Introduction

The cultivated strawberry (Fragaria × ananassa) is economically and commercially important, widely consumed fresh or in processed forms. Fruit quality is fundamental for the acceptance of strawberry cultivars by consumers. Fruit quality includes sensory properties (appearance, texture, taste and aroma), nutritional values, chemical compounds, mechanical properties, and functional properties (Abbott 1999). Specific attention has been paid to fruit sensorial and nutritional traits, which have become major goals in many strawberry breeding programs. Strawberries with intense flavor are characterized by their high titratable acidity, highly soluble solid content (Kader 1991), and strong aromatic properties (Aharoni et al. 2004). In recent years, some breeding programs have considered selecting for nutritional value in addition to sensory properties (Capocasa et al. 2008). Strawberries are best known as an important source of vitamin C. Strawberries contain other bioactive compounds that have physiological effects (Hannum 2004; Seeram 2008) such as antioxidant compounds, which are thought to help reduce risk of some chronic diseases (Demmig-Adams and Adams 2002; Battino et al. 2009). Phenolics are important strawberry antioxidants that strongly influence fruit quality, contributing to both the sensorial-organoleptic attributes and the nutritional value (Espin and Tomas-Barberan 2001). The major class of phenolic compounds in strawberry is represented by the flavonoids (Kähkönen et al. 2001; Määttä-Riihinen et al. 2004; Aaby et al. 2005). Anthocyanins are an important type of flavonoid compound and are responsible for the red color of the berries. They are the best known polyphenolic compounds and quantitatively the most important in strawberry (Clifford 2000; Lopes-da-Silva et al. 2002). One of the major antioxidants in strawberry is the anthocyanin pelargonidin-3-glucoside (Wrolstad et al. 1970; Kalt et al. 1993; Tulipani et al. 2009). The concentration and composition of anthocyanins are thus important for strawberry sensory quality and nutritional value.

Inheritance of some strawberry quality traits such as fruit size and ripening (Galletta and Maas 1990; Coman and Popescu 1997), soluble solid content and sugars, titratable acidity and organic acids (Shaw et al. 1987; Shaw 1988; Ohtsuka et al. 2004), and ascorbic acid (Lundergan and Moore 1975; Sone et al. 2003) has been reported as quantitatively measured traits. Fruit-quality traits are strongly influenced by environmental conditions (Schauer et al. 2008). Molecular markers allow the dissection of such quantitative traits into discrete quantitative trait loci (QTL), which can be located on the strawberry genetic map. Cultivated strawberry is an allo-octoploid that displays disomic pairing (Ashley et al. 2003). Previous studies have been conducted to detect QTL associated with fruit-quality traits in two different octoploid strawberry populations (Lerceteau-Köhler et al. 2012; Zorrilla-Fontanesi et al. 2011a, 2012). Zorrilla-Fontanesi et al. (2011a) detected 28 QTL for fruit-quality traits (4 QTL for total anthocyanin content, 4 QTL for soluble solids content, and 4 QTL for titratable acidity) in the ‘232’ × ‘1392’ F1 population. This population was also employed in a different study to identify QTL for volatile compounds (Zorrilla-Fontanesi et al. 2012). Lerceteau-Köhler et al. (2012) used the map previously developed by Rousseau-Gueutin et al. (2008) to identify QTL for fruit traits related to fruit development, texture, color, anthocyanin, sugar, and organic acid content. They detected a total of 87 unique QTL for the analyzed traits (5 QTL for total anthocyanin content, 3 for soluble solids content and 5 for titratable acidity) in the F1 population derived from the cross ‘Capitola’ × ‘CF1116.’ Some of the fruit-quality traits evaluated (firmness, soluble solid content, titratable acidity, pH and anthocyanins) were common between both studies and were found to map to the same homoeologous groups (HG) in the two mapping populations. QTL for anthocyanins were reported in HG II and HG III. QTL for soluble solids content were detected in HG V and HG VI and QTL for titratable acidity in HG IV and HG V. Also, there were other QTL for these traits that were reported in only one of the populations. The markers linked to the QTL and the phenotypic variation explained by each of them are presented in Online Resource 1. In addition, a non-random distribution of QTL was observed, with clustering of QTL observed in both works. Some of them were forming QTL clusters that represented the same class of traits, i.e., fruit development, color, acidity, or sweetness. Also, some regions with QTL for several types of traits were observed. For example, sugar- and acid-related traits clusters were observed on the linkage group (LG) VIa belonging to the HG VI (Lerceteau-Köhler et al. 2012). Colocation among fruit-quality QTL (titratable acidity, L-ascorbic acid and pH) or between fruit-quality and horticultural QTL was detected in HG III, IV, and V. A cluster involving QTL for titratable acidity, pH, anthocyanin content, and color parameters were reported on LG V-M.2 (Zorrilla-Fontanesi et al. 2012).

Flowering habit is another important trait for strawberry breeding programs. Cultivated strawberries are sold, according to their flowering habit, as two primary types: short-day and dayneutral. Short-day genotypes or “June-bearers,” initiate flower buds either under short-day lengths (<14 h of day length) or at temperatures below 15 °C. These plants flower only once a year and, therefore, have only one harvest period. Dayneutral genotypes are said to be photoperiod insensitive, initiating flowers under any photoperiod conditions as long as temperatures are moderate (below 30/26 °C day/night) (Darrow 1966; Hancock 1999). Because these definitions indicate that the control of the trait is also conditioned by temperature, the terms “once flowering” and “repeat flowering” are more neutral than “short-day” and “dayneutral.” In favorable conditions, repeat-flowering cultivars can produce fruit continuously all summer and into the autumn. Repeat flowering is a highly desirable trait in strawberry breeding programs due to its importance in extending the harvest season in commercial production.

Several mapping approaches have been used to study the genetic control of flowering habit and to identify QTL associated with the trait in cultivated octoploid strawberry (Weebadde et al. 2008; Gaston et al. 2013; Castro et al. 2015). Weebadde et al. (2008) used AFLP primer combinations with a mapping population derived from the cross ‘Tribute’ (repeat flowering) × ‘Honeoye’ (once flowering) to identify QTL associated with repeat flowering. Eight QTL explaining the trait were reported in this population, suggesting that, although repeat flowering is accepted as being controlled by a single dominant (Ahmadi et al. 1990; Shaw and Famula 2005; Sugimoto et al. 2005; Gaston et al. 2013) or recessive gene (Scott 1936) in cultivated strawberry, the regulation of the trait under diverse environmental conditions may be polygenic (Clark 1937; Powers 1954; Ourecky and Slate 1967). Gaston et al. (2013) evaluated “perpetual flowering” as both a qualitatively classified trait and as a quantitatively measured trait in a population obtained from ‘Capitola’ × ‘CF1116.’ They reported a major dominant QTL (named FaPFRU) located on LG4b-f, which affected both perpetual flowering and the production of runners. Recently, Castro et al. (2015) scored the repeat-flowering trait qualitatively (once flowering or repeat flowering) and quantitatively (number of weeks of flowering) using SSR and SCAR markers in the same population used by Weebadde et al. (2008) and mapped the trait to LG IV-T-1 of the ‘Tribute’ map when the population was grown and evaluated in five different locations. When scored quantitatively, QTL for the total number of weeks of flowering in Maryland and California also was detected on LG IV-T-1. EST-SSR markers linked to the repeat-flowering trait were identified for the first time. The transition from anonymous and untranscribed markers to genic and transcribed markers allowed greater opportunity to find direct markers for the genes controlling the repeat-flowering trait and increase the ability to transfer the information from this population to others.

The identification of DNA markers tightly linked to genes or QTL associated with the traits of interest in mapping populations provides efficient tools for the application of marker-assisted selection (MAS) in breeding programs and may help identify the mechanism associated with the traits of interest. Trait-mapping experiments to identify QTL cosegregating with important traits have become popular as a means to help unravel the mechanisms underlying the genetic control of important characteristics in strawberry. Markers associated with a trait when tested in different years, growth sites, generations, and genetic backgrounds are of obvious utility to multiple breeding programs. And, as demonstrated by Weebadde et al. (2008), markers associated with a trait in only one location are equally important if that location is the target of a breeding program. In this study, a genetic linkage map was developed using an F1 population obtained from the cross ‘Delmarvel’ × ‘Selva,’ a cross between a US east-coast once-flowering cultivar known for its good flavor and a west coast repeat-flowering cultivar. The mapping population was evaluated for total anthocyanins, total phenolics, and antioxidant capacity measured as the oxygen absorbance capacity (ORAC), the flavor-related fruit-quality traits of percentage of soluble solid content (SSC) and titratable acidity (TA), and total number of weeks of flowering in order to identify QTL and markers linked to these traits.

Materials and methods

Plant material

A genetic mapping population of 177 F1 individuals was generated from a cross between the two strawberry cultivars ‘Delmarvel’ (from the USDA-ARS at Beltsville, MD) and ‘Selva’ (from the University of California at Davis). The two parental lines were chosen because they represent major divergent breeding populations, and differ in fruit-quality traits and the horticultural trait of repeat flowering. ‘Delmarvel’ flowers only in spring, whereas ‘Selva’ flowers from spring to autumn. The mapping population was grown in a field at the Beltsville Agricultural Research Center in Maryland (USA) during 2006 and 2007 using annual plasticulture production practices. Sixteen mapping progeny died in the field before they could be evaluated for all the phenotypic traits.

Phenotypic evaluation

To evaluate the fruit-quality traits, the mapping population was planted in August 2007 in a field on the North Farm at the Beltsville Agricultural Research Center in Maryland. Three hills with trickle irrigation were formed and covered with black plastic mulch according to the annual hill method of strawberry production. Each hill was considered a block in a randomized complete block experimental design. The parents and population of 177 were propagated so that each parent and progeny genotype could be represented once in each hill, though there were some deaths, and were planted with the position in the hill determined through randomization. Fruit were harvested once from each plant separately in May 2008; this required three harvest dates within 1.5 weeks, and the majority of plants could be harvested on the middle date. Fruit were processed for analysis to measure total anthocyanins, total phenolics, antioxidant capacity measured as the oxygen radical absorbance capacity (ORAC), titratable acids (TA), and soluble solid content (SSC). Fruit preparation and determination of total anthocyanins, total phenolics, ORAC, SSC, and TA were performed as previously described by Lewers et al. (2010).

Data for repeat flowering were taken from unreplicated plants in the field in 2006. Flowering data for the parents and the population were collected weekly from April to October at the Beltsville Agricultural Research Center in Maryland. The trait was scored quantitatively, recording the number of weeks of flowering for the parents and the progeny. Each of the progeny also were evaluated as repeat flowering or once flowering according to whether or not it flowered after a set date, as previously described by Castro et al. (2015). The date 14 June was chosen as the cutoff date, because that was the date after which once-flowering genotypes in a nearby field quit flowering for the spring. Chi-square goodness-of-fit tests to optional segregation ratios were tested with the CHITEST function in Microsoft Excel (Redmond, WA).

To map the repeat-flowering trait qualitatively, we considered progeny flowering only 1 or 2 weeks after the cutoff date as once-flowering genotypes and scored the progeny qualitatively to create a marker. To allow for the possibility that the repeat-flowering allele at the proposed single locus could be either dominant or recessive, and to determine which alleles of linked molecular-marker loci were in coupling or repulsion with the repeat-flowering allele, two markers were created for the trait, one assuming the repeat-flowering allele was dominant, and the other assuming it was recessive. For the marker “RF-dom” the repeat-flowering progeny received a score of ‘1’ and the once-flowering progeny received a score of ‘0.’ Coding in that way assumes that the trait was controlled by a dominant gene. For the marker “RF-rec” the repeat-flowering progeny received a score of ‘0’ and the once-flowering progeny received a score of ‘1.’

Statistical data analysis

The mean values for both parents and each F1 progeny were calculated for each trait and replicate. To evaluate whether the data followed a normal distribution, the Shapiro–Wilk test was performed (Shapiro and Wilk 1965), and global means, ranges, standard deviations (SD) and histograms of frequency were calculated for the trait distributions. A t test was employed to compare the mean values of the parents. Trait segregation was declared transgressive when at least one progeny had a trait value that was higher or lower than that of the highest or lowest parent by at least twice the standard deviation of the parents. For each trait, differences among progeny were tested by analysis of variance (ANOVA). Pearson correlation coefficients were calculated for the different trait values.

DNA isolation and molecular markers analysis

DNA for molecular markers analyses was isolated from young leaves of the parents and F1 mapping population using the DNeasy Plant Mini Kit (Qiagen Inc., Valencia, CA, USA) according to the manufacturer’s instructions.

Primer pairs of SSR and SCAR markers previously reported (Albani et al. 2004; Sargent et al. 2004, 2006, 2007, 2008; Rousseau-Gueutin et al. 2008; Spigler et al. 2008, 2010; Zorrilla-Fontanesi et al. 2011b) were analyzed for polymorphisms in the mapping population. We also analyzed five SSR markers (FxaACA02I08C, FxaAGA02N04C, FxaAGA01H04C, FxaAGA01G05C and FxaAGA21O11C) linked to the everbearing gene (Honjo et al. 2016). For use in parental screens and genetic mapping, the 5′ ends of the forward primers were modified with the addition of the M13 sequence (5′-TGTAAAACGACGGCCAGT-3′), and separate M13 primers were labeled with the fluorescent dyes 6-FAM, VIC, NED and PET (Applied Biosystems, Foster City, CA, USA) (Schuelke 2000). The 5′ ends of the reverse primers were modified by the addition of a short sequence (GT, GTT or GTTT) to facilitate adenylation for more uniform product sizes (Brownstein et al. 1996). PCR amplifications were performed as described by Schuelke (2000). Primer pairs were first tested with the parents, and those that generated amplicons that were polymorphic in the parents were combined by product size and fluorescent dye color into multiplexes of up to four primer pair products after amplification in the mapping population. Samples for electrophoresis were prepared as previously reported by Castro et al. (2015). PCR products were separated by capillary electrophoresis on an ABI3730 DNA Analyzer (Applied Biosystems Inc., Foster City, CA, USA). Allele sizes were determined using GeneMapper 3.5 software (Applied Biosystems) and GENSCAN 500HD (LIZ) as an internal size standard.

Linkage analysis and map construction

Polymorphic fragments were scored as present or absent. Chi-squared tests to determine goodness of fit to an expected segregation ratio of 1:1 (for markers present in only one parent) or 3:1 (for markers present in both parents) were carried out for all markers segregating in the F1 population. Markers deviating significantly from the expected segregation ratios (P < 0.001) were excluded from further analysis. An analysis for cosegregating fragments generated from the same primer pair allowed the identification of allelic pairs segregating in a 1:1:1:1 (genotypes ac, ad, bc, bd for loci in the conformation ab × cd or genotypes ee, eg, ef, fg for loci in the conformation ef × eg) or 1:2:1 (hh, hk, kk for hk × hk loci). Linkage analyses and map construction were performed using JoinMap 4 (van Ooijen 2006). Two separate parental maps were constructed using 1:1, 3:1 and codominant markers and the population coded as CP (for cross pollinated). Then, in order to perform QTL analysis, two independent parental maps were constructed using 1:1 and codominant markers from each parental line translated to a population-type DH (as described on MapQTL user manual). Linkage groups (LGs) were selected using a minimum LOD threshold of 6. The strongest cross-link (SCL) value was used to add ungrouped markers to established LGs. Marker order and map distances were determined using the Kosambi mapping function, the regression mapping algorithm (Stam 1993), a recombination fraction of 0.4, a ripple value of 1, and a jump threshold of 5.0. Linkage maps were drawn using MapChart 2.2 for Windows (Voorrips 2002).

Markers names are followed by the size of the allele in base pairs and by the letters D, S or DS to indicate whether the allele is present in ‘Delmarvel’ (D), ‘Selva’ (S), or in both parents (DS). For codominant markers the two alleles of each parent are shown together. The seven homoeologous groups (HGs) were named I–VII, as the corresponding LGs in the diploid Fragaria reference map (Sargent et al. 2008) followed by a D for ‘Delmarvel’ LGs and by an S for ‘Selva’ LGs. LGs within each HG were numbered arbitrary.

Identification of QTL

QTL analyses were performed on the parental maps using MapQTL 5 (van Ooijen 2004). First, we used a single-marker nonparametric method, the Kruskal–Wallis (KW) test, which is extensively used for QTL analysis (Liu 1998; Collard et al. 2005; Van Ooijen 2004). The KW test is performed on each locus separately without taking the map information into account. It is the simplest method to identify markers linked to QTL. Only markers with a stringent significance level of P ≤ 0.005 were selected as those tightly linked to a QTL. Next, an interval mapping (IM) analysis (Lander and Botstein 1989) was carried out to locate the position of the QTL on each LG on each parental map. The multiple QTL method (MQM; Jansen and Stam 1994), using the most significantly linked markers to detected QTL as cofactors, was applied to control the genetic background for a better location of the QTL. LOD threshold values for QTL significance were estimated with a 1000 permutation test (Churchill and Doerge 1994) for each trait and replicate on each map. QTL with LOD values higher than the genome-wide threshold at P ≤ 0.05 were considered significant. Additionally, QTL with LOD values higher than the individual linkage group threshold at P < 0.01 were reported. The QTL were named by the abbreviation of the trait name followed by the number of the replicate and the name of the linkage group on which the QTL was detected (D, ‘Delmarvel’ and S, ‘Selva’). QTL location and 1-LOD confidence intervals were drawn using MapChart 2.2.

Results

Trait means and distributions

The mean phenotypic values and basic statistical parameters for the two parental lines and the F1 progeny are shown in Table 1. The parental and progeny means of each fruit-quality trait were calculated mathematically by combining the data obtained from the three different replicate blocks of the parents and progeny growing in the field. The parents exhibited significantly different values (P < 0.05) for the six fruit-quality traits analyzed. ‘Delmarvel’ had higher concentration of soluble solids content (SSC), titratable acidity (TA) and the ratio of soluble solids content to titratable acidity (SSC/TA) than ‘Selva.’ ‘Selva’ contained a higher concentration of anthocyanin, phenolic compounds, and ORAC.

Table 1 Phenotypic values of fruit-quality traits and number of weeks of flowering for ‘Delmarvel’ and ‘Selva’ parents and the F1 population
Full size table

Phenotypic variation was found in the F1 progeny for the six traits evaluated (Table 1). The means and ranges of variation between replicates were similar for all the fruit-quality traits. No significant differences (P > 0.05) were found among replications in the F1 progeny for all fruit-quality traits except SSC. Analysis of variance of these traits showed significant differences between the progeny (P < 0.001). The six fruit-quality traits showed continuous variation in the progeny. The variation for total anthocyanins and total phenolics fit a normal distribution, while the variation for the related ORAC trait was skewed toward the lower ‘Delmarvel’ mean (Online Resource 2). The variation for soluble solid content fit a normal distribution, while the variation for titratable acids and the resulting SSC/TA ratio both skewed toward the lower ‘Selva’ mean (Online Resource 2). For each trait, the progeny mean was between the value of the two parents (Table 1). Transgressive segregation was detected for all traits but SSC.

Correlations among traits

Pearson correlation coefficients between the traits are presented in Online Resource 3. The traits, in general, showed similar values in their correlation coefficients over the three replicates. The highest correlation coefficients (0.71–0.92) were found between ORAC, total phenolics, and total anthocyanins. These three traits were somewhat positively correlated with titratable acids. Interestingly, from a breeding standpoint, the titratable acids trait was somewhat correlated positively with soluble solid content (P < 0.001). No significant correlation was found between the fruit-quality traits and the number of weeks of flowering.

Number of weeks of flowering

The parents and the F1 progeny were evaluated for the number of weeks of flowering in the field in 2006. ‘Delmarvel’ flowered only in spring (April–May) whereas ‘Selva’ flowered from spring to autumn (April–October). The total number of weeks of flowering was 3 and 20 for ‘Delmarvel’ and ‘Selva,’ respectively. The trait followed a continuous distribution, but it was not normally distributed according to the Shapiro–Wilk test; it was skewed toward ‘Delmarvel’s fewer weeks of flowering (Online Resource 2). Among the progeny, the total number of weeks of flowering ranged from 3 to 24. There were eight genotypes that flowered as long (20 weeks) or longer than ‘Selva’ and 10 genotypes that flowered for only 3 weeks. In an attempt to score the trait qualitatively, each progeny genotype was classified into the classes repeat-flowering or once-flowering according to whether or not it flowered after a date set, 14 June, which is when once-flowering cultivars and breeding selections in a nearby field quit flowering for the spring. The classification of the progeny into the repeat-flowering or once-flowering classes did not have a clear cutoff in this population. According to the data set selected to classify the progeny as repeat flowering or once flowering, there were 55 progeny that did not flower after the set cutoff date, 9 progeny that flowered for 1 week after the cutoff date, and 12 that flowered for 2 weeks after the cutoff date. The remaining progeny (85 individuals) flowered for more than 2 weeks after the cutoff date.

To determine the number of loci involved and if the progeny that flowered only 1 or 2 weeks after the cutoff date should be treated as repeat flowering or once-flowering genotypes in further testing, Chi-square goodness-of-fit tests were done with both options. If the genotypes that flowered only one or 2 weeks after the cutoff date selected are classified as once flowering, the best fit of the population (85RF:76OF) was to a segregation ratio of 1:1 (χ 2 = 0.503; P = 0.478). The population also fit a segregation ratio of 9RF:7OF (P = 0.377), but this ratio requires that both parents are repeat flowering and involves an epistatic interaction, so this possibility was discounted. If the genotypes that flowered only once or twice after the cutoff date are classified as repeat flowering, the population does not fit a ratio 1:1 (106RF:55OF; χ2 = 16.1; P < 0.0001). With this classification, the population segregation ratio best fits a 5RF:3OF ratio (χ 2 = 0.764; P = 0.382), requiring two loci with an epistatic interaction. One parent could be repeat flowering, and the other once flowering, but we found no other locus in this mapping population that was significantly associated with the trait when mapped with the “weak-repeat-flowering” progeny classified as “repeat flowering.” Therefore, the hypothesis best supported by these segregation data alone is that, in this population evaluated in this location, the “weak-repeat-flowering” progeny should be considered once-flowering types in qualitative mapping of the trait, the trait most likely is controlled by a single locus, and one parent is heterozygous at this locus while the other is homozygous.

Molecular markers and resulting linkage maps

A total of 301 SSR primer pairs were used to genotype the parents and the F1 population. These primers generated 740 polymorphic fragments (2.5 fragments per primer pair). Among them, 245 fragments (33 %) segregated 1:1 for ‘Delmarvel,’ 210 (28 %) segregated 1:1 for ‘Selva’ and 161 (22 %) were present in both parents and segregated 3:1. The remaining fragments (17 %) did not fit single-dose ratios and were not used to construct the linkage maps. After analyzing all possible combinations of cosegregating fragments generated from each primer pair, 64 amplicons produced a total of 25 codominant (1:1:1:1 or 1:2:1) markers that were used, together with dominant heterozygous markers (3:1), to identify homologous linkage groups.

A total of 177 individuals of the F1 population were used to construct the genetic linkage maps of ‘Delmarvel’ and ‘Selva.’ The ‘Delmarvel’ map contained 292 markers distributed in 41 linkage groups with length ranging from 5.6 to 77.6 cM and a total length of 1529.3 cM. The ‘Selva’ map comprised 242 markers distributed in 36 linkage groups with length ranging from 2.3 to 99.1 cM and a total length of 1317.5 cM. The range of number of markers per linkage group was 2 to 20 and 2 to 24 for ‘Delmarvel’ and ‘Selva,’ respectively (Online Resource 4).

Map locations of quantitatively measured traits

QTL analyses were performed for all the seven quantitative traits using MapQTL software in the ‘Delmarvel’ and ‘Selva’ maps generated by JoinMap in DH configuration, so that only two genotypic classes were assumed and modeled. The Kruskal–Wallis test confirmed regions where markers were significantly (P < 0.005) associated with the fruit-quality traits and the total number of weeks of flowering (Table 2). The most significant markers were EMFv104-117D for total anthocyanins, EMFn184-245S for ORAC, CO818147-296S for total phenolics, Fvi20-143S for SSC, ARSFL7-245DS/285DS for TA, BFACT010-240D for SSC/TA, and FxaACA02I8C-145S for total number of weeks of flowering. A total of nine QTL (eight for fruit-quality traits and one for total number of weeks of flowering) with LOD scores greater than or equal to the genome-wide threshold at P ≤ 0.05 were detected using interval mapping (IM) analysis. These QTL are highlighted in bold in Table 2. The LOD threshold values varied from 2.7 to 2.9 for all the traits in both parental maps. These nine significant regions were confirmed by both IM and KW. Several regions were significant according to the KW test but with LOD scores just below the genome-wide threshold yet above the linkage group threshold (P ≤ 0.010). Considering these regions too, a total of 27 QTL for fruit-quality traits and one QTL for total number of weeks of flowering were detected. Sixteen of these QTL were located on the ‘Delmarvel’ map and 12 on the ‘Selva’ map (Table 2; Fig. 1). MQM analysis did not identify any QTL that may have been masked by other QTL. QTL were identified on HGs II, IV, V, VI and VII of the ‘Delmarvel’ × ‘Selva’ maps, ranging from three QTL on HG II, IV and VII to 14 QTL on HG VI. Clusters of QTL for fruit-quality traits were detected on HGs II, V, VI and VII. Analyzing individually the QTL identified per each trait, four QTL were detected for total anthocyanins, five QTL for ORAC, total phenolics, soluble solids content and SSC/TA, three QTL for titratable acidity and one for the total number of weeks of flowering.

Table 2 QTLs detected for total anthocyanins (Antho), antioxidant capacity (ORAC), total phenolics (Pheno), total soluble solids content (SSC), titratable acids (TA), ratio of soluble solids content and titrable acids (SSC/TA) and total number of weeks of flowering (NWF) in the ‘Delmarvel’ × ‘Selva’ population based on Kruskal–Wallis (KW) test and Interval mapping (IM) analysis
Full size table
Fig. 1
figure 1figure 1

Location of QTL associated with total anthocyanins (Antho), total phenolics (Pheno), antioxydant capacity (ORAC), total soluble solids content (SSC), titratable acids (TA), ratio of soluble solids content and titratable acids (SSC/TA) and total number of weeks of flowering (NWF) identified in the ‘Delmarvel’ × ‘Selva’ F1 population. QTL are drawn at the right of each corresponding linkage group. The abbreviations of the trait names are followed by the number of the replicate and by the name of the linkage group in which the QTL was located (e.g., Antho1_II-D-4). Only linkage groups including QTL are presented

Full size image

QTL for fruit-quality traits

The phenotypic variation (R 2) explained by each QTL associated with fruit-quality traits ranged from 4.8 to 10.7 % (Table 2). QTL for total anthocyanins and ORAC colocated in LGs V-S-3, VI-D-3, and VI-D-4. A QTL for total phenolics was detected in V-S-3. Colocations between total anthocyanins and total phenolics also were detected in II-D-4. Additional QTL controlling total phenolics were detected in LG IV-S-1, VII-D-5, and VII-S-1. Colocations of QTL for soluble solids content, titratable acidity and SSC/TA were identified in HG VI. In LGs VI-S-3 and VI-D-3 a QTL for soluble solids content colocated with SSC/TA. A QTL for soluble solids content also was identified in LG V-D-1. QTL controlling titratable acidity were identified in LGs IV-D-3 and VI-D-4, colocalizing with a QTL for SSC/TA (Fig. 1).

QTL for weeks of flowering

For the total number of weeks of flowering, a QTL located on LG IV-S-1 was associated with the markers FxaACA02I8C-145S and ChFaM148-182DS/184DS. FxaACA02I8C-145S was the most significant marker and explained 43.3 % of the phenotypic variation. The presence of the marker FxaACA02I8C-145S was associated with an increase in number of weeks of flowering of 7, i.e., the 145-bp allele from ‘Selva’ was associated with a mean score of 13.2 weeks of flowering, while the individuals without the allele had a mean score of 6.2 weeks. This is somewhat typical for once-flowering varieties grown in Maryland; as flowering start 4 weeks before fruiting, fruiting can last 2–3 weeks and flowers that do not result in fruit may continue to appear when the first fruits are ripe. Other significant QTL associated with the total number of weeks of flowering were not detected in this population, suggesting that the trait is controlled by a single region in this population.

Qualitative mapping of flowering and determination of gene action

Two qualitative markers were created for the repeat-flowering/once-flowering traits, one assuming the repeat-flowering allele was dominant, and the other assuming it was recessive. Both traits were colocated (as alleles) on LG IV-S-1 of the ‘Selva’ map linked to the markers FxaACA02I8C-145S and ChFaM148-182DS/184DS (Fig. 2a), showing that ‘Selva’ is heterozygous for repeat flowering. This is the same location where number of weeks of flowering was mapped as a QTL. If a single-locus controls a trait in a population segregating in a 1:1 ratio, then one parent is heterozygous, and the other is homozygous. The phenotype of the heterozygous parent reveals which allele and phenotype are dominant, and the phenotype of the homozygous parent reveals which allele and phenotype are recessive. Therefore, in this population, because ‘Selva’ is heterozygous for repeat flowering, the qualitative data show that the repeat-flowering allele at this locus is dominant over the once-flowering allele. ‘Selva’ was selected from the University of California at Davis strawberry breeding population and is five generations from the wild strawberry credited as the source of the repeat-flowering trait in many breeding populations. Early-generation testcrosses and segregation ratios that treated repeat flowering qualitatively instead of quantitatively strongly supported control by a dominant allele at a single locus (Ahmadi et al. 1990).

Fig. 2
figure 2

a Map location of the repeat-flowering trait in the ‘Delmarvel’ × ‘Selva’ F1 population. b Markers linked to the repeat-flowering trait in coupling and in repulsion

Full size image

The marker FxaACA02I8C-145S was the closest to the trait at a distance of 14 cM. FxaACA02I8C is the primer name and 145 is the size of the fragment amplified in ‘Selva’ and linked in coupling to the RF-dom allele. The 145 bp fragment was observed only in ‘Selva.’ Most individuals with the 145-bp size allele were repeat flowering, while the individuals without the allele were once flowering (Table 3). This primer also amplified two other fragments (141 and 143 bp size) only from ‘Delmarvel,’ and not linked to the trait, as well as three other fragments (138, 154 and 173 bp size) from both parents and all the progeny and, therefore, were not mapped. The markers linked to the trait in coupling and repulsion are shown in Fig. 2b. The other markers (FxaAGA02N04C, FxaAGA01H04C, FxaAGA01G05C and FxaAGA21O11C) reported by Honjo et al. (2016) were not polymorphic in this population or produced several polymorphic fragments but that were not linked to the trait. For the other markers (ChFaM011, CX661225, F.v.D3) closest to the trait as reported by Castro et al. (2015), no polymorphic fragments were amplified from ‘Selva.’

Table 3 Segregation of the SSR markers linked to the repeat-flowering trait and phenotypes in the ‘Delmarvel’ × ‘Selva’ F1 population
Full size table

Expected segregation ratios for four classes (repeat-flowering and once-flowering progeny with and without the 145-bp allele) were calculated based on the observed recombination rate of 14 % between the trait and the closest marker. When the resulting ratio was tested against actual segregation of the four classes that resulted when the progeny that flowered only one or two weeks after the cutoff date were classified as once-flowering types (P = 0.310) versus repeat-flowering types (P < 0.001), further supporting the classification of these progeny as once-flowering types was observed.

All these results taken together indicate that repeat flowering in this population is controlled by a dominant allele at a single locus. ‘Selva’ is heterozygous for repeat flowering and for the marker FxaACA02I8C whereas ‘Delmarvel’ is homozygous. Unfortunately, the primer pairs amplifying both markers linked in this population to repeat flowering are not linkage group specific, as they amplified additional products that did not link to repeat flowering. The products 141 and 143 amplified by FxaACA02I8C located in the same homeologous group but different linkage groups, LG IV-D-3 and LG IV-D-2 of the ‘Delmarvel’ map. The products 178 and 188 amplified by the marker ChFaM148 were present in both parents and mapped on LG IV-D-4 and IV-S-4 in the ‘Delmarvel’ and ‘Selva’ maps, respectively, the same homeologous group but different linkage groups. ChFaM148 also amplified the fragment 180, which was present only in ‘Delmarvel’ and mapped on LG IV-D-4. Therefore, use of these markers in breeding would require some degree of pedigree information with regard to the amplicons from selected parents.

Discussion

Using markers to assist breeding for important horticultural traits such as fruit quality and extended flowering season could improve the efficiency and reduce the cost for strawberry breeding programs focused on these traits. Identifying QTL and markers for these traits is the first step toward designing marker-assisted-breeding strategies to test for efficiency. In this study, a ‘Delmarvel’ × ‘Selva’ genetic linkage map was constructed to detect QTL associated with fruit-quality traits and number of weeks of flowering in cultivated strawberry.

The ‘Delmarvel’ and ‘Selva’ maps presented here have a total length of 1529.3 cM and 1317.5 cM spanning 41 and 36 linkage groups, exceeding the 28 expected groups for each parent map of F. × ananassa. However, all the LGs have a marker present in other strawberry maps (Rousseau-Gueutin et al. 2008; Spigler et al. 2008, 2010; Sargent et al. 2009, 2012; Zorrilla-Fontanesi et al. 2011a; Lerceteau-Köhler et al. 2012; Castro et al. 2015; Honjo et al. 2016), which permitted map comparison and the assignment of each linkage group to one of the seven HG. The map is composed exclusively of EST-SSR and SSR markers, and therefore, will assist future QTL mapping studies in strawberry.

Antioxidant-related traits

Colocation of QTL for antioxidant-related traits was found on LG II-D-4 (anthocyanins and phenolics), V-S-3 (anthocyanins, phenolics and ORAC), VI-D-3 (anthocyanins and ORAC), VI-D-4 (anthocyanins and ORAC) and VII-S-1 (phenolics and ORAC). QTL for total phenolics were independent in IV-S-1 and in VII-D-5. In a general sense, clustering of QTL may be associated with significant correlations between traits (Paterson et al. 1991). Accordingly, if two traits are correlated, QTL for both traits are often detected in similar regions and may reflect either pleiotropic effects or the presence of closely linked genes independently affecting the two traits. The results obtained using a correlation approach showed high correlation coefficients between total anthocyanins, total phenolics and ORAC, supporting the colocations of these QTL. Also, the correlation between these three traits indicated that there is a strong relationship between each of them, most likely because anthocyanins are a class of phenolics, both of which are major contributors to antioxidant capacity. As previously mentioned by Lewers et al. (2010), the strong association between these traits in blackberry suggests that the simpler, cheaper assays for total phenolics or total anthocyanins may be useful for a more efficient indirect selection of antioxidant capacity. Therefore, the use of these assays for indirect selection for high antioxidant capacity should increase both cost and time efficiency for strawberry breeders.

Identification of markers associated with only one or two of the triad could be interpreted to contradict the conclusions drawn using the more traditional breeding approach of correlation studies to some extent. Rather, it is useful to study correlation for the purpose of determining difficulty in separating traits or determining if a less expensive assay can be used to advance multiple traits together. In later cycles of breeding, as linkages with QTL for multiple correlated traits become fixed, the correlations between traits not linked together as QTL may be lost (such as the two independent QTL for total phenolics), and markers would prove more useful in obtaining further progress.

It is well known that anthocyanins are an important type of flavonoid compound and are responsible for the red color of the berries. In the population employed in this study (‘Delmarvel’ × ‘Selva’), consistent with the findings from other two different mapping populations (‘232 × ‘1392’ and ‘Capitola’ × ‘CF116’), multiple QTL for anthocyanins have been identified suggesting that there are multiple loci controlling anthocyanins in cultivated strawberry. The FaMYB1gene and two other strawberry anthocyanin biosynthesis genes, chalcone isomerase (CHI) and flavanone 3-hydroxylase (F3H) (Deng and Davis 2001), were located on the ‘232’ × ‘1392’ genetic linkage map (Zorrilla-Fontanesi et al. 2011a), but they were not associated with QTL for anthocyanins. These three genes were not mapped in the ‘Delmarvel’ × ‘Selva’ population because no polymorphism was detected. Further work to map candidate genes will be required to determine whether these and/or other candidate genes are underlying the QTL for anthocyanins in strawberry.

Flavor-related traits

Selection for flavor is very important for consumer acceptance. The two most important flavor components for many fruits are sweetness and tartness, measured in this study by percentages of SSC and TA and their ratio (SSC/TA). Total soluble solids and titratable acids were correlated with SSC/TA as expected (across replications, 0.5269 and −0.6167, respectively). Although the total soluble solid contents were significantly and positively correlated with titratable acids, the correlation coefficient value of 0.2452 across replications indicates that selection for high soluble solids can be practiced independent of selection for low acidity. Likewise, other significant and positive correlations between titratable acids and total anthocyanins, total phenolics and ORAC were too low (0.16–0.23) to justify any concern about selecting acidic fruit in an effort to select fruit with high ORAC, anthocyanin or phenolic levels. Similar correlation coefficients between anthocyanins and titratable acidity previously have been reported in strawberry (Zorrilla-Fontanesi et al. 2011a). A correlation-based approach to breeding for the flavor components, SSC and TA, indicates that selection for these would not affect anthocyanins, phenolics, or ORAC. Similarly, selection for SSC can be independent of selection for TA, so that selections with higher SSC and lower TA would be possible from this population.

The comparison between the QTL detected in this study and those identified in different strawberry populations would allow identifying stable QTL across multiple genetic backgrounds and environments. Some of the markers (e.g., Fvi11, EMFv006, ChFaM023 and ChFaM269) associated with fruit-quality traits in previous studies (Zorrilla-Fontanesi et al. 2011a; Lerceteau-Köhler et al. 2012) were mapped in the ‘Delmarvel’ × ‘Selva’ population but they were not associated with the fruit-quality traits. The primer pairs for these markers (except EMFv006) amplified products that mapped to two or more LGs, so that they are not linkage group specific and have only tentative value for assisting in breeding. Many of the markers closely associated with the fruit-quality QTL in this population were not mapped in the previous studies (Zorrilla-Fontanesi et al. 2011a; Lerceteau-Köhler et al. 2012). Therefore, it was not possible to know if the QTL found in this study are the same as those previously reported. However, some QTL have been found in the same HG across the three studies (HG II, HG V, and HG VI for total anthocyanins; HG V and HG VI for SSC), which might indicate that those regions could be associated with the trait in different genetic backgrounds and have a higher chance of being useful in breeding. Some of the QTL identified in this study are located in HG where QTL have not been previously reported (HG VI for TA; HG IV and VII for total phenolics). These regions could be specific to the population employed in this study, although factors such as environmental differences, harvesting and sampling protocols also could lead to our different findings. Linkage maps more saturated with common transferable markers would be necessary to compare QTL positions for these traits in different populations and align the genetic linkage map of the different parents before the QTL results can be developed into practical breeding tools. Even so, for practical application, it is likely that markers for fruit-quality improvement would be specific to the breeding population in use, in other words, very few markers may be of use to multiple breeding programs for improving fruit quality.

Repeat flowering

Apart from fruit-quality traits, repeat flowering is an important focus of several strawberry breeding programs. QTL for repeat flowering have been reported in previous studies (Weebadde et al. 2008; Gaston et al. 2013; Castro et al. 2015) and reported once again in the current study. In addition, the current work maps repeat flowering qualitatively and shows that the repeat-flowering parent is heterozygous for repeat flowering when mapped qualitatively, and, therefore, that the repeat-flowering allele is dominant over the once-flowering allele at this locus. This work also reports which alleles, at all linked loci, are in coupling with repeat flowering and which are in repulsion in the repeat-flowering parent, information critical for use in breeding but very rarely reported. Recently, Honjo et al. (2016) reported similar results using qualitative mapping combined with segregation ratios from crossbreeding and inbreeding of key individuals thought to be either homozygous for the trait or heterozygous. They did not emphasize their determination of gene action as strongly as their finding that the forms of repeat flowering referred to as “dayneutrality” and “everbearing” appear to be controlled by the same locus and are likely the same trait with different names. Like Gaston et al. (2013), Honjo et al. (2016) mistakenly claim that a 1:1 segregation ratio is sufficient evidence that the trait of interest is conferred by a dominant allele, but their crossbreeding and inbreeding evidence for dominance is strong.

Evidence is accumulating that the genomic region identified in the current work is the same region identified in several other works using different germplasm and testing environments. In the current study, one genomic region associated with the total number of weeks of flowering was identified on LG IV-S-1. The genetic marker ChFaM148-182DS/184DS was linked to the QTL on LG IV-S-1 in the ‘Selva’ map and seems to be the same locus as ChFaM148-184T reported in ‘Tribute.’ Honjo et al. (2016) reported that the everbearing locus was linked to the marker FxaACA02I08C-151 and located on the same LG where the repeat-flowering trait was mapped by Castro et al. (2015). In the ‘Delmarvel’ × ‘Selva’ population, the marker FxaACA02I08C-145S was linked to repeat flowering and mapped on LG IV-S-1, indicating that the qualitatively mapped repeat-flowering trait and the QTL for total number of weeks of flowering are in the same LG. FxaACA02I08C-145S seems to be the same locus as FxaACA02I08C-151 reported by Honjo et al. (2016), though the allele size is 6 bp smaller. We have observed that all the product sizes produced by the primers reported by Honjo et al. (2016) are 6 bp smaller in our parents. Those differences are most likely due to the size of the sequences added to the primers for cost-savings and increased repeatability. Therefore, if markers will be employed for screening strawberry cultivars or to compare and align maps between different population we should keep in mind that some differences in the allele sizes may be due to primer redesign. The identification of the repeat-flowering trait in the same LG in different mapping efforts indicates reliability across population and environments, which would be useful for marker-assisted breeding in strawberry. The markers ChFaM148 and FxaACA02I08C could be good candidates for selecting repeat-flowering plants.

In spite of all the evidence supporting a single genomic region common to multiple genotypes and testing environments, the number of loci controlling strawberry repeat flowering may still be argued. Depending on how the “weak-repeat-flowering” progeny (flowering only one or two weeks after the once-flowering/repeat-flowering cutoff date) in this ‘Delmarvel’ × ‘Selva’ population were scored, the segregation ratio could support either a single-locus or two-locus model. ‘Selva’ was selected from the University of California at Davis strawberry breeding population, in which repeat flowering was reported to be conferred by a dominant allele at a single locus (Ahmadi et al. 1990), yet later was attributed to multiple loci (Shaw 2003). In the current study, a second locus unlinked to the primary one could not be identified with either the Kruskal–Wallis test or by QTL LOD significance level. Weebadde et al. (2008) showed evaluation environment can influence the number of individuals in a population that are classified as repeat flowering in a ‘Tibute’ × ‘Honeoye’ population. With this same population and evaluation data, Castro et al. (2015) identified two linked QTL, depending on where, geographically, the population was evaluated. It is possible that the discrepancies observed by many are caused by multiple linked loci in this genomic region that can be discerned only by marker saturation in this region and larger mapping population size. Until then, use of multiple markers in this region is advised for marker-assisted breeding.

It is well known that, although the utility of QTL linked to traits has been demonstrated, linked markers should be tested in different populations due to the potential exclusivity or specificity of markers within the original population used for discovery of the markers. In this study we have employed SSR and EST-SSR markers that have been previously linked to repeat flowering (Castro et al. 2015; Honjo et al. 2016). It is interesting that some of the markers linked to repeat flowering in previous studies, produced a fragment in ‘Selva’ but were not polymorphic in the population. These results indicate that ‘Selva’ could be homozygous for those loci, even though it appears heterozygous for the trait. ‘Delmarvel’ also had some fragments linked to the trait as is the case of the 184-bp fragment of the marker ChFaM148, indicating that some once-flowering cultivars also can have the fragment linked to the trait and can limit the application of the markers. As has been demonstrated many times before, the power of markers in multiple breeding populations should not be overestimated.

These results highlight the importance of having LG highly saturated with markers around the region of interest. The recent advances in F. × ananassa genomics provide a framework from which to identify additional markers and candidate genes to aid in this process. These advances include the public release of the octoploid strawberry genome (Hirakawa et al. 2014) and the development of a high-throughput genotyping tool (Axiom® IStraw90® genotyping array) for the cultivated strawberry (Bassil et al. 2015).