SEARCH

SEARCH BY CITATION

Keywords:

  • bloom date;
  • chilling requirement;
  • endo-dormancy;
  • heat requirement;
  • peach;
  • quantitative trait loci

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
  • Chilling requirement, together with heat requirement, determines the bloom date, which has an impact on the climatic distribution of the genotypes of tree species. The molecular basis of floral bud chilling requirement is poorly understood, despite its importance to the adaptation and production of fruit trees. In addition, the genetic nature of heat requirement and the genetic interrelationships among chilling requirement, heat requirement and bloom date remain unclear.
  • A peach (Prunus persica) F2 population of 378 genotypes developed from two genotypes with contrasting chilling requirements was used for linkage map construction and quantitative trait loci (QTL) mapping. The floral bud chilling and heat requirements of each genotype were evaluated over 2 yr and the bloom date was scored over 4 yr.
  • Twenty QTLs with additive effects were identified for three traits, including one major QTL for chilling requirement and two major QTLs for bloom date. The majority of QTLs colocalized with QTLs for other trait(s). In particular, one genomic region of 2 cM, pleiotropic for the three traits, overlapped with the sequenced peach EVG region.
  • This first report on the QTL mapping of floral bud chilling requirement will facilitate marker-assisted breeding for low chilling requirement cultivars and the map-based cloning of genes controlling chilling requirement. The extensive colocalization of QTLs suggests that there may be one unified temperature sensing and action system regulating chilling requirement, heat requirement and bloom date together

Abbreviations:
AFLP

amplified fragment length polymorphism

ANOVA

analysis of variance

BD

bloom date

CH

chilling hour

CI

confidence interval

CR

chilling requirement

CU

chilling unit

EVR

Evergrowing

G

linkage group

GDH

growing degree hour

HR

heat requirement

LOD

logarithm of the odds

QTL

quantitative trait locus

r

Spearman correlation coefficient

Sh/sh

non-showy/showy flower

SSR

simple sequence repeat

T × E Prunus reference map

almond (cv. Texas) × peach (cv. Earlygold) map

Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Temperate tree species have the ability to cease meristem activity in the fall and establish a dormant state (endo-dormancy or true dormancy) in which the meristem is rendered insensitive to growth-promoting signals before it is released (Rohde & Bhalerao, 2007). The chilling requirement (CR) refers to the duration of low temperatures required for the release of temperate trees from endo-dormancy. CR prevents trees from initiating growth in response to transient warm temperatures, thus avoiding damage by subsequent frost(s) in the late winter or early spring. CR is the result of long-term climatic adaption of genotypes of tree species developed in different regions. Conversely, it limits the climatic distributions of the genotypes of temperate fruit trees (Sherman & Beckman, 2003). CR is the major factor determining the bloom date (BD, also referred to as the flowering time) (Egea et al., 2003; Ruiz et al., 2007; Alburquerque et al., 2008), which is an important agronomic trait affecting the seed and fruit development of temperate fruit tree species. Genotypes with low CR bloom early in cold regions/years and are susceptible to late frost damage (Scorza & Okie, 1990). Genotypes with high CR could suffer inadequate chilling in warm regions/years, resulting in irregular floral and leaf bud break and thus poor fruit set, which is potentially problematic with the current global warming trend (Topp et al., 2008). On the other hand, in temperate fruit tree species, early ripening cultivars are often preferred because of better early market prices for their fruits (Ruiz et al., 2007; Topp et al., 2008). Breeding for earlier BD (often associated with low CR) is one approach to obtaining earlier ripening fruit with adequate size.

Heat requirement (HR) is another factor determining the BD of cultivars of temperate tree species (Richardson et al., 1974; Citadin et al., 2001). It is unclear whether heat accumulation for floral or vegetative bud break starts before or after the release of endo-dormancy. It has also been reported that extended chill (more than CR) results in a reduction in HR of tree buds (Scalabrelli & Couvillon, 1986; Citadin et al., 2001; Harrington et al., 2009). These two issues complicate the quantification of the variation of HR among different genotypes. Currently, the growing degree hour (GDH) model developed by Richardson et al. (1975) is the most widely used (Citadin et al., 2001; Egea et al., 2003; Ruiz et al., 2007; Alburquerque et al., 2008), but it only counts the heat accumulation from endo-dormancy release to full bloom.

Among the three interrelated traits, BD is considered to be quantitatively inherited in most fruit tree species (Anderson & Seeley, 1993), CR is considered to be semi-qualitatively inherited in apple (Malus × domestica Borkh.) (Hauagge & Cummins, 1991), and no study has yet been reported on the genetic nature of HR.

Quantitative trait locus (QTL) mapping results for BD in various genomic regions in Prunus has been reported. Using the terminology of the almond (Prunus amygdalus L.) (cv. Texas) × peach [Prunus persica (L.) Batsch] (cv. Earlygold) map (T × E Prunus reference map) on linkage groups (G), four QTLs on G1, G4, G6 and G7 were detected by Joobeur (1998) in an almond × peach F2 population, two QTLs on G2 and G7 by Dirlewanger et al. (1999) in a peach F2 population, one major gene (Late blooming) on G4 by Ballester et al. (2001) in an almond F1 population, and one QTL on G4 by Verde et al. (2002) in a peach backcross (BC1) population. A candidate gene approach associated only two of ten candidate genes homologous to LEAFY and MADS-box genes in Arabdopsis with two QTLs in almond (Silva et al., 2005), suggesting that direct application of the knowledge of the genetic control of the flowering time of annual plants to perennial tree species may be more complicated than expected.

There have been no reported results on the successful mapping of QTLs associated with CR for floral bud break in temperate tree species. However, two genetic studies have indicated that CR is in control of at least one major gene with dominant low CR allele(s) (Hauagge & Cummins, 1991;Tzonev & Erez, 2003). With regard to HR, almost no genetic studies have been reported. It is even unclear whether HR is an intrinsic characteristic in several fruit tree species (Couvillon & Erez, 1985; Ruiz et al., 2007).

In temperate and subtropical regions, peach is widely grown and economically important. As a proposed tree model species (Abbott et al., 2002), its self-compatibility and short generation cycle (2–3 yr) enable the relatively easy development of true F2 populations and early characterization of floral and seed-related traits. Its diploidy and the availability of a large number of mapped simple sequence repeat (SSR) markers transferable within Prunus greatly facilitate linkage map construction. The small genome size (c. 220 Mbp; B. Sosinski, North Carolina State University, Raleigh, pers. comm.) and extensive genomics/genetics resources available at the Genome Database for Rosaceae website (http://www.bioinfo.wsu.edu/gdr/) enable the map-based cloning and annotation of genes controlling important agronomic traits for tree arboriculture, and the development of markers inside or tightly linked with these genes for marker-assisted breeding applications. However, to achieve these goals, it is critical to have a detailed resolution of the location of the genomic regions (QTLs) harboring these genes.

The major objective of this research was to identify QTLs associated with CR and CR-related traits using a peach F2 population derived from two genotypes with contrasting CR values: the high CR cv. ‘Contender’ and the low CR selection ‘Fla.92-2C’. The F2 progenies segregate in a continuous fashion for a variety of traits, including CR, HR and BD. Utilizing this mapping population, we obtained the first data on the genomic regions (QTLs) determining floral bud CR and HR, and provided the first picture of potential genetic interrelationships among CR, HR and BD in temperate tree species.

Materials and Methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Plant materials

A peach [Prunus persica (L.) Batsch] F2 population with 378 different genotypes was developed at ARS-USDA, Southern Fruit and Tree Nut Research Laboratory (Byron, GA, USA) by crossing two peach genotypes with high and low CR values and selfing the resultant F1 hybrid ‘BY01p6245’. The female grandparent ‘Contender’ is a commercial peach cultivar in the southeastern USA developed by the North Carolina Agricultural Service (Raleigh, NC, USA) and requiring c. 1050 chilling hours (CHs) of CR. The male grandparent ‘Fla.92-2C’ is a selection from the University of Florida’s (Gainesville, FL, USA) low chilling peach breeding program requiring c. 300 CHs of CR. Both grandparents have ‘Candor’ and ‘Pekin’ as distant ancestors in their pedigrees. F2 seeds were stratified, germinated and pot-planted in a glasshouse in 2003 and transplanted to Clemson University’s Musser Fruit Research Center (Seneca, SC, USA) in 2004. Three to four clones of each genotype were made by rooting the shoot cuttings from seedling trees, and planted in a second plot at the same site in 2006. This population segregates for multiple quantitative traits, including CR, HR and BD. It also segregates for ripening date and the qualitative trait non-showy/showy flower (Sh/sh) in a 3 : 1 ratio.

Phenotyping

Chilling requirement (CR)  For deciduous fruit trees, two methods are routinely employed to determine when the CR is fulfilled for blooming. One is to expose the cuttings harvested from different stages to a controlled warm condition for a period of time, with subsequent scoring of the status of floral bud break (Gibson & Reighard, 2002). Another is to measure and compare the weight of floral buds before and after these cuttings are exposed to a warm condition for a period of time (Tabuenca, 1964). Because of the necessity of large-scale rapid screening, the first method was used in this study.

Floral bud CR data for the F2 population obtained in winter 2007/spring 2008 and winter 2008/spring 2009 were designated as CR2008 and CR2009, respectively. Average temperatures in 10 min intervals were continuously recorded by the temperature data loggers placed in the canopy of the experimental trees, starting in the middle of October, when the air temperature drops to below 7.2°C, and ending in late March of the next year. The < 7.2°C (Weinberger, 1950) model was chosen to determine the times to sample the branches and to evaluate the chilling fulfillment. The number of hours below 7.2°C (CHs) was counted. Starting with the time of 300 CHs, the branches of each genotype were harvested approximately every 100 CHs until the time of 1000 CHs (2007/2008) or 1100 CHs (2008/2009). For each genotype, three clones grown in natural field conditions were sampled and three branches (generally longer than 40 cm and populated with floral buds) were taken from each clone. Branch cuttings were placed into 1%‘Floralife (Fresh Flower Food)’ solution (Floralife, Inc., Walterboro, SC, USA) in a glasshouse at 25°C to force floral bud break under a 16 h photoperiod. After 14 d, the progression of floral bud break of the branches was evaluated. A genotype’s CR was considered to be satisfied at a specific sampling time if 50% of the floral buds on the branch cuttings opened (pink stage).

After CR evaluation based on the < 7.2°C model had been completed, CR of each genotype was recalculated based on the 0–7.2°C model (Eggert, 1951), the Utah model (Richardson et al., 1974), the Low Chill model (Gilreath & Buchanan, 1981) and the Dynamic model (Fishman et al., 1987; Erez et al., 1988).

Chilling accumulations calculated by different models on each sampling date in the years (winter/spring) 2007/2008 and 2008/2009 are listed in Table S1 (see Supporting Information).

Heat requirement (HR)  Floral bud HR data for the F2 population obtained in winter 2007/spring 2008 and winter 2008/spring 2009 were designated as HR2008 and HR2009, respectively. HR of each F2 genotype was evaluated with the GDH model developed by Richardson et al. (1975). GDHs for a specific genotype were determined by subtracting 4.5°C (below which no growth or development of peach buds occurs) from the hourly temperature, and accumulating the balance from the time of CR completion to full bloom. Temperatures above 25°C were treated as 25°C because of no extra heat benefit for the tree (Anderson et al., 1986).

Bloom date (BD)  The BD of each F2 genotype was evaluated as the date at which 50% of the floral buds reached the full bloom stage in the spring of 2006, 2007, 2008 and 2009. For each genotype, the whole tree of one clone was observed every 1 or 2 d in the spring to determine BD. BD was recorded and analyzed as the number of days from January 1st to the date of bloom.

Non-showy/showy flower (Sh/sh)  Sh/sh was evaluated in the spring of 2006 as two classes: non-showy (flower with small petals, dominant) and showy (flower with large petals, recessive).

Statistical analysis of phenotypic data

Statistical analyses of the phenotypic data were performed with the Statistical Analysis System (SAS) 9.2 package (SAS Institute Inc., Cary, NC, USA). The ‘UNIVARIATE’ procedure of SAS was used to test for the normality of phenotypic data distributions. The ‘CORR’ procedure of SAS was used to test for correlations between different traits. The Spearman correlation coefficient (r) from the SAS output was chosen because of the non-normal distribution of all traits. The range of ‘r’ was interpreted empirically: the correlation between two variables was considered to be ‘weak’ if ‘r’ = 0–0.3, ‘moderate’ if ‘r’ = 0.31–0.7 and ‘high’ if ‘r’ = 0.71–1.0.

Genotyping

SSR markers  A set of 370 SSRs isolated from different Prunus species was tested for polymorphism in the F2 mapping population using the female grandparent ‘Contender’ and the F1 tree ‘BY01p6245’. The origins and references of these SSRs are listed in Table S2 (see Supporting Information). Segregation analysis was carried out in the entire F2 population for polymorphic SSR markers with clear segregation patterns as outlined in Zhebentyayeva et al. (2003), with preference for those mapped onto the T × E Prunus reference map (Dirlewanger et al., 2004) and peach ‘bin map’ (Howad et al., 2005).

Amplified fragment length polymorphism (AFLP) markers  AFLP marker analysis was essentially performed as outlined in Vos et al. (1995). In total, 206 EcoRI/MseI primer combinations were tested for polymorphism in the F2 population with the female grandparent ‘Contender’ and F1 tree ‘BY01p6245’. Selective amplification was performed using an EcoRI-end primer with two selective nucleotides and an MseI-end primer with three selective nucleotides. Segregation analysis was then carried out in the entire F2 population for the primer combinations showing polymorphisms and clear segregation patterns. Following the manufacturer’s manual, the size of AFLP fragments was determined by the DNA ladders generated from the fmol DNA Cycle Sequencing System (Promega Corporation, Madison, WI, USA). A dominant AFLP marker was named EXXMYYY(a) and a codominant AFLP marker EXXMYYY(a/b), with ‘XX’ being the selective nucleotides for EcoRI-end primers, ‘YYY’ the selective nucleotides for MseI-end primers, and ‘a’ or ‘b’ the number of base pairs of AFLP fragment(s).

Map construction

Genetic mapping of the F2 population was performed using JoinMap 3.0 software (Van Ooijen & Voorrips, 2001). Kosambi’s mapping function was applied for map distance calculation (Kosambi, 1944). Segregation distortion of individual markers was revealed by the chi-squared test of JoinMap. Markers showing skewed segregation (< 0.05) were still utilized for mapping after the verification of the genotypic data. Linkage groups (G) were constructed and marker order was determined using the default parameters of JoinMap. Only marker order and distances generated by the first or second run of mapping were adopted. Finally, the name and orientation of all linkage groups, except G4, were dictated by the T × E Prunus reference map (Dirlewanger et al., 2004) based on the SSR markers shared by two maps (Fig. S1, see Supporting Information). G4 shared only one SSR marker with the Prunus reference map and its orientation was dictated by the peach ‘bin’ map (Howad et al., 2005).

The Sh/sh trait was mapped as a dominant phenotypic marker as it segregates in a 3 : 1 (non-showy : showy) ratio. Generally, SSRs were scored and mapped as codominant markers, and AFLPs as dominant markers. In the case of possible multiloci SSR markers or codominant AFLP markers, all separated PCR bands were first scored as dominant markers and processed by JoinMap 3.0 together with other markers. In dominant scoring, if a pair of PCR bands from the same primer combination was mapped to the same locus, the pair was considered to be allelic, and then rescored and mapped as a codominant marker. The SSR marker names standardized in the Genome Database for Rosaceae website (http://www.bioinfo.wsu.edu/gdr/) were adopted. In the case of multiloci SSR markers amplified with the same pair of primers, a capital letter was added to the end of the marker name for each locus. The selection of letters was consistent with that for the T × E Prunus reference map (Dirlewanger et al., 2004), if these markers had also been mapped on it.

QTL analysis

Composite interval mapping (Jansen & Stam, 1994; Zeng, 1994) was performed using PLABQTL version 1.2bic (Utz & Melchinger, 2006), a QTL mapping software based on a multiple regression approach with flanking markers described by Haley & Knott (1992).

Different years of phenotypic data for the same trait were analyzed separately. Cofactors (markers best accounting for QTL effects) for QTL mapping in each trait were selected by a stepwise regression procedure. A pure additive model for each trait was chosen by fitting phenotypic and marker data with different gene action models (different combinations of additive, dominance and epistatic effects) and selecting the model with the minimal Bayesian Information Criterion value after the ‘final simultaneous fit’ procedure (simultaneous multiple regression using all detected QTLs and their estimated positions). The threshold of the logarithm of the odds (LOD, 2.85) for QTL detection at a genome-wise error rate of 5% was obtained by 1000 iterations of permutation tests for all traits. LOD curves were created by scanning every 1 cM of the genome.

Once all parameters for composite interval mapping were set, the ‘final simultaneous fit’ procedure was carried out again to obtain final estimates of the additive effects for each QTL, the proportion of the phenotypic variance explained by each QTL (partial R2) and the proportion of the phenotypic variance explained by all QTLs with adjustment for the number of QTL terms in the full regression model (adjusted R2) (Hospital et al., 1997). The additive effect is half of the difference between two homozygotes. The allele from the low CR male grandparent of the F2 population was assumed to be superior. If it was actually weaker, a negative additive effect was assessed. The additive effects divided by the phenotypic standard deviation (standardized additive effects) were reported. The partial R2 for each QTL term was calculated as the change in R2 of the regression model with that term removed from the model: partial R2 = [R2 (full model)−R2 (reduced model)]/[1 − R2 (reduced model)]. It should be noted that the denominator of the formula above is different for each partial R2 calculated. Therefore, the partial R2 value will not sum to the adjusted R2 value for the full model (Utz, 2000; Wassom et al., 2008).

The two-way analysis of variance (ANOVA) for the genotype × environment interaction was performed with multiple years of phenotypic data of each trait by the ‘QTL-ANOVA’ procedure of PLABQTL. Broad-sense heritability (H2) and mean squares from different sources (genotypes, genotype × environment, etc.) were reported based on the PLABQTL output. Mean squares from environmental sources were calculated manually according to the method described by Lynch & Walsh (1998).

One- or two-LOD intervals (c. 95% or 99% confidence interval) (Lynch & Walsh, 1998) for QTL detection were reported. The QTL graphs were prepared using MapChart 2.2 (Voorrips, 2002). The QTLs with partial R> 30% were arbitrarily declared as major QTLs.

A detected QTL is named qXXYa-ZZZZ, with ‘XX’ being the trait acronym, ‘Y’ the number of linkage groups on which the QTL was detected, ‘a’ the letter specifying different QTLs if more than one QTL was detected for the same trait on one linkage group, and ‘ZZZZ’ the year in which the trait was phenotyped.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Distribution and correlation analysis of phenotypic data and heritability

Both years (2007/2008 and 2008/2009) of CR data of the F2 population showed bimodal distributions, but the bimodality of CR2008 was more obvious (Fig. 1a,c). Both CR2008 and CR2009 were right skewed, that is low CR genotypes dominate the F2 population. CRs evaluated by the different models highly (or perfectly) correlated with each other (= 1; < 0.001) in both years. The 2 yr of CRs were highly correlated (= 0.723; < 0.001) (Table 1).

image

Figure 1.  Frequency distributions of chilling requirement (CR) and heat requirement (HR) for floral bud break in the Contender × Fla.92-2C peach population. (a, c) CR evaluated in year (winter/spring) 2007/2008 (CR2008) and 2008/2009 (CR2009) with c. 100 chilling hours interval and the < 7.2°C model. (b, d) HR evaluated in year 2007/2008 (HR2008) and 2008/2009 (HR2009) with the growing degree hour (GDH) model.

Download figure to PowerPoint

Table 1.   Spearman correlation coefficients (r) of chilling requirement (CR, < 7.2°C model), heat requirement (HR) and bloom date (BD) in the Contender × Fla.92-2C peach population in different years
 CR2009HR2008HR2009BD2007BD2008BD2009
  1. All correlations are significant (< 0.001), except for that between HR2008 and BD2008 (= 0.793). CR2008 and CR2009, CR data obtained in winter 2007/spring 2008 and winter 2008/spring 2009, respectively; H2008 and HR2009, HR data obtained in winter 2007/spring 2008 and winter 2008/spring 2009, respectively.

CR20080.723−0.653  0.698 
CR2009  −0.820  0.672
HR2008  0.379 −0.014 
HR2009     −0.188
BD2006   0.7380.7350.704
BD2007    0.8310.784
BD2008     0.821

Both years (2007/2008, 2008/2009) of HRs showed single peaks, but skewed distributions (Fig. 1b,d). The 2 yr of HRs were moderately correlated (= 0.379; < 0.001) (Table 1).

All 4 yr of BDs showed multimodal distributions (Fig. 2). The ranges of BDs varied from 16 d (year 2006) to 53 d (year 2007). The distribution of BD was right skewed in year 2006 and left skewed in the other years. The 4 yr of BDs were highly correlated with each other (= 0.704–0.831; < 0.001) (Table 1).

image

Figure 2.  Frequency distributions of bloom dates (BDs) in the Contender × Fla.92-2C peach population scored in year 2006 (a), 2007 (b), 2008 (c) and 2009 (d).

Download figure to PowerPoint

In addition, the Kolmogorov–Smirnov test by the ‘CORR’ procedure of SAS 9.2 also confirmed that the distributions of all three phenotypic traits departed significantly (< 0.01) from normality.

Both years of CRs (< 7.2°C model) were moderately correlated with BDs (= 0.698, 0.672; < 0.001) and moderately or highly correlated with HRs (= −0.653, −0.820; < 0.001); the correlation with BD was positive and that with HR was negative. HRs showed nonsignificant (year 2008, = −0.014, > 0.793) or weak (year 2009, = −0.188, < 0.001) correlations with BDs (Table 1).

The broad-sense heritability (H2) was 79.5% for CR (< 7.2°C model), 54.0% for HR and 85.2% for BD (Table S3, see Supporting Information).

Linkage map

A linkage map composed of 96 SSR markers (six of which were dominant), 30 AFLPs (four of which were codominant) and one phenotypic marker (Sh) was constructed. Markers were organized into eight linkage groups that are consistent with the number of chromosomes in the peach genome. G1 covers the longest genetic distance of 96.4 cM, whereas G3 covers the shortest genetic distance of 51.7 cM. The total map length of 535 cM was established, corresponding to an average interval of 4.2 cM between adjacent markers. As a result of a lack of segregating markers in certain genomic regions, there are three gaps of 24–29 cM in G2, G4 and G5 (Fig. 3). Marker orders in each linkage group were in good agreement with those in the T × E Prunus reference map, with a few minor differences detected. Of the 36 SSRs shared by the two maps, 32 were mapped in the same linkage groups and orders with those on the reference map. Two more SSRs (pchgms3 and CPPCT026) were mapped in the same region in G1, but with a different orientation (Fig. S1). The agreement with the reference map implies high quality for the newly constructed linkage map, and forms a solid basis for further QTL analysis.

image

Figure 3.  Location of quantitative trait loci (QTLs) for chilling requirement (CR), heat requirement (HR) and bloom date (BD) on the Contender × Fla.92-2C peach map. The solid and whisker parts of the vertical bars next to the linkage groups (G) indicate one-logarithm of the odds (LOD) intervals (c. 95% confidence intervals) and two-LOD intervals (c. 99% confidence intervals) of QTLs for different traits, which are differentiated by the styles of the solid parts of the bars: filled black for CR, crosshatch for HR and open for BD. A QTL is named as qXXYa-ZZZZ, with ‘XX’ being the trait acronym, ‘Y’ the number of the linkage group, ‘a’ the letter to specify different QTLs for the same trait in one linkage group (G), and ‘ZZZZ’ the year in which the trait was phenotyped. Markers with names followed by ‘*’ have significantly distorted genotypic ratios (< 0.05). The highlighted fragment of G1 covers the Evergrowing (EVG) locus (Bielenberg et al., 2008).

Download figure to PowerPoint

Most loci (77.8%) exhibited genotype ratios as expected for a segregating F2 population (1 : 2 : 1 for codominant markers or 3 : 1 for dominant markers). Among 28 markers with significantly skewed genotypic ratios (< 0.05), a cluster of 17 was mapped in G1 from 68 cM to the end of the group, with an overrepresentation of the alleles inherited from the low CR male grandparent; the other 11 were randomly distributed onto G1, G2, G3, G6, G7, G8 (Fig. 3).

Mapping QTLs

QTLs for CR  Using the < 7.2°C CR evaluation model, in both years, four QTLs (qCR1a, qCR4b, qCR5, qCR7) were detected in the same or largely overlapping genomic regions and considered to be the same QTLs. Among these, qCR1a and qCR7 showed very prominent effects. qCR1a explained 40.5–44.8% of the phenotypic variance and was declared to be a major QTL. qCR7 explained 17.8–24.9% of the phenotypic variance (Table 2, Fig. 3). Additionally, 4 yr-specific QTLs were detected for CR, explaining 4.2–9.7% of the phenotypic variance (Table 2, Fig. 3).

Table 2.   Quantitative trait loci (QTLs) detected for chilling requirement (CR) calculated by the different CR models with the Contender × Fla.92-2C peach population in different years (winter/spring)
YearQTLG/PosCICofactorLODPart R2 (%)AddR2 (%)
  1. G/Pos, linkage group/QTL position (cM); CI, two-LOD or c. 99% confidence interval (cM); LOD, logarithm of the odds; Part R2 (%), percentage of phenotypic variance explained by one QTL when other QTL effects are fixed; Add, additive QTL effect divided by the SD of the trait value, the male grandparent is assumed to carry the superior QTL allele; R2 (%), percentage of the phenotypic variance explained by all QTLs with the adjustment for the number of QTL terms in the full regression model.

2007/2008
< 7.2°CqCR1a-2008G1/8786–88Pchgms2944.5244.8−0.8355.7
qCR4a-2008G4/74–19ssrPaCITA69.779.7−0.32
qCR4b-2008G4/5441–62AMPA1032.954.1−0.20
qCR5-2008G5/3224–38ssrPaCITA213.794.5−0.20
qCR6-2008G6/4135–43EPPISF0023.294.20.19
qCR7-2008G7/4843–59UDAp-409A16.9517.8−0.46
qCR8-2008G8/5139–54PacC133.604.4−0.20
0–7.2°CqCR1a-2008G1/8786–88Pchgms2944.8742.6−0.8153.9
qCR4a-2008G4/51–12MD205a9.008.9−0.31
qCR4b-2008G4/5441–62AMPA1034.274.4−0.21
qCR5-2008G5/3219–35ETGMCAG(80)3.734.3−0.21
qCR6-2008G6/4135–43EPPISF0023.143.80.18
qCR7-2008G7/4944–60UDAp-409A15.8917.1−0.45
qCR8-2008G8/5141–54PacC134.485.2−0.23
Low ChillqCR1a-2008G1/8786–88Pchgms2946.0843.5−0.8155.0
qCR4a-2008G4/60–12MD205a8.186.4−0.26
qCR4b-2008G4/5041–62AMPA1033.904.0−0.22
qCR5-2008G5/3424–38ssrPaCITA214.346.1−0.23
qCR7-2008G7/4944–57UDAp-409A19.5420.1−0.50
qCR8-2008G8/5140–54PacC134.575.2−0.22
UtahqCR1a-2008G1/8786–88Pchgms2947.6243.6−0.8155.6
qCR4a-2008G4/60–12MD205a8.747.9−0.29
qCR4b-2008G4/5841–62AMPA1034.114.2−0.21
qCR5-2008G5/3424–38ssrPaCITA214.236.0−0.22
qCR7-2008G7/4944–57UDAp-409A20.0420.4−0.50
qCR8-2008G8/5141–54PacC134.815.4−0.23
DynamicqCR1a-2008G1/8786–88Pchgms2949.5444.7−0.8256.3
qCR4a-2008G4/61–12MD205a9.268.2−0.29
qCR4b-2008G4/5841–62AMPA1034.354.5−0.22
qCR5-2008G5/3324–38ssrPaCITA214.436.2−0.23
qCR7-2008G7/4944–57UDAp-409A19.8920.1−0.49
qCR8-2008G8/5141–54PacC134.475.0−0.22
2008/2009
< 7.2°CqCR1d-2009G1/60–13UDA-0536.717.6−0.2954.3
qCR1a-2009G1/8786–88pchgms4018.3740.5−0.78
qCR4b-2009G4/4640–55M12a2.925.9−0.26
qCR5-2009G5/3324–38ssrPaCITA213.964.6−0.20
qCR7-2009G7/4743–51CPPCT03325.6824.9−0.58
0–7.2°CqCR1d-2009G1/60–13UDA-0536.757.7−0.2954.5
qCR1a-2009G1/8786–88pchgms4018.6540.8−0.78
qCR4b-2009G4/4540–55M12a2.915.9−0.25
qCR5-2009G5/3424–38ssrPaCITA214.124.7−0.20
qCR7-2009G7/4743–51CPPCT03325.7625.0−0.58
Low ChillqCR1a-2009G1/8786–88pchgms407.9540.5−0.8150.2
qCR5-2009G5/3424–38ssrPaCITA213.403.3−0.18
qCR7-2009G7/4844–52CPPCT03324.2822.0−0.56
qCR8-2009G8/5236–54PacC132.872.6−0.17
UtahqCR1d-2009G1/50–13UDA-0536.587.9−0.2955.0
qCR1a-2009G1/8786–88pchgms4019.1941.4−0.79
qCR4b-2009G4/4640–55M12a3.555.5−0.25
qCR5-2009G5/3424–38ssrPaCITA214.004.8−0.20
qCR7-2009G7/4844–52CPPCT03325.5225.2−0.58
DynamicqCR1d-2009G1/50–13UDA-0536.478.0−0.2955.9
qCR1a-2009G1/8786–88pchgms4019.0142.3−0.79
qCR4b-2009G4/4640–55M12a3.466.2−0.26
qCR5-2009G5/3424–38ssrPaCITA213.874.6−0.20
qCR7-2009G7/4844–52CPPCT03325.4626.3−0.59
qCR8-2009G8/5336–54PacC133.113.4−0.18

The full regression model for CR QTLs explained 55.7% and 54.3% of the phenotypic variance in each year, respectively (Table 2).

CR2008 calculated by five different CR models was subjected to QTL analysis and yielded very similar results, except that, when the < 7.2°C and 0–7.2°C models were used, one more QTL (qCR6-2008) was detected. qCR6-2008 displayed a minor effect, only explaining 4.2% (< 7.2°C model) or 3.8% (0–7.2°C model) of the phenotypic variance (Table 2). When the other three models were used, LOD peaks in the position of qCR6-2008 were also found. Because the peak values (1.97, 2.42 or 2.67) were lower than the significant LOD threshold of 2.85, it was mis-detected. Besides qCR6-2008, the other six QTLs showed very similar two-LOD confidence intervals (CIs), LOD peak scores and proportions of explained phenotypic variance with all five CR models.

CR2009 calculated by the different CR models also yielded very similar QTL mapping results, except that two minor QTLs (qCR4b-2009 and qCR8-2009) were not consistently detected when different CR models were used (Table 2).

QTLs for HR  In both years, qHR1 was detected in overlapping genomic regions and was considered to be the same QTL. qHR1 explained 7.1% and 11.2% of the phenotypic variance in years 2007/2008 and 2008/2009, respectively. Another QTL was detected only in year 2007/2008, explaining 3.1% of the phenotypic variance (Table 3, Fig. 3). The full regression models for HR QTLs explained 8.6% and 10.7% of the phenotypic variance in years 2007/2008 and 2008/2009, respectively (Table 3).

Table 3.   Quantitative trait loci (QTLs) detected for heat requirement (HR) with the Contender × Fla.92-2C peach population in different years (winter/spring)
YearQTLG/PosCICofactorLODPart R2 (%)AddR2 (%)
  1. G/Pos, linkage group/QTL position (cM); CI, two-LOD or c. 99% confidence interval (cM); LOD, logarithm of the odds; Part R2 (%), percentage of phenotypic variance explained by one QTL when other QTL effects are fixed; Add, additive QTL effect divided by the SD of the trait value, the male grandparent is assumed to carry the superior QTL allele; R2 (%), percentage of the phenotypic variance explained by all QTLs with the adjustment for the number of QTL terms in the full regression model.

2007/2008qHR1-2008G1/8786–89pchgms296.067.10.378.6
qHR8-2008G8/5036–54PacC132.943.10.25 
2008/2009qHR1-2009G1/8786–88Pchgms407.8111.20.4710.7

QTLs for BD  In all 4 yr, four QTLs for BD (qBD1a, qBD2, qBD4 and qBD7a) were detected in the same or largely overlapping genomic regions and considered to be the same QTLs. Among these, qBD1a and qBD7a were two QTLs having very prominent effects. Except for qBD1a in year 2006, both QTLs explained more than 30% of the phenotypic variance in different years, and were declared to be major QTLs. qBD4 also explained a relatively large proportion of the phenotypic variance, ranging from 8.5 to 19.9% (Table 4, Fig. 3). In 2 yr, one QTL (qBD5) was detected in the same genomic region on G5 and also considered to be the same QTL. In addition, 5 yr-specific QTLs were detected for BD, explaining 3.5–12.8% of the phenotypic variance in different years (Table 4, Fig. 3).

Table 4.   Quantitative trait loci (QTLs) detected for bloom date (BD) with the Contender × Fla.92-2C peach population in different years (springs)
YearQTLG/PosCICofactorLODPart R2 (%)AddR2 (%)
  1. G/Pos, linkage group/QTL position (cM); CI, two-LOD or c. 99% confidence interval (cM); LOD, logarithm of the odds; Part R2 (%), percentage of phenotypic variance explained by one QTL when other QTL effects are fixed; Add, additive QTL effect divided by the SD of the trait value, the male grandparent is assumed to carry the superior QTL allele; R2 (%), percentage of the phenotypic variance explained by all QTLs with the adjustment for the number of QTL terms in the full regression model.

2006qBD1b-2006G1/5043–56Pchgms312.7112.8−0.4252.0
qBD1a-2006G1/8786–88Pchgms4015.7815.2−0.45
qBD2-2006G2/2720–31ECAMCCG(99)3.884.5−0.26
qBD4-2006G4/116–28ssrPaCITA66.759.5−0.31
qBD7a-2006G7/4443–47CPPCT03331.2730.4−0.66
2007qBD1c-2007G1/3327–42UDP-0057.8410.0−0.2673.1
qBD1a-2007G1/8786–88Pchgms4011.9549.4−0.73
qBD2-2007G2/3123–37EPPCU4962A3.354.5−0.16
qBD4-2007G4/94–23ssrPaCITA618.1619.6−0.37
qBD7b-2007G7/1813–22CPPCT0223.903.5−0.17
qBD7a-2007G7/4443–47CPPCT03324.9041.5−0.72
2008qBD1d-2008G1/00–1CPPCT10B2.944.0−0.1474.1
qBD1a-2008G1/8786–89Pchgms2934.5654.5−0.77
qBD2-2008G2/2220–31ECAMCCG(99)6.778.6−0.23
qBD4-2008G4/105–24ssrPaCITA616.4319.9−0.36
qBD5-2008G5/2724–38ssrPaCITA213.374.2−0.16
qBD6-2008G6/3534–42UDP-4123.434.00.14
qBD7a-2008G7/4341–44UDAp-46045.9555.2−0.81
2009qBD1a-2009G1/8786–88Pchgms4024.3441.3−0.7658.0
qBD2-2009G2/2320–31ECAMCCG(99)6.527.0−0.26
qBD4-2009G4/218–33ssrPaCITA68.178.5−0.33
qBD5-2009G5/3124–38ssrPaCITA214.916.4−0.24
qBD7a-2009G7/4340–44UDAp-46032.5432.6−0.65

The approximate locations of BD QTLs in the T × E Prunus reference map, detected in this and previous studies in Prunus, are shown in Fig. S2 (see Supporting Information). Among ten BD QTLs detected in this study, four (qBD2, qBD4, qBD7b-2007, qBD7a) have overlapping intervals with previously reported QTLs, with two on G1 (qBD1c-2007, qBD1d-2008) closely flanking a previously reported QTL. The other two QTLs, qBD1a [overlapping with the Evergrowing (EVG) locus] and qBD5, were in similar positions to two QTLs poster-reported by Howad & Arús at the 2007 Plant & Animal Genome XV Conference (not shown in Fig. S2). No QTL found in this study harbors the Late blooming locus (Fig. S2).

The full regression models for BD QTLs explained 52–74.1% of the phenotypic variance in different years (Table 4).

Comparison across traits  Based on one- and two-LOD CIs, all QTLs are illustrated in Fig. 3. Comparison of the QTL CIs indicated that all CR QTLs essentially shared the same or overlapping genomic regions with BD QTLs, except for two with minor effects (qCR4b, qCR8). Among year-recurrent QTLs, one major CR QTL (qCR1a) and one CR QTL with a large effect (qCR7) shared common genomic regions with two major BD QTLs (qBD1a and qBD7a). Four BD QTLs did not have overlapping CIs with any CR QTLs. However, only one (qBD2) of these four is a year-recurrent QTL.

The year-recurrent HR QTL (qHR1, G1/87) shared the same genomic region with one major CR QTL (qCR1a) and one major BD QTL (qBD1a), whereas the year-specific HR QTL (qHR8-2008) only shared the same genomic region with one CR QTL (qCR8-2008).

All QTLs for CR and BD, except for two on G6 (qCR6-2008 and qBD6-2008), had negative additive effects, whereas both QTLs for HR had positive additive effects (Tables 2–4). As QTL alleles inherited from the male grandparent were assumed to be superior when calculating the additive effects, this result could be interpreted as QTL genotypes for BD having the same direction with those for CR, but opposite to those for HR, i.e. QTL alleles from the high CR grandparent favored higher CR and later BD, but lower HR. This was consistent with positive correlations between BD and CR and negative correlations (albeit not significant or weak) between BD and HR (Table 1).

Two QTLs on G6 showing minor effects exhibited exactly the opposite behavior, i.e. QTL alleles from the high CR grandparent favored low CR and earlier BD (Tables 2,4).

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Influence of CR evaluation models on CR QTL mapping

In woody plants, many models have been developed to evaluate the CR for the release of endo-dormancy. Most of these models fall into two categories: CH models and chilling unit (CU) models (Cesaraccio et al., 2004). The CH models count the number of hours at which the air temperature is in a certain range, and assume that all air temperatures in this range are equally effective. The < 7.2°C (Weinberger, 1950) and 0–7.2°C models (Eggert, 1951) are the two models most often used in this category. In the CU models, different weighting factors are assigned to temperatures in different ranges. High temperatures above a limit are considered to reverse the chilling effects of lower temperatures, and negative chill units are assessed for them (Cesaraccio et al., 2004). The Utah model (Richardson et al., 1974) and Low Chill model (Gilreath & Buchanan, 1981) are two popular CU models in temperate regions (Cesaraccio et al., 2004).

The Dynamic model (Fishman et al., 1987; Erez et al., 1988) is a two-step CU model developed for the evaluation of CR of tree species in warm winter regions, such as Israel and California in the USA. It assumes a biochemical basis for endo-dormancy release. The first step produces a reversible intermediate of the substance for endo-dormancy release and the second fixes the intermediate by an irreversible transition. This model can account not only for the apparent negative effect of high temperature, but also the varying effect of the same temperature in different daily temperature cycles (Erez et al., 1988).

We should keep in mind that, because of a lack of knowledge of the biochemical and physiological mechanisms controlling CR, almost all CR models have been developed empirically or statistically to fit the responses (mainly BDs) of tree species to local weather conditions. A model appropriate for one species/genotype growing in one area may not necessarily fit another species/genotype growing in another area. In warm winter regions, the reliability of different CR models is different (Erez et al., 1990). The southeastern USA, in which our peach mapping population is maintained and phenotyped, is a variable warm winter region with potential low or high chilling accumulations in different years. If we choose an inappropriate CR model, the resultant CR phenotypic data may not accurately show the differences among genotypes, and may significantly affect the accuracy of CR QTL mapping. In order to resolve this issue, we evaluated CR based on two CH models (the < 7°C and 0–7.2°C models) and three CU models (the Utah, Low Chill and Dynamic models). CR phenotypic data based on different models were significantly and highly correlated (= 1, < 0.001). This correlation could be caused by a lack of long periods of warm and fluctuating temperatures, so that chilling accumulations based on different models all steadily increased in a similar trend through the two winters (Fig. S3, see Supporting Information). The variable weather also tends to cancel out the differences among different models, e.g. the < 7°C model does not count temperatures above 7°C, but counts subfreezing temperatures, in contrast with the Utah and Low Chill models (Cesaraccio et al., 2004). The high correlations of CR phenotypic data resulted in very similar QTL mapping results. Except for one (year 2007/2008) or two (year 2008/2009) QTLs showing minor effects, QTL positions and magnitudes mapped with these CR data were nearly the same (Table 2). Based on these results, we believe that, at least in the years 2007/2008 and 2008/2009 at the experimental site, the influence of different CR models for CR QTL mapping was minor and the results were reliable.

Genetic control of CR

Previous genetic studies in apple and apricot have indicated the dominance of low CR character resulting from the involvement of at least one (major) dominant gene (Oppenheimer & Slor, 1968; Hauagge & Cummins, 1991; Tzonev & Erez, 2003). At first glance, our research appeared to show the dominance of low CR character as well: low CR genotypes obviously dominate in the F2 mapping population (Fig. 1a,c). However, a pure additive model of gene action best fits the CR phenotypic data, which means that none of the detected QTLs for CR showed significant dominance or even partial dominance favoring low CR alleles. Interestingly, distorted marker genotypic ratios provide a valuable hint to resolve the contradiction in this experiment. A cluster of 17 markers mapped to a large genomic region (68–96.4 cM) in the bottom part of G1 was found to have seriously distorted genotypic ratios favoring the allele from the low CR grandparent. This region covers the CI of qCR1a, a major QTL explaining more than 40% of the phenotypic variance of CR (Fig. 3, Table 2). Apparently, it is the distorted genotypic ratio of the CR major QTL alleles, instead of the dominance of the QTL allele favoring the low CR trait, that causes the phenomenon of low CR dominance in peach. More evidence is needed to determine whether this also occurs in other tree species mentioned above.

It is not clear what causes the distortion of the marker genotypic ratio in this large genomic block. Very likely, this region might harbor the gene(s) controlling important traits, such as gamete fertility, seed formation or seed germination (seed dormancy). The tight linkage of the allele(s) of this(these) gene(s) having better fitness with the allele of the major QTL favoring low CR could explain the contradiction above. Another interesting hypothesis is that possibly both the stratification requirement for seed dormancy breaking and CR for bud dormancy breaking are controlled by a similar set of genes. Therefore, seeds with low stratification requirement germinate more easily, which results in more trees (genotypes) with low CR. However, these hypotheses need to be tested by future studies.

To our knowledge, this is the first successful and comprehensive report on floral bud CR QTL analysis in a perennial tree species. The detection of CR QTLs, especially 2 yr-recurrent QTLs with large effects (qCR1a and qCR7), will not only facilitate the marker-assisted breeding for low CR cultivars, but also pave the way for future fine mapping and map-based cloning of genes controlling CR. The two-LOD CI of the major CR QTL (qCR1a) spans only 2 cM, which overlaps with the peach EVG region (Fig. 3). The peach EVG (Evergrowing, previously known as Evergreen) mutant was originally identified in Mexico. In temperate regions, its terminal apices keep growing until they are killed by subfreezing winter temperatures (Rodriguez et al., 1994). The EVG locus was genetically mapped as a ‘recessive gene’ (Wang et al., 2002; Bielenberg et al., 2008). A 132 kb genomic region around EVG was cloned, sequenced and annotated utilizing the peach ‘Nemared’ BAC library. The mutant harbors a sizable deletion, which spans all or part of four MADS-box genes. Two additional MADS-box genes adjacent to the deletion are also not expressed in the mutant (Bielenberg et al., 2004, 2008). Although it is still unclear whether this nondormant (or very low CR) mutation affects the induction of endo-dormancy or has something to do with CR, the colocalization of a CR major QTL and the sequenced EVG region makes the six identified MADS-box genes promising candidate genes for CR of peach floral buds.

Currently, CR QTL mapping on another peach F2 population derived from two different grandparents and association mapping using peach germplasms with different CR are in progress. With these efforts, we aim to verify and refine the CR QTL regions to better suit the needs of marker-assisted breeding and map-based cloning of important genes for CR.

Colocalization of QTLs for CR, HR and BD

The BD in Prunus is determined by the cultivar’s CR needed to break endo-dormancy, as well as HR (Andrés & Durán, 1999). Although it is known that CR and BD of Prunus are genetically controlled (Anderson & Seeley, 1993;Tzonev & Erez, 2003), genetic characterization has not been reported, and controversy exists as to whether genetic components are involved in HR for bloom in Prunus. It has been shown that prolonged exposure to low temperature reduces HR (Couvillon & Erez, 1985; Citadin et al., 2001). Couvillon & Erez (1985) have pointed out that excessive chilling in several fruit tree species results in 90% of HR variations among different cultivars with different CRs, and there is no actual (genetic) difference in HR for bloom among different cultivars. Okie & Blackburn (2008) confirmed that artificially supplied, incremental chilling dramatically reduced HR for bud break in peach when shoots were underchilled, but they found that the effects diminished when buds received more chilling. Recently, Harrington et al. (2009) have proposed a model whereby, between the critical CR and optimum CR, many combinations of CUs and forcing (heat) units could make bud break possible, implying a possible overlapping period of CR and HR after a tree’s critical CR has been met.

If Couvillon & Erez (1985) are correct in proposing that there are no genetic differences for HR among fruit tree cultivars, in our study, low CR genotypes would be overchilled and require less heat accumulation for bloom than would high CR genotypes. Indeed, low CR genotypes have high HRs for bloom, and we found a significant negative correlation between CR and HR in the mapping population (Table 1). A negative correlation between CR and HR has also been reported in apricot (Ruiz et al., 2007), and some peach selections from Aguascalientes, Mexico have been found to have low CR, but late BD (Scorza & Okie, 1990). These results suggest the existence of different HRs among genotypes/cultivars and a potential genetic contribution to this character.

In our study, HR segregated over a wide range (Fig. 1b,d). The ANOVA for HR indicated a significant genotypic effect (< 0.01) (Table S3). Two QTLs for HR, accounting for 8.6–10.7% of the phenotypic variance, were detected (Table 3). Therefore, we believe that the genetic components play a limited role in determining HR of each genotype in our mapping population.

In our study, the distribution of BD varied dramatically across years (Fig. 2). Both environmental (year) effects and genotype (QTL) × environment interaction effects for BD significantly contributed to the variation of this character (Table S3). The variable chilling and heat accumulations in different years could be the major sources of environmental effects (Figs S3,S4, see Supporting Information). Exactly how the genotype × environment interaction influences BD is unknown. However, very possibly, the variable temperatures interact with different genotypes and affect their CR and HR and, finally, BD, because the genotype × environment interactions for CR and HR are also significant (Table S3).

The extensive overlap of QTL CIs for different traits is illustrated in Fig. 3. Two major BD QTLs (qBD1a and qBD7a) colocalize with one major CR QTL (qCR1a) and one CR QTL with a large effect (qCR7). Moreover, despite the negligible or weak correlations between HR and BD (Table 1), the HR QTL qHR1 colocalizes with the major QTL for CR and BD. Furthermore, among all 20 QTLs for three traits, only three BD QTLs and one CR QTL neither colocalize nor overlap with any QTL for other traits (Fig. 3). These noncolocalized QTLs either explained a small proportion of phenotypic variance or were detected only in 1 yr (Tables 2–4; Fig. 3), implying that the noncolocalization could be a result of a low power of detection for QTLs with minor effects for some traits, unavoidable human errors in phenotyping, or the fact that these QTLs are not real. The colocalization of the majority of the detected QTLs might suggest that, in each colocalization case, the genes regulating different traits are tightly linked together. However, considering the significant correlation of phenotypic data between CR and HR or CR and BD, more probably it suggests the pleiotropy of these QTLs and the existence of one unified temperature sensing and action system, some components of which regulate both CR and HR, and others only regulate CR. The regulation of gene expression in this system should generally guarantee late BDs for high CR cultivars and early BDs for low CR cultivars. It should also upregulate HR for low CR peach cultivars, so that they could be generally protected from flower or fruit damage by late spring frosts.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

The authors thank Wayne Sherman for providing the pollen of ‘Fla.92-2C’, Ignazio Verde, Elisa Vendramin and Werner Howad for providing aliquots and unpublished sequences of SSR primers, Chittaranjan Kole and Marisa Badenes for assistance with data analysis, Laura Georgi for assistance with manuscript preparation, and the many people who participated in the phenotyping of the chilling requirement and bloom date. This research was supported in part by the BARD program US-3746-05R and the Robert and Louise Coker Endowed Chair funds at Clemson University.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
  • Abbott AG, Georgi L, Yvergniaux D, Wang Y, Blenda A, Reighard G, Inigo M, Sosinski B. 2002. Peach: the model genome for Rosaceae. Acta Horticulturae 575: 145155.
  • Alburquerque N, García-Montiel F, Carrillo A, Burgos L. 2008. Chilling and heat requirements of sweet cherry cultivars and the relationship between altitude and the probability of satisfying the chill requirements. Environmental and Experimental Botany 64: 162170.
  • Anderson JL, Seeley SD. 1993. Bloom delay in deciduous fruits. Horticultural Reviews 15: 97144.
  • Anderson JL, Richardson EA, Kesner CD. 1986. Validation of chill unit and flower bud phenology models for ‘Montmorency’ sour cherry. Acta Horticulturae 184: 7178.
  • Andrés MV, Durán JM. 1999. Cold and heat requirements of the apricot tree (Prunus armeniaca L.). The Journal of Horticultural Science and Biotechnology 74: 757761.
  • Ballester J, Company RS, Arús P, De Vicente MC. 2001. Genetic mapping of a major gene delaying blooming time in almond. Plant Breeding 120: 268270.
  • Bielenberg DG, Wang Y, Fan S, Reighard GL, Scorza R, Abbott AG. 2004. A deletion affecting several gene candidates is present in the evergrowing peach mutant. Journal of Heredity 95: 436444.
  • Bielenberg DG, Wang Y, Li Z, Zhebentyayeva T, Fan S, Reighard GL, Scorza R, Abbott AG. 2008. Sequencing and annotation of the evergrowing locus in peach [Prunus persica (L.) Batsch] reveals a cluster of six MADS-box transcription factors as candidate genes for regulation of terminal bud formation. Tree Genetics and Genomes 4: 495507.
  • Cesaraccio C, Spano D, Snyder RL, Ducea P. 2004. Chilling and forcing model to predict bud-burst of crop and forest species. Agricultural and Forest Meteorology 126: 113.
  • Citadin I, Raseira MCB, Herter FG, Baptista da Silva J. 2001. Heat requirement for blooming and leafing in peach. Hortscience 36: 305307.
  • Couvillon GA, Erez A. 1985. Influence of prolonged exposure to chilling temperatures on bud break and heat requirement for bloom of several fruit species. Journal of the American Society for Horticultural Science 110: 4750.
  • Dirlewanger E, Moing A, Rothan C, Svanella L, Pronier V, Guye A, Plomion C, Monet R. 1999. Mapping QTL controlling fruit quality in peach (Prunus persica (L.) Batsch). Theoretical and Applied Genetics 98: 1831.
  • Dirlewanger E, Graziano E, Joobeur T, Garriga-Calderé F, Cosson P, Howad W, Arús P. 2004. Comparative mapping and marker-assisted selection in Rosaceae fruit crops. Proceedings of the National Academy of Sciences, USA 101: 98919896.
  • Egea J, Ortega E, Martínez-Gómez P, Dicenta F. 2003. Chilling and heat requirements of almond cultivars for flowering. Environmental and Experimental Botany 50: 7985.
  • Eggert FP. 1951. A study of rest varieties of apple and in other fruit species grown in New York State. Proceedings of the American Society of Horticultural Science 51: 169178.
  • Erez A, Fishman S, Gat Z, Couvillon GA. 1988. Evaluation of winter climate for breaking bud rest using the dynamic model. Acta Horticulturae 232: 7689.
  • Erez A, Fishman S, Linsley-Noakes GC, Allan P. 1990. The dynamic model for rest completion in peach buds. Acta Horticulturae 276: 165174.
  • Fishman S, Erez A, Couvillon GA. 1987. The temperature dependence of dormancy breaking in plants: mathematical analysis of a two step model involving cooperative transition. Journal of Theoretical Biology 124: 473483.
  • Gibson PG, Reighard GL. 2002. Chilling requirement and postrest heat accumulation in peach trees inoculated with peach latent mosaic viroid. Journal of the American Society for Horticultural Science 127: 333336.
  • Gilreath PR, Buchanan DW. 1981. Rest prediction model for low-chilling Sungold nectarine. Journal of the American Society for Horticultural Science 106: 426429.
  • Haley CS, Knott SA. 1992. A simple regression method for mapping quantitative trait loci in line crosses using flanking markers. Heredity 69: 315324.
  • Harrington CA, Gould PJ, St Clair JB. 2009. Modeling the effects of winter environment on dormancy release of Douglas-fir. Forest Ecology and Management. doi: 10.1016/j.foreco.2009.06.018
  • Hauagge R, Cummins JN. 1991. Genetics of length of dormancy period in Malus vegetative buds. Journal of the American Society for Horticultural Science 116: 121126.
  • Hospital F, Moreau L, Lacoudre F, Charcosset A, Gallais A. 1997. More on the efficiency of marker-assisted selection. Theoretical and Applied Genetics 95: 11811189.
  • Howad W, Yamamoto T, Dirlewanger E, Testolin R, Cosson P, Cipriani G, Monforte AJ, Georgi L, Abbott AG, Arús P. 2005. Mapping with a few plants: using selective mapping for microsatellite saturation of the Prunus reference map. Genetics 171: 13051309.
  • Jansen RC, Stam P. 1994. High resolution of quantitative traits into multiple loci via interval mapping. Genetics 136: 14471455.
  • Joobeur T. 1998. Construccíon de un mapa de marcadores moleculares y análisis genético de caracteres agronónicos en Prunus. PhD thesis, Universtat de Lleida, Spain.
  • Kosambi DD. 1944. The estimation of map distance from recombination values. Annals of Eugenics 12: 172175.
  • Lynch M, Walsh B. 1998. Genetics and analysis of quantitative traits. Sunderland, MA, USA: Sinauer Associates.
  • Okie WR, Blackburn B. 2008. Interaction of chill and heat in peach flower bud dormancy [abstract]. HortScience 43: 1161.
  • Oppenheimer CH, Slor E. 1968. Breeding of apples for a subtropical climate. II. Analysis of two F2 and nine back populations. Theoretical and Applied Genetics 38: 97102.
  • Richardson EA, Seeley SD, Walker DR. 1974. A model for estimating the completion of rest for Redhaven and Elberta peach trees. HortScience 9: 331332.
  • Richardson EA, Seeley SD, Walker DR, Anderson JL, Ashcroft GL. 1975. Pheno-climatography of spring peach bud development. Hortscience 10: 236237.
  • Rodriguez AJ, Sherman WB, Scorza R, Wisniewski M, Okie WR. 1994. ‘‘Evergreen’’ peach, its inheritance and dormant behavior. Journal of the American Society for Horticultural Science 119: 789792.
  • Rohde A, Bhalerao RP. 2007. Plant dormancy in the perennial context. Trends in Plant Science 12: 217223.
  • Ruiz D, Campoy JA, Egea J. 2007. Chilling and heat requirements of apricot cultivars for flowering. Environmental and Experimental Botany 61: 254263.
  • Scalabrelli G, Couvillon GA. 1986. The effect of temperature and bud type on rest completion and the GDH°C requirement for bud break in “Red Haven” peach. Journal of the American Society for Horticultural Science 111: 537540.
  • Scorza R, Okie WR. 1990. Peaches (Prunus Persica L. Batsch). Acta Horticulturae 290: 177231.
  • Sherman WB, Beckman TG. 2003. Climatic adaptation in fruit crops. Acta Horticulturae 622: 411428.
  • Silva C, Gacia-Mas J, Sánchez AM, Arús P, Oliveira MM. 2005. Looking into flowering time in almond (Prunus dulcis (Mill) D. A. Webb): the candidate gene approach. Theoretical and Applied Genetics 110: 959968.
  • Tabuenca MC. 1964. Necesidades de frõÂo invernal de variedades de albaricoquero, melocotonero y peral. Annales Aula Dei 7: 113132.
  • Topp BL, Sherman WB, Raseira MCB. 2008. Low-chill cultivar development. In: LayneDR, BassiD, eds. The peach botany, production and uses. Wallingford, Oxfordshire, UK: CABI, 106138.
  • Tzonev R, Erez A. 2003. Inheritance of chilling requirement for dormancy completion in apricot vegetative buds. Acta Horticulturae 622: 429436.
  • Utz HF. 2000. PLABQTL frequently asked questions. Available at: https://www.uni-hohenheim.de/plantbreeding/software/plabqtl/pq_faq.txt.
  • Utz HF, Melchinger AE. 2006. PLABQTL: software for QTL analysis with composite interval mapping. Available at: https://www.uni-hohenheim.de/plantbreeding/software/.
  • Van Ooijen JW, Voorrips RE. 2001. JoinMap® 3.0, software for the calculation of genetic linkage maps. Wageningen, the Netherlands: Plant Research International.
  • Verde I, Quarta R, Cedrola C, Dettori MT. 2002. QTL analysis of agronomic traits in a BC1 peach population. Acta Horticulturae 92: 291297.
  • Voorrips RE. 2002. MapChart: software for the graphical presentation of linkage maps and QTL. Journal of Heredity 93: 7778.
  • Vos P, Hogers R, Bleeker M, Reijans M, Van De Lee T, Hornes M, Frijters A, Pot J, Peleman J, Kuiper M et al. 1995. AFLP: a new technique for DNA fingerprinting. Nucleic Acids Research 23: 44074414.
  • Wang Y, Georgi LL, Reighard GL, Scorza R, Abbott AG. 2002. Genetic mapping of the evergrowing gene in peach [Prunus persica (L.) Batsch]. The Journal of Heredity 93: 352358.
  • Wassom JJ, Wong JC, Martinez E, King JJ, DeBaene J, Hotchkiss JR, Mikkilineni V, Bohn MO, Rocheford TR. 2008. QTL associated with maize kernel oil, protein, and starch concentrations; kernel mass; and grain yield in Illinois high oil × B73 backcross-derived lines. Crop Science 48: 243252.
  • Weinberger JH. 1950. Chilling requirements of peach varieties. Proceedings of the American Society of Horticultural Science 56: 122128.
  • Zeng ZB. 1994. Precision mapping of quantitative trait loci. Genetics 136: 14571468.
  • Zhebentyayeva TN, Reighard GL, Gorina VM, Abbott AG. 2003. Simple sequence repeat (SSR) analysis for assessment of genetic variability in apricot germplasm. Theoretical and Applied Genetics 106: 435444.

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
FilenameFormatSizeDescription
NPH_3119_sm_FigS1.pdf24KSupporting info item
NPH_3119_sm_FigS2.pdf22KSupporting info item
NPH_3119_sm_FigS3.pdf15KSupporting info item
NPH_3119_sm_FigS4.pdf9KSupporting info item
NPH_3119_sm_Rev_figurelegends.doc30KSupporting info item
NPH_3119_sm_TablesS1-S3.doc106KSupporting info item