ELUCIDATING THE GENETIC MECHANISMS OF FLOWERING AND THE REPRESSION 

OF FLORAL INITIATION BY FRUIT IN APPLE (MALUS × DOMESTICA BORKH.) 

 

By 

 

Christopher Charles Gottschalk 

 
 
 
 
 
 
 
 
 
 

 
 
 
 
 
 

 

 

A DISSERTATION 

 

Submitted to  

Michigan State University 

in partial fulfillment of the requirements 

for the degree of 

 

Plant Breeding, Genetics, and Biotechnology - Horticulture - Doctor of Philosophy 

 

2020 

 

 

ELUCIDATING THE GENETIC MECHANISMS OF FLOWERING AND THE REPRESSION 

OF FLORAL INITIATION BY FRUIT IN APPLE (MALUS × DOMESTICA BORKH.) 

ABSTRACT 

 

 

By 

 

Christopher Charles Gottschalk 

 

Many tree fruit crops exhibit yearly cyclical fluctuations in flowering and fruiting, 

including apple (Malus × domestica Borkh.). This phenomenon, known as biennial bearing, is an 

intriguing biological problem, as well as a production limitation of many high-value apple 

cultivars. The current hypotheses to explain biennial bearing focus on the repressive effects of 

fruit on the initiation of floral primordia, which would develop into flowers the following year. 

This floral repressive response could be the result of nutrient competition between the spur apex 

and the strong sink of the developing fruit, or flowering-inhibitory gibberellins (GAs) that are 

produced in the seeds of the developing fruit and exported to the spur apex. However, the 

molecular mechanism by which fruit load and/or GAs represses floral initiation is unknown. 

The first aim of my dissertation was to identify the genes involved in the floral initiation 

pathway. Utilizing transcripts assembled from a transcriptome of the biennial apple cultivar 

Honeycrisp and the known flowering genes from Arabidopsis, I identified a comprehensive list 

of flowering-related homologs based on sequence homology, phylogenetic relationship, and 

syntenic organization. The second aim was to characterize the expression of the flowering-

related homologs during the period of floral initiation under crop load conditions that either 

promote or repress flowering. Homologs of AGL24/SVP, AP1, FT, LFY, and SPLs were strongly 

expressed in apices of floral-induced trees, supporting their presumed role as floral promoters. In 

contrast, a homolog of the floral repressor TERMINAL FLOWER 1 (MdTFL1-2), was strongly 

 

 

up-regulated in apices of fruit-bearing, floral-repressed trees. Additionally, genes involved in the 

deactivation of bioactive GAs were strongly up-regulated in the floral-repressed trees. 

The timing of floral initiation is also known to vary between cultivars in apple. However, 

the diversity in flowering gene regulation that underlies this variation is unknown. The third aim 

of my dissertation was to investigate diversity of gene expression in six different apple species 

and cultivars. Generally, the expression profiles of the flowering-related genes were similar, 

suggesting a widely conserved mechanism within the genus. However, a few key genes involved 

in the regulation of floral initiation and development exhibited differential expression. For 

example, during the floral initiation period, five of the six genotypes were found to have 

differentially expressed MdTFL1s. This result implicates a potential role for MdTFL1s in 

determining the timing of floral initiation in addition to a role in repressing flowering in response 

to crop load. 

Plant growth regulators (PGRs) offer the potential to control flowering and biennial 

bearing in established apple plantings. My final aim was to evaluate and improve the efficacy of 

foliar-applied GA to repress flowering in apple. This study included the evaluation of application 

timings, GA types and application concentrations, and cultivar-specificity. The most successful 

PGR trial used GA3 to inhibit flowering in 'Honeycrisp' when a strong return bloom was 

expected. As a result of the decreased flowering and fruit load, the harvested fruit in the year 

following GA application exhibited higher fruit-quality. The results from this collection of 

studies provide insight into the molecular control of flowering and biennial bearing while 

demonstrating a practical approach to managing flowering in a high-value apple cultivar. 

 

 

Copyright by 
CHRISTOPHER CHARLES GOTTSCHALK 
2020

 

 

Dedicated to my loving parents Dawn & Gerard Gottschalk. I also wish to thank my very 

influential grandparents: Charles & Nancy Curnow, and Charles & Eve Gottschalk.

 
 

v 

ACKNOWLEDGEMENTS 

 

 

I would like to especially thank my major professor, Dr. Steve van Nocker, for his 

expectational mentorship and for providing me the opportunity to pursue my graduate education. 

He gave me, a student that did not excel in their undergraduate education, a chance to prove 

themselves and obtain a Ph.D. He also allowed me endless opportunities to participate in 

impactful and interesting research, including the pursuit of my own research interests. In 

particular, I want to acknowledge his willingness to guide and mentor me in the construction of a 

long-term study to identify cider apples for production in Michigan.  

There are many other mentors and scientists I would like to thank for their assistance in 

my education. My committee members of Drs. M. John Bukovac, Mitch McGrath, and Paolo 

Sabbatini, helped me develop a more thorough understanding of horticultural principles, plant 

physiology, and genetics. I would also like to thank the countless times Drs. Joseph Hill, 

Courtney Hollender, Ben Mansfeld, Shujun Oh, and Joshua VanderWiede came to my 

assistance. I am also forever grateful for all the collaboration from my counterpart in the lab - 

Songwen Zhang. We spent many hours working closely together and I am forever grateful for 

her collaboration. Lastly, I what to thank my fellow graduate students and friends for their 

support throughout my education: Patrick Abeli, Kristin Andersen, Kathleen Bitter-Rhodes, 

Charity Goeckeritz, and Emily Pawa.  

 

 
 

 
vi 

TABLE OF CONTENTS 

 
 

LIST OF TABLES .................................................................................................................x 
 
LIST OF FIGURES ...............................................................................................................xi 
 
CHAPTER ONE ....................................................................................................................1 
Literature Review...................................................................................................................1 
Introduction ....................................................................................................................2 
Background and significance .........................................................................................4 
Hypotheses of biennial bearing ......................................................................................6 
Management of biennial bearing ....................................................................................8 
Genetics of floral initiation .............................................................................................10 
Influence of crop load on flowering genes .....................................................................13 
Hypothesis of different floral initiation timings and biennial bearing ...........................14 
Rationale .........................................................................................................................15 
REFERENCES ......................................................................................................................21 

 

 
CHAPTER TWO ...................................................................................................................29 
Genetic mechanisms associated with floral initiation and the repressive effect of fruit on 
flowering in apple (Malus × domestica Borkh.) ....................................................................29 
ABSTRACT ...................................................................................................................30 
INTRODUCTION ..........................................................................................................31 
MATERIALS AND METHODS ...................................................................................33 
Plant materials, growth conditions and field experimental design ...........................33 
Nucleic acid preparation, sequencing, and data analyses .........................................34 
Reference-based transcriptome assembly.................................................................35 
Identification of novel ‘Honeycrisp’ reference-based transcripts ............................35 
De novo assembly of unmapped reads .....................................................................36 
General transcriptome annotation.............................................................................37 
Data accessibility ......................................................................................................37 
Identification of apple flowering genes ....................................................................37 
Gene expression analysis ..........................................................................................39 
TaqMan®️ qRT-PCR ................................................................................................39 
RESULTS AND DISCUSSION ....................................................................................40 
Effects of reducing fruit load on floral initiation......................................................40 
Transcriptome assembly and characterization..........................................................40 
Identification of flowering gene homologs ..............................................................42 
Transcriptional analysis of the apple shoot apex during the floral transition...........45 
Transcriptional response to the presence of a fruit load ...........................................47 
MdTFL1-2 expression profiling and identification of co-expressed genes ..............48 
Expression profiles of GA2ox and GA20ox genes ....................................................50 
Divergent expression patterns of homeologous gene pairs ......................................51 
REFERENCES ......................................................................................................................62 

 
 

vii 

 

CHAPTER THREE ...............................................................................................................69 
Molecular basis of diversity in floral initiation and development in Malus ..........................69 
ABSTRACT ...................................................................................................................70 
INTRODUCTION ..........................................................................................................71 
METHODS .....................................................................................................................73 
  Plant material and sample collection .........................................................................73 
  RNA extraction and sequencing ................................................................................74 
  Transcriptome assembly and analysis ........................................................................74 
  Analysis of expression for landmark flowering genes ...............................................75 
RESULTS .......................................................................................................................77 
  Sequencing and transcriptome assembly results ........................................................77 
  Determining floral genes expressed state ..................................................................79 
  Expression patterns of flowering homologs in July ...................................................80 
  Expression patterns of flowering homologs in October ............................................81 
DISCUSSION ................................................................................................................82 

REFERENCES ......................................................................................................................92   
 
 
CHAPTER FOUR ..................................................................................................................98 
Molecular identification and transcriptional patterning of FLOWERING LOCUS T genes  
in Apple (Malus × domestica)................................................................................................98 
ABSTRACT ...................................................................................................................99 
INTRODUCTION ..........................................................................................................100 
METHODS .....................................................................................................................102 
  Plant material and tissue sampling .............................................................................102 
  MdFT sequencing retrieval and analysis....................................................................103 
  Expression analysis ....................................................................................................104 
  Statistical analysis ......................................................................................................105 
RESULTS AND DISCUSSION ....................................................................................105 
Identification of FT-like sequences ...........................................................................105 
 
  Tissue-specific expression of MdFT1 ........................................................................106 
  Diurnal expression of MdFTs in bourse shoot leaves ................................................107
Seasonal expression patterns of MdFTs .....................................................................108 
CONCLUSIONS ............................................................................................................109 
REFERENCES ......................................................................................................................121 

 

 

 
CHAPTER FIVE ...................................................................................................................125 
Efficacy of foliar-applied GA in controlling flowering .........................................................125 
ABSTRACT ...................................................................................................................126 
INTRODUCTION ..........................................................................................................127 
METHODS .....................................................................................................................129 
  2016 PGR Experiments..............................................................................................129 
  2017 PGR Experiments..............................................................................................130 
  2018 PGR Experiments..............................................................................................131 
Fruit Yield and Quality Experiments .........................................................................132 

 

 

viii 

 

  Statistics .....................................................................................................................133 
RESULTS  ......................................................................................................................133 
Flowering response ....................................................................................................133 
Effects of GA on fruit maturity, size, and quality......................................................135 
Flowering in Honeycrisp two years after GA3 treatment...........................................135 
DISCUSSION ................................................................................................................136 
REFERENCES ......................................................................................................................147 

 

 

 

 

GENERAL CONCLUSIONS ................................................................................................151 

 

ix 
 

LIST OF TABLES 

 
 

Table 2.1. Sequencing and transcriptome assembly statistics ...............................................53 
 
Table 2.2. Identified homologs of Arabidopsis flowering genes in apple .............................54 
 
Table 3.1. Tissue collections for anatomical (D) and transcriptome (T) analyses from  
selected early and late-blooming Malus species from the USDA ARS PGRU Malus  
germplasm collection .............................................................................................................85 
 
Table 3.2. Identified homologs of Arabidopsis floral initiation and development genes in  
apple .......................................................................................................................................86 
 
Table 3.3. Uniquely expressed states of flowering homologs ...............................................88 
 
Table 5.1. Commercial GA products tested in 2016 for effectiveness at manipulating floral 
initiation .................................................................................................................................138 
 
Table 5.2. Application dates during the 2017 season, the target GDD50 measurement, and the  
actual GDD50 measurement ...................................................................................................139 
 
Table 5.3. 2018 PGR application trial design ........................................................................140 
 
 
 

 

 

x 
 

 

LIST OF FIGURES 

Figure 1.1. Apple shoot architecture ......................................................................................18 
 
Figure 1.2. Potential roles of fruit in repression of flowering in apple ..................................19 
 
Figure 1.3. Simplified model of the molecular-genetic pathway that regulates floral  
initiation in Arabidopsis .........................................................................................................20 
 
Figure 2.1. Effect of flower removal on subsequent-year flowering .....................................55 
 
Figure 2.2. Characterization of the reference-mapped assembled 'Honeycrisp'  
transcriptome..........................................................................................................................56   
 
Figure 2.3. Analysis of genomic collinearity and phylogeny for 15 flowering gene  
families ...................................................................................................................................57 
 
Figure 2.4. Expression patterns of flowering gene homologs in flowering-induced apices ..58 
 
Figure 2.5. Heatmap of differential expression of genes between thinned and non-thinned 
 apices across the study period ...............................................................................................59 
 
Figure 2.6. Expression profiles of MdTFL1-1 and MdTFL1-2 calculated from qPCR  
(upper panels) and RNAseq (lower panels) ...........................................................................60 
 
Figure 2.7. Heatmap of expression of homeologous gene pairs under various fruit load  
conditions and the resulting fold-change between conditions across the study period .........61 
 
Figure 3.1. Intersections of expressed flowering genes in the transcriptomes of each  
accession ................................................................................................................................89 
 
Figure 3.2. Heatmap of the expression of flowering genes in July........................................90 
 
Figure 3.3. Heatmap of the expression of flowering genes in October .................................91 
 
Figure 4.1. Genomic position of MdFT1 in the GDDH13 genome and corresponding  
alignments with the two FT genes assembled in the GDR genome ......................................111 
 
Figure 4.2. Alignment between the genomic sequence for MdFT1 (MD12G1262000),  
segment of the de novo assembled FT-like transcript TRINITY_DN25413_c6.gi (labelled as 
TRINITY_DN25413), and a de novo assembled FT DNA sequence contig .........................112 
 
Figure 4.3. Tissue and developmental stage specific expression of MdFT1 in Malus fusca .113 
 

 

 

xi 
 

Figure 4.4. Percent return bloom in cultivars that received blossom thinning treatments.....114 
 
Figure 4.5. MdFT1 relative expression in the bourse leaves of nonthinned and thinned  
shoots of two apple cultivars..................................................................................................115 
 
Figure 4.6. MdFT2 relative expression in the bourse leaves of nonthinned and thinned  
shoots of two apple cultivars..................................................................................................116 
 
Figure 4.7. Clustering of MdFT homologs expression profiles over 24 hours from trees  
thinned or nonthinned of flowers ...........................................................................................117 
 
Figure 4.8. MdFT1 relative seasonal expression in the bourse leaves of nonthinned and  
thinned shoots of two apple cultivars.....................................................................................118 
 
Figure 4.9. MdFT2 relative seasonal expression in the bourse leaves of nonthinned and  
thinned shoots of two apple cultivars.....................................................................................119 
 
Figure 4.10. Clustering of MdFT homologs expression profiles over the season from trees 
thinned or nonthinned of flowers ...........................................................................................120 
 
Figure 5.1. The percent flowering spurs in 2017 following application of different GAs to  
fruit thinned trees in 2016. .....................................................................................................141 
 
Figure 5.2. The percent flowering spurs in 2018 following application of GA3 to  
thinned trees in 2017. .............................................................................................................142 
 
Figure 5.3. Flowering response in 2019 following application of different GAs 
to floral thinned Honeycrisp trees in 2018 .............................................................................143 
 
Figure 5.4. Yield and maturity metrics from the 2019 harvest following application of  
GA3 to floral thinned Honeycrisp trees in 2018.....................................................................144 
 
Figure 5.5. Fruit quality measurements from the 2019 harvest following application of  
GA3 to floral thinned Honeycrisp trees in 2018.....................................................................145 
 
Figure 5.6. Flowering response in 2020 following application of GA3 to floral thinned 
Honeycrisp trees in 2018 .......................................................................................................146 
 
 

 

 
 

 

xii 

 

 

 
 

CHAPTER ONE 

Literature Review 

 

1 

Introduction 

Many fruit trees such as apple, pear, avocado, olive, and orange exhibit yearly cyclical 

fluctuations in flowering and fruiting. This phenomenon, termed 'alternate' or 'biennial' bearing, 

is an intriguing biological problem as well as a production limitation for the tree fruit industry. It 

is generally accepted that developing fruit inhibits the initiation of flowers, which would 

otherwise complete development and form fruit the following year. Two current hypotheses to 

explain biennial bearing focus on (I) nutrient competition between the shoot apex and the 

subtending fruit, and (II) flower-inhibitory gibberellins (GAs) produced in the developing seeds 

of the fruit and exported to the spur apex. How this physiology might be driven by or influence 

the expression of genes that control flowering remains relatively unexplored. In higher plants yet 

studied, flowering involves the activation of the meristem identity genes LEAFY (LFY) and 

APETELA 1 (AP1) in the shoot apex, via the transmissible product of the FLOWERING LOCUS 

T (FT) gene produced in the leaves. A gene related to FT, TERMINAL FLOWERING 1 (TFL1), 

may act in opposition to FT to inhibit floral initiation. Recent studies in several tree fruit species 

suggested that the effects of both crop load and floral-repressive GAs on floral initiation may be 

ultimately mediated by activity of apple genes homologous to FT and/or TFL1. However, this 

has not been rigorously established, due both to the incomplete characterization of these and 

other genes in the apple (Malus × domestica Borkh.) genome, and insufficient knowledge of the 

potential relationships between the underlying physiology and the activity of these genes. 

The aim of this dissertation is to identify and describe the molecular and physiological 

mechanism(s) by which developing fruit inhibit floral initiation in apple. The first step was to 

rigorously identify flowering-related genes based on homology with known flowering genes 

from other plants. Phylogenetic techniques were used to identify the number and genomic 

 
 

2 

organization of the apple homologs of FT, TFL1, LEAFY and AP1, as well as other genes with 

presumed roles in floral initiation, using a recently released, high-quality sequence of the apple 

genome and exhaustive transcriptional data from the apple shoot apex. The second step was to 

characterize the variability in timing of floral initiation and development within the Malus 

genus. The third step was to evaluate expression of key flower inducting genes under varying 

crop load conditions. We used previously generated transcriptional data of the shoot apex and 

target high-resolution expression assays of leaf tissues to construct an atlas of expression for the 

identified flowering homologs. The fourth and final step was to evaluate the use of plant growth 

regulators (PGRs) to control floral initiation and, in turn, biennial bearing. 

I hypothesize that developing fruit directly or indirectly repress expression of the FT gene 

in the leaf, leading to a loss of the transmissible floral promotion signal at the apex, and that FT 

transmission is dependent on the export of photosynthate from the leaves. I further hypothesize 

that developing fruit promotes expression of TFL1 in the shoot apex, repressing any positive 

effect of FT. Finally, I hypothesize that the diversity of apple cultivars to bear annually vs 

biennially is partially due to the specific timing of floral initiation, in addition to allelic variation 

within the FT, TFL1, or other genes that may confer reduced expression and/or activity of these 

genes. The analyses conducted here provided an updated index of genes that may play central 

roles in flowering in apple, as well as establish a genetic framework to explain the influence of 

crop load on floral initiation. The results of this study may lead to new approaches to control 

biennial bearing in existing plantings and/or provide ideas of how to manipulate flowering, while 

facilitating the development of improved apple cultivars that are less prone to biennial bearing.  

 

 

 

3 
 

Background and significance 

Fruit plays a vital role in the human diet by providing vitamins, sugars, and others 

nutrients (Seymour et al. 2013). Tree fruit, such as apples, contain some of the highest levels of 

dietary components (phytonutrients), such as quercetin glycosides, vitamin C, procyanidin, and 

chlorogenic acid (Boyer and Liu 2004). The cultivated apple is the third highest produced fruit 

crop worldwide and is the third most valuable fruit crop in the United States (FOA 2014; USDA 

2017). However, consistent and sustainable apple production is limited by several aspects, 

ranging from pre-harvest tree management to post-harvest fruit storage. Many of the obstacles 

associated with production limitations are increasing in significance as a result of climate 

change. This limitations include spring frost events, emerging biotic and abiotic pressures, and 

shifting international trade policies (Cannell and Smith 1986; Lavee 2007). 

The control of flowering is a major subject of research for improving commercial apple 

production due to the importance of precise timing of flowering on crop yield and quality. Three 

major aspects of flowering – juvenility/precocity, seasonal bloom timing, and biennial bearing - 

have special significance to apple production. Apple, like other woody perennial plants, 

transitions through a prolonged, 'juvenile' phase characterized by the inability to flower. This 

phase can last to over ten years when plants are grown from seed (Visser 1964). Following the 

juvenile phase, the plant transitions to the adult phase, in which flowering can be triggered by 

environmental cues such as seasonal temperatures and inductive photoperiod (Corbesier and 

Coupland 2005). Juvenility is a major limitation for the genetic improvement of apple, because 

breeders must wait several years after making a cross before the resulting progeny plants have 

acquired the competency to flower. However, juvenility also impacts the commercial production 

of apple. When adult-phase shoots are grafted onto rootstocks, as is the common practice for 

 

4 
 

clonal propagation of apple, the shoot passes through a juvenile-like phase before resuming 

flowering. The ability of grafted plants to quickly flower and produce a crop is termed precocity, 

and is a major area of research investigating the interaction between rootstock and grafted scion. 

In general, rootstocks that limit growth of the scion (‘dwarfing rootstocks’) provide the greatest 

precocity, but this effect is variable among rootstock genotypes (Webster et al. 1985), and 

growers must often wait several years before full production of a new planting is attained.  

The second major aspect of flowering impacting apple production is the timing of spring 

bloom. Bloom timing is important for four primary reasons. First, apples are pollinated by 

insects, cultivars that bloom very early in the season - when conditions are still relatively cool - 

may not be adequately pollinated (McGregor 1976; Free 1993). Second, flowers can be easily 

damaged by early-season frost and freezes (Cannell and Smith 1986; Rodrigo 2000). Third, 

apples are typically self-incompatible, and so it is imperative to have two or more cultivars 

blooming synchronously (Lewis and Vincent 1909). Lastly, genotypes that bloom relatively late 

may be more prone to infection by the fire blight pathogen Erwinia amylovora, which prefers hot 

and humid conditions and easily infects susceptible cultivars through the flowers (Jones 

1992). The timing of bloom in the spring is difficult to control as it is strongly influenced by both 

environment (temperature and/or photoperiod) and genetics (specific genetic alleles within the 

flowering and dormancy pathway) (Lawson et al. 1995; Liebhard et al. 2003; Celton et al. 2011; 

Gottschalk et al. 2013; Kurokura et al. 2013). 

 

The third major aspect of flowering that has special significance for apple production is 

biennial bearing. Biennial bearing is the tendency to flower and produce fruit every other year 

(Butler 1917; Monselise and Goldschmidt 1982). This trait is mediated at the point of floral 

initiation, and is characterized by profuse flowering in a given year (the “on-year”), followed by 

 

5 
 

minimal flowering the subsequent year (the “off-year”) (Jonkers 1979; Monselise and 

Goldschmidt 1982). The resulting variability in production creates instabilities in consumer 

supplies and income for producers (Williams and Edgerton 1974). Generally, the presence of 

developing fruit on spurs inhibits floral initiation within the adjacent bourse shoot (Fig. 1.1). 

Because apple flowers in a two year cycle, with flowers initiated in the first year and completing 

development (bloom) in the second year, this inhibition results in lack of flowers and fruit the 

year following a heavy crop load.  

Hypotheses of biennial bearing 

There are two popular hypotheses to explain the physiology that underlies biennial 

bearing. The first states that the strong sink strength established by developing fruit during an on-

year diverts photosynthates and nutrients away from the shoot apex of the developing spur, 

where floral initiation would occur (Fig. 1.2). The resulting low concentration of photosynthate 

in the spur apex then leads to a vegetative fate determination (Wardlaw 1990). In contrast, during 

an off-year, photosynthates accumulate to relatively high levels, promoting floral initiation. The 

observation that a high crop load is associated with diminished shoot growth (Quinlan and 

Preston 1971) demonstrates that there is a competitive interaction between the sink strengths of 

fruit and bourse shoots, and lends some support to this hypothesis. In addition, it has been shown 

that floral initiation can be inhibited by removal of bourse shoot leaves, a condition that limits 

photosynthate availability (Fulford 1966b, Elsysy et al. 2019). In contrast, girdling the branches 

of mature apple trees, which should result in photosynthate accumulation in the girdled branch 

and corresponding spur apices, is associated with increased floral initiation (Dennis and 

Edgerton 1966). These studies manipulated the source and sink balance, suggesting that 

 

6 
 

photosynthates play a major role in floral initiation. However, they failed to uncouple the 

potential role of ‘Florigen’, which is also produced in leaves.  

The second hypothesis states that excessive floral-repressive GAs are produced in 

developing seeds during fruit development and are exported up to the spur apex (Fig. 1.2). This 

hypothesis is supported by studies in the cultivar 'Spencer Seedless’, which naturally produces 

parthenocarpic (seedless) fruit but can be induced to produce seeds by hand pollination. Under 

natural conditions, when seedless fruit is produced, this variety tends to bear annually. However, 

when fruit was manipulated to produce seeds, reduced floral initiation was observed (Neilsen and 

Dennis 1999). Developing seeds are known to be a rich source of GAs, a class of phytohormone 

that generally promotes stem elongation and leaf expansion (Olszewski et al. 2002). Apple seeds 

have been found to contain large quantities of two forms of bioactive GAs, GA4 and GA7 

(Luckwill et al. 1969). However, because fruit growth is stimulated by seeds via GAs, it has not 

been clear whether the floral-repressive effects are mediated directly by the GAs or indirectly 

through the increased sink strength of the larger fruit. Application of exogenous GA can inhibit 

flower initiation in apple if the GA is applied during the anticipated period of floral initiation; 

this effect is independent of the presence of fruit, suggesting that it influences flowering more 

directly (Guttridge 1962; Schmidt et al. 2009). In order for seed-produced GAs to directly inhibit  

flowering, they would need to be translocated to the site of initiation, the shoot apex (Fig. 1.2). 

This translocation of GA would be counter to the expected strong flow of solutes and nutrients to 

the fruit (i.e. through the xylem and phloem tissues), transmission of GAs from seed to apex 

seems unlikely. Direct involvement of transmissible GAs from the fruit also seems unlikely 

because developing leaves, which are present in both fruiting and non-fruiting bourse shoots, are 

also known to generate GAs (Grauslund 1972). An argument against transmission of active GAs 

 

7 
 

is several GA2 OXIDASE genes, which catabolize active GAs, are strongly expressed in the base 

of the fruit pedicel and thus are expected to form a barrier to GA diffusion (Songwen Zhang and 

Steve van Nocker, unpublished).  

The tendency to bear annually or biennially is an inherent characteristic of specific 

cultivars, indicating a strong genetic basis for this trait. For example, numerous cultivars (e.g. 

Fuji and Honeycrisp) are known to be biennial, whereas others (e.g. Gala) typically bear 

annually (Pellerin et al. 2011). However, environmental and physiological conditions can 

influence biennial bearing indirectly through effects on flowering and/or crop load. Factors that 

reduce crop load, such as a hard freeze during bloom or poor pollination, can moderate the 

amplitude of the bearing cycle, while favorable conditions that promote high crop load, can 

exacerbate the effect. 

Management of biennial bearing 

The impact of biennial bearing on production has typically been reduced through the 

thinning of excess flowers or fruit during on-years (Downing 1900; Dennis 2000; Tromp 2000). 

Fruit thinning is the primary commercial practice to address subsequent-year flowering (referred 

to as ‘return bloom’) but also to maximize fruit size and quality (Link 2000). In regards to 

biennial bearing, fruit thinning is only effective at promoting a return bloom when carried out 

before a specific stage of fruit development (Byers and Carbaugh 2002). As a result, commercial 

apple producers will conduct thinning numerous times across the growing season, addressing the 

effects of biennial bearing and fruit quality improvement separately.  

The application of GA formulations to trees in their off-year has potential to reduce 

overcropping and the need for thinning. For example, application of GA3 to five-year old ‘Fuji’ 

trees during the anticipated period of floral initiation resulted in a strong (~55%) reduction in 

 

8 
 

return bloom (Zhang et al. 2016). Bertelsen et al. (2002) reported decreases of ~10 - 20% in the 

number of flowering buds in ‘Pacific Rose’ following GA3 or GA4+7 application during an off 

year. Bertelsen and Tustin (2002), found that GA3 applied two weeks after full bloom led to a 

60% decrease in flowering the following year. Similarly, McArtney and Li (1998) reported that 

both GA3 and GA7 inhibited floral buds on ‘Braeburn’ by 28 and 38%, respectively. However, 

repression of return bloom by application of GAs has yet to be implemented as a standard 

practice, at least in the U.S. This is most likely due to the fact that many other trials have had 

variable, and often ineffective results, high price of GAs, and the potential loss of a crop by a 

frost event. For example, Schmidt et al. (2009) found that application of GA4+7 to ‘Cameo’ and 

‘Fuji’ trees in the off-year caused only a small reduction (5% to 30%) in flowering. Year-to-year 

variation was also reported, where GA4 applications to the same group of ‘Golden Delicious’ 

trees for four consecutive years resulted in decreased flowering only in one year, with no effect 

or even increases in flowering the remaining years (Greene 1993).  

In addition to GAs, synthetic GA biosynthesis inhibitors have been used in attempts to 

manipulate flowering. The expectation is that, if GAs repress flowering, compounds that 

interfere with GA biosynthesis or signaling should promote flowering. This hypothesis has been 

supported through evidence obtained from PGR trials of GA inhibitors. The application of 

paclobutrazol (PAC) to ‘Fuji’ in an on year increased flowering by ~25% in the subsequent year 

(Zhang et al. 2016). PAC, although approved as a PGR for floriculture applications, has yet to be 

approved for use in apple production by the US EPA. Another inhibitor, Daminozide (Alar), has 

also been used successfully to promote return bloom (Edgerton and Hoffman 1965; Rogers and 

Thompson 1968; Looney 1969). However, the use of Alar in apple is now restricted due to its 

potential to be a human carcinogen (EPA 1989).  

 

9 
 

Other PGRs have been found to influence flowering in apple, mostly with inconsistent 

results. 6-benzylaminopurine, a synthetic cytokinin, was used in combination with Alar on 

blooming branches of ‘Baldwin’, ‘Delicious’, and ‘McIntosh’ resulting in a significant increase 

to flowering the subsequent year (McLaughlin and Greene 1991). Ethephon, a compound that 

breaks down in plant tissues to produce ethylene, reduced flowering after an off-year (Bukovac 

et al. 2006). However, ethephon also increased flowering when a low bloom density was 

expected after an on-year (Schmidt et al 2009). Naphthaleneacetic acid (NAA), a synthetic auxin, 

increased return bloom under some conditions but was variable in efficacy over different 

cultivars and application intervals (McArtney et al. 2007). Combining NAA and Ethephon 

resulted in a significant increase in floral initiation over thinning alone in biennial bearing ‘York’ 

(McArtney et al. 2013). In many of these studies it is not clear whether the increase in bloom was 

due to the PGR application or due to fruit thinning induced by the PGRs. 

Genetics of floral initiation 

Ultimately, the mechanism underlying biennial bearing must be governed by the 

molecular-genetic pathways controlling floral initiation. A working model for the flowering 

genetic network in apple can be proposed based on the known activities and interactions of 

flowering genes in the research reference plant Arabidopsis thaliana and the known functional 

conservation of these genes across plant species (Corbesier and Coupland 2005) (Fig. 1.3). In 

Arabidopsis, the GIGANTEA (GI) gene mediates output from a circadian clock and cooperates 

with light quality (wavelength) to regulate the diurnal expression of CONSTANS (CO). CO 

promotes expression of the FT gene under inductive photoperiods, and, in conjunction with the 

gibberellin pathway, also activates SUPPRESSOR OF CONSTANS 1 (SOC1). SQUAMOSA 

PROMOTER BINDING PROTEIN-LIKE (SPL) genes act as additional promoters of FT, SOC1, 

 

10 
 

and the floral meristem identity gene AP1. SPL gene expression is post-transcriptionally 

repressed by a class of microRNA, MIR156/157, during the very short juvenile-like phase. FT 

expression is primarily repressed by FLOWERING LOCUS C (FLC), in the absence of 

vernalizing cold and, thus, mediates the promotive effects of cold temperatures on flowering 

(vernalization). FT and SOC1 cooperatively activate AP1 and LFY, which together direct floral 

initiation and flower formation. The TERMINAL FLOWER 1 (TFL1) gene represses LFY and 

AP1 in the center of the inflorescence meristem, thus preventing formation of a terminal flower 

and enabling indeterminate flower formation. This effect is the result of TFL1 competing with 

FT in forming a transcriptional regulator complex with FLOWERING LOCUS D. 

Functional homologs of these genes have been identified in many plant species. GI has 

been extensively studied in rice for its roles as a major regulator of the circadian rhythm and 

promoter of additional flowering genes (Matsuzaki et al. 2015; Hayama et al. 2002). CO has 

been studied to understand how sequence and functional divergence has given rise to the varied 

flowering times as a response to photoperiod across diverse plant species (Griffiths et al. 2003; 

Campoli et al. 2012). In poplar, two copies of the FT gene have been identified (Böhlenius et al., 

2006). FT1 promotes reproductive onset in response to winter temperatures, whereas FT2 

promotes vegetative growth during warm, long days of the growing season (Hsu et al. 2011). 

Poplar also contains two copies of TFL1, PopCEN1 and PopCEN2. Both TFL1 orthologs control 

shoot meristem identity by repressing the developmental transition to mature flowering shoots 

(Mohamed et al., 2010). Developmental regulation of these TFL1 orthologs was found to be 

distinct; PopCEN1 was expressed in shoot tips and vegetative buds, whereas PopCEN2 was 

expressed in the stem, leaf blade, petiole and immature inflorescence (Mohamed et al. 2010). 

Silencing of both PopCEN paralogs by RNAi was associated with flowering after only two years 

 

11 
 

of growth, whereas non-transgenic plants took five years. The catkins of the CEN RNAi lines 

were branch-like in appearance suggesting the apex underwent a reversion towards a vegetative 

identity during development (Mohamed et al. 2010). A SOC1 ortholog has been found in both 

rice and citrus, and is functionally conserved with Arabidopsis (Lee et al. 2004; Tan and Swain 

2007). In rice, a suite of AP1 orthologs have been identified and characterized for their role in 

meristem identity (Kobayashi et al. 2012). In poplar, a potential functional homolog of AP1 has 

been identified as a downstream effector of FT2 in the flowering pathway (Hsu et al. 2006). 

Orthologs of LFY have been extensively studied in a wide range of plants from tomato to pines, 

where they are conserved meristem identity genes (Mellerowicz et al. 1998; Molinero-Rosales et 

al. 1999). 

In apple, several of the known Arabidopsis flowering genes have clear structural 

homologs and exhibit conserved function (Kotoda et al. 2000; Kotoda et al. 2002; Wada et al. 

2002; Kotoda et al. 2010; Trankner et al. 2010; Guitton et al. 2012, 2016). An exception is FLC; 

although apple contains several genes showing some sequence homology with FLC, a definitive 

phylogenetic relationship between these genes and FLC is still lacking (Porto et al. 2015; Kumar 

et al. 2016; Peace et al. 2019). In apple, the existence of two FT-like homologs has been reported 

(Kotoda et al. 2010). One FT homolog was expressed diurnally in leaves where its expression 

increased throughout the day and peaked at night (Trankner et al. 2010). Constitutive expression 

of one of these apple FT genes led to an early flowering phenotype in transgenic Arabidopsis, 

poplar and apple (Trankner et al. 2010). Apple FT protein has also been implicated in 

transcriptionally regulating a number of different pathways besides flowering, such as cell 

growth and organ development (Mimida et al. 2011a). Two TERMINAL FLOWER 1 (TFL1) 

homologs, MdTFL1-1 and MdTFL1-2, have been identified in apple (Mimida et al. 2009). 

 

12 
 

Knock-down experiments targeting MdTFL1-1 using RNAi resulted in decreased time to flower, 

reduced vegetative growth and singular terminal flowers (Flachowsky et al. 2012). This suggests 

that, at least, MdTFL1-1 is operating as a floral repressor and governor of inflorescence 

architecture in apple.  

Influence of crop load on flowering genes  

Research conducted in other perennial fruit trees species on the molecular genetics of 

biennial bearing suggests mechanism(s) that may also operate in apple. In biennial avocado and 

citrus, endogenous FT expression in leaves during the off-year was markedly higher than that in 

the on-year, suggesting that crop load mediates its expression (Munoz-Fambuena et al. 2011; Ziv 

et al. 2014). However in two studies with biennial apples, the level of FT expression in the leaves 

showed no difference between on- and off-years (Kittikorn et al. 2011; Haberman et al. 2016). 

MdTFL1 expression has been reported to be higher in apical buds in trees carrying a heavy crop 

versus trees that were subjected to thinning (Kittikorn et al. 2011). Haberman et al. (2016) found 

that MdTFL1-1 expression dramatically decreased during the course of the growing season, and 

this decrease occurred earlier in the season in on-year trees than in off-year trees, with MdTFL1-

2 significantly increasing in expression later in the season in the on-year trees. This suggests that 

MdTFL1-1 might play a role in maintaining the apex in a vegetative state during the early 

growing season, whereas MdTFL1-2 might repress flowering in response to high crop load. 

During off-years, citrus trees showed increased expression of multiple SPL genes in the shoot 

apex (Wu and Poethig 2006; Shalom et al. 2012). Similarly, in apple, two SPL genes were 

expressed to higher levels in apices of trees that were thinned of fruit, relative to non-thinned 

trees (Guitton et al. 2016). This suggests that the influence of crop load is upstream of these 

genes. 

 

13 
 

If the tendency to bear biennially can be attributed to floral-repressive signals originating 

in the fruit and acting at the shoot apex, then annual bearing could be explained in a number of 

ways. For example, the repressive signal may be absent or weaker in annual cultivars. 

Alternatively, the signal may not be routed correctly to the shoot apex. Another possibility is that 

the signal is transmitted to the shoot apex, but that the apex is not programmed to respond to this 

signal. There is also the possibility that floral initiation depends on coincidence of a fruit-

produced signal and a signal-sensitive phase of apex development, and that this is lost in annual 

cultivars. 

Hypothesis of different floral initiation timings and biennial bearing  

I hypothesize that, in biennial cultivars, the period of initiation overlaps with a period of 

fruit induced influencing of floral promoters/repressors. With annual cultivars, their floral 

initiation period might fall outside the range of time in which fruit influences the floral 

promoters/repressors activity. Previous research has identified highly variable timings for floral 

initiation with variation observed within the same tree, across seasons, and between cultivars 

(McArtney et al. 2001; Foster et al. 2003; Hoover et al. 2004; Dadpour et al. 2011). As a point of 

reference for this variation, the cultivar ‘Gala’ has been featured as a prominent cultivar of study 

for floral initiation. The onset of initiation has been seen to range from as early 72 days after full 

bloom (DAFB) to as late as 99 DAFB in the following season (McArtney et al. 2001). Another 

report, from the same orchard, found ranges of initiation between 96 to 109 DAFB in ‘Gala’ 

trees (Foster et al. 2003). However, all of these measurements are based on documented 

morphological changes observed in the structure of the meristem. 

The morphological changes that is indicate floral initiation, as described by Foster et al. 

(2003), begins with Stage 1 where the apex is flat (vegetative) but is undergoing an increase in 

 

14 
 

diameter. This increase in diameter signifies a commitment to floral development. At Stage 2, 

the apex is no longer flat and appears “domed” shaped, indicating that its identity has 

transitioned to an reproductive meristem. Identification of associated changes in gene expression 

at this critical period of time between Stages 1 and 2 could identify the molecular signature of 

floral initiation and is currently unknown. Foster et al. (2003) also reported observing Stage 1 

apices as early as -3 DAFB and as late as 96 DAFB, and 100% percent of the Stage 1 meristems 

continued through Stage 3, when floral buds begin to form. This suggests that competency to 

transition can extend for a long period of the growing season. This potential long competency 

period does not seem to completely align with what is known about how the presence of fruit 

affects floral initiation, especially in regards to the effective periods of fruit thinning to promote 

bloom. As reviewed by Jonkers (1979), thinning to promote a return bloom is only effective 

through the first month following bloom (~30 DAFB). These observations taken together could 

indicate that the potential for a floral initiation (or repression) signal occurs earlier in the season, 

before the time in which morphological signs of initiation are evident.  

Rationale  

Although the phenomenon of biennial bearing has been studied for more than 50 years at 

the physiological level, there have been few studies at the molecular and genetic levels. Research 

based in fundamental gene regulation has the potential to quickly advance our understanding of 

biennial bearing because genetic pathways of flowering have already been drafted in apple. 

Previous research on the molecular genetics of flowering in apple have been limited by the use of 

a poor quality reference genome and low-resolution molecular techniques. The experiments 

following utilized a new high quality reference genome and high-resolution methods to survey 

the apex and leaves of the bourse shoot for effects of crop load on floral initiation.  

 

15 
 

A prerequisite step to understand why fruit suppresses flowering in apple is to identify 

flowering-control genes in the apple genome, and document their expression profiles. Previous 

studies have already recognized homologs of AP1, FT, LFY, and TFL1, and have taken 

preliminary steps to understand their expression and function (Kotoda et al. 2000, 2010; Wada et 

al. 2002; Mimida et al. 2011b; Flachowsky et al. 2012). However, these studies were obfuscated 

by the presence of homeologous genes and/or allelic variants, and the use of molecular 

techniques such as quantitative PCR and microarrays that could not discriminate among them. 

Although more recent work has begun to employ direct transcriptome sequencing, the analysis to 

this date was hindered by the poor quality of the previous (pre-2017) apple draft genome 

(Velasco et al. 2010). A recent release of a new high quality hybrid apple genome (Daccord et al. 

2017) coupled with high-resolution RNA-sequencing can provide for an improved census of the 

flowering genetic network and estimation of expression patterns of these genes. 

The FT and TFL1 genes are the focus of the following experiments, given the primary 

importance of the FT gene and its structural and functional conservation. Two FT paralogs were 

reported in apple (Kotoda et al. 2010). Expression of FT1 was reported to be highest in shoot 

apices transitioning to flowering, whereas expression of FT2 was highest in floral organs. Both 

paralogs exhibited only very low expression levels in mature leaves (Kotoda et al. 2010). If FT 

function is conserved between Arabidopsis and apple, strong expression in apple leaves would be 

expected. Kotoda et al. (2010) does not specify the developmental stage, age, or type (spur, 

bourse shoot or elongated shoot) of the mature leaves that were analyzed, and this is an 

important detail because previous studies showed that various leaf types have different strengths 

of influence on flowering (Fulford 1965, 1966a, 1966b, Elsysy et al. 2019). In contrast, the study 

by Haberman et al. (2016) analyzed those leaves that were expected to influence flowering the 

 

16 
 

most, but did not find a difference in FT expression under varying crops loads. However this 

study used PCR primers that were not expected to discriminate between the potential FT 

paralogs. Expressing apple FT to high levels in transgenic apple resulted in early flowering, 

suggesting that FT has potential to promote flowering under natural conditions (Flachowsky et 

al. 2012; Trankner et al. 2010). However, these experiments were done in juvenile trees, so it is 

not known whether the influence of FT could be blocked by fruit in adult trees. In addition, the 

dynamics of FT trafficking from leaf to apex has yet to be explored in apple. 

TFL1, a regulator of shoot architecture in Arabidopsis, has been implicated as a potential 

floral repressor in apple (Flachowsky et al. 2012). TFL1 and FT are homologous to one another, 

and one possibility is that these two genes work antagonistically. TFL1 expression in the apex 

was reported to increase in spring, peak in June, and then decline through August (Mimida et al. 

2011b). This period of expression overlaps with the anticipated period of floral initiation. TFL1 

was also found to be strongly expressed in the apex of juvenile trees (Mimida et al. 2011b), 

potentially serving as a mechanism to maintain vegetative growth. Apple TFL1 was claimed to 

be a floral repressor, because when TFL1 was silenced in transgenic apple lines the plant 

flowered early (Flachowsky et al. 2012). However, early flowering was also observed in 

Arabidopsis when Arabidopsis TFL1 was silenced, even though the primary role of Arabidopsis 

TFL1 is to maintain indeterminate inflorescence architecture. Taken together, this evidence is not 

sufficient to distinguish between roles in flowering or inflorescence structure. It has been 

reported that TFL1 is expressed to higher levels in apices of fruited trees, relative to thinned trees 

(Kittikorn et al. 2011; Haberman et al. 2016). However, these studies also used low-resolution 

techniques that were not expected to discriminate between TFL1 and paralogous genes. These 

studies also did not consider other potential floral promoters or repressors. 

 

17 
 

 
Figure 1.1. Apple shoot architecture. (Top) In the spring, spurs release from dormancy 

 

expanding the spur leaves and flowers that were initiated the prior season. (Bottom) After fruit 

set, the bourse shoot apex begins to grow vegetatively producing bourse shoot leaves while the 

spur leaves are maintained. 

 

 
 

 

18 

Figure 1.2. Potential roles of fruit in repression of flowering in apple. (Left panel) Nutrient 

 

competition. In the presence of developing fruit (bottom), photosynthates are transported from 

their site of production in the leaf into the fruit, and the deficit of photosynthates in the apex 

suppresses floral initiation. When no fruit is present (top) the photosynthates are instead 

transported to the apex, allowing for floral initiation. (Right panel) GA-mediated. In the presence 

of developing fruit (top) GA produced by the developing seeds is transferred from the fruit to the 

apex where it suppresses floral initiation. When no fruit is present (bottom) there is not enough 

endogenously produced GA to repress floral initiation.  

 

 

 

19 
 

Figure 1.3. Simplified model of the molecular-genetic pathway that regulates floral 

 

initiation in Arabidopsis. Inductive photoperiods promotes the production of FT in the leaf, 

which is transported to the shoot apex. In the shoot apex, FT activates flowering genes such as 

LEAFY and AP1 cooperatively with the FD protein. TFL1 acts as a competitive antagonist to FT. 

For a detailed description of the individual role of these genes, see text. 

 

 

 

20 
 

 

 

REFERENCES 

 

21 
 

REFERENCES 

 
 

Bertelsen MG, Tustin DS and Waagepetersen RP. 2002. Effects of GA3 and GA4+7 on early bud 

development of apple. J Hort Sci and Biotech 77(1):83-90. 

Bertelsen MG and Tustin DS. 2002. Suppression of flower bud formation in light cropping trees 

of ‘Pacific Rose’ apple using gibberellin sprays. J Hort Sci and Biotech 77(6): 753-757. 

Böhlenius H, Huang T, Charbonnel-Campaa L, Brunner AM, Jansson S, Strauss SH and Nilsson 

O. 2006. CO/FT regulatory module controls timing of flowering and seasonal growth 
cessation in trees. Science 312:1040-1043. 

Boyer J and Liu RH. 2004. Apple phytochemicals and their health benefits. Nutrition Journal 

3(5). 

Bukovac MJ, Sabbatini P and Schwallier PG. 2006. Modifying Alternate Bearing of Spur-Type 

'Delicious' Apple with Ethephon. HortScience 41(7):1606-1611. 

Butler O. 1917. On the cause of alternate bearing in apple. Bulletin of the Torrey Botanical Club 

44(2):85-96. 

Byers RE and Carbaugh DH. 2002. Effects of Thinning Time on Yield, Fruit Size, and Return 

Bloom of ‘York’ and ‘Golden Delicious’ Apple Trees. J. of Tree Fruit Production 3(1): 55-
62. 

Campoli C, Drosse B, Searle I, Coupland G and Korff M. 2012. Functional characterisation of 

HvCO1, the barley (Hordeum vulgare) flowering time ortholog of CONSTANS. Plant J 
69(5):868-880. 

Cannell MGR and Smith RI. 1986. Climatic warming, spring budburst and frost damage on trees. 

J Applied Ecology 23:177-191. 

Celton JM, Martinez S, Jammes MJ, Bechti A, Salvi S, Legave JM, and Costes E. 2011. 

Deciphering the genetic determinism of bud phenology in apple progenies: a new insight 
into chilling and heat requirement effects on flowering dates and positional candidate 
genes. New Phytologist 192:378-392. 

Corbesier L and Coupland G. 2005. Photoperiodic flowering of Arabidopsis: integrating genetic 

and physiological approaches to characterization of the floral stimulus. Plant, Cell and 
Environment 28:54-66. 

Daccord N, Celton JM, Linsmith G, Becker C, Choisne N, Schijlen E, van de Geest H, Bianco L, 
Micheletti D, Velasco R, Di Pierro EA, Gouzy J, Rees JG, Guérif P, Muranty H, Durel CE, 
Laurens F, Lespinasse Y, Gaillard S, Aubourg S, Quesneville H, Weigel D, van de Weg E, 
Troggie M, and Bucher E. 2017. High-qualty de novo assembly of the apple genome and 
methylome dynamics of early fruit development. Nature Genetics 49:1099-1106. 

Dadpour MR, Movafeghi A, Grigorian W, and Omidi Y. 2011. Determination of floral initiation 

in Malus domestica: A novel morphogenetic approach. Biologia Plantarum 55(2): 243-
252.  

Dennis FG. 2000. The history of fruit thinning. Plant Growth Regulation 31(1):1-16. 

 
 

22 

Dennis FG and Edgerton LH. 1966. Effects of gibberellins and ringing upon apple fruit 

development and flower bud formation. Prco Am Soc Hort Sci 88:14-24. 

Downing AJ. 1900. The fruits and fruit trees of America. Wiley, New York. 

Edgerton LJ and Hoffman MB. 1965. Some physiological response of apple to N-dimethyl 
amino succinamic acid and other growth regulators. Proc Am Soc Hort Sci 86:28-36. 

EPA. 1989. Daminozide (Alar) Pesticide Canceled for Food Use. November 7, 1989. 

https://www.epa.gov/aboutepa/daminozide-alar-pesticide-canceled-food-uses 

Flachowsky H, Szankowski I, Waidmann S, Peil A, Tranker C, and Hanke MV. 2012. The 

MdTFL1 gene of apple (Malus × domestica Borkh.) reduces vegetative growth and 
generation time. Tree Physiology 32; 1288-1301.  

FOA. 2008. Food and Agriculture Organization of the United Nations, FAOSTAT database 

(FAOSTAT, 2008), http://faostat.fao.org/. 

Foster T, Johnston R and Seleznyova A. 2003. A Morphological and Quantitative 

Characterization of Early Floral Development in Apple (Malus × domestica Borkh.). 
Annals of Botany 92:199-206. 

Free JB. 1993. Insect pollination of crops, 2nd ed. Academic, London. 

Fulford RM. 1965. The morphogenesis of apple buds. I. The activity of the apical meristem. 

Annal of Botany 29:167-180. 

Fulford RM. 1966a. The morphogenesis of apple buds. III. The inception of flower. Annal of 

Botany 30:207-219. 

Fulford RM 1966b. The morphogenesis of apple buds. IV. The effect of fruit. Annal of Botany 

30:597-606. 

Gottschalk C and van Nocker S. 2013. Diversity in Seasonal Bloom Time and Floral 

Development among Apple Species and Hybrids. J Amer Soc Hort Sci 138(5):367-374. 

Greene DW. 1993. Effects of GA4 and GA7 on flower bud formation and russet development on 

apple. J Hort Sci 68: 171-176. 

Griffiths S, Dunford RP, Coupland G and Laurie DA. 2003. The evolution of CONSTANS-like 

gene families in barley, rice and Arabidopsis. Plant Physiol 131(4):1855-1867. 

Grauslund J. 1972. Gibberellins in diffusates from shoots of apple trees. Physiologia Plantarum 

27:65-70. 

Guitton B, Kelner JJ, Velasco R, Gardiner SE, Chagne D and Costes E. 2012. Genetic Control of 

biennial bearing in apple. J Exp Botany 63(1):131-149. 

Guitton B, Kelner JJ, Celton JM, Sabau X, Renou JP, Chagne D and Costes E. 2016. Analysis of 

transcripts differentially expressed between fruited and deflowered 'Gala' adult trees: a 
contribution to biennial bearing understanding in apple. BMC Plant Bio 16:55. 

Guttridge CG. 1962. Inhibition of Fruit-Bud Formation in Apple with Gibberellic Acid. Nature 

196:1008. 

 

23 
 

Haberman A, Ackerman M, Crane O, Kelner JJ, Costes E and Samach A. 2016. Different 
flowering response to various fruit loads in apple cultivars correlates with degree of 
transcript reaccumulation of a TFL1-encoding gene. The Plant J 87(2): 161-173.  

Hayama R, Izawa T and Shimamoto K. 2002. Isolation of Rice Genes Possibly Involved in the 

Photoperiodic Control of Flowering by a Fluorescent Differential Display Method. Plant & 
Cell Physiology 43(5):494-504. 

Hoover E, De Silva N, McArtney S, and Hirst P. 2004. Bud development and floral 

morphogenesis in four apple cultivars. J Hort Sci Biotech 79(6): 981-984. 

Hsu CY, Liu Y, Luthe DA, and Yuceer C. 2006. Poplar FT2 Shortens the Juvenile Phase and 

Promotes Seasonal Flowering. Plant Cell 18(8):1846-1861. 

Hsu CY, Adams JP, Kim HJ, No K, Ma CP, Strauss SH, Drnevich J, Vandervelde L, Ellis JD, 

Rice BM, Wickett N, Gunter LE, Tuskan GA, Brunner AM, Page GP, Barakat A, Carlson 
JE, dePamphilis CW, Luthe DS and Yuceer C. 2011. Flowering Locus T duplication 
coordinates reproductive and vegetative growth in perennial poplar. PNAS 108(26):10756-
10761. 

Jones AL. 1992. Evaluation of the Computer Model MARYBLYT for Predicting Fire Blight 

Blossom Infection on Apple in Michigan. Plant Disease 76(4): 344-347. 

Jonkers H. 1979. Biennial bearing in apple and pear: a literature survey. Scientia Horticulturae 

11:303-317. 

Kittikorn M, Okawa K, Ohara H, Kotoda N, Wada M, Yokoyama M, Ifuku O, Yoshida S, and 

Kondo S. 2011. Effects of fruit load, shading, and 9,10-ketol-octadecadienoic acid 
(KODA) application on MdTFL1 and MDFT1 genes in apple buds. Plant Growth 
Regulator 64:75-81. 

Kobayashi K, Yasuno N, Sato Y, Yoda M, Yamazaki R, Kimizu M, Yoshida H, Nagamura Y, 
and Kyozuka J. 2012. Inflorescence meristem identity in rice is specified by overlapping 
functions of three AP1/FUL-like MADS box genes and PAP2, a SEPALLLATA MADS 
box gene. Plant Cell 24(5):1848-59. 

Kotoda N, Wada M, Komori S, Kidou S, Abe K, Masuda T and Soejima J. 2000. Expression 
Pattern of Homologues of Floral Meristem Identity Genes LFY and AP1 during Flower 
Development in Apple. J Amer Soc Hort Sci 125(4):398-403. 

Kotoda N, Wada M, Kusaba S, Kano-Murakami Y, Masuda T and Soejima J. 2002. 

Overexpression of MdMADS5, and APETALA1-Like gene of apple, causes early flowering 
in transgenic Arabidopsis. Plant Sci 162:679-687. 

Kotoda N, Hayashi H, Suzuki M, Igarashi M, Hatsuyama Y, Kidou S, Igasaki T, Nishiguchi M, 

Yano K, Shimizu T, Takahashi S, Iwanami H, Moriya S and Abe K. 2010. Molecular 
Characterization of FLOWERING LOCUS T-Like Genes of Apple (Malus × domestica 
Borkh.). Plant Cell Physiol 51(4):561-575. 

Kumar G, Arya P, Gupta K, Randhawa V, Acharya V and Singh AK. 2016. Comparative 

phylogenetic analysis and transcriptional profiling of MADS-box gene family identified 
DAM and FLC-like genes in apple (Malus × domestica). Sci Reports 6:20695. 

 

24 
 

Kurokura T, Mimida N, Battey NH and Hytönen T. 2013. The regulation of seasonal flowering 

in the Rosaceae. J Exp Bot 64 (14): 4131-4141. 

Lawson DM, Hemmat M, and Weeden NF. 1995. The use of molecular markers to analyze the 

inheritance of Morphological and developmental traits in apple. J Amer Soc Hort Sci 
120(3): 532-537. 

Lavee S. 2007. Biennial Bearing in Olive (Olea europaea). Annales, Series Historia Naturalis 

17(1):101-112. 

Lee SY, Kim J, Han JJ, Han MJ and An GH. 2004. Functional analyses of the flowering time 

gene OsMADS50, the putative SUPPRESSOR of OVEREXPRESSION of CO1/AGAMOUS-
like 20 (SOC1/AGL20) ortholog in rice. Plant J38(5):754-764. 

Lewis C.I. and Vincent C.C. Pollination of the Apple. 1909. Oregon Agricultural College 

Experiment Station Bulletin no. 104. 

Liebhard R, Kellerhals M, Pfammatter W, Jertmini M, and Gessler C. 2003. Mapping 

quantitative physiological traits in apple (Malus × domestica Borkh.). Plant Molecular 
Bio52:511-526. 

Link H. 2000. Significance of flower and fruit thinning on fruit quality. Plant Growth Regulation 

31:18-26. 

Looney NE. 1969. Importance of time and concentration of annual applications of succinic acid 
2,2-dimethyl hydrazide (Alar) to pome fruits. I. McIntosh and Spartan apples. Can J Plant 
Sci 49:489-494. 

Luckwill LC, Weaver P and MacMillan J. 1969. Gibberellins and other growth hormones in 

apple seeds. J Hort Sci 44:413-424. 

Matsuzaki J, Kawahara Y and Izawa T. 2015. Punctual Transcriptional Regulation by the Rice 

Circadian Clock under Fluctuating Field Conditions. Plant Cell 27(3):633-648. 

McArtney S, and Li S. 1998. Selective Inhibition of Flowering on ‘Braeburn’ Apple Treees with 

Gibberellins. HortScience 33(4): 699-700. 

McArtney S, Hoover E, Hirst P and Booking IR. 2001. Seasonal variation in the onset and 

duration of flower development in ‘Royal Gala’ apple buds. J Hort Sci Biotech 76(5): 536-
540. 

McArtney S, Unrath D, Obermiller JD and Green A. 2007. Naphthaleneacetic Acid, Ethephon, 
and Gibberellin A4 + A7 Have Variable Effects on Flesh Firmness and Return Bloom of 
Apple. HortTechnology 17(1):32-38. 

McArtney S, Greene D, Schmidt T and Yuan R. 2013. Naphthaleneacetic acid and Ethephon Are 

Florigenic in the Biennial Apple Cultivars Golden Delicious and York Imperial. 
HortScience 48(6):742-746. 

McGregor SE. 1976. Insect Pollination of Cultivated Crop Plants. United States Department of 

Agriculture. 

McLaughlin JM and Greene DW. 1991. Fruit and Hormones Influence Flowering of Apple. I. 

Effects of Hormones. J Amer Soc Hort Sci 166(3):446-449. 

 

25 
 

Mellerowicz EJ, Horgan K, Walden A, Coker A and Walter C. 1998. PRFLL – a Pinus radiata 

homologue of FLORICAULA and LEAFY is expressed in buds containing vegetative shoot 
and undifferentiated male cone primordia. Planta 206(4): 619-629. 

Mimida N, Kotoda N, Ueda T, Igarashi M, Hatsuyama Y, Iwanami H, Moriya S and Kazuyuki 

A. 2009. Four TFL1/CEN-Like Genes on Distinct Linkage Groups Show Different 
Expression Patterns to Regulate Vegetative and Reproductive Development in Apple 
(Malus × domestica BorkH.). Plant Cell Physiol 50(2):394-412. 

Mimida N, Kidou SI, Iwanami H, Moriya S, Abe K, Voogd C, Varkonyi-Gasic E, Kotoda N, 
Nasholm T. 2011a. Apple FLOWERING LOCUS T proteins interact with transcription 
factors implicated in cell growth and organ development. Tree Physiology 31(5): 555-566. 

Mimida N, Ureshino A, Tanka N, Shigeta N, Sato N, Moriya-Tanaka Y, Iwanami H, Honda C, 

Suzuki A, Komori S and Wada M. 2011b. Expression patterns of several floral gene during 
flower initiation in apical buds of apple (Malus x domestic Borkh.) revealed by in situ 
hybridization. Plant Cell Rep 30:1485-1492. 

Mohamed R, Wang CT, Ma C, Shevchenko O, Dye SJ, Puzey JR, Etherington E, Sheng XY, 

Meilan R, Strauss SH and Brunner AM. 2010. Populus CEN/TFL1 regulates first onset of 
flowering, axillary meristem identity and dormancy release in Populus. Plant J 62(4):647-
688. 

Molinero-Rosales N, Jamilena M, Zurita S, Gomez P, Capel J and Lozano R. 1999. 

FALSIFLORA, the tomato orthologue of FLORICAULA and LEAFY, controls flowering 
time and floral meristem identity. Plant J 20(6):685-693. 

Monselise SP and Goldschmidt EE. 1982. Alternate bearing in fruit trees. Hort Rev 4:128-173. 

Munoz-Fambuena N, Mesejo C, Gonzalez-Max MC, Primo-Millo E, Agusti M and Iglesias DJ. 

2011. Fruit regulates seasonal expression of flowering genes in alternate-bearing 'Moncada' 
mandarin. Annals of Botany  108:511-519. 

Neilsen JC and Dennis FG. 1999. Effects of seed number, fruit removal, bourse shoot length and 

crop density on flowering in 'Spencer Seedless' apple. Acta Hort 527:137-146. 

Olszewski N, Sun T, and Gubler F. 2002. Gibberellin Signaling: Biosynthesis, Catabolism, and 

Response Pathways. The Plant Cell S61-S80. 

Pellerin BP, Buszard D, Iron D, Embree CG, Marini RP, Nichols DS, Neilsen GH and Neilsen 

D. 2011. A Theory of Blossom Thinning to Consider Maximum Annual Flower Bud 
Numbers on Biennial Apple Trees. HortScience 46(1):40-42.  

Porto DP, Brueneau M, Perini P, Anzanello R, Renou JP, dos Santos HP, Failho FB and Revers 

LF. 2015. Transcription profiling of the chilling requirement for bud break in apples: a 
putative role for FLC-like genes. J. of Experimental Botany 66(9):2659-2672. 

Quinlan JD and Preston AP. 1971. Influence of shoot competition on fruit retention and cropping 

of apple trees. J Hort Sci Biot 46:525-534. 

Rodrigo J. 2000. Spring frosts in deciduous fruit trees – morphological damage and flower 

hardiness. Scientia Horticulturae 85:155-173. 

 

26 
 

Rogers BL and Thompson AH. 1968. Growth and fruiting response of young apple and pear 

trees to annual applications of succinic acid 2,2-dimethyl hydrazide. Proc Am Soc Hort Sci 
93: 16-24. 

Schmidt T, Elfving DC, Mcferson JR and Whiting MD. 2009. Crop Load Overwhelms Effects of 
Gibberellic Acid and Ethephon on Floral Initiation in Apple. HortScience 44(7):1900-1906. 

Seymour GB, Ostergaard L, Chapman NH, Knapp S and Martin C. 2013. Fruit Development and 

Ripening. Annual Rev of Plant Bio 64:219-241. 

Shalom L, Samuels S, Zur N, Shlizerman L, Zemach H, Weissberg M, Ophir R, Blumwald E and 

Sadka A. 2012. Alternate Bearing in Citrus: Changes in the Expression of Flowering 
Control Genes and in Global Gene Expression in On- versus Off-Crop Trees. PloS One 
7(10). 

Tan FC and Swain SM. 2007. Functional characterization of AP1, SOC1 and WUS homologues 

from citrus (Citrus sinensis). Physiologia Plantarum 131(3):481-495. 

Trankner C, Lehmann S, Hoenicka H, Hanke M, Fladung M, Lenhardt D, Dunemann F, Gau A, 
Schlangen K, Malnoy M and Flachowsky H. 2010. Over-expression of an FT-homologous 
gene of apple induces early flowering in annual and perennial plants. Planta 232:1309-
1324. 

Tromp J. 2000. Flower-bud formation in pome fruits as affected by fruit thinning. Plant Growth 

Regulation 31:27-34. 

USDA. 2017. National Agricultural Statistics Service: Noncitrus Fruits and Nuts 2016 Summary. 

Washington, DC: U.S. Department of Agriculture. 

Velasco R, Zharkikh A, Affourtit J, Dhingra A, Cestaro A, Kalyanaraman A, Fontana P, 

Bhatnagar SK, Troggio M, Pruss D, Salvi S, Pindo M, Baldi P, Castelletti S, Cavaiuolo M, 
Coppola G, Costa F, Cova V, Dal Ri A, Goremykin V, Komjanc M, Longhi S, Magnago P, 
Malacarne G, Malnoy M, Micheletti D, Moretto M, Perazzolli M, Si-Ammour A, Vezzulli 
S, Zini E, Eldredge G, Fitzgerald LM, Gutin N, Lanchbury J, Macalma T, Mitchell JT, 
Reid J, Wardell B, Kodira C, Chen Z, Desany B, Niazi F, Palmer M, Koepke T, Jiwan D, 
Schaeffer S, Krishnan V, Wu C, Chu VT, King ST, Vick J, Tao Q, Mraz A, Stormo A, 
Stormo K, Bogden R, Ederle D, Stella A, Vecchietti A, Kater MM, Masiero S, Lasserre P, 
Lespinasse Y, Allan AC, Bus V, Chagne D, Crowhurst RN, Gleave AP, Lavezzo E, 
Fawcett JA, Proost S, Rouze P, Sterck L, Toppo S, Lazzari B, Hellens RP, Durel CE, Gutin 
A, Bumgarner RE, Gardiner SE, Skolnick M, Egholm M, Van de Peer Y, Salamini F and 
Viola R. 2010. The genome of the domesticated apple (Malus × domestica Borkh.). Nature 
Genetics 42:833–839. 

Visser T. 1964. Juvenile Phase and Growth of Apple and Pear Seedlings. Euphytica 1964:119-

129. 

Wada M, Cao Q, Kotoda N, Soejima J and Masuda T. 2002. Apple has two orthologues of 
FLORICAULA/LEAFY involved in flowering. Plant Molecular Biology 49:567-577. 

Wardlaw IF. 1990. The control of carbon partitioning in plants. New Phytologist 116:341-381. 

 

27 
 

Webster AD, Oehl VH, Jackson JE, and Jones OP. 1985. The orchard establishment, growth and 

precocity of four micropropagated apple scion cultivars. J Hort Sci 60:169-180. 

Williams MW and Edgerton LJ. 1974. Biennial bearing of apple trees. Proc XIX Intern Hort 

Congr Warsaw 3:343-352. 

Wu G and Poethig RS. 2006. Temporal Regulation of Shoot Development in Arabidopsis 

Thaliana by Mir156 and Its Target SPL3. Development 133(18):3539-3547. 

Zhang S, Zhang D, Fan S, Du L, Shen Y, Xing L, Li Y, Ma J, and Han M. 2016. Effect of 

exogenous hormones, and flowering-associated genes in ‘Fuji’ apple (Malus domestica 
Borkh.) Plant Physio Biochem 107: 178-186. 

Ziv D, Zviran T, Zezak O, Samach A and Irihimovitch V. 2014. Expression Profiling of 

FLOWERING LOCUS T-like Gene in Alternate Bearing 'Hass' Avocado Trees Suggests a 
Role of PaFT in Avocado Flower Induction. PloS One 9(10). 

 

 

 

28 
 

CHAPTER TWO 

Genetic mechanisms associated with floral initiation and the repressive effect of fruit on 

flowering in apple (Malus × domestica Borkh.) 

 

 

 

 

Submitted.

 
 

29 

ABSTRACT 

Many apple cultivars are subject to biennial fluctuations in flowering and fruiting. It is 

believed that this phenomenon is caused by a repressive effect of developing fruit on the 

initiation of flowers in the apex of proximal bourse shoots. However, the genetic pathways of 

floral initiation are incompletely described in apple, and the biological nature of floral repression 

by fruit is currently unknown. In this study, we characterized the transcriptional landscape of 

bourse shoot apices in the biennial cultivar, 'Honeycrisp', during the period of floral initiation, in 

trees bearing a high fruit load and in trees without fruit. Trees with high fruit load produced 

almost exclusively vegetative growth in the subsequent year, whereas the trees without fruit 

produced flowers on the majority of the potential flowering nodes. Using RNA-based sequence 

data, we documented gene expression at high resolution, identifying >11,000 transcripts that had 

not been previously annotated, and characterized expression profiles associated with vegetative 

growth and flowering. We also conducted a census of genes related to known flowering genes, 

organized the phylogenetic and syntenic relationships of these genes, and compared expression 

among homeologs. Several genes closely related to AP1, FT, FUL, LFY, and SPLs were more 

strongly expressed in apices from non-bearing, floral-determined trees, consistent with their 

presumed floral-promotive roles. In contrast, a homolog of TFL1 exhibited strong and persistent 

up-regulation only in apices from bearing, vegetative-determined trees, suggesting a role in floral 

repression. Additionally, we identified four GIBBERELLIC ACID (GA) 2 OXIDASE genes that 

were expressed to relatively high levels in apices from bearing trees. These results define the 

flowering-related transcriptional landscape in apple, and strongly support previous studies 

implicating both gibberellins and TFL1 as key components in repression of flowering by fruit. 

 

 
 

 

30 

INTRODUCTION 

In many tree fruits and nuts, flowering follows a biennial cycle, with maximal and 

minimal flowering alternating yearly [1-3]. This phenomenon, termed biennial (alternate) 

bearing, is both an intriguing biological phenomenon and a significant limitation for the 

production of many horticultural crops. In commercial (domesticated) apple, similar to many 

other tree fruit species, flowering spans two growing seasons. In the first growing season, floral 

meristems initiate at the tips of condensed shoots called bourse shoots [4-7]. The floral 

meristems develop during the remainder of the growing season and arrest in a partially 

developed state before the winter dormant period. In early spring of the subsequent growing 

season, flowers complete development, culminating in bloom shortly after release from 

dormancy. This two-year cycle leads to an overlap between the period of fruit development 

(from the previous season's flowers) and the period of floral initiation (current season). At least 

for domesticated apple, it is generally acknowledged that the presence of developing fruit 

inhibits floral initiation within the adjacent bourse shoot. Several ideas have been offered to 

explain how developing fruit might repress floral initiation. For example, gibberellins (GAs) 

have been shown to repress floral initiation in apple [8], and as developing fruit contain 

relatively high concentrations of gibberellins [9], it is thought that diffusion of GAs from the 

fruit to shoot apex could underlie floral repression [8]. It has also been hypothesized the biennial 

bearing results from diversion of photosynthate from the apex to the developing fruit due to the 

potentially higher sink strength of the fruit [3,10]. 

The repressive effect of fruit should ultimately be reflected in expression of floral-

promotive genes at the shoot apex. Previous studies have identified putative molecular 

components of the flowering pathway in apple based on apparent homology with well-studied 

 

31 
 

flowering genes such as APETALA 1 (AP1), LEAFY (LFY), SUPPRESSOR OF CONSTANS 

(SOC1), FRUITFULL (FUL), FLOWERING LOCUS T (FT), and TERMINAL FLOWER 1 

(TFL1). Homologs of the floral promoters LFY (designated MdAFL2), FUL 

(MdFUL/MdMADS2), and FT (MdFT1) were reported to exhibit increased expression within the 

apex during the anticipated period of floral induction [11-18]. In contrast, two homologs of the 

floral repressor TFL1 (MdTFL1-1/2) were reported to exhibit rapidly decreasing expression 

either prior to or during the floral induction and initiation period [13-14,16-19]. During floral 

initiation, MdFUL, as well as homologs of the floral promoters AP1 (MdAP1a and MdAP1b), 

and SOC1 (MdSOC1), exhibited increasing expression [14,16-18,20]. Following floral initiation, 

the expression of homologs of FT (MdFT2), LFY (MdAFL1), and AP1 (MdAP1) were either 

maintained at a relatively high level or were further increased coinciding with floral development 

[11,13-14,16-18,20]. During floral development and thereafter, MdTFL1-2 expression increased 

[14,18-19]. Various homologs of AP1, LFY, FT, and TFL1 have been additionally examined for 

flowering function by manipulating their expression in transgenic Arabidopsis or apple [11-

12,16, 19; 21-24].  

Relatively few studies have investigated the effect of fruit load on the expression of 

specific, presumed flowering genes in apple or other tree fruit species. In biennial-bearing 

avocado and citrus cultivars, FT-like genes were expressed in fully developed adult leaves during 

the period of floral initiation, and this expression was found to be significantly higher in the 

bearing year [25-26]. However, in the biennial-bearing apple cultivar ‘Red Delicious’, MdFT1 

expression in the leaves was found to be similar between non-bearing and bearing years [18]. On 

the other hand, a PCR-based study suggested that the expression of a MdFT2 was higher in 

apical buds of apple trees carrying a high fruit load compared with trees with no fruit [17,27]. 

 

32 
 

Haberman et al. (2016) reported that MdTFL1-1 expression in bourse shoot apices decreased 

during the course of the growing season, and that the decrease in MdTFL1-1 expression was 

more rapid in trees carrying a high fruit load compared to low fruit load. In addition, they found 

that MdTFL1-2 expression significantly increased relatively late in the season, but only in the 

high fruit load trees. This pattern was interpreted as suggesting that MdTFL1-1 might play a role 

in maintaining the apex in a vegetative state early in the growing season, whereas MdTFL1-2 

might repress flowering in response to high fruit load [18].  

Although these previous studies documenting gene expression in the apex have provided 

a solid blueprint for the advanced molecular study of flowering and alternate bearing in apple, 

they have focused on a limited number of anticipated landmark genes. In this study, as a 

subsequent step to understanding the genetic basis of floral repression by fruit in apple, we 

carried out an extensive census of flowering-related genes, a comprehensive analysis of gene 

expression in the bourse shoot apex during the transition to floral initiation, and evaluation of the 

effect of fruit load on the expression of flowering-related genes. Ultimately, this work should 

provide a deeper understanding of the endogenous mechanism(s) responsible for floral initiation 

and alternate bearing. This, in turn, may facilitate the development of approaches to control 

flowering in commercial operations, and the development of new cultivars less prone to alternate 

bearing.  

MATERIALS AND METHODS 

Plant materials, growth conditions and field experimental design 

Field experiments were conducted at the Michigan State University (MSU) Clarksville 

Research Center (Field: 42°52'28.91"N, 85°16'8.15"W – Station: 42°52'24.20"N, 

85°15'30.81"W) located in Clarksville, MI. Trees were managed in accordance with standard 

 

33 
 

commercial practices for disease, insect, and weed control. ‘Honeycrisp' trees had been 

established for five years as grafts on Nic 29® rootstocks. The date of full bloom was defined as 

the date in which the maximum numbers of flowers were open but had not reached anthesis. Six 

trees were chosen that showed at least 80% bloom density, defined as the percentage of nodes on 

one-year-old shoots that showed flower clusters. For each tree, six branches, each between 4 and 

6 cm diameter at the base, were selected and randomly assigned for apex collection dates (five 

branches) or for observation of flowering the following spring (one branch). Plants were 

randomly assigned as three replicate pairs, with each pair comprising one plant that was 

subjected to removal of all flowers, and one plant that was left untouched. Collections were 

made at 2 days after full bloom (DAFB), 15-17 DAFB, 35-38 DAFB, 49-52 DAFB, and 72-75 

DAFB. On each collection date, dominant buds immediately subtending the position of flower 

clusters or former cluster position, or the apex of actively growing shoots originating from this 

position, were removed using a razor blade, immediately frozen in liquid N2, and transferred to 

storage at -80°C. 

Nucleic acid preparation, sequencing, and data analyses 

RNA was isolated from frozen apex samples using the method of Gasic et al. (2004) with 

the exception that spermine was substituted for spermidine in the extraction buffer, followed by a 

final 'clean-up' step using a commercial kit (RNeasy Mini; QIAGEN, Germantown, MD). RNA 

quality and quantification was analyzed by the use of a Nanodrop 2000c (Thermo Scientific, 

Waltham, MA) and electrophoresis (2100 Bioanalyzer; Agilent, Santa Clara, CA). Library 

preparation and sequencing used the Illumina (San Diego, CA) platform and TruSeq platform 

with 101-b paired-end protocols, starting with 1 ug of total RNA from each sample. The raw 

sequence files were processed with fastq-mcf [29] using the parameters -t 0.10 -p 15 -l 20 -q 25 

 

34 
 

to remove adapter sequences, very short reads, and terminal bases with a Phred score below 25. 

The number of read pairs generated is shown in S1 Table. 

Reference-based transcriptome assembly 

Sequence reads were aligned to v1.1 of the GDDH13 reference sequence [30] and splice 

junctions were identified using the program HISAT2 (v.2.1.0) invoking the --dta-cufflinks and --

un-conc-gz options [31]. The --un-conc-gz option was invoked to capture reads that failed to 

map to the reference genome. HISAT2 was operated using the default maximum and minimum 

mismatch penalties of six and two, respectively. Alignment metrics are shown in S1 Table. 

Transcript models were assembled using StringTie (v.1.3.3) using default parameters, including 

the -G option for use of a reference annotation as described [32-33]. Transcript models generated 

for each sample library were reduced to a consensus set of transcript models using the StringTie 

–merge function. The program Cuffquant (v.2.2.1; included in the Cufflinks suite [34]) was then 

used to calculate sequence read counts for each transcript model, and significant differentially 

expressed genes and isoforms were identified by the use of Cuffdiff [35]. Metrics for assessing 

read mapping and transcriptome assembly were obtained using RNA-SeQC (v.1.1.8 [36]) and 

GFF utilities suite [37], respectively.  

Identification of novel ‘Honeycrisp’ reference-based transcripts 

Novel transcripts contained within the reference-based transcriptome were identified by 

comparing the reference genome gene models (retrieved from 

https://iris.angers.inra.fr/gddh13/the-apple-genome-downloads.html as 

gene_models_20170606.gff3) and the ‘Honeycrisp’ reference-based transcript models (S1 File) 

using the gffcompare (v.0.9.12) software package within the GFF utilities [37]. The resulting 

annotated gtf file was filtered for classification codes associated with non-isoform-like transcript 

 

35 
 

features and/or assembly errors (classification codes: e, i, o, u, x, y) and removal of transcripts 

with lengths <200b. This subset of transcripts was then analyzed for protein-coding capacity 

using the software programs CPC2 (beta version [38]), PLEK (v1.2 [39]), and CPAT (v.1.2.4 

[40]). For CPC2 and CPAT, the coding potential probability was set to ≥0.5 to assign a transcript 

as coding and ≤0.5 as noncoding/ambiguous. For PLEK, the coding or noncoding/ambiguous 

determination was assigned by the program’s default parameters. The final coding definition of a 

transcript was based on an agreement between at least two of the programs. Detailed transcript 

information can be found in S2 - S4 Tables. 

De novo assembly of unmapped reads 

Reads that were unmapped by HISAT2 were assembled into contiguous sequences using 

the Trinity de novo assembler with default settings [41]. The resulting FASTA file (S2 File) 

containing the de novo assembled transcripts was then used as an input to construct consensus 

gene models using the python program Trinity_gene_splice_modeler.py provided by the Trinity 

suite (S3 File). The python script produced a consensus FASTA file containing gene models and 

a corresponding GTF file. The unmapped read files were then realigned to the consensus gene 

FASTA file using HISAT2 invoking the --dta-cufflinks options and using the previously 

generated GTF file. Alignments were then processed through the same Cufflinks pipeline used in 

the referenced-based transcriptome assembly. The initial output comprised ~250,000 sequences 

corresponding to ~92,000 distinct loci. Because most output sequences appeared to be 

sequencing or assembly artifacts, we limited further consideration to contigs representing 

putative transcripts that were likely to be strongly expressed (upper 10th percentile based on 

FPKM, and expressed in at least three samples) and that had coding potential (determined as 

described above for novel reference-based transcript models) (S5 - S7 Tables).  

 

36 
 

General transcriptome annotation 

Transcript sequences were annotated based on sequence homology to Arabidopsis open 

reading frame translations (TAIR10; TAIR10_pep_20101214_updated 2012-04-16, [42]) using 

the BLASTx module from NCBI [43] with an Expect (E)-value cutoff of 1e-11. Homologous 

sequences were then used as queries to identify similar transcripts within the ‘Honeycrisp’ 

transcriptome, using the tBLASTx module. Gene model sequences generated from the de novo 

assembly were annotated by aligning sequences to the nr NCBI database (downloaded on 2018-

09-18, [44]) and the ‘Honeycrisp’ reference-based transcriptome using the BLASTx and 

BLASTn modules, respectively. A minimum E-value of 1e-10 and a max_target_seqs of 1 were 

used.  

Data accessibility 

Raw sequence libraries can be downloaded from the National Center of Biotechnology 

Information Short Read Archive under biosample SAMN04239699. Our constructed reference-

based transcriptome annotation (S1 File), de novo transcript FASTA and annotation (S2 and S3 

Files), phylogenies of 125 flowering genes, and differential expression data files (S4 and S5 

Files) can be retrieved from the Dryad repository https://doi.org/10.5061/dryad.fn2z34tr5. 

Computation protocols used in this study can be retrieved from the Protocols.io repository 

https://dx.doi.org/10.17504/protocols.io.bp54mq8w. 

Identification of apple flowering genes 

To identify potential homologs of flowering genes, we indexed genes from Arabidopsis 

(TAIR10) annotated with potential roles in flowering: Gene Ontology terms 0048438 ('floral 

whorl development'), 0009908 ('flower development'), 0009910 ('negative regulation of flower 

development'), 0009911 ('positive regulation of flower development'), 0048578 ('positive 

 

37 
 

regulation of long-day photoperiodism, flowering'), 0010220 ('positive regulation of 

vernalization response'), 0009909 ('regulation of flower development'), 0048510 ('regulation of 

timing of transition from vegetative to reproductive phase'), 0010321 ('regulation of vegetative 

phase change'), 0010228 ('vegetative to reproductive phase transition of meristem'), 0010048 

('vernalization response'), and 0010093 (‘specification of floral organ identity’). This set of 437 

genes was manually curated to omit those without strong functional evidence for a direct role in 

flowering. The curated subset contained 180 genes. Conceptual translations of the corresponding 

representative gene models were obtained from TAIR 

(TAIR10_pep_20110103_representative_gene_model_updated) and used as queries to search 

open reading frame translations of our mapped-assembled and de novo-assembled transcript 

models (BLASTp) using an E-value cutoff of 1e-12. The open reading frame translations of the 

Honeycrisp assembled transcript models were identified using TransDecoder 

(v5.5.0;https://github.com/TransDecoder). All identified transcript translations were then used as 

queries to search the Arabidopsis representative gene model translations. Those transcripts that 

reciprocally identified their original Arabidopsis query were defined as reciprocal homologs. For 

phylogenetic analyses of the 16 intensively studied flowering gene families, we considered only 

the 25 highest-scoring apple transcript translations and only the 25 highest-scoring Arabidopsis 

gene translations identified with each apple sequence query. Phylogenetic trees were then 

constructed using the ETE3 toolkit (v.3.1.1) build function invoking the standard_fasttree 

workflow under default settings [45-47]. Collinearity among identified flowering genes was 

performed using the MCScanX toolkit following the manual’s instructions and the use of default 

parameters [48]. Graphics to illustrate the collinear relationships between homeologous 

 

38 
 

chromosomes and flowering genes identified by MCScan X were generated using Circos (v.0.69-

6) program package [49]. 

Gene expression analysis 

Estimated expression levels for homologs of flowering-related genes/transcripts were 

obtained from Cuffnorm and Cuffdiff output. Heat maps were created and expression profiles 

were clustered using R statistical software (v.3.5.2 [50]) and the CummeRbund (v.2.24.0 [51]) 

package. Expression profiles of homologous flowering-related genes that exhibited significant 

changes in expression were clustered using a K-means approach by the csCluster command of 

CummeRbund. Cluster expression pattern was then defined by the general trend of the modal 

expression pattern. Venn diagram of differentially expressed genes was created using an online 

tool (http://bioinformatics.psb.ugent.be/webtools/Venn/). Co-expressed gene modules were 

identified using the weighted gene co-expression network analysis (WGCNA) R package [52], 

following the analysis methodology outlined by Zhang and Horvath (2005) and using normalized 

gene expression (FPKM) as calculated by cufflinks.  

TaqMan®️ qRT-PCR 

Confirmation of MdTFL1 gene expression was determined using a two-step quantitative 

polymerase chain reaction (qRT-PCR). Primers and probes were designed from sequences 

assembled in our transcriptome and aligned to apple nucleotide sequences maintained by the 

NCBI (taxid: 3750). The primers and probes were designed in a previous study [54] and were 

based on the specificity to selected target sequence and overlapped of an exon junction (S8 

Table). An apple homolog of ACTIN served as an internal control. The reactions were performed 

using an Agilent Technologies Stratagene Mx3005P (Santa Clara, CA) qPCR machine with 

cDNA derived from the RNA samples prepared for RNA-seq. Each reaction consisted 

 

39 
 

TaqManTM Gene Expression Master Mix (10 μl), 5x diluted cDNA template (2 μl), forward (1 

μl) and reverse primers (1 μl), and probe (1 μl) for ACTIN, the primer-probe assay for the gene 

of interest (1 μl), and ddH2O (4 μl). The thermal profile for TaqManTM assay followed the 

instructions provided with the Agilent machine. 

RESULTS AND DISCUSSION 

Effects of reducing fruit load on floral initiation 

As a physiological and molecular model for biennial bearing in apple, we focused on the 

popular commercial cultivar ‘Honeycrisp’, which can exhibit extreme biennial tendency under 

production conditions [55]. At full bloom in early spring, we selected three paired sets of trees 

with high bloom density and removed all flowers from one tree from each pair. This floral 

thinning treatment had a strong effect on initiation of new flowers, as evidenced by observed 

bloom density in the spring of the second year of the study (Fig 2.1A). Those trees that were 

thinned of flowers produced floral shoots at an average of ~52% (range 39-82%) of potential 

flowering nodes, whereas the non-thinned control trees produced almost exclusively vegetative 

shoots (Fig 2.1B).  

Transcriptome assembly and characterization  

Based on anatomical characterization of the bourse shoot meristem in the apple cultivar 

used in this study, floral meristems began to be initiated approximately two weeks after full 

bloom (data not shown). This is consistent with observations of the bourse shoot meristem in 

other apple cultivars [4-7]. Based on this, we sampled the bourse shoot apex from the thinned 

and non-thinned trees at approximately 2, 15, 35, 50, and 70 DAFB. Dissected apices were 

subjected to high-throughput RNA-based sequencing, yielding a total of ~390 million paired 

reads. These were aligned to a recently published reference genome sequence (GDDH13 v.1.1) 

 

40 
 

assembled from a doubled-haploid individual generated from ‘Golden Delicious’ [30]. We 

obtained a mean alignment rate of 90.1%, with 96.3% of the aligned reads mapping within 

annotated intragenic regions (S1 Table). Transcript models were then assembled from aligned 

reads using the StringTie transcript assembler [33]. 

The recent availability of a high-quality apple genome sequence and exhaustive depth of 

our transcriptional data provided the opportunity to document genes expressed in the apple 

bourse shoot apex with high-resolution and accuracy. Our reference-based transcriptome 

assembly cataloged a total of 104,690 transcripts arising from 58,452 loci (Table 2.1; S1 File). 

This extends considerably the previously annotated gene content of the GDDH13 genome, which 

was based on nine RNA-seq libraries representing diverse structures, including the shoot apex, 

along with cDNAs and expressed sequence tags (ESTs) cataloged in NCBI databases. Our 

sequence and assembly results complemented the reference annotation with the identification of 

an additional 11,264 novel transcriptional models and 39,227 'Honeycrisp'-specific isoforms of 

annotated transcripts (~10.8% and ~37.5% of the assembled transcripts, respectively; Fig 2.2A). 

These novel transcripts comprised 23,034 novel exons and originated from 8,753 previously 

unidentified loci. We further characterized these novel transcripts in terms of length, expression 

level, coding potential, genomic organization, and homology with known, expressed genes (Fig 

2.2B; S2 - S4 Tables).  

The majority of these transcripts (81.2%; 9,096) were expressed (FPKM > 1; TPM 1.19 - 

1.72; Fig 2.2B; S3 Table). Of those expressed transcripts, 63.7% were predicted to encode 

proteins. About 46% of the expressed-coding transcripts were located in previously annotated 

intergenic regions. The remaining 56% showed some positional overlap with previously 

annotated genes (Fig 2.2B). In total, 73.8% (8,310) of the novel transcripts showed significant 

 

41 
 

(E-value < 1e-10) nucleotide sequence homology to previously cataloged, expressed genes from 

Malus spp. (Fig 2.2C). These genes included 163 distinct loci encoding the M. floribunda HcrVf-

like and M. x domestica Rvi15 apple scab (Venturia inaequalis) resistance genes, and 159 loci 

encoding the M. x robusta fire blight (Erwinia amylovora) resistance genes. A total of 1,304 

reference-mapped transcripts exhibited no significant homology to any sequence cataloged in the 

NCBI nt database (S2 Table). 

Reads that did not align with the reference genome may represent sequence from 

uncharted segments of the apple genome including extrachromosomal DNAs or loci that are 

extremely diverged between GDDH13 and 'Honeycrisp', or may be derived from exogenous 

biota. We assembled unmapped reads de novo into contiguous sequences (S2 and S3 Files) (see 

Methods), and evaluated the potential of the contigs to represent authentic apple transcripts. A 

total of 5,542 potential transcripts, representing 4,737 gene models, showed apparent expression 

values >100 FPKM in at least three of the sequencing libraries. About 39% of this subset of de 

novo transcripts were predicted to encode proteins (S5 Table; S1 Fig A). About 75% of the 5,542 

strongly expressed potential transcripts displayed significant homology to cataloged Malus 

sequences, and another 15% to sequences from related Rosaceae genera (S5 Table; S1 Fig B). 

Identification of flowering gene homologs 

Although genes with anticipated roles in flowering have previously been identified in 

apple, there has often been confusion and conflicting reports regarding gene identity, copy 

number, and expression pattern. This is most likely due to the existence of closely related 

orthologs for some of these genes, the heterogeneous and paleo-allopolyploid nature of the apple 

genome, and the inability of some previous approaches to discriminate among closely related 

sequences. The ~40 billion bases of transcriptional sequence data from the shoot apex analyzed 

 

42 
 

in this project, as well as our identification of novel genes, provided the opportunity to resolve 

gene identities and estimate orthologous relationships. We identified a set of 180 Arabidopsis 

genes with flowering-related annotations (see Methods), and searched the combined GDDH13 / 

'Honeycrisp' transcriptome for expressed sequences with significant homology (E-value < 1e-

12). In each case, the open reading frame translation from the primary designated transcript of 

the Arabidopsis gene was used to query the primary translations from both the annotated and 

novel reference-based transcriptional models, as well as the ORF-containing de novo 

transcriptional models. The highest-scoring, matching sequences were then used reciprocally to 

query a comprehensive database of open reading frame translations from Arabidopsis. Using this 

approach, we identified a total of 321 apple counterparts to 125 Arabidopsis genes. For further 

discussion, we refer to this collection as 'flowering gene homologs'. Three of the identified apple 

genes had not previously been annotated in the GDDH13 reference genome (S9 Table). 

At least 106 of the 125 Arabidopsis flowering genes had multiple homologs. Previous 

research indicates that genes in apple generally exist as duplicates as a result of an ancient 

whole-genome duplication [56]. We analyzed genomic synteny for all of the 125 flowering gene 

families (S9 Table; Fig 2.3). Based on chromosomal positions, a simple genome duplication 

appears to have contributed to family expansion for at least 86 of these 106 Arabidopsis genes, 

and tandem duplication contributed to expansion for at least 14 (S10 Table). 

We identified 55 Arabidopsis flowering-related genes lacking a reciprocal homolog in the 

combined reference shoot apex transcriptome. These unrepresented genes included several 

functioning in the Arabidopsis vernalization-response pathway, including FLOWERING LOCUS 

C (FLC) and its sibling MADS AFFECTING FLOWERING (MAF), FRIGIDA (FRI), and 

VERNALIZATION INSENSITIVE 3 (VIN3). This result is consistent with the apparently cold-

 

43 
 

independent initiation of flowers during the summer period in apple [57]. The gene previously 

described as an FLC homolog (MD09G1009100) by Takeuchi et al. (2018) and Nishiyama et al. 

(2019) was found to be not closely related to FLC in our study (S2 Fig [d] and not shown). Other 

Arabidopsis flowering genes without clear apple representatives included 16 additional members 

of the AGAMOUS-like (AGL) MADS-box gene superfamily. These results are consistent with 

the observed rapid evolution and diversification of the large MADS-box genes observed in apple 

and other plants [60-61]. 

We focused further study on a subset of flowering genes that have been intensively 

studied both in Arabidopsis and other plants [62]. Apart from FLC, this subset included AGL24, 

AP1, FD, FUL, FT, LFY, SOC1, SPL3, SPL4, SPL5, SPL9, SPL15, SVP, TSF, and TFL1. (Table 

2.2; S2 Fig [a-n]; Fig 2.3). In Arabidopsis, FT is transcribed in the leaves along with its paralog 

TSF and translocated to the apex[Andrés and Coupland 2012]. In the apex, FT forms a complex 

with FD which activates transcription of AP1 and SPL3/4/5 directly [62]. The FD/FT complex 

also indirectly activates the expression of FUL and SOC1 [62]. In addition, FUL and SOC1 

expression is reinforced by SPL9/15 [63]. This collective network promotes a phase change 

within the apex leading to floral initiation. SOC1 and AGL24 form a positive-feedback loop, 

promoting one another's expression along with promoting LFY expression [64]. LFY expression 

is also directly promoted by SPL3/4/5 and indirectly by AP1 establishing floral meristem identity 

[63]. Negative regulators of this process are SVP and TFL1. SVP represses FT expression, 

whereas TFL1 competes with FT for complex formation with FD [62].  

We reconstructed phylogenies for these genes, including the most homologous genes 

from both Arabidopsis and apple, and generated un-rooted trees (Fig 2.3). Apple genes related to 

AGL24/SVP, AP1/FUL, FD, FT/TFL1, LFY, SOC1, SPL4/5, and SPL9/15 were included in well-

 

44 
 

defined (>90% bootstrap replicates) clades. The majority of these apple genes existed as pairs on 

homeologous chromosomes, as anticipated. Additional homologs likely resulted from tandem 

duplication, as evidenced by their close proximity (e.g., the AGL24/SVP clade pair 

MD15G1384500/MD15G1384600). Apple was previously found to contain two homologs of FT, 

one positioned on Chr. 4 (MdFT2) and the other on Chr. 12 (MdFT1) [16]. The GDDH13 

genome contains only MdFT1. We assembled sequence reads that did not map to the GDDH13 

genome (see Methods) and were able to identify a MdFT2-like transcript (FT-like de novo) 

(Table 2.2; S2 Fig [a-n]; Fig 2.3). This result suggests that the GDDH13 genome sequence is 

incomplete for Chr. 4 or that the GDDH13 doubled-haploid genotype lacks MdFT2. As 

anticipated from the reciprocal homology results (above), individual members of the FLC/MAF 

family showed no specific phylogenetic relationships with apple genes, although a group of three 

apple genes were often (88% of bootstrap replicates) placed into a clade with the FLC/MAF 

family (S2 Fig).  

Transcriptional analysis of the apple shoot apex during the floral transition 

To gain insight into genetic pathway(s) associated with flowering in apple, we examined 

changes in gene expression occurring in the bourse shoot apex in the set of flowering-induced 

(thinned) trees spanning 2 DAFB to 70 DAFB. Because an appreciable fraction (~48%) of the 

apices did not initiate flowers during the year of this study (Fig 2.1), this set of genes represent 

those associated with vegetative apex activity (i.e. continued production of leaf primordia), as 

well as the transition to flowering. We identified a total of 12,661 reference-mapped genes, 

including ~100 flowering gene homologs, that exhibited significant changes in expression in at 

least one pairwise comparison among the five developmental stages evaluated.  

 

45 
 

To define transcriptional programs potentially involving the ~100 flowering gene 

homologs, we clustered their expression profiles using a K-means approach (k = 5) (Fig 2.4). 

Clusters 1-3 represented genes that showed decreases in expression at some point during the 

period, consistent with a floral repressive role, or expression largely limited to vegetative phase. 

Cluster 1 genes (n = 18) generally showed a strong decrease in expression at the earliest studied 

interval, between 2 and 15 DAFB, with little or no expression change at later time points. This 

cluster included a homolog of FD (Fig 2.4). In contrast, Cluster 2 (n = 19) genes showed 

progressively decreasing expression over the course of the season. This cluster included 

MdSOC1a, as well as homologs of AGL24/SVP, SPL4/5, and SPL9/15. Most of the genes in 

Cluster 3 (n = 10) showed a strong decrease in expression at the earliest studied interval, between 

2 and 15 DAFB, and continued decreasing expression at the later time points. This cluster 

included both MdTFL1-1 and MdTFL1-2, although we noted that MdTFL1-1 was upregulated 

between 2 and 15 DAFB (Fig 2.4). The strong decrease in expression of these two TFL1 

homologs during the anticipated period for floral initiation has previously been reported [13-

14,16-19,65-66].  

Genes in Clusters 4 and 5 showed generally increasing expression across the entire study 

period, suggesting promotive roles in flowering or expression domains linked with the floral 

phase. Cluster 4 genes (n = 48) showed steadily increasing expression across the period. These 

included MdFT1, as well as homologs of AGL24/SVP, SPL3/4/5, and AP1/FUL. Cluster 5 

contained only five genes, and these were characterized by a generally more substantial increase 

in expression over the season. This cluster included MdAFL1, as well as a homolog of AP1/FUL 

and two homologs of AGL24/SVP. The increasing expression of MdAFL1 and the AP1/FUL 

homolog reflects the increased expression of their counterparts during flowering in Arabidopsis. 

 

46 
 

Expression of the two AGL24/SVP homologs was analogous with that of AGL24 in Arabidopsis 

during the transition to a reproductive meristem [62].  

Transcriptional response to the presence of a fruit load 

To identify genetic mechanisms that may be specifically involved in the repression of 

flowering by developing fruit, we compared gene expression between apices from the thinned 

(flowering-induced) and non-thinned (non-induced) trees at each time point over the study. At 

the 15 and 35 DAFB sampling times, fruit had reached ~10 mm and ~20 mm in diameter, 

respectively. At 50 DAFB, fruit had reached ~ 30 mm in diameter, and at 70 DAFB, fruit was 

~40 mm in diameter. At 70 DAFB, seeds and embryos were still immature, but had reached their 

final size. Fruit and seed reached maturity at ~120-130 DAFB (not shown). We identified a total 

of 6,595 genes that were differentially expressed between the two conditions at one or more time 

points. Of these, 55 were included in the defined set of flowering gene homologs (S11 Table; Fig 

2.5).  

K-means clustering identified five modal expression patterns. Genes in Cluster 1 were 

generally expressed to higher levels in non-thinned apices at later time points (50 and 70 DAFB) 

and thus could represent downstream floral repressors or genes expressed in the vegetative 

tissues of the apex. This expression pattern was exemplified by the AGL24/SVP homolog 

MD15G1384600 (Fig 2.4). This result suggests that the function of MD15G1384600 could be 

similar to SVP in maintaining vegetative identity [62]. 

Cluster 2 genes were generally expressed to higher levels in thinned apices at the earliest 

time points (2 and 15 DAFB) and could represent early flowering promoters. An example 

included in this cluster is the AP1/FUL-related gene, MD06G1204400. Genes in Clusters 3 and 4 

showed generally increasing expression in thinned apices, relative to non-thinned apices, 

 

47 
 

throughout the study period. Cluster 3 (higher expression in non-thinned apices only at the 

earliest time points) could represent early flowering repressors or genes expressed early in the 

vegetative tissues. This cluster contained the SPL4/5 homolog, MD03G1230600. Cluster 4 

(higher expression in thinned apices at the latest time point) might represent genes acting as 

promoters late in flowering, including floral development, or genes expressed in floral tissues. 

This cluster contained homologs of SPL3/4/5 and SPL9/15, as well as the AP1/FUL homolog 

MdMADS2.1 (Fig 2.5), which we had also found to increase in absolute expression over the 

season (Fig 2.4). Cluster 4 additionally included MdTFL1-1, which we had also found to show a 

strong decrease in absolute expression after 15 DAFB (Fig 2.4). This suggests that the presence 

of fruit promotes the seasonal decrease in expression of MdTFL1-1, as previously observed by 

Haberman et al. (2016). 

The final cluster, Cluster 5, contained a small group of genes (76) that showed greatly 

reduced expression in thinned apices, relative to non-thinned apices, at 15 DAFB and 70 DAFB 

(Fig 2.5). MdTFL1-2 was the sole flowering gene homolog included in this group. Like 

MdTFL1-1, MdTFL1-2 showed a decrease in expression throughout the season in flowering-

induced apices, and the observed differential expression pattern suggests that the presence of 

fruit counteracts this seasonal decrease. This was also previously observed by Haberman et al. 

(2016). 

MdTFL1-2 expression profiling and identification of co-expressed genes 

This expression pattern of MdTFL1-2 as reported previously by other groups, and here 

determined by RNA-seq, suggests this could be a key gene in regulating floral repression in the 

presence of fruit on the bourse shoot. We carried out qRT-PCR to quantify relative expression of 

both MdTFL1-1 and MdTFL1-2 to confirm the expression trend observed in the RNA-seq results. 

 

48 
 

The results were generally consistent between the two approaches (Fig 2.6). Haberman et al. 

(2016) previously reported that MdTFL1-2 increased in expression between ~30 and ~60 DAFB 

in fruit bearing spurs. Our observations are distinct from those of Haberman et al. (2016), as our 

results indicate that fruit promotes significant increased expression of MdTFL1-2 as early as 15 

DAFB. This early seasonal expression of MdTFL1-2 overlaps with the period of floral 

induction/initiation in apple [4-7], and argues for a direct role for MdTFL1-2 in repressing floral 

initiation, rather than a conceivable function in governing inflorescence architecture once 

initiation has occurred [54].  

MdTFL1-2 is expected to act in transcriptional regulation of flowering. Genes expressed 

similarly with MdTFL1-2 (i.e., more strongly in non-thinned apices) could represent upstream 

promoters of MdTFL1-2 expression or downstream positive targets. Considering only flowering 

gene homologs, in addition to the AGL24/SVP homolog MD15G1384600, these included 

homologs of AGL16, BLH8, EFM, GPRI1, LSN, and one homolog of RAP2.7 (Fig 2.5). 

Conversely, genes expressed in a reciprocal manner to MdTFL1-2 (i.e., less strongly in non-

thinned apices), could represent upstream repressors of MdTFL1-2 expression or downstream 

negative targets. This included homologs of AGL6, MdMADS2.1, RAP2.7, and various SPLs (Fig 

2.5). In Arabidopsis, AP1 represses TFL1 expression, and this finding is consistent with a 

conserved function of MdMADS2.1 in apple [67].  

We also searched the subset of genes assigned to Cluster 5 for other potential upstream 

promoters or downstream regulatory targets of MdTFL1-2 (Fig 2.5; S12 Table). Of the 75 other 

genes assigned to this cluster, four would encode transcriptional regulator-like proteins. These 

included MD05G1203300, a homolog of FERTILIZATION INDEPENDENT ENDOSPERM 

(FIE). FIE encodes a component of a POLYCOMB REPRESSOR COMPLEX 2 (PRC2) protein 

 

49 
 

that represses flowering and floral development in Arabidopsis [68]. In our study, this FIE 

homolog was expressed to higher levels in the non-thinned apices, relative to thinned, from 15 

DAFB thru 70 DAFB (S3 Fig).  

We employed a second, independent method to identify genes that could represent 

upstream promoters of MdTFL1-2 or downstream targets through the identification of co-

expression networks using the weighted gene correlation network analysis (WGCNA) approach 

[52]. This resolved 28 modules of co-expressed genes, with MdTFL1-2 assigned to a module 

containing 200 genes (Fig 2.5C; S13 Table). This module contained homologs of APL, EFM, 

SPL3, and ATH1 (S14 Table). We also identified a module with a strong negative correlation to 

MdTFL1-2’s module (Fig 2.5C; S15 Table). Here, we identified a total of 93 genes, including a 

distinct homolog of ATH1 (S15 Table).  

Expression profiles of GA2ox and GA20ox genes 

In our previous study of the mechanisms of the repression of flowering by GAs in apple, 

we found that MdTFL1-2 was rapidly (within 2 days) upregulated in the shoot apex in response 

to exogenous GA4+7 [54]. In that study, we also found that exogenous GA resulted in the rapid 

upregulation of four genes classified as GA2 OXIDASE (GA2ox). Interestingly, all of the four 

GA2ox genes were included in the set of 6,595 genes differentially expressed in response to fruit 

load. A heat map of expression of these and additional GA2ox genes identified by Zhang et al. 

(2019) is shown in Fig 2.5E. The four GA2ox genes identified as differentially expressed shared 

a general pattern of higher expression in the thinned, relative to non-thinned, apices very early in 

the season (2 DAFB). Interestingly, as the season progressed, these genes showed higher 

expression in the non-thinned apices. This is consistent with previous studies by Guitton et al. 

(2016) and Habermann et al. (2016) showing that two of these four genes, MD05G1207000 and 

 

50 
 

MD10G1194100, were expressed to higher levels in non-thinned samples at a similar sampling 

date (48 DAFB) as in our study. If cellular GA levels promote expression of these GA2ox genes, 

then the strong shift to higher expression in non-thinned apices could reflect increased GA levels 

in the apex, potentially driven by the presence of fruit. 

In the previous study [54] we also documented that exogenous GA resulted in rapid 

downregulation of several genes encoding GA20 OXIDASES (GA20ox), which participate in GA 

biosynthesis and are recognized to be subject to feedback repression in many contexts in various 

plants. Here, we observed that the GA20ox homolog MD01G1192100 was expressed to 

relatively higher levels in thinned apices at all time points (Fig 2.5B). Habermann et al. (2016) 

also reported higher expression of specific GA20ox homologs in thinned apices, including 

MD01G1192100. Thus, this observation might reflect lowered levels of bioactive GAs in thinned 

apices. 

Divergent expression patterns of homeologous gene pairs 

The differential regulation of the apple TFL1-1 and TFL1-2 genes is an interesting 

example of functional divergence of ancestrally related genes. Although expression of many of 

the key flowering gene homologs could not be reliably estimated, we identified several 

additional cases in which apparent gene duplication and/or gene family expansion was associated 

with distinctions in expression (Fig 2.7). For example, the AP1/FUL homeologs 

MD06G1204400 and MD14G1215700 showed distinct absolute expression patterns across the 

season, with MD06G1204400 increasing strikingly and MD14G1215700 remaining relatively 

constant. In contrast, the SPL4/5 gene MD03G1230600 exhibited a strong decrease in expression 

as the season progressed, while expression of the homeologous MD11G1251800 stayed 

 

51 
 

relatively constant. A third example was the FD homolog MD15G1230800, which was strongly 

increased at later time points, while its homeolog MD02G1125100 was not (Fig 2.7). 

Expression of related genes was also differentially influenced by fruit load in several cases. 

Besides TFL1-1/TFL1-2, MdAFL1 was more strongly expressed in the non-thinned apices at 

several time points, whereas MdAFL2 was more weakly expressed. The distinctions in 

expression of these homologs of the intensively studied flowering genes underscores the 

importance of thoroughly indexing the genomic content and rigorously establishing phylogenetic 

relationships. Future characterization of function of these gene pairs can provide novel insight 

into the conserved and/or divergent genetic mechanism(s) that underlie their role in flowering in 

apple. 

 

52 
 

Table 2.1. Sequencing and transcriptome assembly statistics. 
 

Measured Statistic 

Bases Sequenced 

Total Sequence Reads 

Value 

39,201,481,600 

392,014,816 

Mean Overall Read Mapping Rate 

90.10% 

Total Number of Transcript Models 

104,690 

Total Number of Genes 

Novel Exons 

Novel Introns 

Novel Loci 

Average Transcripts per Loci 

58,452 

23,034 

11,152 

8,753 

3.4 

Average Transcript Length (bases) 

1,617 

 

 

 

 

 

 

 

 

 
 

53 

Table 2.2 Identified homologs of Arabidopsis flowering genes in apple. 

 

Gene Family 

Reference Locus  Clade/Alias 

Citation 

AGL24 
SVP 

AP1 
CAL 
FUL 
AGL79 

FD 
BZIP27 

  

  

  

  

  

AP1 
CAL 

  

  

  

  

  

  

FT 
TSF 
AT5G62040 
ATC 
TFL1 

FT 
TSF 

TFL1 
ATC 

  

  

  

  

  

  

  

  

  

  

  

  

  

  

  

MD01G1038600 

 AGL24/SVP 

MD08G1197300 

 AGL24/SVP 

MD15G1313200 

 AGL24/SVP 

MD15G1384500 

 AGL24/SVP 

MD15G1384600 

 AGL24/SVP 

  

  

  

  

  

MD13G1059200 

  MdMADS5 

Kotoda et al. 2010 

MD16G1058500 

 AP1 

MD06G1204400 

 AP1/FUL fam 

  

  

MD14G1215700  MdMADS2.1 

Cevik et al. 2010 

MD17G1065500 

 AP1/FUL fam 

MD02G1125100 

FD 

MD15G1008300 

FD 

MD15G1240800 

FD 

  

  

  

  

MD12G1262000  MdFT1 

Kotoda et al. 2010 

FT-like de novo 

MdFT2 

ATC  MD03G1143000 

 ATC 

  

  

  

  

  

  

  

  

  

  

  

  

  

  

  

  

  

  

  

  

  

  

  

SPL4 
SPL5 

  

  

  

  

  

  

  

LFY 

SOC1 

SPL3 
SPL4 
SPL5 

SPL9 
SPL15 

 

MD11G1163500 

 ATC 

MD12G1023900  MdTFL1-1 

MD14G1021100  MdTFL1-2 

MD01G1198400 

FT/TFL1 fam 

  

  

Kotoda and Wada 2005; 
Hättasch et al 2008; 
Mimida et al. 2009 
  

MD07G1265900 

FT/TFL1 fam 

  

MD06G1129500  MdAFL1 

Wada et al. 2002 

MD14G1146700  MdAFL2 

MD02G1197400  MdSOC1a 

Kotoda et al. 2010 

MD07G1123600  MdSOC1b 

MD03G1230600 

 SPL4/5 

MD11G1251800 

 SPL4/5 

MD05G1312300 

 SPL3/4/5 

MD09G1244200 

 SPL3/4/5 

MD10G1291800 

 SPL3/4/5 

MD17G1236000 

 SPL3/4/5 

MD12G1060000 

 SPL9/15 

MD12G1060200 

 SPL9/15 

MD14G1060200 

 SPL9/15 

  

  

  

  

  

  

  

  

  

54 
 

 

 

Figure 2.1. Effect of flower removal on subsequent-year flowering. At full bloom, trees were 

thinned of flowers or were left non-thinned, and the fraction of flower-bearing shoots, relative to 

total shoots arising from spur structures, was evaluated the following spring. (A) Visual 

estimation of floral density. The rating scale extends from 0% (no obvious flowers) to 100% 

(abundant flowers). Visual density was estimated by two, independent, trained observers 

(correlation p value < 0.05). (B) Quantification based on sampling a minimum of 40 spur shoots 

designated for evaluation prior to flower removal. Graphs show the results from three biological 

replicates (R1-R3). 

 

 

 

55 
 

 

Figure 2.2. Characterization of the reference-mapped assembled 'Honeycrisp' 

transcriptome. (A) Proportions of the assigned organization of transcripts assembled from reads 

that mapped to the reference genome. (B) Upper pie chart illustrates the proportion of novel 

(non-isoform) transcripts that were expressed (FPKM > 1) and were either predicted to be 

protein coding or noncoding/ambiguous. Lower pie chart further characterizes the expressed and 

coding novel transcripts by their genomic organization. (C) The distribution of assigned 

BLASTn match results. 

 

 

 

56 
 

Figure 2.3. Analysis of genomic collinearity and phylogeny for 15 flowering gene families. 

The circle plot depicts the position of flowering genes in the apple genome and their collinear 

relationships (red links). The background grey links represent the complete genomic collinearity 

within the apple genome. 

 

 

 

 

57 
 

 

Figure 2.4. Expression patterns of flowering gene homologs in flowering-induced apices. A) 

Expression plots for each of the five clusters. The average expression pattern is represented by a 

black line. B) Heatmap and K-means clustering of expression values sorted by cluster 

assignment for the flowering gene homologs that exhibited significant changes in expression 

across the study period.  

 

 

 

58 
 

 

Figure 2.5. Heatmap of differential expression of genes between thinned and non-thinned 

apices across the study period. A) K-means clusters of all differentially expressed genes 

between the two conditions at each time point. B) Heatmap of the fold-difference in expression 

of flowering gene homologs at each sample date. Homologs of the intensively studied flowering 

genes are indicated in bold text. C) Dendrogram of co-expressed gene modules and a heatmap of 

the correlation between modules. D) Overlap among differentially expressed genes at each 

sample date. E) Heatmap of fold change in expression of GA2ox homologs.  

 

 

 

59 
 

Figure 2.6. Expression profiles of MdTFL1-1 and MdTFL1-2 calculated from qPCR (upper 

panels) and RNAseq (lower panels). The correlation between the relative expression and 

FPKM values for MdTFL1-1 and MdTFL1-2 had R2 values of 0.97 and 0.99, respectively. Triple 

asterisks (***) indicate a q value <0.001. NT = Non-thinned and T = Thinned 

 

 

 

 

60 
 

Figure 2.7. Heatmap of expression of homeologous gene pairs under various crop load 

conditions and the resulting fold-change between conditions across the study period. 

 

 

 

 

61 
 

REFERENCES 

 
 

62 

REFERENCES 

 

 

1.  Butler O. On the cause of alternate bearing in apple. Bulletin of the Torrey Botanical Club. 

1917;44(2): 85-96. 
 

2.  Jonkers H. Biennial bearing in apple and pear: a literature survey. Sci Hortic. 1979;11: 303-

317. 
 

3.  Monselise SP, Goldschmidt EE. Alternate bearing in fruit trees. In: Janick J, editor. 

Horticultural Reviews. Westport, CT: AVI Publishing Company Inc; 1982. pp.128-173. 
 

4.  McArtney S, Hoover E, Hirst P, Booking IR. Seasonal variation in the onset and duration of 

flower development in ‘Royal Gala’ apple buds. J Hort Sci Biotech. 2001;76(5): 536-540. 
 

5.  Foster T, Johnston R, Seleznyova A. A morphological and quantitative characterization of 

early floral development in Apple (Malus x domesitca Borhk.). Ann Bot. 2003;92: 199-206. 
 

6.  Hoover E, De Silva N, McArtney S, Hirst P. Bud development and floral morphogenesis in 

four apple cultivars. J Hort Sci Biotech. 2004;79(6): 981-984. 
 

7.  Dadpour MR, Movafeghi A, Grigorian W, Omidi Y. Determination of floral initiation in 

Malus domestica: A novel morphogenetic approach. Biol Plant. 2011;55(2): 243-252. 
 

8.  Dennis FG, Neilsen JC. Physiological factors affecting biennial bearing in tree fruit: the role 

of seeds in apple. HortTechnology. 1999;9(3): 317-322. 
 

9.  Luckwill LC, Weaver P, MacMillan J. Gibberellins and other growth hormones in apple 

seeds. J Hort Sci. 1969;44: 413–424. 
 

10. Wardlaw IF. The control of carbon partitioning in plants. New Phytol. 1990;116: 341-381. 
 
11. Wada N, Cao Q, Kotoda N, Soejima J, Masuda T. Apple has two orthologues of 

FLORICAULA/LEAFY involved in flowering. Plant Mol Bio. 2002;49: 567-577. 

 
12. Kotoda N, Wada M, Kusaba S, Kano-Murakami Y, Masuda T, Soejima J. Overexpression of 

MdMADS5, an APETALA1-like gene of apple, causes early flowering in transgenic 
Arabidopsis. Plant Sci. 2002;162(5): 679-687.  

 
13. Kotoda N, Wada M. MdTFL1, a TFL1-line gene of apple, retards the transition from the 

vegetative to reproductive phase in transgenic Arabidopsis. Plant Sci. 2005;168(1): 95-104. 

 
14. Hättasch C, Flachowsky H, Kapturska D, Hanke M. Isolation of flowering genes and 

seasonal changes in their transcript levels related to flower induction and initiation in apple 
(Malus domestica). Tree Physiol. 2008;28: 1459-1466. 

 

 
 

63 

15. Cevik V, Ryder CD, Popovich A, Manning K, King Gj, Seymour GB. A FRUITFUL-like 
gene is associated with genetic variation for fruit flesh firmness in apple (Malus domestica 
Borkh.). Tree Genet Genomes. 2010;6: 271-279. 

 
16. Kotoda N, Hayashi H, Suzuki M, Igarashi M, Hatsuyama Y, Kidou S, et al. Molecular 

characterization of FLOWERING LOCUS T-Like gene in apple (Malus × domestica 
Borkh.). Plant Cell Physiol. 2010;51(4): 561-575. 

 
17. Guitton B, Kelner JJ, Celton JM, Sabau X, Renou JP, Chagné D, et al. Analysis of transcripts 

differentially expressed between fruited and deflowered ‘Gala’ adult trees: a contribution to 
biennial bearing understanding in apple. BMC Plant Biol. 2016;16: 55. 

 
18. Haberman A, Ackerman M, Crane O, Kelner JJ, Costes E, Samach A. Different flowering 

response to various fruit loads in apple cultivars correlates with degree of transcript 
reaccumulation of a TFL1-encoding gene. Plant J. 2016;87: 161-173. 

 
19. Mimida N, Kotoda N, Ueda T, Igarashi M, Hatsuyama Y, Iwanami H, et al. Four 

TFL1/CEN-like genes on distinct linkage groups show different expression patterns to 
regulate vegetative and reproductive development in apple (Malus × domestica Borkh.). 
Plant Cell Physiol. 2009;50(2): 394-412. 

 
20. Kotoda N, Wada M, Komori S, Kidou S, Abe K, Masuda T, et al. Expression pattern of 

homologues of floral meristem identity Genes LFY and AP1 during flower development in 
apple. J Amer Soc Hort Sci. 2000;125(4): 398-403. 

 
21. Li WM, Tao Y, Yao YX, Hao YJ, You CX. Ectopic over-expression of two apple Flowering 
Locus T homologues, MdFT1 and MdFT2, reduces juvenile phase in Arabidopsis. Biol Plant. 
2010;54(4): 639-646. 

 
22. Tränkner C, Lehmann S, Hoenicka H, Hanke M, Fladung M, Lenhardt D, et al. Over-
expression of an FT-homologous gene of apple induces early flowering in annual and 
perennial plants. Planta. 2010;232(6): 1309-1324. 

 
23. Mimida N, Kidou S, Iwanami H, Moriya S, Abe K, Voogd C, et al. Apple FLOWERING 
LOCUS T proteins interact with transcription factors implicated in cell growth and organ 
development. Tree Physiol. 2011;31(5): 555-566. 

 
24. Flachowsky H, Szankowski I, Waidmann S, Peil A, Tranker C, Hanke MV. The MdTFL1 
gene of apple (Malus × domestica Borkh.) reduces vegetative growth and generation time. 
Tree Physiol. 2012;32: 1288-1301.  

 
25. Munoz-Fambuena N, Mesejo C, Gonzalez-Max MC, Primo-Millo E, Agusti M, Iglesias DJ. 

Fruit regulates seasonal expression of flowering genes in alternate-bearing 'Moncada' 
mandarin. Ann Bot. 2011;108: 511-519. 

 

 

64 
 

26. Ziv D, Zviran T, Zezak O, Samach A, Irihimovitch V. Expression profiling of FLOWERING 

LOCUS T-like gene in alternate bearing 'Hass' avocado trees suggests a role of PaFT in 
avocado flower induction. PloS One. 2014;9(10): e110613. 

 
27. Kittikorn M, Okawa K, Ohara H, Kotoda N, Wada M, Yokoyama M, et al. Effects of fruit 
load, shading, and 9,10-ketol-octadecadienoic acid (KODA) application on MdTFL1 and 
MdFT1 genes in apple buds. Plant Growth Regul. 2011;64: 75-81. 

 
28. Gasic K, Hernandez A, Korban SS. RNA extraction from different apple tissues rich in 

polyphenols and polysaccharides for cDNA library construction. Plant Mol Bio Rep. 
2004;22: 437a-437g. 

 
29. Aronesty E. ea-utils : "Command-line tools for processing biological sequencing data"; 

https://github.com/ExpressionAnalysis/ea-utils 2011. 

 
30. Daccord N, Celton JM, Linsmith G, Becker C, Choisne N, Schijlen E, et al. High-quality de-

novo assembly of the apple genome and methylome dynamics of early fruit development. 
Nat Genet. 2017;49: 1099-1106. 

 
31. Kim D, Langmead B, Salzberg SL. HISAT: a fast spliced aligner with low memory 

requirements. Nat Methods. 2015;12: 357-360. 

 
32. Pertea M, Pertea GM, Antonescu CM, Chang TC, Mendell JT, Salzberg SL. StringTie 
enables improved reconstruction of a transcriptome from RNA-seq reads. Nat Biotech. 
2015;33: 290-295. 

 
33. Pertea M, Kim D, Pertea GM, Leek JT, Salzberg SL. Transcript-level expression analysis of 

RNA-seq experiments with HISAT, StringTie and Ballgown. Nat Protoc. 2016;11: 1650-
1667. 

 
34. Trapnell C, Roberts A, Goff L, Pertea G, Kim D, Kelley DR, et al. Differential gene and 
transcript expression analysis of RNA-seq experiments with TopHat and Cufflinks. Nat 
Protoc. 2012;7: 562-578. 

 
35. Trapnell C, Williams B, Pertea G, Mortazavi A, Kwan G, van Baren J, et al. Transcript 
assembly and quantification by RNA-Seq reveals unannotated transcripts and isoform 
switching during cell differentiation. Nat Biotech. 2010;28(5): 511-515. 

 
36. Deluca DS, Levin JZ, Sivachenko A, Fennell T, Nazaire MD, Williams C, et al. RNA-SeQC: 
RNA-seq metrics for quality control and process optimization. Bioinformatics. 2012;28(11): 
1530-1532. 

 
37. Pertea G, Pertea M. GFF Utilities: GffRead and GffCompare. F1000 Research. 2020;9: 304. 
 

 

65 
 

38. Yu-Jian K, De-Chang Y, Lei K, Mei H, Yu-Qi M, Liping W, et al. CPC2: a fast and accurate 

coding potential calculator based on sequence intrinsic features. Nucleic Acids Res. 
2017;45(W1): W12–W16. 

 
39. Li A, Zhang J, Zhou Z. PLEK: a tool for predicting long non-coding RNAs and messenger 

RNAs based on an improved k-mer scheme. BMC bioinformatics. 2014;15: 311. 

 
40. Wang L, Park HJ, Dasari S, Wang S, Kocher JP, Li W. CPAT: Coding-potential assessment 
tool using an alignment-free logistic regression model. Nucleic Acids Res. 2013;41(6): e74. 

 
41. Haas BJ, Papanicolaou A, Yassour M, Grabherr M, Blood PD, Bowden J, et al. De novo 

transcript sequence reconstruction from RNA-seq using the Trinity platform for reference 
generation and analysis. Nat Protoc. 2013;8(8): 1494-1512. 

 
42. Berardini TZ, Reiser L, Li D, Mezheritsky Y, Muller R, Strait E, et al. The Arabidopsis 

Information Resource: Making and mining the “gold standard” annotated reference plant 
genome. Genesis. 2015;53: 474-485. 

 
43. Camacho C, Coulouris G, Avagyan V, Ma N, Papadopoulos J, Bealer K, et al. BLAST+: 

architecture and applications. BMC Bioinformatics 2008; 10:421. 

 
44. Geer LY, Marchler-Bauer A, Geer RC, Han L, He J, He S, et al. The NCBI BioSystems 

database. Nucleic Acids Res. 2010;38: D492-D496. 

 
45. Huerta-Cepas J, Serra F, Bork P. ETE 3: Reconstruction, analysis and visualization of 

phylogenomic data. Mol Biol Evol. 2016;33(6): 1635-1638.  

 
46. Sievers F, Wilm A, Dineen D, Gibson TJ, Karplus K, Li W, et al. Fast, scalable generation of 

high-quality protein multiple sequence alignments using Clustal Omega. Mol Syst Biol. 
2011;7: 539.  

 
47. Price MN, Dehal PS, Arkin AP. FastTree 2 - approximately maximum-likelihood trees for 

large alignments. PLoS One 2010;5(3): e9490. 

 
48. Wang Y, Tang H, DeBarry JD, Tan X, Li J, Wang X, et al. MCScanX: a toolkit for detection 

and evolutionary analysis of gene synteny and collinearity. Nucleic Acids Res. 2012;40(7): 
e49. 

 
49. Krzywinski MI, Schein JE, Birol I, Connors J, Gascoyne R, Horsman D, et al. Circos: An 

information aesthetic for comparative genomics. Genome Res. 2009;19: 1639-1645. 

 
50. R Core Team. R: a language and environment for statistical computing. R Foundation for 

Statistical Computing, Vienna, Austria. www. R-project.org. 2013. 

 
51. Goff L, Trapnell C, Kelley D. cummeRbund: Analysis, exploration, manipulation, and 

visualization of Cufflinks high-throughput sequencing data. 2013;R package version 2.22.0. 

 

66 
 

52. Langfelder P, Horvath S. WGCNA: an R package for weighted correlation network analysis. 

BMC Bioinformatics. 2008;9: 559. 

 
53. Zhang B, Horvath S. A General Framework for Weighted Gene Co-Expression Network 

Analysis. Stat Appl Genet Mol. 2005;4(1): 17. 

 
54. Zhang S, Gottschalk C, van Nocker S. Genetic mechanisms in the repression of flowering by 

gibberellins in apple (Malus × domestica Borkh.). BMC Genomics. 2019; 20:747. 

 
55. Embree CG, Myra MTD, Nichols DS, Wright AH. Effect of blossom density and crop load 
on growth, fruit quality, and return bloom in ‘Honeycrisp’ apple. HortScience. 2007;42(7): 
1622-1625. 

 
56. Velasco R, Zharkikh A, Affourtit J, Dhingra A, Cestaro A, Kalyanaraman A, et al. The 

genome of the domesticated apple (Malus × domestica Borkh.). Nat Genet. 2010;42(10): 
833–839.  

 
57. Peace CP, Bianco L, Troggio M, van de Weg E, Howard NP, Cornille A, et al. Apple whole 

genome sequences: recent advances and new prospects. Hortic Res. 2019;6: 59. 

 
58. Takeuchi T, Matsushita MC, Nishiyama S, Yamane H, Banno K, Tao R. RNA-sequencing 

analysis identifies genes associated with chilling-mediated endodormancy release in apple. J 
Am Soc Hort Sci. 2018;143(3): 194-206. 

 
59. Nishiyama S, Matsushita MC, Yamane H, Honda C, Okada K, Tamada Y, et al. Functional 

and expressional analysis of apple FLC-like in relation to dormancy progress and flower bud 
development. Tree Physiol. 2019;tpz111. 

 
60. Ng M, Yanofsky MF. Function and evolution of the plant MADS-box gene family. Nat Rev 

Genet. 2001;2: 186-195. 

 
61. Tian Y, Dong Q, Ji Z, Chi F, Cong P, Zhou Z. Genome-wide identification and analysis of 

the MADS-box gene family in apple. Gene. 2015;555(2): 277-290. 

 
62. Andrés F, Coupland G. The genetic basis of flowering responses to seasonal cues. Nat Rev 

Genet. 2012;13: 627-639. 

 
63. Xu M, Hu T, Zhao J, Park M, Earley KW, et al. Developmental functions of miR156-

regulated SQUAMOSA PROMOTER BINDING PROTEIN-LIKE (SPL) genes in 
Arabidopsis thaliana. PLoS Genet. 2016;12(8): e1006263. 

 
64. Liu C, Chen H, Er HL, Soo HM, Kumar PP, et al. Direct interaction of AGL24 and SOC1 

integrates flowering signals in Arabidopsis. Development. 2008;135: 1481-1491. 

 
65. Bradley D, Ratcliffe O, Vincent C, Carpenter R, Coen E. Inflorescence Commitment and 

Architecture in Arabidopsis. Science. 1997;275(5296): 80-83. 

 

67 
 

 
66. Ratcliffe OJ, Bradley DJ, Coen ES. Separation of the shoot and floral identity in Arabidopsis. 

Development. 1999;126:1109-1120. 

 
67. Goslin K, Zheng B, Serrano-Mislata A, Rae L, Ryan PT, Kwaśniewska K, et al. 

Transcription factor interplay between LEAFY and APETALA/CAULIFLOWER during 
floral initiation. Plant Physiol. 2017;174(2):1097-1109. 

 
68. Kinoshita T, Harada JJ, Goldberg RB, Fischer RL. Polycomb repression of flowering during 

early plant development. PNAS. 2001;98(24):14156-14161. 

 

 

 

68 
 

Molecular basis of diversity in floral initiation and development in Malus

CHAPTER THREE 

 
 

69 

ABSTRACT 

 

An important limitation to commercial apple production is freezing temperatures during 

the spring blooming period which can damage floral organs. Members of the Malus genus, 

including the domesticated apple, wild species, and hybrids all bloom in spring, and exhibit wide 

variation in spring bloom date. Although variation in seasonal bloom time is thought to be 

strongly influenced by genotypic differences in the chilling requirement, contributions of other 

factors have not been well studied. Here, we documented the seasonal timing of floral initiation 

and early floral development for three extreme early-blooming and three extreme late-blooming 

apple accessions through transcriptional profiling. The generated transcriptomic data also 

provided an opportunity to thoroughly identify homologs of floral initiation and developmental 

genes within Malus. The floral initiation genes were found to be well-conserved between 

accessions, whereas the floral development genes were less conserved. Transcript profiling 

identified shared differential regulation of the MdTFL1 genes between five of the six accessions 

in July, which is the anticipated peak timing of floral initiation. This result suggests MdTFL1 has 

a role in regulating the onset of floral initiation. In October, the expression of floral development 

homologs was relatively consistent across cultivars, except for SEP1/2 and SEP3. These two 

genes were highly expressed in one accession of M. angustifolia relative to the other accessions. 

This result correlates with M. angustifolia's delayed floral development, where the anatomical 

dissections indicated the accession had just entered the sepal initiation stage. Taken together, 

these results suggest that genotypic variation in floral initiation and early development differs 

among the members of the Malus genus. Our results also indicate that floral initiation timing, 

rate of development, and stage at which the apices enter dormancy has no correlation with the 

respective spring bloom time. 

 
 

70 

INTRODUCTION 

The domesticated apple (Malus × domestica Borkh.) is an important source of nutrients 

and energy throughout the temperate regions of the world. Similar to other perennial tree fruits, 

domesticated apple exhibits a two-season flowering cycle. Floral meristems are initiated and 

begin to develop during the first growing season, remain dormant during the winter, and 

complete development and bloom the following spring. Commercial apple cultivars and wild 

Malus species can exhibit a wide range of spring bloom times with some cases spanning greater 

than 20 days (Gottschalk and van Nocker, 2013). This natural variation in bloom time can have 

negative impacts on commercial apple production. For example, apple typically outcrosses, thus 

growers need to select cultivars that bloom synchronously to ensure adequate cross-pollination 

(Dennis, 2003; Ramírez and Davenport, 2013). In addition, cultivars that bloom late in the spring 

are prone to the devastating disease fire blight, as the bacterial pathogen agent, Erwinia 

amylovora enters the plant through open flowers in warm and humid conditions (Spotte et al., 

1976; Jones, 1992). Lastly, early-blooming apple cultivars are susceptible to freezing injury from 

‘spring frosts' that occur in temperate growing regions (Cannell and Smith, 1986; Rodrigo, 2000; 

Campoy et al., 2011). Spring freezing injury also impacts other tree fruit crops, and is a major 

determinant of production for Prunus species. Tree fruit crop loss due to spring freeze events is 

becoming of greater concern as climate change further exacerbates extreme and unpredictable 

weather patterns (Cannell and Smith 1986; Legave et al., 2007; Unterberger et al., 2018). These 

production considerations underscore the importance of developing improved cultivars with 

bloom times tailored to specific environments. 

In apple, bloom time is conditioned by both genotype and environment (Hauagge and 

Cummins, 1991; Lawson et al., 1995; Liebhard et al., 2003; Celton et al., 2011; Gottschalk and 

 

71 
 

van Nocker, 2013). Numerous quantitative trait loci (QTL) influencing bloom time have been 

identified in apple (Lawson et al., 1995; Liebhard et al., 2003; Celton et al., 2011). The identified 

loci were generally distinct between different studies, and none were found to have a major 

effect. In other Rosaceous fruit trees, including peach (Prunus persica (L.) Batsch), cherry 

(Prunus cerasus L.), and almond (Prunus dulcis (Mill.) D. A. Wade), large-effect QTLs for 

bloom time have been identified (Ballester et al., 2001; Fan et al., 2010; Sánchez-Pérez et al., 

2012; Castede et al., 2014; Cai et al., 2018). The high degree of synteny among Prunus species 

and the colocalization of major QTLs on linkage groups 1 and 4 suggest there may be common 

mechanisms controlling bloom time in these species (Dirlewanger et al., 2004; Fan et al., 2010).  

In domesticated apple, the inflorescence is initiated terminally on shoots early in the 

growing season. The meristem differentiates into a terminal and lateral floral meristems with 

subtending bracts and bractlets, respectively. Floral meristems continue to develop throughout 

the summer, and arrest in an incompletely developed state prior to winter dormancy. Following 

an extended period of chilling, warm temperatures in early spring promote final development 

(Foster et al., 2003). At the developmental level, the seasonal time of spring bloom could be 

conditioned by the timing of floral meristem initiation, rate of development prior to and during 

winter, amount of chilling required for exit from dormancy, and the rate of development in 

spring. The timing of floral initiation has been reported to be variable among M. x domestica 

cultivars (McArtney et al., 2001; Foster et al., 2003; Hoover et al., 2004; Dadoup et al., 2011). 

Wide natural variation also exists for chilling requirements (Hauagge and Cummins, 1991). 

Although chilling requirement obviously plays an important role in bloom time in apple, the 

extent to which other factors contribute to bloom time has not been addressed. Furthermore, 

 

72 
 

there is only limited information on natural variation in floral initiation and development among 

apple cultivars and Malus species. 

We previously documented natural diversity in spring bloom time among nearly 1,800 

distinct Malus accessions, comprising 31 wild species, and >1,000 domestic cultivars and 

hybrids, maintained at the USDA Agricultural Research Service Plant Genetic Resources Unit 

(Geneva, NY) (Gottschalk and van Nocker, 2013). This resulted in the identification of subsets 

of accessions with extreme early or extreme late bloom times. In the following study, we 

combined anatomical observations and transcriptional profiling studies for six representative, 

from the extreme-early or extreme-late bloom groups to investigate the potential relationship 

between bloom time, timing of floral initiation, and rate of development. Representative apices 

were collected at seasonal timings that were anticipated to represent periods of floral initiation 

and development. Two collection dates that correspond to the periods of floral initiation and 

development were also transcriptional profiled to identify transcriptional differences between the 

genotypes.  

METHODS 

Plant material and sample collection 

Shoot apices were randomly collected from six apple accessions the belong to extreme 

late and early bloom groups (Gottschalk and van Nocker 2013). The specimen trees were located 

at the USDA Agricultural Research Service Plant Genetic Resources Unit (USDA ARS PGRU) 

in Geneva, NY (Table 3.1). Apex collections occurred on July 25th, August 28th, October 5th, 

and November 8th, 2018, as well as on February 26th, 2019.  Bud scales, when present, were 

removed. Pools of 25-35 apices/buds were collected for RNA extraction and were immediately 

frozen in liquid nitrogen and stored at -80℃.  

 

73 
 

RNA extraction and sequencing 

Frozen apices were ground into a fine powder using a mortar and pestle under liquid 

nitrogen, and RNA was extracted from the frozen ground tissue using a CTAB-based extraction 

method (Gasic et al. 2004). Spermidine was substituted for spermine in the extraction buffer. 

Extracted RNA was then further purified using a commercial kit (RNeasy Plant; Qiagen). 

Purified RNA was assessed for quality using 1.2% formaldehyde gel electrophoresis, and 

quantified by spectroscopy (NanoDrop 2000c, ThermoScientific, Waltham, MA USA). RNA 

libraries were prepared and sequenced by Novogene (Sacramento, CA USA) on the Illumina 

HiSeq platform generating paired-end, 150-base reads. The sequencing generated a total of ~1.63 

billion reads, with an average of ~25.9 million reads per library. 

Transcriptome assembly and analysis 

Computational quality filtering of raw sequences was performed by Novogene and 

included removal of adapter sequence, reads containing Ns >10% of read length, and reads with 

Phred quality (Q)-scores ≤5. Filtered reads were then aligned to a draft Malus × domestica 

genome sequence generated from a doubled haploid individual (GDDH13v.1.1; Daccord et al., 

2017) using the HISAT2 (v.2.1.0) aligner with the -dta-cufflinks parameter (Kim et al., 2015). 

Transcript models were constructed using the alignment files and the transcriptome assembler 

StringTie (v.1.3.3) with default parameters (Pertea et al., 2015). The StringTie -G option was 

invoked to guide the construction based on externally supplied transcriptomes, which consisted 

of annotated transcripts from the GDDH13v.1.1 reference genome supplemented with additional 

novel loci identified in a previous study of flowering in apple (Gottschalk et al. Chapter 2). 

Transcriptome comparisons were conducted using the GFFcompare feature within the GFF 

Utilities package (v.0.11.2; Pertea and Pertea, 2020). Determination of a gene to be in an 

 

74 
 

expressed state or basally expressed was calculated using a hierarchical Bayesian mixture model 

using FPKM values retrieved from the output of Cuffnorm (v.2.2.1 Goff et al., 2019) and 

executed through the R (v.3.6.2; R Core Team 2013) program Zigzag (v.0.1.0; Thompson et al. 

2020). Genes with a calculated probability ≥0.8 were determined to be in an expressed state. 

Plots generated to display the intersections of expressed genes between accessions were 

generated using the R program UpSet (v.1.4.0; Lex et al., 2014). Differential expression analysis 

used the multifactor comparisons scheme in the R program edgeR (v.3.11; McCarthy et al., 

2012) with raw gene counts that were retrieved from denormalized Cuffnorm outputs. 

Differentially expressed genes (DEGs) were identified with FDR adjusted P-value <0.05 and a 

log fold change of >2 or <-2 for each comparison.  

Analysis of expression for landmark flowering genes 

Previously identified apple homologs of floral initiation genes were identified as 

described by Gottschalk et al. (Chapter 2). In brief, we indexed genes GO term annotations 

related to potential roles in flowering from Arabidopsis (TAIR10):GO terms 0048438 ('floral 

whorl development'), 0009908 ('flower development'), 0009910 ('negative regulation of flower 

development'), 0009911 ('positive regulation of flower development'), 0048578 ('positive 

regulation of long-day photoperiodism, flowering'), 0010220 ('positive regulation of 

vernalization response'), 0009909 ('regulation of flower development'), 0048510 ('regulation of 

timing of transition from vegetative to reproductive phase'), 0010321 ('regulation of vegetative 

phase change'), 0010228 ('vegetative to reproductive phase transition of meristem'), 0010048 

('vernalization response'), and 0010093 (‘specification of floral organ identity’). This data set 

contained 437 genes which were then manually curated to only those genes that have been 

intensively studied for roles in the flowering pathway. This subset included AG, AGL24, AP1-

 

75 
 

AP3, FD, FLC, FT, FUL,  LFY, LSN, PI, SEP1-SEP4, SOC1, SPL3-SPL5, SPL9, SPL15, SUP, 

SVP, TFL1, TSF, and UFO. Conceptual translations of the corresponding representative gene 

models were obtained from TAIR 

(TAIR10_pep_20110103_representative_gene_model_updated) and used as queries to search 

genotype-specific protein databases for each apple genotype in this study, using BLASTp and an 

expect (E) value cutoff of 1e-12 (Camacho et al., 2008).  

For each of the six genotypes, protein databases were generated using the following 

procedure. A consensus set of transcript models were collected from a combined transcriptome 

representing the July and October datasets, as well as a single library of pooled RNA from June, 

August, and November for each genotype separately. The resulting transcript model annotations 

were then merged with the GDDH13 reference gene model annotation 

(gene_models_20170606.gtf) using StringTie -merge option. Nucleotide FASTA sequences were 

then extracted from the merged transcriptome using the GFF utilities package (Pertea and Pertea, 

2020). For transcript models corresponding to annotated transcripts in GDDH13, a single 

corresponding GDDH13 protein model was adopted. For transcript models assembled de novo 

by StringTie that were not organized within GDDH13 reference loci had all their corresponding 

transcript sequences extracted and retained for analysis. The extracted nucleotide sequences were 

then processed by TranDecoder (v.5.5.0; https://github.com/TransDecoder) to construct open 

reading frame translations. TransDecoder often predicts multiple open reading frame products 

from a single transcript model, which resulted in redundancy for non-reference de novo transcript 

models. In this case, all translations were retained for each genotype-specific protein sequence 

dataset.  

 

76 
 

For each genotype-specific protein dataset, individual peptide sequences were used as a 

query to search against all Arabidopsis representative gene model translations using Blastp. The 

top 50 returned blast hits were then used as queries to blast back into the apple peptide sequences 

(i.e. reciprocal BLAST). The 50 highest-scoring apple transcript translations and the 25 highest-

scoring Arabidopsis gene translations identified with each apple sequence query were then used 

for phylogenetic tree construction. Phylogenetic trees were constructed using the ETE3 toolkit 

(v.3.1.1) build function invoking the standard_fasttree workflow under default settings (Huerta-

Cepas et al., 2016; Sievers et al., 2011; Price et al., 2010). Representative flowering homologs 

were then identified through manual curation of each phylogenetic tree.  

RESULTS 

Sequencing and transcriptome assembly results 

To determine if potential distinctions in seasonal floral initiation and development exist 

and may influence spring bloom, we carried out transcriptional profiling of the developing shoot 

apices focusing on the expression of flowering-related genes. Apices collected on July 25th, and 

Oct 5th, representing anticipated stages of early and late floral developmental, were subjected to 

RNA extraction and sequencing. For each accession per time point, between 5.7 and 10 high-

quality gigabases of sequence were generated. Sequence reads were then aligned to a draft Malus 

genome sequence generated from a doubled-haploid clone of 'Golden Delicious' (GDDH13; 

Daccord et al., 2017), and transcripts were modeled based on established computational 

approaches. As anticipated, given that the reference sequence GDDH13 was derived from M. × 

domestica, sequences generated from the M. × domestica accessions 'Anna' and 'Koningszuur', 

and M. sylvestris exhibited the highest alignment rates (>90%). The remaining wild species 

exhibited lower alignment rates, with M. angustifolia having the lowest (~74%). The similarly 

 

77 
 

high alignment rates of the M. × domestica cultivars and M. sylvestris are indicative of the 

domestication history of M. × domestica, of which M. sylvestris is a progenitor (Velasco et al. 

2010). The lower alignment rates of the wild species reflects their evolutionary distance from M. 

× domestica (Nikiforova et al., 2013).  

In a previous study, we defined a clear subset of genes associated with floral initiation 

(Gottschalk et al., Chapter 2). The previous census of flowering genes was incomplete as it did 

not catalog genes involved in floral development. To identify genes with possible conserved 

function in the development of floral meristems and floral organ primordia, we selected a subset 

of well-studied floral development genes in Arabidopsis and identified their homologs in these 

Malus transcriptomes. We identified apple genes related to AG, SHP1/2, AP2, AP3, LSN, 

PI/TOE3, SEP1/2, SEP3, SUP, and UFO that had significant blast homology and belonged to 

well-defined (>90% bootstrapped) clades for each accession. Seven of the nine families of floral 

development genes were found to contain pairs of apple homologs (Table 3.2). This result was 

anticipated due to the paleopolyploid origins of apple (Velasco et al., 2010). We identified the 

families of AP3/PI and LSN to have undergone an expansion event, with each family containing 

three homologous copies (Table 3.2). Of note, we identified a novel copy of an FD homolog 

unique to M. orthocarpa (Table 3.2). Additionally, SUP homologs were absent from most 

accessions except M. angustifolia PI589789 (two copies), and ‘Koningszuur’ (one copy) (Table 

3.2). Both M. angustifolia accessions lacked homologs of AG (Table 3.2). Generally, the floral 

initiation genes were more conserved across the transcriptomes of the accessions than floral 

development genes.  

 

 

 

78 
 

Determining floral genes expressed state 

To identify molecular pathways associated with floral initiation and development we 

tracked the expression for a set of genes that we previously identified as indices for floral 

initiation in M.× domestica cultivar 'Honeycrisp' (Gottschalk et al., Chapter 2) and the genes we 

identified here with potential roles in flower development. In total, we identified 93 flowering 

genes of interest. To infer the probability of their active expression, we utilized a Bayesian 

statistical model to compare gene expression across each of the six assembled transcriptomes 

independently. With this information, we then assign an expression state classification to all 93 

flowering-associated homologs per accession. Any gene with a probability for expression of ≥0.8 

was deemed as being in an “expressed state” as opposed to a “basal expression”. Of the 93 genes 

surveyed, ten were found to be considered as basally expressed in all six accessions. These genes 

included homologs of AG, AGL12, AP3/PI, SEP1/2, SEP3, SHP1/2, SUP, UFO, and the novel 

FD transcript identified in M. orthocarpa. We also identified 19 genes that were deemed as 

expressed in one or more accessions but not in all six accessions (Table 3.3). In this subset, 

homologs of known roles in the promotion or repression of flowering in apple were identified. 

These homologs included the floral promoters MdAFL1, MdMADS2.1, and MdMADS5 and the 

floral repressors MdTFL1-1 and MdTFL1-2. Of the expressed genes, a total of 67 were shared 

across all the accessions (Fig 3.1). Also none of the expressed genes were uniquely associated 

with either bloom time group or were associated with accessions of the same species (M. 

angustifolias and M. × domestica accessions). M. orthocarpa and the two M. angustifolias were 

found to have a six uniquely expressed genes. For M. orthocarpa, the uniquely expressed genes 

were homologs of AG, AP1, SEP1/2, and UFO. For the two M. angustifolias, the uniquely 

expressed genes were homologs of SUP (PI 589763) and MdTFL1-1 (PI 613880). 

 

79 
 

Expression patterns of flowering homologs in July 

With the subset of expressed flowering genes identified, we then evaluated their 

expression levels across accessions during the July sample date (Fig 3.2). Hierarchical clustering 

was performed on the six accessions, resulting in two clades that contained both M. angustifolia 

accessions and a clade containing the other four accessions. The two M. x domestica accessions 

also clustered together within the larger of the two clades. Differential gene expression was then 

calculated for comparisons between each accession at the July sample date. The two M. 

angustifolia accessions generally exhibited a similar expression profiles, but noticeable 

expression differences were observed with two homologs of AGL24/SVP, a homolog of SHP1/2, 

and MdTFL1-1. These four genes were also found to meet our threshold to be classified as 

differentially expressed (DE) between the two genotypes. The two AGL24/SVP homologs were 

more highly expressed in M. angustifolia PI 589763, whereas the SHP1/2 homolog and 

MdTFL1-1 were more highly expressed in M. angustifolia PI 613880. The cluster analysis 

identified a separation of M. orthocarpa from a clade containing M. sylvestris and the two M. x 

domestica cultivars. The homologs that differentiated M. orthocarpa were AGL24/SVP, FUL, 

LSNs, MdAFL1, MdFT1, MdMADS2.1, MdTFL1s, RAP2.7, and SPL4/5. However, the only 

differentially expressed homolog shared between M. orthocarpa, M. sylvestris, and two M. x 

domestica accessions was MdMADS5. MdMADS5 was expressed more highly in the M. 

orthocarpa relative to the other three accessions. However, MdMADS5 was predicted to not be in 

an “expressed state” in M. orthocarpa, M. sylvestris, and two M. x domestica accessions (Table 

3.3). In the subclade containing M. sylvestris and M. x domestica accessions, the expression 

profiles were strikingly similar. Although, COL2 and MdTFL1-2 exhibited distinct expression 

profiles among the accessions and were found to be DE. In this comparison, COL2 was more 

 

80 
 

highly expressed in the two M. x domestica accessions and MdTFL1-2 was more highly 

expressed in M. sylvestris. The only DE homologs between the two M. x domestica accessions 

were ANP1 and MdTFL1-2. ANP1 was more highly expressed in ‘Anna’ and MdTFL1-2 was 

more highly expressed by ‘Koningszuur’.  

Expression patterns of flowering homologs in October 

It was anticipated that by October most reproductive apices had initiated flowers and 

were then undergoing differentiation. Transcriptional profiling was conducted from this sample 

date using the same core set of flowering-related genes. Hierarchical clustering applied to the 

expression profiles for the accession resulting in the formation of two distinct clades (Fig 3.3). 

The first clade contained M. orthocarpa, M. sylvestris, and ‘Koningszuur’, and the second clade 

contained the two M. angustifolia accessions and ‘Anna’. Differential gene expression was then 

calculated between each accession. Within the first clade, there were no DE homologs that could 

explained the separation of M. orthocarpa, M. sylvestris, and ‘Koningszuur’. In the second clade, 

both homologs AGs, a homologs of AP3/PI, and SUP were found to be DE between ‘Anna’ and 

the two M. angustifolia accessions. However, both of the AGs homologs were previously 

identified as being “basally expressed” in all three accessions. The classification of the AG loci 

as being DE is most likely an artifact of extreme low read coverage at those loci. In this 

comparison, AP3/PI homolog was more highly expressed in ‘Anna’ relative to the M. 

angustifolia accessions. The SUP homolog was more highly expressed in the two M. angustifolia 

accessions relative to Anna. Between the two M. angustifolia accessions, homologs of AGL16, 

SHP1/2, both UFOs, and MdTFL1-1 were found to be DE. Here, expression of the AGL16 

homolog and MdTFL1-1 was higher in M. angustifolia PI 589763. The homologs of SHP1/2 and 

two UFOs were more highly expressed in M. angustifolia PI 613880. Additionally, when 

 

81 
 

comparing the two three-member clades against one another no DE homologs were identified. 

Although, when interpreting the expression values for the SEP1/2 and SEP3 homologs a 

distinction between the two clades is observed (Fig 3.3). When inspecting the DE results, we 

found DE in these two genes were associated with only M. angustifolia PI 613880 compared to 

the three members of the other clade (M. orthocarpa, M. sylvestris, and ‘Koningszuur’).  

DISCUSSION 

The transcriptomes that we assembled suggested a well-conserved exome across the 

genus, even considering the evolutionary distance of some of the species from Malus × 

domestica. In particular, M. angustifolia is the most distant relative to M. x domestica included in 

our analysis and it yielded read alignments of ~74% (Nikiforova et al., 2013). In comparison, the 

more related M. sylvestris exhibited an alignment rate (>90%) similar to those observed with the 

M. x domestica accessions (Velasco et al. 2010; Nikiforova et al., 2013). Without individual 

genome sequences available for each species, we cannot definitively measure expression as a 

considerable percentage of reads fail to map to the reference genome. This limitation can be 

attributed to potential sequence divergence of these wild species and potential variations in gene 

copy number.  

The phylogenetic analysis of flowering genes among these accessions identified many 

conserved homologs of floral initiation associated genes. In contrast, the floral development 

genes were less conserved. For example, homologs of AG, SHP1/2, SEP3, and SUP were 

inconsistently identified between the accessions. Generally, the two M. x domestica accessions 

studied contained homologs of these genes whereas, they were absent in the wild species. This 

suggests that there is a divergence in gene content or appreciable sequence variation of these 

floral development genes within the Malus genus. We also identified an expansion of the AP3/PI 

 

82 
 

and LSN gene families from the anticipated two copy organization to three paralogs. The 

expansion of these families could result in neo- or subfunctionalization.  

The expression profiling of the floral initiation and developmental genes illustrates a 

relatively well-conserved flowering pathway between the accessions. The major difference at the 

July sample date was the DE of the MdTFL1s, which were identified as being DE between the 

M. angustifolia accessions, M. sylvestris, and the two M. x domestica cultivars. MdTFL1s have 

been previously reported to be associated with and function as floral repressors in apple (Kotoda 

and Wada, 2005; Hättasch et al., 200; Mimida et al., 2009; Guitton et al., 2016; Haberman et al., 

2016; Gottschalk et al., Chapter 2). The finding that MdTFL1s are DE between these accessions 

suggests that MdTFL1 could have a role in the determining the onset of floral initiation. The 

MdTFL1s generally exhibited lower expression at the October sample date, which is consistent 

with previous finding that MdTFL1 expression decreases throughout the period of floral 

initiation (Gottschalk et al. Chapter 2). We also found no consistent DE homologs between the 

accessions at the October sampling date. This suggests a diverse transcriptome during the floral 

developmental stage. However, the most striking difference was the DE of the SEP1/2 and SEP3 

homologs associated with M. angustifolia PI 613880 compared to M. orthocarpa, M. sylvestris, 

and ‘Koningszuur’. These genes are required for the development of the B and C floral organ 

whorls (Pelaz et al., 2000). The finding that SEP1/2 and SEP3 are DE at this time point suggests 

that M. angustifolia PI 613880 is developing its B and C whorls, whereas the other accessions 

have either already developed their whorls or have not yet reached that stage of development 

(data not shown). 

The results obtained from this study also indicate that the timing of floral initiation and 

pace of floral development do not influence spring bloom times. The transcriptomes analyzed 

 

83 
 

exhibited gene expression profiles that were strikingly similar both in and between spring bloom 

time groups. Furthermore, none of the flowering genes were differentially expressed and 

associated with a respective bloom group. The differences in expression that were identified only 

support the previously identified variation of the flowering processes in M. x domestica 

(McArtney et al., 2001; Foster et al., 2003; Hoover et al., 2004; Dadoup et al., 2011). However, 

this results indicate the variation extends beyond M. x domestica and exists within the genus at-

large. Ultimately, the findings we present here support a conclusion that the flowering processes 

in apple are not a determining factor nor influential in spring bloom time difference.  

 

 

 

84 
 

Table 3.1. Tissue collections for anatomical (D) and transcriptome (T) analyses from 

selected early and late-blooming Malus species from the USDA ARS PGRU Malus 

germplasm collection. 

 
Cultivar 
name 

Collection Dates and Purpose of Tissue 

 
PI 
number  7/25/18  8/28/18  10/5/18  11/8/18  2/26/19 

 

 

 
  

early 

Bloom Group 
Extreme 
bloom 
M. orthocarpa 
M. sylvestris 
M. x domestica   Anna 
Extreme 
bloom 
 
M. angustifolia 
589763 
M. angustifolia 
613880 
M. x domestica   Koningszuur  188517 

 
589392 
633824 
280400 

late 

 
 
 

 
 
 

 

A, T 
A, T 
A, T 

 

A, T 
A, T 
A, T 

 
A 
A 
A 

 
A 
A 
A 

 

A, T 
A, T 
A, T 

 

A, T 
A, T 
A, T 

 
A 
A 
A 

 
A 
A 
A 

 
A 
A 
A 

 
A 
A 
A 

 

85 
 

Table 3.2. Identified homologs of Arabidopsis floral initiation and development genes in 

apple. “+” indicates the presence of the homolog in the respective transcriptome. NA indicates 

the absence of the homolog in the respective transcriptome. 

   

 

Presence in Transcriptome 

M. 

M. 

angustifolia  

angustifolia 

M. 

M. 

589763 

613880  Anna  Koningszuur 

orthocarpa 

slyvestris 

MD15G1384600  AGL24/SVP 

NA 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Family 

AGL24 
SVP 

AP1 
CAL 
FUL 
AGL79 

FD 
BZIP27 

FT 
TSF 

TFL1 
ATC 

LFY 

SOC1 

 

 

 

 

 

 

AP1 
CAL 

 

 

 

 

 

 

 

 

 

 

 

 

Locus 

Clade 
Assignment 

MD01G1038600  AGL24/SVP 

MD08G1197300  AGL24/SVP 

MD15G1313200  AGL24/SVP 

MD15G1384500  AGL24/SVP 

MD13G1059200  MdMADS5 

MD16G1058500  AP1 

MD06G1204400  FUL 

MD14G1215700  MdMADS2.1 

MD02G1125100  FD 

MD15G1008300  FD 

MD15G1240800  FD 

MSTRG.33471.1  FD 

  MD12G1262000  MdFT1 
 

MD12G1023900    MdTFL1-1 

MD14G1021100    MdTFL1-2 

MD06G1129500  MdAFL1 

MD14G1146700  MdAFL2 

MD02G1197400  MdSOC1a 

MD07G1123600  MdSOC1b 

 

 

 

 

 

 

SPL3   
 

MD05G1312300  SPL3 

MD10G1291800  SPL3 

MD03G1230600  SPL4/5 

MD11G1251800  SPL4/5 

MD09G1244200  SPL4/5 

MD17G1236000  SPL4/5 

MD12G1060000  SPL9/15 

MD12G1060200  SPL9/15 

 

 

 

 

 

 

SPL3 
SPL4 
SPL5 

SPL4 
SPL5 

SPL9 
SPL15 

 

 

 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

NA 

NA 

NA 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

NA 

NA 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

NA 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

NA 

NA 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

NA 

+ 

NA 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

NA 

NA 

+ 

+ 

+ 

+ 

+ 

+ 

86 
 

Table 3.2. (cont’d) 
 

 

 

 

 

 

 

 

 

 

 

 

 

 

AG 

SHP1 
SHP2 

 

 

 

 

 

 

 

SEP1 
SEP2 

SEP3 

 

 

AG 
SHP1 
SHP2 

AP2 
TOE3 

AP3 
PI 

LSN 

SEP1 
SEP2 
SEP3 
SEP4 

SUP 
ZPF11 

UFO 

 

MD14G1060200 

SPL9/15 

MD05G1293700 

MD10G1271000 

  AG 
  AG 

MD09G1155200 

SHP1/2 

MD17G1141300 

SHP1/3 

MD02G1176000 

AP2/TOE3 

MD15G1286400 

AP2/TOE3 

MD02G1136500 

MD08G1021300 

MD15G1250200 

MD04G1138100 

MD07G1205600 

MD12G1153600 

 
  MD09G1073900 
  MD17G1065400 

MD13G1121500 

MD16G1121800 

MD01G1139000 

MD07G1203300 

MD05G1246800 

MD10G1227400 

 

 

 

 

 

 

 

AP3/PI 

AP3/PI 

AP3/PI 

LSN 

LSN 

LSN 

SEP1/2 

SEP1/2 

SEP3 

SEP3 

SUP 

SUP 

UFO 

UFO 

+ 

NA 

NA 

NA 

+ 

+ 

+ 

+ 

+ 

+ 

NA 

NA 

NA 

NA 

+ 

+ 

+ 

+ 

NA 

NA 

+ 

+ 

+ 

+ 

+ 

+ 

NA 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

NA 

+ 

NA 

NA 

NA 

+ 

NA 

+ 

+ 

NA 

+ 

+ 

+ 

+ 

+ 

+ 

NA 

+ 

+ 

NA  NA 

NA 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

NA 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

+ 

NA  NA 

NA 

+ 

+ 

+ 

+ 

+ 

NA 

NA 

NA 

+ 

NA 

NA 

+ 

+ 

 

87 
 

 
Table 3.3. Uniquely expressed states of flowering homologs. NE – predicted to be expressed 

at a basal level or not expressed. 

Expression Probability 

Gene ID 

Flowering 
Homolog 

M. 

M. 

M. 

M. 

angustifolia 

angustifolia 

sylvestris 

orthocarpa  Koningszuur  Anna 

589763 

613880 

MD10G1271000 

AG 

NE 

Expressed 

NE 

NE 

MD08G1206400 

AGL16 

Expressed  Expressed  Expressed  Expressed 

MD16G1058500 

AP1 

NE 

Expressed 

NE 

NE 

MD15G1008300 

FD 

Expressed  Expressed  Expressed  Expressed 

NE 

NE 

NE 

NE 

NE 

NE 

NE 

Expressed 

MD01G1192100  GA20OX2  Expressed  Expressed  Expressed 

NE 

Expressed 

Expressed 

MD06G1129500  MdAFL1 

NE 

Expressed  Expressed  Expressed  Expressed 

Expressed 

MD14G1215700  MdMADS2.1  Expressed  Expressed  Expressed  Expressed  Expressed 

NE 

NE 

Expressed 

NE 

Expressed 

Expressed 

NE 

NE 

NE 

NE 

NE 

MD13G1059200  MdMADS5 

MD12G1023900  MdTFL1-1 

NE 

NE 

MD14G1021100  MdTFL1-2  Expressed 

Expressed 

NE 

NE 

NE 

NE 

Expressed 

MD09G1073900 

SEP1/2 

NE 

Expressed 

NE 

MD13G1121500 

SEP3 

Expressed  Expressed  Expressed 

NE 

NE 

NE 

NE 

NE 

MD10G1067700 

SHL1 

Expressed  Expressed  Expressed  Expressed  Expressed 

Expressed 

MD09G1155200 

SHP1/2 

Expressed  Expressed  Expressed  Expressed 

NE 

Expressed 

MD13G1120300 

SPL2 

Expressed  Expressed  Expressed  Expressed  Expressed 

Expressed 

MD05G1312300 

SPL3 

Expressed 

MD01G1139000 

SUP 

NE 

NE 

NE 

Expressed  Expressed 

NE 

NE 

NE 

Expressed 

NE 

NE 

MD14G1204700 

TOC1 

Expressed  Expressed  Expressed  Expressed  Expressed 

Expressed 

MD10G1227400 

UFO 

NE 

Expressed 

NE 

NE 

NE 

NE 

 

 

 

88 
 

 

Figure 3.1. Intersections of expressed flowering genes in the transcriptomes of each 

accession. A) Total number of flowering genes that are expressed in each accession's 

transcriptome. B) The intersection of the number of expressed flowering genes between 

accessions. Each column of the histogram represents the number of genes that correspond with 

the intersection displayed as dots connected by lines below the X-axis.  

 

 

 

89 
 

 

Figure 3.2. Heatmap of the expression of flowering genes in July. Homolog symbol as 

displayed on the left side of the heatmap and gene names on the right. Gene names that represent 

a locus containing more than one gene are identified by the two potential integers listed within 

brackets. General expression trends across all genes were utilized to perform hierarchical clusters 

of similarity between accessions.  

 

 

 

90 
 

 

Figure 3.3. Heatmap of the expression of flowering genes in October. Homolog symbol as 

displayed on the left side of the heatmap and gene names on the right. Gene names that represent 

a locus containing more than one gene are identified by the two potential integers listed within 

brackets. General expression trends across all genes were utilized to perform hierarchical clusters 

of similarity between accessions. 

 

91 
 

REFERENCES

 

92 
 

REFERENCES 

 

 
Andres F and Coupland G. 2012. The genetic basis of flowering in response to seasonal cues. 
Nat Rev Genet 13(9): 627-639. 
 
Ballester J, Socias I Company R, Arus P, and De Vicente MC. (2001). Genetic mapping of a 

major gene delaying blooming time in almond. Plant Breeding, 120(3): 268–270. 

 
Blažek J, Paprštein F, and Křelinová J. (2015). Heritability of flowering time within apple 

progeny. Czech Journal of Genetics and Plant Breeding, 51(1), 23–28. 

 
Buban T and Faust M. (1982). Flower bud induction in apple trees: internal control and 

differentiation. Horticultural Reviews, 4(6), 174–203.  

 
Cai L, Stegmeir T, Sebolt A, Zheng C, Bink MCAM, and Iezzoni A. (2018). Identification of 
bloom date QTLs and haplotype analysis in tetraploid sour cherry (Prunus cerasus). Tree 
Genetics and Genomes, 14(2): 1-11. 

 
Camacho C, Coulouris G, Avagyan V, Ma N, Papadopoulos J, Bealer K, et al. BLAST+: 

architecture and applications. BMC Bioinformatics 2008; 10:421. 

 
Campoy J, Ruiz D, and Ega J. 2011. Dormancy in temperate fruit trees in a global warming 

context: A review. Scientia Horticulturae 130(2): 357-372. 

 
Cannell MGR and Smith RI. 1986. Climatic warming, spring budburst and frost damage on trees. 

J Applied Ecology 23:177-191. 

 
Castede S, Campoy A, Garcia J, Dantec L, Lafargue M, et al. (2014). Genetic determinism of 

phenological traits highly affected by climate change in Prunus avium: flowering date 
dissected into chilling and heat requirements. New Phytologist, 202: 703–715. 

 
Celton JM, Martinez S, Jammes MJ, Bechti A, Salvi S, et al. 2011. Deciphering the genetic 

determinism of bud phenology in apple progenies: a new insight into chilling and heat 
requirement effects on flowering dates and positional candidate genes. New Phytologist 
192:378-392. 

 
Daccord N, Celton J, Linsmith G, Becker G, Choisne N, et al. 2017. High-quality de novo 

assembly of the apple genome and methylome dynamics of early fruit development. Nature 
Genetics 49:1099-1106. 

 
Dennis F. 2003 Flowering, pollination and fruit set and development. Apples: botany, production 

and uses. Ed. D.C. Ferree and I.J. Warrington. Cambridge: CABI Publishing: 153-166. 

 

 

93 
 

Dirlewanger E, Graziano E, Joobeur T, Garriga-Caldere F, Cosson P, et al. 2004. Comparative 
mapping and marker-assisted selection in Rosaceae fruit crops. Proceedings of the National 
Academy of Sciences, 101(26): 9891–9896. 

 
Doukhanina EV, Chen S, van der Zalm E, Godzik A, Reed J, and Dickman MB. 2006. 

Identification and Functional Characterization of the BAG Protein Family in Arabidopsis 
thaliana. J Biological Chem 281(27): 18793-18801. 

 
Falavigna VS, Porto DD, Buffon V, Margis-Pinheiro M, Pasquali G, and Revers LF. 2014. 

Differential Transcriptional Profiles of Dormancy-Related Genes in Apple Buds. Plant Mol 
Biol Rep 32:796-813. 

 
Fan S, Bielenberg DG, Zhebentyayeva TN, Reighard GL, Okie WR, et al. 2010. Mapping 

quantitative trait loci associated with chilling requirement, heat requirement and bloom date 
in peach (Prunus persica). New Phytologist, 185: 917–930. 

 
Foster T, Johnston R, and Seleznyova A. 2003. A Morphological and Quantitative 

Characterization of Early Floral Development in Apple (Malus × domestica Borkh.). Annals 
of Botany 92:199-206. 

 
Gasic K, Hernandez A, and Korban SS. 2004. RNA Extraction From Different Apple Tissues 
Rich in Polyphenols and Polysaccharides for cDNA Library Construction. Plant Mol Biol 
Rep 22: 437a-437g. 

 
Goff L, Trapnell C, and Kelley D. 2019. cummeRbund: Analysis, exploration, manipulation, and 

visualization of Cufflinks high-throughput sequencing data. R package version 2.24.0. 

 
Gómez-Mena C, Piñeiro M, Franco-Zorrilla JM, Salinas J, Couland G, and Martínez-Zapater JM. 
2001. Early bolting in short days: An Arabidopsis Mutation That Causes Early Flowering and 
Partially Suppresses the Floral Phenotype of LEAFY. Plant Cell 13: 1011-1024.  

 
Gottschalk C and van Nocker S. 2013. Diversity in Seasonal Bloom Time and Floral 

Development among Apple Species and Hybrids. J Amer Soc Hort Sci 138(5):367-374. 

 
Gottschalk C, Zhang S, Schwallier P, Rogers S, Bukovac MJ, and van Nocker S. Genetic 

mechanisms associated with the repressive effect of fruit on floral initiation in apple (Malus 
× domestica Borkh). Submitted 

 
Gou D, Zhang Z, Wang X, Han X, Wei B, et al. 2015. The WRKY Transcription Factor 

WRKY71/EXB1 Controls Shoot Branching by Transcriptionally Regulating RAX Genes in 
Arabidopsis. Plant Cell 27: 3112-3127. 

 
Grainger J. 1939. Studies upon the time of flowering of plants: anatomical, floristic and 

phenological aspects of the problem. Annals of Applied Biology 26, 684–704. 

 

 

94 
 

Guitton B, Kelner JJ, Celton JM, Sabau X, Renou JP, Chagné D, et al. Analysis of transcripts 

differentially expressed between fruited and deflowered ‘Gala’ adult trees: a contribution to 
biennial bearing understanding in apple. BMC Plant Biol. 2016;16: 55. 

 
Hauagge R and Cummins JN. 1991. Seasonal Variation in Intensity of Bud Dormancy in Apple 

Cultivars and Related Malus Species. J Amer Soc Hort Sci 116(1): 107-115. 

 
Haas B. TransDecoder (find coding regions within transcripts). http://github.com/TransDecoder. 

Accessed September 22, 2020. 

 
Haberman A, Ackerman M, Crane O, Kelner JJ, Costes E, Samach A. Different flowering 

response to various fruit loads in apple cultivars correlates with degree of transcript 
reaccumulation of a TFL1-encoding gene. Plant J. 2016;87: 161-173. 

 
Hättasch C, Flachowsky H, Kapturska D, Hanke M. Isolation of flowering genes and seasonal 
changes in their transcript levels related to flower induction and initiation in apple (Malus 
domestica). Tree Physiol. 2008;28: 1459-1466. 

 
Huerta-Cepas J, Serra F, Bork P. ETE 3: Reconstruction, analysis and visualization of 

phylogenomic data. Mol Biol Evol. 2016;33(6): 1635-1638.  

 
Hoover E, De Silva N, McArtney S, and Hirst P. 2004. Bud development and floral 

morphogenesis in four apple cultivars. J Hort Sci Biotech 79(6): 981-984. 

 
Jones AL. 1992. Evaluation of the Computer Model MARYBLYT for Predicting Fire Blight 

Blossom Infaction on Apple in Michigan. Plant Disease 76(4): 344-347. 

 
Kim D, Langmead B, and Salzberg SL. 2015. HISAT: a gast spliced aligner with low memory 

requirements. Nature Methods 12: 357-360. 

 
Kotoda N and Wada M. MdTFL1, a TFL1-line gene of apple, retards the transition from the 

vegetative to reproductive phase in transgenic Arabidopsis. Plant Sci. 2005;168(1): 95- 628 
104. 

 
Lawson DM, Hemmat M, and Weeden NF. 1995. The use of molecular markers to analyze the 

inheritance of morphological and developmental traits in apple. J Amer Soc Hort Sci 120(3): 
532-537. 

 
Legave JM, Farrera I, Almeras T, and Calleja M. 2007. Selecting models of apple flowering 

times and understanding how global warming has had an impact on this trait. J Hort Sci and 
Biotech 83(1): 76-84. 

 
Lex A, Gehlenborg N, Strobelt H, Vuillemot R, and Pfister H. 2014. UpSet: Visualization of 

Intersecting Sets. IEEE Trans Vis Comput Graph 20(12):1983-1992. 

 

 

95 
 

Liebhard R, Kellerhals M, Pfammatter W, Jertmini M, and Gessler C. 2003. Mapping 

quantitative physiological traits in apple (Malus × domestica Borkh.). Plant Molecular Bio 
52:511-526. 

 
Kiyosue T and Wada M. 2001. LKP1 (LOV kelch protein 1): a factor involved in the regulation 

of flowering time in Arabidopsis. Plant J 23(6): 807-815. 

 
Kotoda N, Wada M, Komori S, Kiduo S, Abe K, et al. 2000. Expression Pattern of Homologues 

of Floral Meristem Identity Genes LFY and AP1 during Flower Development in Apple. J 
Amer Soc Hort Sci 125(4): 398-403. 

 
McArtney S, Hoover E, Hirst P, and Booking IR. 2001. Seasonal variation in the onset and 

duration of flower development in ‘Royal Gala’ apple buds. J Hort Sci Biotech 76(5): 536-
540. 

 
McCarthy DJ, Chen Y, and Smyth GK. 2012. Differential expression analysis of multifactor 

RNA-seq experiments with respect to biological variation. Nucleic Acids Research 40(10): 
4288-4297. 

 
Mimida N, Kotoda N, Ueda T, Igarashi M, Hatsuyama Y, Iwanami H, et al. Four TFL1/CEN-

like genes on distinct linkage groups show different expression patterns to regulate vegetative 
and reproductive development in apple (Malus × domestica Borkh.). Plant Cell Physiol. 
2009;50(2): 394-412. 

 
Nikiforova SV, Cavalieri D, Velasco R, and Goremykin V. 2013. Phylogenetic analysis of 47 

chloroplast genomes clarifies the contribution of wild species to the domesticated apple 
maternal line. Molecular Biology and Evolution 30(8): 1751-1760. 

 
Pelaz S, Ditta GS, Baumann E, Wiseman E, and Yanofsky MF. 2000. B and C floral organ 

identity functions require SEPALLATA MADS-box genes. Nature 405(6783): 200-203. 

 
Pertea M, Pertea GM, Antonescu CM, Chang TC, Medell JT, and Salzberg SL. 2015. StringTie 

enables improved reconstruction of a transcriptome from RNA-seq reads. Nature Biotech 33: 
290-295. 

 
Pertea G and Pertea M. 2020. GFF Utilities: GFFRead and GFFCompare. F1000 Research 9: 

304. 

 
Piñeiro M, Gómez-Mena C, Schaffer R, Martínez-Zapater JM, and Coupland G. 2003. EARLY 
BOLTING IN SHORT DAYS Is Related to Chromatin Remodeling Factors and REgulates 
Flowering in Arabidopsis by Repressing FT. Plant Cell 15: 1552-1562. 

 
Price MN, Dehal PS, Arkin AP. FastTree 2 - approximately maximum-likelihood trees for large 

alignments. PLoS One 2010;5(3): e9490. 

 
 

 

 

96 
 

R Core Team. R: a language and environment for statistical computing. R Foundation for 

Statistical Computing, Vienna, Austria. www. R-project.org. 2013. 

 
Ramírez F and Davenport TL. 2013. Apple pollination: A review. Scientia Horticulturae 162: 

188-203.  

 
Rodrigo J. 2000. Spring frosts in deciduous fruit trees - morphological damage and flower 

hardiness. Scientia Horticulturae 85(3): 155-173. 

 
Sánchez-Pérez R, Dicenta F, and Martínez-Gómez P. (2012). Inheritance of chilling and heat 

requirements for flowering in almond and QTL analysis. Tree Genetics & Genomes, 8: 379–
389. 

 
Sievers F, Wilm A, Dineen D, Gibson TJ, Karplus K, Li W, et al. Fast, scalable generation of 

high-quality protein multiple sequence alignments using Clustal Omega. Mol Syst Biol. 
2011;7: 539.  

 
Spotte RA, Stang EJ, and Ferree DC. 1976. Effect of overtree misting for bloom delay on 

incidence of fire blight in apple. Plant Dis. Rep. 60: 329–330. 

 
Spurr AR. 1969. A low-viscosity epoxy resin embedding medium for electron microscopy. 

Journal of Ultrastructure Research, (26), 31–43. 

 
Thompson A, May MR, Moore BR, and Kopp A. 2020. A hierarchical bayesian mixture model 

for inferring the expressed state of genes in transcriptomes. PNAS 117(32): 19339-19346. 

 
Trapnell C, Williams B, Pertea G, Mortazavi A, Kwan G, et al. 2012. Differential gene and 

transcript expression analysis of RNA-seq experiments with TopHat and Cufflinks. Nature 
Protocols 7: 562-578. 

 
Unterberger C, Brunner L, Nabernegg S, Steininger KW, Steiner AK, et al. 2018. Spring frost 

risk for regional apple production under a warmer climate. PLoS One 13(7):e0200201. 

 
Velasco R, Zharkikh A, Affourtit J, Dhingra A, Cestaro A, Kalyanaraman A, et al. The genome 
of the domesticated apple (Malus × domestica Borkh.). Nat Genet. 2010;42(10): 833–839.  

 

 

 

97 
 

CHAPTER FOUR 

Molecular identification and transcriptional patterning of FLOWERING LOCUS T genes 

in Apple (Malus × domestica)

 

98 
 

ABSTRACT 

The identification and characterization of Flowering locus T (FT) homologs in plants 

have played a critical role in understanding the evolution of unique flowering traits. Apple 

(Malus × domestica Borkh.) exhibits many flowering traits that are critical to production, 

including variation in bloom time and biennial bearing tendencies. Apple contains two presumed 

homologous copies of FT, MdFT1 and MdFT2, of which MdFT1 has been thoroughly identified 

and characterized. MdFT2 has remained relatively uncharacterized due to an elusive identity. In 

this study, I identify a transcriptional product that exhibits conserved FT sequence and structure 

and a distinctive sequence from that of MdFT1. MdFT1 expression was compared across a wide 

range of tissues and developmental stages using a gene expression atlas that resulted in the 

documentation that MdFT1 is predominately expressed in apex and rachis tissues. MdFT2 

quantification was unsuccessful due to its absence from the gene expression atlas. Using paralog-

specific probes, I conducted expression profiling of MdFT1 and the MdFT2 transcripts to 

characterize the daily and seasonal expression patterns. Both MdFTs were identified to be 

expressed in bourse shoot leaves but did not respond to floral inductive conditions. MdFT1 daily 

expression was diurnal, a conserved pattern with Arabidopsis, and was cultivar dependent. 

Seasonal expression patterns and response to crop load conditions were also found to be cultivar 

dependent. However, expression patterns of both MdFTs were not differentially expressed in 

response to crop load conditions that are floral promotive. These results indicate that MdFT 

could have limited function in the bourse shoot leaves. 

 

 

 

99 
 

INTRODUCTION 

The mechanism of floral initiation remained elusive until the identification and functional 

characterization of FLOWERING LOCUS T (FT) (Andrés and Coupland 2012). In Arabidopsis, 

FT is expressed in companion cells of the phloem within leaves and its translated product is 

loaded into the phloem stream for translocation to the apex (Corbesier et al. 2007, Andrés and 

Coupland 2012). In the apex, FT forms a transcription regulator complex with FLOWERING 

LOCUS D, which promotes the expression of meristem identity genes such as APETALA 1 (Abe 

et al. 2005; Wigge et al. 2005). Thus, FT functions as the primary integrator of the flowering 

signal from the leaves to the apex. To ensure flowering occurs during optimal conditions, a 

diverse network of environmental and genetic mechanisms regulate the expression of FT. The 

primary mechanisms for Arabidopsis regulation of FT is by photoperiod and temperature. Under 

long-day conditions, CONSTANS (CO) and its translated peptide gradually accumulate 

throughout the day, reaching a peak 10-14 hours after daybreak (Sawa et al. 2007; Song et al. 

2012; Golembeski et al. 2014). The accumulation of CO during these long-days leads to the 

promotion of FT, which triggers the transition to flowering (Song et al. 2012; Andrés and 

Coupland 2012; Golembeski et al. 2014). This CO-driven mechanism results in FT exhibiting a 

diurnal expression pattern when conditions are suitable for flowering. Under unfavorable 

conditions, such as short-days, CO is unable to accumulate due to its ubiquitination and resulting 

degradation, and transcription repression of the CO locus (Andrés and Coupland 2012; 

Golembeski et al. 2014). Temperature also regulates FT by the vernalization response that is 

conveyed by the FLOWERING LOCUS C (FLC) gene (Searle et al. 2006). Adequate cool 

temperatures will down regulate the expression of FLC, relieving its transcriptional repression of 

 

100 

 

FT. However, FLC mediated response is associated with a requirement for vernalization and 

many plants can initiation flowers without this requirement including apple.  

The documentation of the conservation of the FT genes and their functions is critical to 

our understanding the diverse flowering process that are exhibited by other annual and perennial 

plants. In apple (Malus × domestica Borkh.), two FT genes have been identified MdFT1 and 

MdFT2 (Kotoda et al. 2010). MdFT1 was found to be positioned on Chromosome (Chr.) 12 in 

both of the primary reference genomes (Velasco et al. 2010; Foster et al. 2014; Daccord et al. 

2017). The second copy of FT, MdFT2, was mapped onto Chr. 4 (Kotoda et al. 2010). However, 

the two primary reference genomes have incompletely described the genomic coordinates and 

sequence of MdFT2. This difficulty to resolve the genomic position of MdFT2 could be 

attributed to the striking similarity of the paralogs. Previous studies have reported that 

the FT paralogs have a >94% conserved coding sequencing identity (Kotoda et al. 2010). As a 

result of this high similarity, the misidentification of alleles from one MdFT gene as separate loci 

have occurred (Tränkner et al. 2011). This complexity to properly identify MdFT2 has also 

resulted in MdFT1 getting more attention related to the characterization of its function.  

MdFT1 is predominantly expressed in the apical meristems of vegetative and 

reproductive shoots (Kotoda et al. 2010). The peak expression periods of MdFT1 in the apex is 

during the anticipated periods of floral induction and initiation (Hättasch et al. 2008; Kotoda et 

al. 2010; Mimida et al 2011; Habermann et al. 2018; Guitton et al. 2016; Gottschalk et al. 

Chapter 2). The presence of fruit on the shoot, a floral repressive mechanism in some apple 

cultivars, has been reported to not influence MdFT1 expression in the corresponding shoot apex 

(Guitton et al. 2016, Habermann et al. 2018, Gottschalk et al. Chapter 2). In contrast, the 

presumed MdFT2 was found to be expressed in the apex, but its predominant expression is in 

 

101 

 

tissues of the floral organs (Kotoda et al. 2010). Seasonal expression patterns of MdFT2 are 

closely associated with the period of floral development late in the growing seasons and 

following dormancy release prior to spring bloom (Kotoda et al. 2010).  

Little is known if MdFTs exhibit the conserved daily expression pattern during floral 

induction that are characteristic of other long-day plants such as Arabidopsis. Additionally, many 

of these previous reports have relied on relative expression measures under narrow sets of 

conditions and tissues. In this study, I investigated the identity of the MdFTs using 

transcriptional models constructed in prior chapters of this dissertation. Using a gene expression 

atlas, I quantify the expression of MdFT1 across a diverse set of tissues and developmental 

stages. Lastly, I investigated the daily and season expression patterns of MdFT, while addressing 

cultivar-specific expression in the bourse shoot leaves. 

METHODS 

Plant material and tissue sampling 

Tissue collection was performed on trees maintained at the Michigan State University 

Clarksville Research Center (CRC) (Field: 42°52'24"N, 85°16'05"W – Station: 42°52'24"N, 

85°15'30"W) in Clarksville, Michigan. All trees were 10 years old or older and grown in a 

vertical axis high-density system using standard commercial practices. The specific strains of the 

cultivars and rootstocks used are as follows: Gala “Brookfield” on Budagovsky 9 (Bud9), 

Honeycrisp on Bud9, and Red Delicious “Schlet Spur” on M26. Ten trees with similar, high 

bloom density were selected, and all blossoms were removed by hand at the full bloom stage 

(early May 2016). For 'Honeycrisp' and 'Gala', three additional trees with high bloom density 

were selected as non-thinned controls. Individual bourse shoots were tagged at full bloom, and 

were collected at dusk on the following time intervals: 2-3, 12-13, 26-27, 39-40, 55-56, 68-69, 

 

102 

 

and 83-84 days after full bloom (DAFB). For each biological replicate tree, ten tagged bourse 

shoots were collected per time point. The first mature leaf was excised from each collected 

bourse shoot and frozen in liquid N2. To analyze daily gene expression, six ‘Honeycrisp’ and 

‘Red Delicious’ trees with similar, high bloom density were selected at full bloom. Three of the 

six trees per cultivar were then thinned of blossoms by hand, while the other three trees remained 

nonthinned. Individual bourse shoots were collected at three-hour intervals over the course of 

one 24 hour period on June 13th, 2016 (33 DAFB). This date was selected as it represents a 

midpoint of the anticipated period of floral initiation. The first mature leaf was excised from each 

collected bourse shoot and frozen in liquid N2. 

MdFT sequence retrieval and analysis 

All coding sequences were retrieved from the Genomic Database for Rosaceae (GDR) 

using the JBrowse (Buels et al. 2016) software. The de novo assembled MdFT2-like sequence 

was retrieved from previously published works (Gottschalk et al. Chapter 2). In short, the de 

novo transcripts were assembled from reads that failed to align to the GDDH13 v.1.1 reference 

genome sequence (Daccord et al. 2017). Alignments were performed using the HISAT2 aligner 

(v.2.1.0) invoking the --dta-cufflinks and --un-conc-gz options (Kim et al. 2015). Unaligned 

reads were then assembled into contiguous sequences using the Trinity de novo assembler using 

default settings (Haas et al. 2013). The de novo contigs were then constructed into gene models 

using the python program Trinity_gene_splice_modeler.py from the Trinity suite. Genomic 

sequences from Honeycrisp were obtained from an in-house dataset of Illumina sequence reads. 

Blastn was then used with an Expect (E)-value cutoff of 1e-12 to identify sequences that aligned 

to the de novo assembled FT transcript TRINITY_DN25413_c6_g1 (Camacho et al. 2008). 

 

103 

 

These sequences were then assembled into a de novo genomic sequence using the SPAdes 

assembler (v.3.11.1) with default assembly options (Bankevich et al. 2012). 

Expression analysis 

Total RNA was extracted from bourse shoot leaves using the Gasic et al. (2004) 

extraction protocol with the modification of using spermine instead of spermidine in the 

extraction buffer. Each extraction utilized 1 mm transverse section of five or more leaves per 

replicate extraction. For the seasonal expression experiment, two replicates per time point and 

treatment per cultivar were conducted. For the daily expression experiment, three replicates per 

time point per treatment per cultivar were used. Extracted RNA was then purified using a 

commercial clean-up kit (RNeasy Mini; QIAGEN, Germantown, MD). RNA quality was 

assessed using a 1.2% formaldehyde-agarose gels and quantified using a Nanodrop 2000c 

(Thermo Fisher Scientific, Waltham, MA). Gene expression was measured using a two-step 

quantitative real-time PCR. Previously designed primers and probes were used (Zhang et al. 

2019). Each primer set was designed to be specific to selected MdFT sequences and overlapped 

with an exon junction. Specifically, MdFT2 primers were designed using the de novo assembled 

transcript TRINITY_DN25413_c6_g1. A homolog of MdACTIN served as an internal control 

when calculating relative expression. We used an Agilent Technologies Stratagene Mx3005P 

(Santa Clara, CA) qPCR machine with cDNA derived from the extracted RNA. Each reaction 

was conducted using a mix of TaqMan®️ (Thermo Fisher Scientific, Waltham, MA) Gene 

Expression Master Mix (10 μl), 5x diluted cDNA template (2 μl), forward (1 μl) and reverse 

primers (1 μl), and probe (1 μl) for ACTIN, the primer-probe assay for the gene of interest (1 μl), 

and ddH2O (4 μl).  

 

 

 

104 

 

Statistical analysis 

Statistics were conducted using R (v.3.6.2) (R Core Team 2017) inside the R Studio suite 

(v.1.1.423) (Rstudio Team 2015). R data analysis packages used in this study were ggplot2, 

ggpubr, and tidyverse (Wickham 2016, Wickham et al. 2019). 

RESULTS AND DISCUSSION 

Identification of FT-like sequences 

With the release of the first apple genome and a subsequent high-quality genome derived 

from a double haploid individual, identification of FT-like sequences through sequence 

homology identification became possible (Velasco et al. 2010, Daccord et al. 2017). The first 

genome release (GDR v3.1.1; Velasco et al. 2010), had two annotated FT-like genes, 

MDP0000128821/MDP0000139278 and MDP0000132050 (Velasco et al. 2010), positioned on 

Chr. 4 and 12, respectively. The locus on Chr. 12 was previously annotated as MdFT1 and 

exhibited an expected gene structure with four exons (Kotoda et al. 2010). The locus on Chr. 4 

was previously annotated as MdFT2 and its structure is less clear. This locus contained five 

predicted peptides with FT-like sequence but was annotated as separate genes. A later revised 

reference transcriptome annotated MdFT as a single transcript (MDP0000139278) that exhibited 

the conserved four exon structure as MdFT1 (Fig 4.1) (Kotoda et al. 2010; Jung et al. 2019). In 

the subsequent high-quality genome (GDDH13 v.1.1), only MdFT1 (MD12G1262000) was 

present in the genome assembly and featured overlapping alignments with both previously 

assembled FT genes (Fig 4.1) (Daccord et al. 2017). 

To resolve the missing MdFT2, sequence reads from a previous apple transcriptome 

study that failed to map to the GDDH13 reference genome were de novo assembled (Gottschalk 

et al. Chapter 2). Within those assembled contigs, one identified gene contained FT-like 

 

105 

 

nucleotide sequence and translated into an FT-like peptide (TRINITY_DN25413_c6_g1). 

However, the gene appeared chimeric as it contained a second transcript positioned in the 5’ 

direction to the FT-like sequence. To gain further evidence that the chimeric assembly is actual 

and not an artifact of the assembly process, we performed an independent de novo assembly of 

DNA sequence reads. We utilized a library of high-quality DNA sequences generated from the 

cultivar Honeycrisp and filtered the read libraries for sequences that contained FT-like homology 

using Blastn. These filtered reads were then assembled into eight contigs and aligned with the de 

novo assembled FT-like transcript. The alignments indicated that one of the assembled contigs 

shared  >90% sequence identity to the FT-like region of the de novo transcript previously 

assembled (Fig 4.2). These assembled sequences also exhibited 100% translated identity to an 

NCBI cataloged MdFT2 protein in the -1 frame. These results suggest that the de novo assembled 

transcript has a corresponding genomic sequence. However, these results do not establish that 

MdFT2 is a separate locus as opposed to a divergent allele of MdFT1. From this point forward, 

the de novo FT-like sequence will be referred to as MdFT2. 

Tissue-specific expression of MdFT1 

In Arabidopsis, FT is expressed under promotive photoperiodic conditions in the phloem 

companion cells within the leaf (Corbesier et al. 2007, Andrés and Coupland 2012). The 

resulting peptide, FT, is then translocated from the leaf to the apex via the phloem stream, upon 

reaching the apex it triggers floral initiation (Corbesier et al. 2007, Andrés and Coupland 2012). 

To document if FT expression patterns are conserved between Arabidopsis and apple, we 

evaluated tissue-specific expression patterns using a gene expression browser constructed from 

an apple relative, Malus fusca (Gottschalk et al. in prep). MdFT1 was found to be expressed in 

rachis and the shoot apical meristem (Fig 4.3). MdFT1 expression was also found in the 

 

106 

 

cotyledons and leaves of seedlings.  Little or no expression of MdFT1 was observed in the 

leaves. However, this dataset did not assess gene expression in leaves from specific positions on 

the shoot or spur which could explain the absent expression in leaves. Unfortunately, MdFT2 

was absent in that dataset resulting our in ability to evaluate for tissue-specific expression.  

Diurnal expression of MdFTs in bourse shoot leaves 

Bourse shoot leaves are critical to facilitate floral initiation in apple, suggesting that a 

possible floral signal originates in those tissues (Fulford 1960; Elsysy and Hirst 2017, 2019). 

Moreover, FT in Arabidopsis is produced and exported from leaf tissues. To examine if FT 

regulation is conserved in apple, I evaluated the expression of MdFTs in the bourse shoot leaves. 

To induce flowering, blossoms were thinned from representative trees of ‘Honeycrisp’, ‘Red 

Delicious’ and ‘Gala’ to promote flowering. The thinning of blossoms was effective (p < 0.05) at 

promoting a return bloom in all three cultivars (Fig 4.4). This result suggests that spurs thinned 

of flowers initiated blossoms for the subsequent season, whereas the nonthinned remained 

predominately vegetative in identity. RNA was extracted from the first mature bourse shoot leaf 

over 24 hours at three-hour intervals from thinned and nonthinned spurs of ‘Honeycrisp’ and 

‘Red Delicious’. Using MdFT paralog-specific probes, relative expression was quantified in 

these samples. For MdFT1, peak expression was observed during the night time hours in 

‘Honeycrisp’ indicative of a diurnal pattern (Fig 4.5). In comparing the effect of crop load, 

MdFT1 in ‘Honeycrisp’ exhibited a stronger amplitude during its oscillation of expression in the 

thinned compared to the nonthinned (Fig 4.5 - left panels). In contrast, expression of MdFT1 in 

‘Red Delicious’ peaked around 7:00 PM in the thinned spurs (Fig 4.5 - bottom right panel). A 

oscillating pattern of expression in thinned ‘Red Delicious’ is evident but the amplitude was 

lower than in ‘Honeycrisp’ and it does not exhibit a diurnal pattern. However, no significant 

 

107 

 

differences were found between the expression MdFTs in the thinned compared to nonthinned 

leaves, for either cultivar at any time point. 

For MdFT2, ‘Honeycrisp’ exhibited two peaks of expression in the nonthinned spurs (Fig 

4.6 - top left panel). The peaks occurred between 10:00 AM and 1:00 PM and at 10:00 PM. In 

the thinned spurs, MdFT2 had a single peak at 4:00 PM (Fig 4.6 - bottom left panel). In ‘Red 

Delicious’, MdFT2 had a single peak at 4:00 PM under both crop load conditions similar to what 

was observed in the thinned ‘Honeycrisp’ leaves (Fig 4.6 - right two panels). However, the 

amplitude of the peak in the thinned spurs was more pronounced than nonthinned. No significant 

differences were found between the expression values of the thinned compared to nonthinned, 

for either cultivar at any time point. 

I also investigated whether the expression pattern over time differed between cultivars 

and treatment for each of the MdFTs. The expression profiles were clustered using a Euclidian 

distance calculation to find similarities between cultivars and treatment (Fig 4.7). Differences in 

MdFT1 expression profiles appear to be primarily associated with the cultivars, as they clustered 

together. For MdFT2, the clustering found that nonthinned Honeycrisp formed a clearly defined 

clade separate from the other expression profiles. The clade that contained the three other 

profiles of MdFT2 were separated by cultivar. This result suggests that MdFT2 expression 

profiles are distinct in a crop load and cultivar-specific fashion.  

Seasonal expression patterns of MdFTs 

In other perennial tree fruit, such as Citrus, differential expression of FT-like genes in 

leaves was associated with biennial bearing (Munoz-Fambuena et al. 2011; Ziv et al. 2014). I 

investigated if this finding was conserved in apple by measuring expression of the two MdFT 

paralogs in the bourse shoot leaves of two apple cultivars of opposing biennial bearing habits 

 

108 

 

throughout the growing season (Fig 4.8). ‘Honeycrisp’ is a biennial bearer whereas ‘Gala’ is an 

annual bearer (Embree et al. 2007, Elsysy and Hirst 2017). MdFT1 exhibited higher expression 

in the nonthinned trees early in the growing season for ‘Honeycrisp’, with the expression 

decreasing after a peak around 39/40 DAFB. Although the seasonal expression pattens appear to 

differ, the differences between the thinned and nonthinned were found to not be significant. This 

result could be attributed to the low relative expression values and limited replications. In ‘Gala’, 

MdFT1 expression patterning between the thinned and nonthinned was strikingly similar with 

relatively little change of the season. MdFT2 in ‘Honeycrisp’ exhibited a responsive (p < 0.01) to 

thinning, with lower expression of MdFT2 in the thinned trees compared to the than nonthinned 

(Fig 4.9). However, this difference in expression was associated with the nonthinned treatment 

which repressed floral initiation. MdFT2 in ‘Gala’ exhibited no difference in expression between 

crop load treatments. When clustering the expression profiles of MdFT1, the clades were 

separated by crop load treatments (Fig 4.10). This result is indicative that MdFT1 is more closely 

associated with the treatment than cultivar. The inverse situation was true for MdFT2 where 

cultivar was the separating factor.  

CONCLUSIONS 

This study provides a greater understanding of the expression patterning of apple’s FT 

homologs. Our identification of FT-like sequences in apple provides confirming results for 

MdFT1 but MdFT2 is still not confirmed due to cryptic genomic position. The de novo 

assembled transcript that is FT-like exhibits different sequence and expression patterning than 

MdFT1 suggesting its either MdFT2 or MdFT1 has unique allele-specific expression. Tissue- and 

developmental-specific expression was documented for MdFT1 and was found to be primarily 

localized to the apex and rachis. The localization of MdFT1 expression to the apex, suggests that 

 

109 

 

the FT signaling pathway could also be divergent in apple. MdFT1 expression in bourse leaves 

was found to exhibit a diurnal pattern only for ‘Honeycrisp’ regardless of crop load conditions.  

However, this diurnal expression was not consistent for ‘Red Delicious’ where its peak 

expression occurred during the late evening. ‘Red Delicious’ MdFT1 also exhibited no 

significant differences in expression in response to crop load. MdFT2, on the other hand, 

exhibited no conserved diurnal pattern for either cultivar. Seasonal expression patterns of MdFT1 

did not exhibit a significant response to the floral inductive thinning. In contrast, MdFT2 

exhibited an apparent increase in expression in the nonthinned trees of ‘Honeycrisp’. Although, 

this increase in expression was associated with the nonthinned treatment which repressed 

flowering. Thus, crop load does not appear to affect the daily or seasonal expression of MdFT in 

the bourse shoot leaves. In conclusion, apple’s FT homologs exhibit distinct expression patterns 

that are associated with genotypes, but are not influenced by crop load. MdFTs most likely do 

not function in a similar manner as Arabidopsis FT within leaf structures. This conclusion is also 

supported by previous works that identified a crop load specific response of MdFT1 in the apex 

(Gottschalk et al. Chapter 2).  

 

 

 

110 

 

 

Figure 4.1. Genomic position of MdFT1 in the GDDH13 genome and corresponding 

alignments with the two FT genes assembled in the GDR genome. 

 

 

 

111 

 

Figure 4.2. Alignment between the genomic sequence for MdFT1 (MD12G1262000), 

 

segment of the de novo assembled FT-like transcript TRINITY_DN25413_c6.g1 (labelled 

as TRINITY_DN25413), and a de novo assembled FT DNA sequence contig. 

 

 

 

112 

 

Figure 4.3. Tissue and developmental stage specific expression of MdFT1 in Malus fusca. 

 

 

 

 

113 

 

Nonthinned 

Thinned 

Nonthinned 

Thinned 

Nonthinned 

Thinned 

 
Figure 4.4. Percent return bloom in cultivars that received blossom thinning treatments. A) 

Honeycrisp, B) Red Delicious, C) Gala. 

 

 

 

114 

 

Figure 4.5. MdFT1 relative expression in the bourse leaves of nonthinned and thinned 

 

shoots of two apple cultivars. Yellow shade is the estimated day light sampling hours. 

Generalized trend line and confidence interval are shown (grey shading). 

 

 

 

115 

 

Figure 4.6. MdFT2 relative expression in the bourse leaves of nonthinned and thinned 

 

shoots of two apple cultivars. Yellow shade is the estimated day light sampling hours. 

Generalized trend line and confidence interval are shown (grey shading). 

 

 

 

116 

 

Figure 4.7. Clustering of MdFT homologs expression profiles over 24 hours from trees 

 

thinned or nonthinned of flowers.  

 

 

 

117 

 

Figure 4.8. MdFT1 relative seasonal expression in the bourse leaves of nonthinned and 

 

thinned shoots of two apple cultivars. Generalized trend line and confidence interval are shown 

(grey shading). 

 

 

 

118 

 

Figure 4.9. MdFT2 relative seasonal expression in the bourse leaves of nonthinned and 

thinned shoots of two apple cultivars. Generalized trend line and confidence interval are shown 

 

(grey shading). 

 

 

 

119 

 

Figure 4.10. Clustering of MdFT homologs expression profiles over the season from trees 

 

thinned or nonthinned of flowers.  

 

 

120 

 

REFERENCES

 

121 

 

REFERENCES 

 
 
Abe M, Kobayashi Y, Yamamoto S, Daimon Y, Yamaguichi A, et al. FD, a bZIP Protein 

Mediating Signals from the Floral Pathway Integrator FT at the Shoot Apex. Science 2005 
309(5737): 1052-1056. 

 
Andrés F and Coupland G. The genetic basis of flowering responses to seasonal cues. Nature 

Reviews Genetics 2012 13:627-639. 

 
Bankevich A, Nurk S, Antipov D, Gurevich AA, Dvorkin M, et al. SPAdes: A new genome 

assembly algorithm and its applications to single-cell sequencing. J Comput Biol 2012 19(5): 
455-477. 

 
Buels R, Yao E, Diesh CM, Hayes RD, Munoz-Torres M, et al. JBrowse: a dynamic web 

platform for genome visualization and analysis. BMC Genome Biol 2016 16:66. 

 
Camacho C, Coulouris G, Avagyan V, Ma N, Papadopoulos J, Bealer K, et al. BLAST+: 

architecture and applications. BMC Bioinformatics 2008; 10:421. 

 
Corbesier L, Vincent C, Jang S, Fornara F, Fan Q, et al. FT Protein Movement Contributes to 

Long-Distance Signaling in Floral Induction of Arabidopsis. Science 2007 316(5827): 1030-
1033. 

 
Daccord N, Celton JM, Linsmith G, Becker C, Choisne N, et al. High-quality de novo assembly 

of the apple genome and methylome dynamics of early fruit development. Nature Genetics 
2017 49:1099-1106. 

 
Elsysy M and Hirst P. Molecular Basis of Flower Formation in Apple Caused by Defoliation and 

Gibberellins. J Amer Soc Hort Sci 2019 144(6): 414-419. 

 
Elsysy M and Hirst P. The role of spur leaves, bourse leaves, and fruit on local flower formation 

in apple: an approach to understanding biennial bearing. HortScience 2017 52(9): 1229-1232. 

 
Embree CG, Myra MTD, Nichols DS, and Wright AH. Effect on blossom density and crop load 

on growth, fruit quality, and return bloom in ‘Honeycrisp’ apple. Hortscience 2007 
42(7):1622-1625. 

 
Fulford RM. The use of defoliating sprays for the control of biennial bearing in apples. 

HortScience 1960 35: 202-213. 

 
Gasic K, Hernandez A, Korban SS. RNA extraction from different apple tissues rich in 

polyphenols and polysaccharides for cDNA library construction. Plant Mol Bio Rep. 2004 
22: 437a-437g. 

 

 

122 

 

Golembeski GS, Kinmonth-Schultz HA, Song YH, and Imaizumi T. Photoperiodic flowering 

regulation in Arabidopsis thaliana. Adv Bot Res 2014 72: 1-28. 

 
Gottschalk C, Xie Y, Flores M, Rogers S, Guan Q, and van Nocker S. Gene expression atlas of 

development in a wild apple species. in prep 

 
Gottschalk C, Zhang S, Schwallier P, Rogers S, Bukovac MJ, and van Nocker S. Genetic 

mechanisms associated with the repressive effect of fruit on floral initiation in apple (Malus 
× domestica Borkh). Submitted to PLoS One 

 
Guitton B, Kelner JJ, Celton JM, Sabau X, Renou JP, Chagne D and Costes E. Analysis of 
transcripts differentially expressed between fruited and deflowered 'Gala' adult trees: a 
contribution to biennial bearing understanding in apple. BMC Plant Bio 2016 16:55. 

 
Haberman A, Ackerman M, Crane O, Kelner JJ, Costes E and Samach A. Different flowering 

response to various fruit loads in apple cultivars correlates with degree of transcript 
reaccumulation of a TFL1-encoding gene. The Plant J 2018 87(2): 161-173. 

 
Haas BJ, Papanicolaou A, Yassour M, Grabherr M, Blood PD, Bowden J, et al. De novo 

transcript sequence reconstruction from RNA-seq using the Trinity platform for reference 
generation and analysis. Nat Protoc. 2013 8(8): 1494-1512. 

 
Hättasch C, Flachowsky H, Kapturska D, and Hanke MV. Isolation of flowering genes and 

seasonal changes in their transcript levels related to flower induction and initiation in apple 
(Malus domestica). Tree Physio 2008 28(10): 1459-1466. 

 
Jung S, Lee T, Cheng CH, Buble K, Zheng P, et al. 15 years of GDR: new data and functionality 

in the Genome Database for Rosaceae. Nucleic Acids Res 2019 47:D1137–D1145.  

 
Kim D, Langmead B, and Salzberg SL. HISAT: a fast spliced aligner with low memory 

requirements. Nat Methods. 2015 12: 357-360. 

 
Kotoda N, Hayashi H, Suzuki M, Igarashi M, Hatsuyama Y, et al. Molecular Characterization of 

FLOWERING LOCUS T-Like Genes of Apple (Malus × domestica Borkh.). Plant Cell 
Physiol 2010 51(4):561-575. 

 
Mimida N, Ureshino A, Tanka N, Shigeta N, et al. Expression patterns of several floral gene 

during flower initiation in apical buds of apple (Malus x domestic Borkh.) revealed by in situ 
hybridization. Plant Cell Rep 2011 30:1485-1492. 

 
Munoz-Fambuena N, Mesejo C, Gonzalez-Max MC, Primo-Millo E, Agusti M and Iglesias DJ. 

Fruit regulates seasonal expression of flowering genes in alternate-bearing 'Moncada' 
mandarin. Annals of Botany 2011 108:511-519. 

 
R Core Team. R: a language and environment for statistical computing. R Foundation for 

Statistical Computing, Vienna, Austria. www. R-project.org. 2013. 

 

123 

 

RStudio Team. 2015. RStudio: Integrated Development for R. RStudio, Inc., Boston, MA URL 

http://www.rstudio.com/. 

 
Sawa M, Nusinow DA, Kay SA, and Imaizumi T. FKF1 and GIGANTEA complex formation is 

required for day-length measurement in Arabidopsis. Science 2007 318(5848): 261-265. 

 
Searle I, He Y, Turck F, Vincent C, Fornara F, et al. The transcription factor FLC confers a 

flowering response to vernalization by repressing meristem competence and systemic 
signaling in Arabidopsis. Gene and Development 2006 20: 898-912. 

 
Song YH, Smith RW, To BJ, Millar AJ, and Imaizumi T. FKF1 conveys crucial timing 
information for constans stabilization in the photoperiodic flowering. Science 2012 
336(6084: 1045-1049. 

 
Tränkner C, Lehmann S, Hoenicka H, Hanke MV, Fladung M, et al. Note added in proof to: 

Over-expression of an FT-homologous gene of apple induces early flowering in annual and 
perennial plants. Planta 2011 233: 217-218. 

 
Velasco R, Zharkikh A, Affourtit J, Dhingra A, Cestaro A, et al. The genome of the domesticated 

apple (Malus × domestica Borkh.). Nature Genetics 2010 42:833–839. 

 
Wickham H. ggplot2: Elegant graphics for data analysis. Springer-Verlag, New York. 2016. 
 
Wickham H, Averick M, Bryan J, Chang W, D’Agostino L et al. Welcome to the tidyverse. J 

Open Source Software 2019 4(43): 1686. 

 
Wigge PA, Kim MC, Jaeger KE, Busch W, Schmid M, et al. Integration of Spatial and Temporal 

Information During Floral Induction in Arabidopsis. Science 2005 309(5737): 1056-1059. 

 
Zhang S, Zhang D, Fan S, Du L, Shen Y, Xing L, Li Y, Ma J, and Han M. Effect of exogenous 

hormones, and flowering-associated genes in ‘Fuji’ apple (Malus domestica Borkh.) Plant 
Physio Biochem 2016 107: 178-186. 

 
Ziv D, Zviran T, Zezak O, Samach A and Irihimovitch V. Expression Profiling of FLOWERING 

LOCUS T-like Gene in Alternate Bearing 'Hass' Avocado Trees Suggests a Role of PaFT in 
Avocado Flower Induction. PloS One 2014 9(10). 

 

 

 

124 

 

CHAPTER FIVE 

Efficacy of foliar-applied GA in controlling flowering

 

125 

 

 

ABSTRACT 

 

The precise control of flowering is a critical aspect of the production management for 

biennial bearing susceptible apple cultivars. Traditional methods to control flowering have relied 

on the use of flower and fruit thinning, which are imprecise and can result in undesirable off-

target effects. An alternative approach is the use of plant growth regulators (PGRs) to influence 

the endogenous hormone pathways that control flowering. However, PGRs have failed to 

become a standard practice due to unpredictability and occasional ineffectiveness in eliciting a 

desirable response. In this study, I evaluated the use of exogenous foliar-applied gibberellic acid 

(GA) PGRs in controlling flowering in apple (Malus  domestica Borkh.) across three 

consecutive seasons. This study aimed to identify the most effective formulation of GA, timing 

of application, application rate, and cultivar-specificity. Two commercial available bioactive 

GAs, G3 and GA4+7, were evaluated for efficacy in five different apple cultivars. GA3 exhibited 

the most promising results with applications in two of the three years resulting in the repression 

of flowering in the biennial cultivar Honeycrisp. In the one year that did not result in repression 

of flowering, a promotive effect was observed. The repressive responses to GA3 were most 

effective when applied to low fruit load trees between 27 and 57 DAFB, which overlaps with the 

anticipated period of floral initiation in apple. The strongest repressive response was associated 

with the highest concentration of GA3 evaluated here and yielded a reduction of 50% in the 

number of flowering spurs in the subsequent season. Additionally, the fruit harvest from the high 

concentration treated trees during the subsequent season exhibited earlier maturity and higher 

overall fruit quality without sacrificing the average yield per tree. GA3 application during the 

summer in low-bearing years offers a potential PGR strategy to provide more consistent 

production in Honeycrisp with minimal loss in the quality of harvested fruit.  

 

 

126 

 

INTRODUCTION 

Tree fruit growers have long sought an effective approach to manipulate flowering for the 

purpose of managing crop production and quality. The control of flowering is particularly 

important for many cultivars of apple (Malus  domestica Borkh.) that are prone to biennial (or 

alternate) bearing. In biennial cultivars, over- or under-cropping can drastically affect the amount 

of return bloom and the resulting crop load in the subsequent year (Jonkers 1979, Monselise and 

Goldschmidt 1982). Furthermore, many apple-growing regions are subject to blossom-killing 

spring frosts which can further amplify biennial bearing cycles.  

Commercial apple producers typically manage biennial bearing through adjustments to 

the current year’s crop load by manual or chemical thinning of flowers or fruit, prior to or during 

the early anticipated period of floral initiation during the current season (Dennis 2000). Although 

thinning increases return bloom and fruit size, when applied at appropriate times, those effects 

are not always associated with an increase in crop value (Davis et al. 2004). For example, 

chemical thinners can result in undesirable off-target effects such as foliar phytotoxicity and fruit 

russeting (Kon et al. 2018). The insecurities associated with chemical thinning practices have 

created a need to explore alternative management options. A promising alternative is the use of 

summer applications of plant growth regulator (PGR) to influence the endogenous mechanism(s) 

governing floral initiation.  

The effects of exogenous application of PGRs on flowering in fruit trees has been the 

subject of numerous previous investigations (reviewed by Miller 1988 and Ionescu et al. 2017). 

Gibberellic acid (GA), in particular, has shown promise for inhibiting flowering when applied 

during the period of floral initiation in low bearing years. In apple, this anticipated period of 

floral initiation begins shortly after bloom, when vegetative meristems are competent to 

 

127 

 

transition to a reproductive meristem (Foster et al. 2003). The cessation of this competency 

period occurs roughly 100 days after full bloom (DAFB) (Foster et al. 2003). Guttridge (1962), 

was the first to report that the application of GA in low concentrations of 10 to 50 parts per 

million (ppm) could inhibit flowering in apple cultivars up to 40% compared to controls when 

applied during this period. However, the reported response was not consistent across all six 

cultivars evaluated (Guttridge 1962). Dennis and Edgerton (1966) evaluated the efficacy of GA3 

vs GA4+7 and found GA4+7 to be more effective at inhibiting flowering. This result further 

supported by work by Tromp (1982), who also reported that applications made at full bloom 

were effective at inhibiting flowering using GA3, GA7, or GA4+7. Tromp (1982) also reported 

that GA4+7 was the most effective type of GA when applied at full bloom or later. Other reports 

have found GA3 to be effective at inhibiting flowering when applied after full bloom, which 

contradicts some findings reported by Tromp (1982) (Luckwill and Silva 1979, Bertelsen and 

Tustin 2002). Many other previous reports have found GA to be effective at inhibiting flowering 

at various application times and concentrations (Bertelsen et al. 2002, McArtney 1994, 

McArtney and Li 1998, McArtney et al. 2007, Schmidt et al. 2009, Schmidt et al. 2010, Tromp 

1987, Zhang et al. 2016). However, many of these studies on GA’s efficacy have been 

undermined by inconsistent responses between cultivars, timings, and application rates.   

Tromp (1987) evaluated four different bioactive GAs for effects on flowering. They 

reported that GA3 and GA4 did not influence flowering, a result that is contradictory to much of 

the previous reported effects. Moreover, Looney et al. (1985) demonstrated that GA4 can 

increase flowering the subsequent year under particular circumstances. These inconsistent and 

contrasting responses to GA applications have limited its viability as a commercial approach to 

controlling flowering. In addition, many of the early works involving GA utilized applications at 

 

128 

 

full bloom, which can lead to the development of parthenocarpic fruit (Bukovac 1963, Galimba 

et al. 2019). Parthenocarpy can lead to changes in fruit size and quality (Galimba et al. 2019). 

Identifying a late-spring or summer application timing, in which GA is still an effective method 

to controlling flowering, would offset these concerns with bloom time applications.  

In the following study, the efficacy of foliar-applied GAs were tested for controlling 

flowering in Honeycrisp and other high-value apple cultivars. This study aimed is to document 

the efficacy of early summer time applications, different GAs and application concentrations, 

inter-seasonal responses, and cultivar-specificity. In addition, the evaluation of fruit yield, 

maturity, and quality were assessed in trees that exhibited a significant response to GA.     

METHODS 

2016 PGR Experiments 

All experiments were conducted at the Michigan State University Clarksville Research 

Center (CRC) (Field: 42°52'24"N, 85°16'05"W – Station: 42°52'24"N, 85°15'30"W) in 

Clarksville, Michigan. Trees were grown as a vertical axis with trellis support. All trees were 

greater than 10 years old and were randomly assigned to treatment dates or as controls. All PGR 

applications were made using a four-gallon pump-action backpack sprayer (GroundWork, 

Distributed by Tractor Supply Company, Brentwood, TN) with an application rate of 1 L of 

solution applied to all foliar surfaces on a per tree basis. After each application, spray tanks were 

triple rinsed before use in another experiment. Trees selected for the GA treatments and their 

controls underwent thinning of fruit 30 - 32 days after full bloom (DAFB), once fruit set was 

established.  

For the evaluation of GA4+7, we selected five apple cultivars that are commonly grown 

for commercial production in Michigan. The five cultivars selected were Red Delicious, Fuji, 

 

129 

 

Jonagold, Gala, and Honeycrisp. Twenty trees per cultivar, except Jonagold which had 16, were 

randomly assigned to one of four possible treatment application dates or to serve as controls. 

Using this design, three or four trees were assigned as replicates per treatment. Application were 

made at weekly intervals starting at 36 DAFB and continued through 57 DAFB (Table 5.1). For 

the evaluation of GA4+7, commercially available Provide 10SG (Valent Biosciences, Libertyville, 

IL) at 188 ppm plus 0.1% Regulaid surfactant (Kalo, Overland Park, KS) was used. For 

evaluation of GA3, 40 Honeycrisp trees were selected and randomly assigned to the same 

treatment dates as the GA4+7 experiment. Spray applications of GA3 were made using 

commercially available ProGibb 4% (Valent Biosciences, Libertyville, IL) at 200 ppm plus 0.1% 

Regulaid. All control trees received water plus 0.1% surfactant application on 36 DAFB.  

Assessment of the effect on flowering was conducted during bloom (tight cluster stage) 

of the following season. For each treatment date and PGR, a minimal of three representative 

trees were selected for flowering spur counting. Within each of the three selected trees, three 

random branches were selected and the total spurs and flowering spurs were counted on each 

branch.  

2017 PGR Experiments 

Field experiments were conducted at two different locations during the 2017 season. The 

first was at the CRC, in which the same orchards used in 2016 were used again in 2017. The 

second location was a commercial orchard located in Sparta, MI (Field: 43°06'13"N, 

85°41'57"W). Applications performed at the CRC were conducted using the same equipment as 

described for the 2016 experiments. The commercial orchard applications were carried out using 

a commercial airblast sprayer (Blueline, Yakima, WA).  

 

130 

 

ProGibb was evaluated for efficacy at both field locations. At the CRC, four cultivars 

(Fuji, Gala, Honeycrisp, and Red Delicious) were selected for use, while two cultivars 

(Honeycrisp and Jonagold) were selected at the commercial orchard. For the CRC trials, 30 trees 

per cultivar were selected and hand-thinned of fruit prior to PGR application. Trees were 

grouped into blocks of five replicate trees and assigned to a specific treatment date or as a 

control. Applications of ProGibb at 400 ppm plus 0.1% Regulaid were made on a ~100 Growing 

Degree Days base 50 (GDD50) intervals, starting at 400 and continuing until 800 GDD50 (Table 

5.2). Control trees received an application of a mix of water and 0.1% Regulaid at 400 GDD50. 

The commercial orchard had received a late spring frost during bloom resulting in an estimated 

10 - 20% decrease in viable blossoms. Severely affected frost thinned rows were selected for use 

in new PGR trials. Two rows of each cultivar were assigned to one treatment date and a third 

row assigned as a control. ProGibb at 400 ppm plus 0.1% Regulaid applications were made at 

~450 and ~550 GDD50. Control trees received no applications. 

Flowering data was collected again during bloom (tight cluster stage) in the following 

spring. At the CRC, every tree used in the experiment had three randomly selected branches used 

in counting total flowering and vegetative spurs. At the commercial orchard, ten randomly 

selected trees within the row were used in counting total flowering and vegetative spurs. In a 

similar fashion as the CRC experiment, three randomly selected branches from each of the ten 

trees were used.    

2018 PGR Experiments 

All field experiments were conducted at the CRC during the 2018 season and spray 

applications were performed using the same equipment as described for the 2016 experiments. 

Honeycrisp was selected as the focus cultivar for all trials during this season (Table 5.3). A 

 

131 

 

planting of third leaf trees was selected for evaluation of ProGibb at 400 ppm plus 0.1% 

Regulaid. Trees were planted in a high-density system consisting of 84 trees per row split into 

three blocks of 28 trees. All trees were thinned of flowers or fruit within two weeks after bloom. 

ProGibb was applied to three replicate blocks of five trees each per treatment date. PGR 

applications were applied at two-week intervals starting at 27 DAFB and continuing until 84 

DAFB. Control trees received an application of water plus 0.1% Regulaid at 27 DAFB. In 

addition, ProGibb at 200 ppm and Provide at 100, 200, and 500 ppm plus 0.1% Regulaid were 

applied at 29 DAFB as a single timed treatment to three blocks of five trees each. One replication 

of the Provide at 200 ppm treatment was located in a separate row of thinned Honeycrisp trees 

used previously in 2016/17 experiments. For the Provide 500 ppm experiment, four floral/fruit 

thinned trees were assigned per treatment replicate. Application dates were the same as the 

previously discussed ProGibb 400 ppm experiment. The evaluation of flowering was carried out 

in the same fashion as described for the 2017 experiments. An additional assessment of visual 

bloom was determined by a trained profession using a rating scale of 1 (absent of flowers) to 10 

(every spur is flowering).  

Fruit Yield and Quality Experiments 

For yield measurements, all fruit was harvested from each replicate tree from each of 

three treatment blocks during the commercial harvest date for Honeycrisp (September 15th, 

2019). Harvested fruit was weighed (Model 4010 Scale Pelouze, Bridgeview, IL) and the yield 

recorded. Trunk diameter measurements were recorded for each tree in the experiment with a 

digital caliper (DKC-8050 GlowGeek, Guangdong, China) and recorded. For fruit quality, three 

representative fruit were collected from five replicate trees from each of the three treatment 

blocks. The fruit samples were then subjected to commercial quality assessment at Michigan 

 

132 

 

State University’s Ridge Apple Quality Lab to be processed using a commercial apple sorting 

machine (Compac Spectrum System, Auckland, NZ). Firmness was measured using a 

penetrometer (QA Supplies, Norfolk, VA) mounted onto a moveable stand. Brix measurements 

were taken with a pocket brix-acidity meter (PAL-BX|ACID5, ATAGO, Tokyo, Japan). 

Statistics 

All statistics were performed using R (v.3.6.2) (R Core Team 2017) within the R Studio 

suite (v.1.1.423) (RStudio Team 2015). Plots were generated using the ggplots2 (Wickham 2016) 

and ggpubr packages.  

RESULTS 

Flowering response 

In 2016, ProGibb (GA3, 200 ppm) and Provide (GA4+7, 188 ppm) were evaluated for their 

effectiveness at specific summer application timings that correspond to the anticipated period of 

floral initiation on fruit thinned trees. ProGibb was found to exhibit a repressive effect on return 

bloom in Honeycrisp, with effectiveness increasing at progressively later treatment dates (Figure 

5.1A). The inhibition of flowering reached a statistically significant threshold when GA3 was 

applied at 50 and 57 DAFB, with the later date having a stronger effect. However, the 57 DAFB 

GA3 treatment exhibited greater variability in the flowering response. GA4+7 also exhibited a 

inhibitory effect on flowering in Honeycrisp but only when applied at 57 DAFB (Fig 5.1B). 

However, this inhibitory response was not statistically significant. For the four other high-value 

apple cultivars, the only significant effect on flowering in response to GA4+7  was observed in 

Red Delicious (Fig 5.1C-F). Here, GA4+7 promoted a return bloom when applied at 36 DAFB 

(Figure 5.1D).  

 

133 

 

Based on the significant effect observed with GA3 applied in 2016, its evaluation was 

expanded during the 2017 season to include three additional cultivars. Additionally, a second 

location was used to evaluate an on-farm approach to the application of GA3. GA3 was found to 

have no inhibitory effect on flowering in any of the cultivars used at the CRC (Fig 5.2A-D). In 

contrast, Honeycrisp, Fuji, and Red Delicious exhibited an increase in flowering in response to 

the application of GA3 (Fig 5.2A-C). Honeycrisp exhibited the strongest promotive response, 

with an increase in flowering of ~23% over the control at the final treatment date of 53 DAFB 

(Fig 5.2A). An increase in flowering of 15% and 12% were observed in ‘Fuji’ and ‘Red 

Delicious’, respectively (Fig 5.2B-C). ‘Gala’ exhibited no response to the application of GA3 

(Fig 5.2D). GA3 was applied at the commercial orchard to full rows of frost-thinned 

‘Honeycrisp’ and ‘Jonagold’ trees. A slight decrease in return bloom at the earlier application 

date was observed in Honeycrisp but the response did not to a significant threshold (Fig 5.2E). In 

contrast, ‘Jonagold’ exhibited a significant decrease in flowering at both treatment dates (Fig 

5.2F). 

In 2018, GA3 and GA4+7 were reevaluated for effects on flowering in Honeycrisp due to 

the inconsistent results observed during the prior two seasons. For each GA, the number of 

application concentrations and timings were increased (Table 5.3). GA3, when applied at 400 

ppm, was found to be significantly effective at inhibiting flowering when applied at 27 DAFB, 

while also inhibiting flowering at 39 and 56 DAFB, but not significantly (Fig 5.3A-B). This 

result was consistent using both a quantitative count of flowering spurs per branch and a 

qualitative visual rating by a trained professional (Fig 5.3A-B). The 27 DAFB treatment resulted 

in the inhibition of flowering at greater than 50% in the subsequent season compared to the 

control (Fig 5.3A). All of the GA4+7 treatments and the GA3 200 and 300 ppm treatments were 

 

134 

 

found to be ineffective at inhibiting flowering in the spring of 2019 (Fig 5.3C-F). However, the 

GA4+7 500 ppm treatment significantly increased flowering when applied at 39 and 56 DAFB 

(Fig 5.3C). Those promotive effects were less than a 20% increase over the controls.  

Effects of GA on fruit maturity, size, and quality  

In response to the flowering inhibitory effect of GA3 in the 2018 trials, fruit maturity and 

quality was evaluated during the subsequent season’s harvest. The GA3 treatment that resulted in 

inhibited flowering were found to not have an effect on total yield per tree (Fig 5.4A). However, 

when the yield was normalized to trunk diameter (yield efficiency) a significant decrease was 

observed in response to the GA3 application at 27 DAFB (Fig 5.4B). A decrease in the firmness 

was also observed with the treated trees (Fig 5.4C). Juice extracted from the fruit of the treated 

trees was found to have a higher sugar content (Fig 5.4D).  

To evaluate if the GA3 treatment resulted in larger, high-quality Honeycrisp, a 

commercial fruit sorting machine was used to quantify fruit-quality metrics. Firstly, the GA3 

treatment resulted in a significant increase in fruit size relative to the control (Fig 5.5A-C). The 

mass, length, and width were all significantly higher in the fruit of the treated trees regardless of 

application date. A significant increase in the percent red + pink peel coloration was also 

observed in the fruit of the treated trees (Fig 5.5D). Moreover, a higher percentage of fruit that 

met a “high-quality” and “fresh-market grade”, as determined by a commercial sorting machine 

algorithm, were associated with the GA3 treatments (Fig 5.5E-F).  

Flowering in Honeycrisp two years after GA3 treatment 

In the spring of 2020, the trees used in the 2018 experiment were evaluated for variations 

in flowering following a year of no additional treatment or crop load management. ‘Honeycrisp’ 

is known to be an extremely biennial cultivar requiring extensive crop load management 

 

135 

 

(Embree et al. 2007). Thus, it was anticipated that the reduced flowering in 2019 of the treated 

trees would result in greater flowering in 2020. The trees treated in 2018 with GA3 at 27 DAFB 

exhibited a return bloom in the spring of 2020 (Fig 5.6). The 27 DAFB treatment exhibited a 

statistically significant average >25% flowering spurs compared to the control, which was 

relatively absent of bloom (Fig 5.6A). The average flowering rating for the 27 DAFB was also 

statistically significant, with an average rating greater than six (Fig 5.6B).  

DISCUSSION 

In general, the results presented in this study reaffirmed many of the previously reported 

variability in the response to the application of GAs. Both promotive and inhibitory effects on 

flowering were found in response to foliar GA application, regardless of the GA type, timing of 

application, and cultivar. However, the most promising and consistent results was the use of GA3 

to inhibit flowering in the cultivar Honeycrisp. Here, an inhibition of flowering was observed in 

two out of the three years. Although a promising result, the most effective timing of application 

of GA3 varied from year-to-year. This result underscores the need for more long-term studies on 

timing of application.  

The results from the GA3 application in 2018 was the most significant inhibitory effect 

observed. It was also found that the 27 DAFB GA3 treatment did not significantly decrease total 

yield in the following season. However, there was a significant reduction in yield efficiency 

measurement, showing that there was less fruit per cm2 of the trunk cross-sectional area of the 

treated trees. The treated trees were also found to produce larger fruit, which could explain the 

finding that total yield was similar, but the yield efficiency differed. In general, the treated fruit 

were found to be of higher quality compared to the controls. Taken together, these results 

 

136 

 

suggest GA3 could offer a useful practice to controlling flowering in the biennial cultivar 

Honeycrisp when applied at 400 ppm at 27 DAFB.  

To evaluate the potential of adaption of a GA3 management strategy over chemical 

thinning the economics need to be compared. Applying GA3 at a concentration of 400 ppm at a 

rate of .67L/tree will cost $420/acre for a moderately high-density orchard of 1,210 trees/acre 

(12’ x 3’ spacing). At this concentration, a grower can expect to obtain between 70 and 80% 

fresh-market quality graded fruit ($15 - 21/bushel). In comparison, using a blossom thinner will 

cost $40/acre per application with two to three applications expected, for a total cost of $80 - 

$120/acre for the same sized orchard. Using chemical thinners and manual fruit thinning, a 

grower can anticipate obtaining an average of 80% fresh-market quality graded fruit. The GA3 

approach is about 3.5 times more costly than traditional thinning. However, the additional cost of 

manual or chemical fruit thinning to establish appropriate crop loads could raise the costs of 

production closer to that of the GA3 approach.  

 

137 

 

Table 5.1. Commercial GA products tested in 2016 for effectiveness at manipulating floral 
initiation. 
 

Product Name 
(GA type) 

GA 
Concentration 

Application Timing 
(DAFB) 

2016 Visual Bloom 
Density 

ProGibb (GA3) 

200 ppm 

36, 43, 50, 57 

Low 

Provide (GA4+7) 

188 ppm 

36, 43, 50, 57 

Low 

 

 

 

 

138 

 

Table 5.2. Application dates during the 2017 season, the target GDD50 measurement, and 
the actual GDD50 measurement. 

 

Application Dates 

Application Date  5/27/17  6/06/17  6/12/17  6/16/17  6/20/17 

GDD50 Target 

400 

500 

600 

700 

800 

GDD50 Measured  386.8 

507.3 

618.2 

750.4 

806.9 

DAFB 

 

29 

39 

 

45 

49 

53 

 

139 

 

Table 5.3. 2018 PGR application trial design. 

PGR 

GA 

Application 

Application 

Trees/Treatment 

Blocks/Treatment 

Product 

Concentration 

Interval 

Timings 

Block 

Date 

Total 
Trees 

Number of 
Replicate 

Replicate 

75 

18 

15 

60 

15 

15 

Total 
Trees 

15 

 

 

 

18 

12 

ProGibb 

400 ppm 

2 weeks 

4 WAFB – 
12 WAFB 

 

 

300 ppm 

single 

4 WAFB 

application 

200 ppm 

single 

6 WAFB 

application 

Provide 

500 ppm 

2 weeks 

4 WAFB – 
12 WAFB 

 

 

200 ppm 

single 

6 WAFB 

application 

100 ppm 

single 

6 WAFB 

application 

5 

6 

5 

4 

5 

5 

3 

3 

3 

3 

3 

3 

Number of 
Replicate 

Replicate 

Trees/Treatment 

Blocks/Treatment 

Block 

5 

 

 

 

6 

4 

Date 

3 

 

 

 

3 

3 

Control 

Group and 
Associated 
Treatments 

Control 

Groups 

Control 1 

ProGibb 400 

 

 

 

ProGibb 200 

Provide 200 

Provide 100 

Control 2 

ProGibb 300 

Control 3 

Provide 500 

 
 

 

 

 

 

 

 

 
 

 

 

 

 

 

 

140 

 

 

  

  

  

 

Figure 5.1. The percent flowering spurs in 2017 following application of different GAs to 

 

fruit thinned trees in 2016. A) GA3 applied to Honeycrisp. B) GA4+7 applied to Honeycrisp. C) 

GA4+7 applied to Fuji. D) GA4+7 applied to Red Delicious. E) GA4+7 applied to Gala. F) GA4+7 

applied to Jonagold. Black circles represent outliers in the dataset. Student’s t-test , *P < 0.05, ns 

= not significant 

 

141 

 

 

Figure 5.2. The percent flowering spurs in 2018 following application of GA3 to thinned 

trees in 2017. A) GA3 applied to fruit-thinned Honeycrisp. B) GA3 applied to fruit-thinned Fuji. 

C) GA3 applied to fruit-thinned Red Delicious. D) GA3 applied to fruit-thinned Gala. E) GA3 

applied to frost-thinned Honeycrisp in a commercial orchard. F) GA3 applied to frost-thinned 

Jonagold in a commercial orchard. Black circles represent outliers in the dataset. Student’s t-test 

, **P < 0.01, *P < 0.05, ns = not significant 

 

142 

 

 

Figure 5.3. Flowering response in 2019 following application of different GAs to floral 

thinned Honeycrisp trees in 2018. A) Percent flowering spurs following application of GA3 

400 ppm. B) Visual flower rating following application of GA3 400 ppm. C) Percent flowering 

spurs following application of GA4+7 500 ppm. D) Visual flower rating following application of 

GA4+7 500 ppm. E) Percent flowering spurs following application of GA3 200 ppm F) Percent 

flowering spurs following application of different GA4+7 treatments 39 DAFB. Black circles 

represent outliers in the dataset. Student’s t-test, **** P<= 0.0001, ***P < 0.001, *P < 0.05, ns = 

not significant 

 

143 

 

Figure 5.4. Yield and maturity metrics from the 2019 harvest following application of GA3 

to floral thinned Honeycrisp trees in 2018. A) Yield per tree. B) Yield per trunk cross sectional 

area. C) Fruit penetration force upon harvest. D) Sugar content. Black circles represent outliers 

in the dataset. Student’s t-test, **P < 0.01, *P < 0.05, ns = not significant 

 

 

144 

 

 

 

Figure 5.5. Fruit quality measurements from the 2019 harvest following application of GA3 

to floral thinned Honeycrisp trees in 2018. A) Mass. B) Major diameter. C) Minor diameter. 

D) Percent of skin coloration of Red + Pink. E) Frequency of fruit grades. F) Frequency of use 

classification. Black circles represent outliers in the dataset. Student’s t-test, **** P<= 0.0001, 

** P<= 0.01, * P< 0.05, ns = not significant 

 

145 

 

Figure 5.6. Flowering response in 2020 following application of GA3 to floral thinned 

Honeycrisp trees in 2018. A) Percent flowering spurs following application of GA3 400 ppm. 

B) Visual flower rating following application of GA3 400 ppm. Black circles represent outliers in 

 

the dataset. Student’s t-test, **** P<= 0.0001, *** P< 0.001, ns = not significant 

 

 

146 

 

REFERENCES

 

147 

 

REFERENCES 

 
 

Bertelsen MG and Tustin DS. 2002. Suppression of flower bud formation in light cropping trees 
of ‘Pacific Rose’ apple using gibberellin sprays. J. Hort. Sci. and Biotech. 77(6):753-757. 
 
Bertelsen MG,  Tustin DS, and Waagepetersen RP. 2002. Effects of GA3 and GA4+7 on early bud 

development of apple. J. Hort. Sci. and Biotech. 77(1):83-90. 

 
Bukovac, MJ. 1963. Induction of parthenocarpic growth of apple fruits with gibberellins A3 and 

A4. Botanical Gazzette 124(3): 191-195. 

 
Bukovac MJ, Sabbatini P, and Schwallier PG. 2006. Modifying Alternate Bearing of Spur-Type 

'Delicious' Apple with Ethephon. HortScience 41(7):1606-1611. 

 
Davis K, Stover E, and Wirth F. 2004. Economics of Fruit Thinning: A Review Focusing on 

Apple and Citrus HortTechnology 14(2):282-289. 

 
Dennis FG. 2000. The history of fruit thinning. Plant Growth Regulation 31:1-16. 
 
Dennis FG, and LJ Edgerton. 1966. Effects of gibberellins and ringing upon apple fruit 

development and floral bud formation. Proc Amer Soci Hort Sci 88: 14-24. 

 
Embree CG, Myra MD, Nichols DS, and Wright A. 2007. Effect of Blossom Density and Crop 

Load on Growth, Fruit Quality, and Return Bloom in ‘Honeycrisp’ Apple. HortScience 42(7): 
1622-1625. 

 
Foster T, Johnston R, and Seleznyova A. 2003. A morphological and quantitative 

characterization of early floral development in apple (Malus × domestica Borkh.). Annals of 
Botany 92(2): 199-206. 

 
Galimba KD, Bullock DG, Dardick C, Liu Z, and Callahan AM. 2019. Gibberellic acid induced 

parthenocarpic ‘Honeycrisp’ apples (Malus domestica) exhibit reduced ovary width and lower 
acidity. Horticulture Research 6:41.  

 
Guttridge CG. 1962. Inhibition of Fruit-Bud Formation in Apple with Gibberellic Acid.  Nature 

196: 1008. 

 
Ionescu IA, Møller BL, and Sánchez-Pérez R. 2017. Chemical control of flowering time. J. Exp. 

Bot. 68(3):369-382. 

 
Jonkers H. 1979. Biennial Bearing in Apple and Pear: A literature survey. Scientia Horticulturae 

11:303-317. 

 

 

148 

 

Kon TM, Schupp JR, Yoder KS, Combs LD, and Schupp MA. 2018. Comparison of chemical 
blossom thinners using ‘Golden Delicious’ and ‘Gala’ pollen tube growth models as timing 
aids. HortScience 53(8): 1143-1151. 

 
Luckwill LC and Silva JM. 1979. The effects of daminozide and gibberellic acid on flowering 
initiation, growth and fruiting of apple cv. Golden Delicious. J. Horticultural Sci. 54(3):217-
223. 

 
Looney NE, Pharis RP, and Noma M. 1985. Promotion of flowering in apple trees with 

gibberellin A4 and C-3 epi-gibberellin A4. Planta 165:292-294. 

 
McArtney SJ. 1994. Exogenous gibberellin affects biennial bearing and the fruit shape of 
‘Braeburn’ apple. New Zealand Journal of Crop and Horticulture Science 22:343-346. 

 
McArtney SJ and Li S. 1998. Selective Inhibition of Flowering on ‘Braeburn’ Apple Trees with 

Gibberellins. HortScience 33(4):699-700. 

 
McArtney S, Unrath D, Obermiller JD and Green A. 2007. Naphthaleneacetic Acid, Ethephon, 

and Gibberellin A4 + A7 Have Variable Effects on Flesh Firmness and Return Bloom of 
Apple. HortTechnology 17(1):32-38.  

 
Miller SS. 1988. Plant bioregulators in apple and pear culture. Hort. Rev. 10:309-401.  
 
Monselise S and Goldschmidt E. 1982. Alternate bearing in fruit trees. Horticultural Reviews 

4:128-173. 

 
R Core Team. 2017. R: A language and environment for statistical computing. R Foundation for 

Statistical Computing, Vienna, Austria. URL https://www.R-project.org/. 

 
RStudio Team. 2015. RStudio: Integrated Development for R. RStudio, Inc., Boston, MA URL 

http://www.rstudio.com/. 

 
Schmidt T, Elfving DC, Mcferson JR and Whiting MD. 2009. Crop Load Overwhelms Effects of 

Gibberellic Acid and Ethephon on Floral Initiation in Apple. HortScience 44(7):1900-1906. 

 
Schmidt T, McFerson J, Elfving DC, and Whiting M. 2010. Practical Gibberellic Acid Programs 

for Mitigation of Biennial Bearing in Apple. Proc. XIth IS on Plant Bioregulators in Fruit 
Production (Ed. Costa G) Acta Hort. 884:663-670. 

 
Tromp J. 1982. Flower-bud formation in apple as affected by various gibberellins. J. 

Horticultural Sci. 57(3):277-282. 

 
Tromp J. 1987. Growth and flower-bud formation in apple as affected by paclobutrazol, 

daminozide, and tree orientation in combination with various gibberellins J. Horticultural Sci. 
62(4):433-440. 

 

 

149 

 

Wickham H. 2016. ggplot2: Elegant graphics for data analysis. Springer-Verlag, New York.  
 
Zhang S, Zhang D, Fan S, Du L, Shen Y, Xing L, Li Y, Ma J, and Han M. 2016. Effects of 

exogenous GA3 and its inhibitor paclobutrazol on floral formation, endogenous hormones, and 
flowering-associated genes in ‘Fuji’ apple (Malus domestica Borkh.). Plant Physiol Biochem. 
107:178-186. 

 

 

 

150 

 

GENERAL CONCLUSIONS

 

151 

 

This body of work provides a deeper understanding of the molecular mechanisms that 

underlie floral initiation in apple (Malus × domestica Borhk.) and how crop load interacts with 

that pathway to inhibit flowering. The second chapter begins with a census to identify homologs 

of flowering-related genes in apple. I used sequence homology, phylogenetic relationship, and 

syntenic organization to identify apple homologs of highly-studied flowering-related gene 

families. These gene families included AP1, AGL24/SVP, FT/TFL1, FUL, LFY, SOC1, and the 

SPLs. The identified apple homologs were then measured for transcriptional changes in the spur 

apex over the period of floral initiation in non-bearing, flowering-induced trees. Several of the 

homologs exhibited increases in expression over the sampling period, providing evidence of a 

presumed role in floral initiation. I also investigated how the presence of a crop load would 

impact the expression of these genes over the same period. A homolog of TFL1, MdTFL1-2, 

exhibited a strong and persistent upregulation in the apices of bearing trees compared to the non-

bearing. Further investigations into genes that were co-expressed with MdTFL1-2 identified 

potential upstream promoters or downstream targets. These included homologs of FIE, APL, 

EFM, SPL3, and ATH1. A final major finding from this chapter was that genes involved in 

gibberellic acid (GA) biosynthesis and deactivation were differentially expressed between the 

two crop load conditions. In particular, four GA2ox genes that function to deactivate bioactive 

GAs were upregulated by the presence of a crop load. In contrast, several of the GA20ox genes 

that are involved in GA biosynthesis were down-regulated by a crop load.  The results presented 

in this chapter suggest that MdTFL1-2 is associated with the repression of flowering in response 

to crop load. Additionally, the differential regulation of genes involved with the GA pathway 

could indicate a strong GA signal at the spur apex of bearing trees. 

 

152 

 

The third chapter of this dissertation explored the molecular diversity of floral initiation 

and development in apple species and cultivars. Previous research had identified a wide variation 

in spring bloom times of accessions maintained in the USDA Malus collection. Although the 

dormancy requirements of apple are known to play an important role in dictating spring bloom 

time, little attention has been given to documenting other factors that could contribute to the 

variations in spring bloom time. Six accessions from the USDA collection that represent extreme 

early- and late-blooming phenotypes were selected for transcriptional profiling at seasonal 

timings associated with floral initiation and development. In addition to the transcriptional 

profiling, a second census of flowering-related genes was conducted using the high-resolution 

transcriptomes of these Malus accessions. The floral initiation genes were found to be highly 

conserved across the genus, whereas the floral development genes were less conserved. The 

transcriptional profiling identified MdTFL1 genes were differentially expressed among five of 

the six accessions during the presumed period of floral initiation. This differentially regulation of 

MdTFL1 suggests a possible role in the onset of floral initiation in addition to its role in 

repressing flowering in response to a crop load. During the presumed period of floral 

development, the transcriptomes of the six accessions were relatively similar. Of note, homologs 

of SEP1/2 and SEP3 were differentially expressed in one accession that was found to be 

underdeveloped compared to three of the more developmentally advanced accessions. Although 

differences in expression were identified, these differences were not associated with the 

differences in spring bloom times. Based on these results, it can be concluded floral initiation 

and development do not contribute to the variation observed in spring bloom time.  

The fourth chapter of this dissertation investigated the identity of MdFTs and 

characterized their expression. In Arabidopsis, FT is diurnally expressed in the leaf tissues and 

 

153 

 

its translated product is transported to the apex to trigger flowering. I investigated if Arabidopsis 

FT expression was conserved in apple leaves. In other tree fruit species, FT was identified to be 

differentially expressed in the leaves in response to crop load. This result suggests FT has a role 

in the mechanism of biennial bearing. Thus, I also investigated if crop load influences the 

expression patterning of MdFT in leaves. I confirmed that a de novo assembled MdFT2 from the 

second chapter exists in apple as I was able to assemble corresponding DNA sequences that 

shared identity. However, I cannot conclusively determine that the de novo assembled MdFT2 is 

a separate gene or just an allele of MdFT1. To identify tissue-specific expression of MdFT, I 

utilized a gene expression atlas constructed from 80+ different tissues and developmental stages 

of a wild apple species, Malus fusca. MdFT1 was found to be expressed primarily in the apex 

and rachis tissues. I was unable to identify MdFT2 tissue-specific expression, as the de novo 

assembled transcript is absent from the expression atlas. This tissue-specificity of expression 

differs from Arabidopsis as it has little to no expression in the leaves. However, the atlas fails to 

differentiate different leaf types that make up the shoot architecture of apple. Using probes 

specific to the MdFT1 and the de novo MdFT2, I measured the expression of the two MdFTs in 

bourse shoot leaves. These specific leaves are necessary for the induction of flowering in apple, 

suggesting they have a potential role in the flowering pathway. First, I quantified expression over 

24 hours to identify if a diurnal expression pattern was conserved under floral inductive 

conditions. MdFT1 did exhibit a diurnal expression pattern but it was cultivar dependent. In 

addition, there was no significant difference in expression of either MdFT to floral inductive 

conditions (absence vs presence of a crop load). The finding that crop load did not affect the 

expression of either gene was also consistent when investigating for seasonal expression 

differences. The results from this chapter identified a possible MdFT2 gene and demonstrated 

 

154 

 

that MdFT expression is not well-conserved between apple and Arabidopsis. Moreover, MdFT 

was not influenced by crop load in the bourse leaves suggesting that the biennial bearing 

mechanism that is hypothesized for other tree fruit species is not conserved in apple. 

The fifth chapter investigated the efficacy of exogenously applied GAs in managing 

flowering. I demonstrated that Honeycrisp, a high-value and biennial bearing susceptible 

cultivar, was responsive to applications of GA3. Two out of three years, GA was effective in 

repressing the bloom in Honeycrisp when a strong return bloom was expected. As a result of the 

repression of flowering, the treated trees produced larger, higher-quality, and earlier maturing 

fruit, without sacrificing total yield. These results suggest that GA offers a promising strategy to 

control flowering in Honeycrisp. However, the repression of flowering by GA was only observed 

in two out of the three years in this trial, which illustrates the unpredictability associated with 

foliar applications of plant growth regulators. When comparing the costs of a GA regiment vs 

traditional fruit thinning, the GA regiment is about 3.5 times more expensive, which is another 

contributing factor that limits on its adoption as a commercial practice.  

The studies presented in this dissertation highlight the key role that MdTFL1 has in 

inhibiting flowering in apple. In particular, the association between a crop load and the high 

expression of MdTFL1-2 implicates it as the molecular mechanism causing biennial bearing in 

apple. In a previous study from our lab group, MdTFL1-2 was identified as being upregulated in 

response to exogenous applications of GA. This previous finding and my finding that crop load 

influences the expression of GA2ox genes in the apex suggest that there is a strong GA signal in 

spur apex of high crop load trees. This strong GA signal could be the driving force behind the 

high expression of MdTFL1-2 and the resulting repression of flowering in biennial cultivars. 

However, the mechanism by which fruit upregulate the possible GA content in the apex is 

 

155 

 

unknown. Additionally, the work presented in this dissertation does not address the potential 

competitive relationship between the developing fruit/seeds and the shoot apex for macro- or 

micronutrients, and/or sugars. Further experimentation is needed to explore how nutrient 

competition and biennial bearing are associated at the molecular level and if this competition is 

mutually exclusive from the GA theory. In conclusion, the information I present in this 

dissertation illuminates an association between MdTFL1, GA, and crop load as a molecular 

mechanism that drives biennial bearing tendency in apple.  

 
 

 

156