Genetic analyses of bolting in bulb onion (Allium cepa L.)
- First Online:
- Cite this article as:
- Baldwin, S., Revanna, R., Pither-Joyce, M. et al. Theor Appl Genet (2014) 127: 535. doi:10.1007/s00122-013-2232-4
We present the first evidence for a QTL conditioning an adaptive trait in bulb onion, and the first linkage and population genetics analyses of candidate genes involved in photoperiod and vernalization physiology.
Economic production of bulb onion (Allium cepa L.) requires adaptation to photoperiod and temperature such that a bulb is formed in the first year and a flowering umbel in the second. ‘Bolting’, or premature flowering before bulb maturation, is an undesirable trait strongly selected against by breeders during adaptation of germplasm. To identify genome regions associated with adaptive traits we conducted linkage mapping and population genetic analyses of candidate genes, and QTL analysis of bolting using a low-density linkage map. We performed tagged amplicon sequencing of ten candidate genes, including the FT-like gene family, in eight diverse populations to identify polymorphisms and seek evidence of differentiation. Low nucleotide diversity and negative estimates of Tajima’s D were observed for most genes, consistent with purifying selection. Significant population differentiation was observed only in AcFT2 and AcSOC1. Selective genotyping in a large ‘Nasik Red × CUDH2150’ F2 family revealed genome regions on chromosomes 1, 3 and 6 associated (LOD > 3) with bolting. Validation genotyping of two F2 families grown in two environments confirmed that a QTL on chromosome 1, which we designate AcBlt1, consistently conditions bolting susceptibility in this cross. The chromosome 3 region, which coincides with a functionally characterised acid invertase, was not associated with bolting in other environments, but showed significant association with bulb sucrose content in this and other mapping pedigrees. These putative QTL and candidate genes were placed on the onion map, enabling future comparative studies of adaptive traits.
Cleaved amplified polymorphic sequence
Alien monosomic addition line
Onion and shallot (Allium cepa L.) are staple functional foods central to most diets, grown from tropical regions to high latitudes. Economic production of dry bulb onion and shallot requires cultivars to be biennial. They must be adapted to local photoperiod and temperature such that a sufficiently large bulb is formed in the growing season without premature transition to flowering (‘bolting’; Kamenetsky and Rabinowitch 2002). Bulb onions are broadly classified into ‘short-day’ (SD) and ‘long-day’ (LD) types depending on the length of day required to initiate bulb formation. In terms of flowering adaptation, all onions produce flowers in the spring. True shallot of species A. cepa is a horticultural type displaying a partial to complete tendency toward asexual reproduction. It is presumed that the shallot phenotype has been selected for practicality of production in tropical regions where florigenesis is compromised by want of vernalization and fungal disease pressure on floral structures (Rabinowitch and Kamenetsky 2002).
There is an extensive literature concerning the physiology of flowering in onion, reviewed most recently by Brewster (2008). Once onion plants reach a critical weight or leaf number, they can be induced to initiate inflorescences by exposure to low temperatures. The critical size and temperature requirements for this vernalization exhibit wide genetic variation. Depending on the conditions and genetic background, there may be ‘competition’ between bulb and inflorescence development (Van Kampen 1970), such that inflorescence development may be suppressed. It has been demonstrated that a vernalization period is all that is required to initiate flowering from onion bulbs, but that temperature and photoperiod affect the time taken for eventual inflorescence appearance and floret opening (Khokhar et al. 2007). De-vernalization in onion is commercially exploited by high-temperature (>25 °C) treatment of cold-stored onion ‘sets’ prior to transplanting in springtime.
The stringent requirement for adaptation of onion cultivars to environment for producing marketable crop places major constraints on breeding and exploitation of germplasm. A deeper understanding of the genetics of temperature and photoperiod adaptation could inform strategies for genetic resource exploitation and development, and provide insights into domestication and dissemination of onion over the past millennia. Most importantly, functional markers to major genes conditioning adaptation could enable more efficient introgression of desirable traits such as disease resistance from wide crosses, as well as enabling seed lot quality control.
Physiological and genetic studies have implicated numerous genes and pathways in control of flowering in Arabidopsis and other species (Matsoukas et al. 2012). Components of the photoperiodic floral induction pathway are conserved between dicots and monocots (e.g. cereals such as rice and barley): for example, the circadian clock is needed to measure day length and downstream components such as Constans (CO; known as Heading date 1/HD1 in rice) play important roles in the photoperiodic flowering of Arabidopsis and rice (Tsuji et al. 2011). The vernalization pathway is less well-conserved and is believed to have evolved separately in grasses and dicots (Greenup et al. 2009). The photoperiodic and vernalization pathways converge on a small number of floral pathway integrators in all plants, including Suppressor of overexpression of constans 1 (SOC1) (Hepworth et al. 2002), and Flowering locus T (FT) genes (HD3a and RiceFT-LIKE1/RFT1 in rice and FT1 in wheat and barley).
Although the extensive genetic and physiological studies of grasses are often assumed to generalise to all monocots, studies of adaptive trait genetics in other monocot clades are very limited. Onion is in the order Asparagales, which diverged 122 m years ago from the monocot grasses (Janssen and Bremer 2004). Therefore, the genetic mechanisms controlling cereal flowering may differ in the Asparagales. Although physiological and genomic studies of flowering are very limited in Asparagales, several recent studies have described transcriptional and functional studies in orchid genera (Chang et al. 2011; Ding et al. 2013; Liang et al. 2012; Xiang et al. 2012). In onion, a period of vernalization was found to be all that was required to initiate flowering from onion bulbs, but that temperature and photoperiod affected the time taken for eventual inflorescence appearance and floret opening (Khokhar et al. 2007). Taylor et al. (2010) hypothesised that components of the photoperiodic pathway might be involved in the induction of onion bulbing. While they found that circadian clock genes appear to have a conserved function based on their expression, whether they are involved in onion bulbing was not established. We recently reported functional characterisation of the FT-like gene family in onion (Lee et al. 2013), identifying three members with distinctive expression patterns associated with bulbing and floral initiation. This work indicated that two antagonistically acting FT-like genes (AcFT1 and AcFT4) are involved in the photoperiodic induction of onion bulbing, while another FT (AcFT2) is involved in the vernalization responsive initiation of flowering (Lee et al. 2013).
Genetic analysis in numerous species has frequently revealed associations between allelic variation in conserved components of these pathways and integrators with flowering time. Natural variation in FT-like genes has been associated with flowering time in dicots such as sunflower and Arabidopsis (Blackman et al. 2010; Laurie et al. 2011), and monocot grasses such as ryegrass (Skøt et al. 2011) and rice (Kojima et al. 2002). Similar associations have been reported for components of the well-conserved photoperiodic pathway such as CO/HD1 in rice (Takahashi et al. 2009).
In this study, we have taken a genetic approach as the first step in understanding the genetic architecture of premature bolting (flowering) in onion. The only reported genetic analysis of onion bolting is that of Hyun et al. (2009) who examined an F2 family segregating for time to flowering using proteomic methods. This revealed that in the family examined, late flowering segregated as a dominant character, and although they identified proteomic differences among the parents they did not examine segregation of these with bolting phenotype. Physiological and genomic studies in onion have been complicated by a lack of freely available reference lines, combined with very limited genomics resources. Recent transcriptome sequencing and mapping studies based on homozygous doubled haploid (DH) lines and RNASEQ have greatly expanded scope for reproducible and comparative approaches (Baldwin et al. 2012b; Duangjit et al. 2013) as well as enabling global searches for candidate genes. We recently functionally characterised members of the FT-like gene family in onion (Lee et al. 2013). In the current study, we sought to place these and other candidate genes on the onion genetic map and to search for QTL conditioning bolting in a wide onion cross (Baldwin et al. 2012b). This revealed that a QTL located on onion chromosome one plays a major role in conditioning bolting in this cross.
Materials and methods
Onion populations used in this study
Population barcode number
NCBI BioSample accession
Type and latitude of origin
Citation or link
LD 42 N
Alan et al. (2004)
SD 20 N
SD 23 N
EU Allium Genebank
‘Criolla molonga’ Guatemalan landrace
SD 14 N
EU Allium Genebank
‘Colorada de conservar’
LD 43 N
EU Allium Genebank
HRIGRU 9874 http://ealldb.ipk-gatersleben.de/
LD 41 N
EU Allium Genebank
HRIGRU 11567 http://ealldb.ipk-gatersleben.de/
‘Texas Grano 438’
Origin SD 26 N, selected at 37 N
Seminis vegetables seeds, Plant and Food Research
McCallum et al. (2006a)
LD 43 N
University of Wisconsin
F2 families (‘2262’ and ‘G2320’; Baldwin et al. 2012b) were derived by self-pollinating individual F1 plants from the cross between the doubled haploid (DH) onion line ‘CUDH2150’ and the heterozygous landrace ‘Nasik Red’. Families were increased further by cage-pollinating clonal topsets from the original F1 umbels. These families were sown in spring in 2008/9 (sowing date 25 Sept 2008) and 2011/12 (sowing date 9 Sept 2011) seasons at Lat 42° S near Christchurch, New Zealand and in 2011/12 (sowing date 29 August 2011) at Lat 37° S near Pukekohe, New Zealand. A plant was scored as bolting if it exhibited a clearly elongated flower scape and expanded umbel at harvest in early March. Bulb carbohydrate analysis of these populations has been described previously (Revanna et al. 2013).
DNA extraction and molecular marker analysis
DNA was isolated from F2 plants of the CUDH2150 × Nasik Red family ‘2262’ in the 2008/9 season from fresh leaf material or freeze-dried bulb tissue of bolting and non-bolting plants as described previously (Baldwin et al. 2012b; McCallum et al. 2006a). For validation in the 2012 season, template DNA was sampled directly from bulbs by stabbing fleshy scales with a 200 μl pipette tip (Axygen). The tip was then placed into a 50 μl volume solution containing 1 × buffer Gold and 1 μL prepGEM reagent (prepGEM™ Tissue kit; ZyGEM Corp, Hamilton, New Zealand). This was incubated for 15 min at 75 °C with a 5 min hold at 95 °C. An aliquot of the solution was then diluted in half with TE buffer (10 mM Tris and 0.1 mM EDTA) and the remainder stored at −20 °C. SSR, CAPS and HRM marker analyses were carried out as described previously (Baldwin et al. 2012b) but the HRM amplification conditions were modified to include a touchdown PCR. The modified conditions were: 95 °C for 15 min followed by seven cycles of 94 °C for 30 s, 62 °C for 30 s (reducing by 1 °C per cycle), and 72 °C for 15 s. Another 45 cycles were carried out at 94 °C for 30 s, 55 °C for 30 s and 72 °C for 15 s. Then, a final hold temperature of 25 °C for 2 min.
Primer pairs used for amplicon re-sequencing of flowering candidate genes
Primer pairs used and nucleotide diversity of genomic loci from parallel-tagged sequencing of candidate genes
Amplicon size bp
CUDH2150 reference assembly Genbank accession
Reference cDNA Genbank accession
Reads from the seven heterozygous populations were mapped onto the CUDH2150 assemblies using BWA_SW (Li and Durbin 2010; http://www.ncbi.nlm.nih.gov/pubmed/19451168). SNPs were called using samtools mpileup (Li et al. 2009), patterns of nucleotide diversity were determined using PoPoolation (Kofler et al. 2011a) and allele frequencies were compared among genes and populations using PoPoolation2 (Kofler et al. 2011b). A reproducible workflow containing scripts and all settings for read mapping, analysis and visualisation can be accessed at GitHub in the repository https://github.com/cfljam/onion_PTS_workflow. Variant effects were determined using snpEff (Cingolani et al. 2012). Primer sets for indel, HRM or CAPS markers were designed to variants of interest as described previously (Baldwin et al. 2012b).
Candidate gene mapping
Candidate genes that were not segregating in the F2 population ‘2262’ were tested on the ‘W202A × Texas Grano 438’ mapping population (McCallum et al. 2006a, b) or on sets of A. fistulosum–A. cepa alien monosomic addition lines (AMALs) developed by Shigyo and colleagues (1996). Linkage maps were produced using JoinMap v4 (Van Ooijen 2006) as described previously (Baldwin et al. 2012b).
Genetic map locations of flowering candidate gene sequences in onion
Map position (cM)
W202A × Texas Grano
Nasik Red × CUDH2150 (2262)
ACP675 and ACP676
CAPS-AluI and RsaI
W202A × Texas Grano
W202A × Texas Grano
Nasik Red × CUDH2150 (2262)
ACI046 and ACP681
InDel and CAPS-AluI
W202A × Texas Grano
Nasik Red × CUDH2150 (2262)
Candidate gene identification and re-sequencing
PCR-based markers designed to selected variants using Galaxy-based tools (Baldwin et al. 2012b) were either placed on the onion genetic map or assigned to physical chromosomes where mapping was not possible using AMAL (Table 4). The fixation of several candidate loci in the inbred line ‘W202A’ enabled mapping of several variants in the ‘W202A × Texas Grano 438’ mapping population. Candidate genes tested were present on chromosomes 1–6 with chromosome 6 containing three candidates including two FT-like genes (AcFT4 and AcFT5). Using molecular markers all six onion FT-like genes were assigned to the chromosomes of onion along with other major candidates. Map position could not be assigned for AcFT4 along chromosome 6 because none of the markers were polymorphic in the populations tested. Few of the candidates could be mapped in the ‘Nasik Red × CUDH2150’ F2 population.
Segregation of bolting phenotype in mapping populations
Frequencies of bolting observed at harvest in progeny of two F2 families from the ‘CUDH2150 × Nasik Red’ cross grown in three environments
10 % (50/500a)
31 % (184/598)
10 % (29/277)
25 % (144/578)
12 % (54/453
Molecular marker and QTL analysis
Genetic map positions and estimated variance explained by putative QTL associated with bolting in the ‘2262’ population phenotyped in the 2008/9 season at Lincoln
% variance explained
Genotype counts and Chi-square goodness-of-fit tests for marker ACP267 linked to the QTL AcBlt1 over seasons and populations
χ2 2 df
‘Nasik red’ homozygote
‘Nasik red’ homozygote
On chromosome 6, homozygous ‘Nasik Red’ genotypes for marker ACP395 bolted (14/14). Comparison with the ‘W202A × Texas Grano438’ linkage map would put this marker closest to the candidate genes AcCO-like and AcFT5, which would both be distal to the end of the ‘Nasik Red × DH2150’ chromosome 6 linkage group (Fig. 4). Possible interactions between the loci were detected and tested but the 2008/9 sample size was too limited to draw any accurate conclusions.
Based on expression and complementation studies, we previously concluded that there is strong evidence of a functional role in regulation of bulbing and flowering of three of the six FT-like genes we identified in onion, which we designated AcFT1, AcFT2 and AcFT4 (Lee et al. 2013). We hypothesised that AcFT1 and AcFT4, respectively, promote and repress bulbing, and that AcFT2 may be the key signal for florigenesis induced in vernalized meristems. Interactions among these FT-like genes may condition the developmental transitions among the juvenile, ‘thermophase’, and ‘competition’ phases of onion floral development described in the classical model of Van Kampen (1970). The evidence of population differentiation identified for AcSOC1 with the re-sequencing data is interesting as this could not be placed on the ‘Nasik Red × DH2150’ linkage map but was physically located to chromosome 1 and hence, remains a candidate for the AcBlt1 locus. SOC1 is one of six flowering time genes within the days to heading QTL (dth1.1) interval on chromosome 1 in rice (Thomson et al. 2006). MAS was used to produce sub-introgression lines (SILs) containing one or more of these six candidate genes from Oryza rufipogon in an O. Sativa background. Only the SIL with an introgression containing SOC1 and a Flowering time locus T (FT-L8) was consistently early in all experiments (Maas et al. 2010). In onion, the association of AcBlt1with bulb weight in the 2008/9 trial suggests that AcBlt1 may condition variability in the ‘competition’ phase of floral development (Brewster 2008; Van Kampen 1970), so that bulb formation is favoured over inflorescence development.
Our observation that the AcSuc locus conditions variation in bulb sucrose levels and that it may be associated with bolting in some environments suggests that further investigation of its role in adaptive physiology is worthwhile. Physiological studies have demonstrated that acid invertase activity in onion is very strongly regulated by environment. Photoperiodic regulation of acid and neutral invertases was reported by Lercari (1982), and Benkeblia et al. (2004) reported that acid invertase activity in bulbs of cv ‘Tienshan’ is higher at 10 °C storage compared to 20 °C. Further indirect support for a functional role of acid invertase in bolting is suggested by the origins of the tulip acid invertase cDNA clone used by Vijn et al. (1998) to clone the onion gene corresponding to API89. This clone was shown to encode an acid invertase that was strongly up-regulated by cold and shown to be involved in driving flower stalk elongation (Balk and de Boer 1999). Correlation between carbon physiology traits including invertase activities and flowering date have been reported in maize (Thévenot et al. 2005). Based on the reported effects of environmental factors on onion acid invertase, we would expect that our strategy of using large, non-replicated progenies would be inefficient for validating the effects of any locus conditioning genetic variation in this activity. This is emphasised by the large variability for sucrose content observed among F2 bulbs of ‘Nasik Red × CUDH2150’.
The QTL AcBlt1 had the largest estimated effect on bolting explaining 17 % of the additive variation and this is likely to have been over-estimated given the selective genotyping strategy used in this study. Buckler et al. (2009) detected 29–59 additive QTL for flowering time traits in maize. It was suggested that this could be explained by the out-crossing nature of maize compared to the other self-pollinating model species such as Arabidopsis and rice, and that small-effect QTLs may have been advantageous to ensure that offspring where likely to have overlapping flowering times. Interestingly the putative QTL coinciding with AcSuc on chromosome 3 had a dominant effect. This would be difficult to fix in out-breeding crops such as onion without MAS.
The ‘Nasik Red × CUDH2150’ populations employed in this research are the products of a cross between genetically distant (Baldwin et al. 2012a; McCallum et al. 2008) and divergently adapted parent lines from SD and LD backgrounds. Generating F2 onion populations by mass-pollination of F1 bulbils, induced spontaneously or by cytokinin treatment (Andrew 1951; Thomas 1972), allows evaluation of large samples across multiple environments. The principal disadvantage of this strategy is that it does not permit comparison of replicated combinations of the same allelic combinations over environments, such as those conducted in studies of bulb composition (McCallum et al. 2006b). However, developing a sufficient number of inbred or backcross lines from such a cross would be highly challenging due to the severe impact of adaptive loci on bulb storage and seed production. The more limited population sizes practical using inbred lines limit the power to detect epistatic interactions, and also dominance variance, which is likely to be of great significance in adaptive and yield traits (Wardyn et al. 2007). Availability of parent lines from public sources, gene-based maps and falling costs of genotyping mean that our findings may be reproduced and tested by other researchers. Developing a deeper understanding of the genetic architecture of onion adaptation will require use of many such crosses among landrace and domesticated germplasm, as well as diverse panels phenotyped across environments such as that previously conducted in the tropic regions by Currah and Proctor (1990).
Comparative mapping with the ‘W202A × Texas Grano’ population map and the placement of candidates on this map are important as these lines are from genetic backgrounds widely used in breeding. An obvious extension to this research would be an association mapping strategy using the candidate gene sequences and markers identified from the QTL analysis in more diverse SD and LD germplasm. The action of AcBlt1 and AcSuc in developmental physiology could be determined by analyzing sets of progeny classified by QTL-linked marker genotype and then subjected to different environmental regimes under controlled environments or different planting dates. Given the difference in flowering habits across the vegetable Alliums, comparative mapping with shallot (generally asexually propagated) and closely related Allium such as Japanese bunching onion (A. fistulosum) and garlic (clonally propagated) would also help further understand the major genes involved in controlling flowering in the Asparagales.
Utilisation of exotic germplasm for breeding and selection by breeders to introduce novel traits is severely hampered by bolting. Although it is an easy trait to visually select, dominance could mask inferior genotypes and slow the adaptation. MAS for a recessive trait would be highly desirable for breeders trying to utilise crosses between mal-adapted and elite, well-adapted lines. The application of the marker associations detected in this research could be tested in larger breeding collections or genebanks to assess the predictive values for MAS programmes.
This research was funded by the NZ Ministry for Business, Innovation and Employment. We gratefully acknowledge provision of germplasm by the EU Allium Genebank (University Of Warwick) and Cornell University Department of Plant Breeding and of monosomic alien addition line DNA samples by Masayoshi Shigyo (Yamaguchi University, Japan). Field trialling support was provided by Enza-Zaden NZ Ltd, Allium Solutions Ltd., Seminis NZ Ltd and FruitFed Supplies. We thank Zygem Corp (Hamilton, New Zealand) for provision of reagents.
Conflict of interest
The authors declare that they have no conflict of interest.