 808-808.

Vorsa, N., & Rowland, L. J. (1997). Estimation of 2n megagametophyte heterozygosity in a diploid blueberry (Vaccinium darrowii Camp) clone using RAPDs. Journal of Heredity, 88, 423-426.

Welch, J. E. (1962). Linkage in autotetraploid maize. Genetics, 47, 367.

Wellensiek, S. J. (1962). Shortening the breeding-cycle. Euphytica, 11(1), 5-10.

108

Wu, R., Gallo-Meagher, M., Littell, R. C., & Zeng, Z. B. (2001). A general polyploid model for analyzing gene segregation in outcrossing tetraploid species. Genetics, 159, 869-882.

Yousef, G. G., Brown, A. F., Funakoshi, Y., Mbeunkui, F., Grace, M. H., Ballington, J. R., & Lila, M. A. (2013). Efficient quantification of the health-relevant anthocyanin and phenolic acid profiles in commercial cultivars and breeding selections of blueberries (Vaccinium spp.). Journal of Agricultural and Food Chemistry, 61, 4806-4815.

109

BIOGRAPHICAL SKETCH

Catherine Cellon was born in Boca Raton, Florida. She is the only child of Helen and Larry Cellon. She grew up a loyal Gator fan and came to the University of Florida in

2009. She graduated from the University of Florida with a degree in Horticultural

Sciences. She started working towards her master of science at the Department of

Horticultural Sciences at the University of Florida in August 2013. Catherine wants to continue working in breeding fruit crops.

110