Waxy (left) vs. glossy (right) foliage on F2 progenies from the cross of B9885 × B8667 of onion.
Epistatic interaction between genotypes on chromosome 1 (isotig31382_1702) vs. chromosome 8 (isotig20266_1040) affecting amounts of hentriacontanone-16 in the F2 family from B9897 × B5351 of onion. A indicates allele from the female parent (B9897) and B is the allele from the male parent (B5351).
Click on author name to view affiliation information
Onion (Allium cepa) plants with lower amounts of epicuticular waxes on foliage suffer less damage from the insect pest Thrips tabaci (onion thrips). Glossy onion accumulates significantly less epicuticular wax compared with wild-type “waxy” onion, and a single recessive locus (gl) has been proposed to condition this phenotype. Genetic analyses of types and amounts of epicuticular waxes were completed using two segregating families from the cross of the glossy inbreds B9885 and B9897 (both originally selected from the onion cultivar White Persian) with waxy inbred B8667 and semiglossy (intermediate amounts of waxes) inbred B5351, respectively. F2 progenies were grown in greenhouses and scored visually for foliar phenotypes, and amounts and types of epicuticular waxes were determined using gas chromatography-mass spectrometry (GCMS). For one F2 family from the cross of glossy B9885 by waxy B8667, visually scored glossy vs. waxy foliage fit a 1:3 ratio and the phenotype mapped to chromosome 8 of onion. This same region on chromosome 8 was significantly associated with amounts of the ketone hentriacontanone-16 (H16) and fatty alcohols 1-octacosanol (Oct1) and 1-triacontanol (Tri1). Visually scored F2 progeny from the cross of glossy B9897 × semiglossy B5351 did not fit expected models for one or two recessive loci. Significant quantitative trait loci (QTL) were revealed on chromosomes 5 and 8 controlling amounts of H16. Epistasis was detected between regions on chromosomes 1 and 8, and a 100-fold increase of H16 was conditioned by homozygous genotypes for the B5351 region on chromosome 1 and the B9885 region on chromosome 8. The three QTL model explained 41% of the phenotypic variation for amounts of H16 at logarithm of odds of 16.6. Amounts of Oct1 and Tri1 in the B9897 × B5351 family were associated with a major QTL on chromosome 1, explaining 37% to 46% of the phenotypic variation, respectively. This research demonstrates that glossy foliage of ‘White Persian’ onion is conditioned by a recessive locus on chromosome 8 for which we propose the name glwp. These results are important for selection of onion with unique profiles of epicuticular waxes to reduce losses resulting from onion thrips.
Waxes accumulate on the surface of plant leaves and are important for avoidance of abiotic stresses such as aerial desiccation (Cameron et al., 2006; Kim et al., 2007), water deprivation (Aharoni et al., 2004; Bourdenx et al., 2011), and damaging wavelengths of light (Holmes and Keiller, 2002). Wax composition varies among plants (Bernard and Joubès, 2013; Jenks et al., 1995), and in arabidopsis (Arabidopsis thaliana), the predominant waxes are C29 alkanes, ketones, and secondary alcohols (Lee and Suh, 2013). In arabidopsis, the eceriferum (cer) mutants have brighter green leaves compared with wild-type plants and have been used to study wax biosynthesis. Production of leaf waxes starts in the leucoplast with synthesis of C16 to C18 fatty acyl-coenzymes A (CoAs) by long-chain acyl-CoA synthetase 1 (LACS1) and 2 (LACS2) enzymes (Lü et al., 2009), follow by carbon elongation by the fatty acid elongase (FAE) complex to produce very long-chain (>C18) fatty acyl-CoAs (VLFA-CoAs) (Yeats and Rose, 2013). The FAE complex in arabidopsis consists of four enzymes that extend fatty acids by two carbons per cycle: β-keto-acyl-CoA synthase (KCS), β-ketoacyl-CoA reductase, 3-hydroxyacyl-CoA dehydratase, and trans2,3-enoyl-reductase (Fiebig et al., 2000; Haslam et al., 2012, 2015). The KCS genes have shown substrate specificity for acyl-CoAs of different carbon chain length (Haslam and Kunst, 2013; Samuels et al., 2008). Hydrolysis of VLFA-CoAs by a thioesterase generates very long-chain fatty acids that can enter two pathways: the acyl-reduction pathway, forming primary alcohols (Rowland et al., 2006) and wax esters (Li et al., 2008); and the decarbonylation pathway, which forms secondary alcohols, aldehydes, alkanes, and ketones (Samuels et al., 2008).
Wax coverage and composition are important for biotic interactions providing stimuli that attract or repel insects (Eigenbrode and Espelie, 1995). The glossy phenotype of onion has light-green leaf color, accumulates significantly less epicuticular waxes on its foliage compared with wild-type “waxy” onion, and suffers less damage by onion thrips (Damon et al., 2014; Jones et al., 1934; Molenaar, 1984; Munaiz et al., 2019). The main epicuticular wax on the leaves of wild-type onion and leek (Allium ampeloprasum) is the ketone H16, followed by lower amounts of fatty alcohols and alkanes (Damon et al., 2014; Rhee et al., 1998). Scanning electron microscopy of epicuticular waxes on onion leaves revealed crystals associated with the phenotypic appearance of leaves (Damon et al., 2014; Molenaar, 1984; Yang et al., 2017). Specific wax constituents may form crystals on the leaf surface, and different components produce unique shapes (Gülz et al., 1992; Jenks et al., 1995). Visual crystals are produced on the surface of waxy onion leaves when the amount of H16 represents more than two thirds the total wax load, and they are responsible for the blue-green foliar color of waxy onion (Damon et al., 2014; Molenaar, 1984).
Jones et al. (1944) crossed glossy and waxy plants from the onion cultivar Australian Brown and completed genetic studies revealing that a single locus (gl) controlled glossy foliage. Crosses involving glossy plants from the onion cultivar White Persian (Jones et al., 1934) did not fit a single recessive model because of too few glossy segregants (Jones et al., 1944). Molenaar (1984) studied the glossy phenotype from ‘White Persian’ and proposed that it was conditioned by a single recessive locus, which he assumed to be gl. However, Jones and Mann (1963, p. 81) reported that crosses between glossy plants from ‘Australian Brown’ and ‘White Persian’ produced waxy hybrids, indicating that different loci condition the glossy phenotype in these two populations. In bunching onion (A. fistulosum), the glossy phenotype is conditioned by a single recessive locus (Yang et al., 2017). The goal of this study was to determine the genetic basis of the glossy phenotype from ‘White Persian’, which has been shown to suffer less damage by onion thrips in numerous studies (Damon et al., 2014; Jones et al., 1934; Molenaar, 1984; Munaiz et al., 2019).
Glossy inbreds B9885 [U.S. Department of Agriculture (USDA) plant introduction (PI) 546303] and B9897 (PI 546305) of onion were developed by backcrossing glossy foliage from ‘White Persian’ (Jones et al., 1934) to long-day storage germplasm with yellow bulb color and were released in 1983 by the USDA and the experiment stations of Iowa, Michigan, and Wisconsin (Goldman et al., 2001; Molenaar, 1984). Inbred B8667 has waxy foliage and red bulbs (Havey and Bohanec, 2007); inbred B5351 has semiglossy foliage (Damon et al., 2014). Seeds of these inbreds were planted in field plots at the Dean Kincaid Farm (Palmyra, WI) under normal production conditions and were scored visually to confirm foliar phenotypes. Bulbs were harvested at maturity, stored for 6 months at 7 °C, and planted in a greenhouse in a soilless medium (Metro-Mix; Sun Gro Horticulture, Agawam, MA). Plants were grown under 14-h days and temperatures of 27 °C days and 22 °C nights, and fertilized every week with 20N–8.7P–16.6K (Peters Professional, Allentown, PA). Leaves were sampled for analyses of epicuticular waxes by GCMS as described later.
Glossy B9885 was crossed as the female parent with B8667, and hybrids identified by waxy foliage and red bulbs. Individual hybrid plants were self-pollinated to produce segregating F2 families. Seeds from F2 families were planted in field plots as described earlier and were scored visually for glossy vs. waxy foliage. Bulbs from one F2 family were harvested at maturity, stored for 90 d at 7 °C, planted, and grown in a greenhouse as described previously. A second family was developed by crossing glossy B9897 as the female with semiglossy B5351. Hybrids were identified by waxy foliage and self-pollinated. Seeds from the parental inbreds and one F2 family were planted in a greenhouse as described previously, and the plants were grown for 8 weeks as described earlier. For both families, foliage was scored visually as waxy vs. nonwaxy and sampled using GCMS as described later.
Leaf tissue was collected from at least five plants of the parental inbreds and each segregating progeny, frozen in liquid nitrogen, lyophilized, and stored at –20 °C. Samples were ground by adding metal spheres and shaking for 45 s until the leaf tissue was pulverized. Genomic DNA was extracted (NucleoSpin Plant II Midi Kit; Macherey-Nagel, Düren, Germany), and purity and concentration were evaluated spectrophotometrically (NanoDrop; Thermo Fisher Scientific, Waltham, MA).
DNAs from 93 F2 progenies of the cross B9885 × B8667 were genotyped using the 1692 SNP array (Havey and Ghavami, 2018). Goodness-of-fits to the expected 1:2:1 segregation for codominant markers were tested using chi-square analyses, and the linkage map was created using JoinMap 4.0 (Van Ooijen, 2006) with regression mapping and the Kosambi function. DNAs of 98 F2 plants from the B9897 × B5351 cross were genotyped for 232 SNPs (Duangjit et al., 2013) using the KASPar assay (LGC Genomics, Boston, MA). Genotypic scores were verified using SNP viewer 2 version 4.0.0 software (LGC Genomics). Segregation analysis and construction of the linkage map were completed as described earlier. Linkage groups from both maps were assigned to chromosomes using common SNPs that segregated in the mapping populations OH1 × 5225, BYG15-23 × AC43, or DehyA × B5351 (Damon and Havey, 2014; Duangjit et al., 2013).
Two samples from one fully extended leaf were collected from five plants of the parental inbreds and from each F2 progeny from the B9885 × B8667 family at 45 d after planting of bulbs in a greenhouse. For the B9897 × B5351 family, two samples from one fully extended leaf from each F2 progeny were collected 8 weeks after planting of seed in the greenhouse. For all samplings, the chosen leaf was fully extended and equidistant between the oldest and youngest leaves. Individual leaf samples were weighed, and docosane (Sigma-Aldrich, St. Louis, MO) dissolved in high-performance liquid chromatography (HPLC) reagent-grade chloroform (Fisher Scientific, Hampton, NH) was added to leaf pieces at a rate of 100 µg·g–1 fresh weight. Each leaf segment was then submerged in HPLC-grade chloroform (Fisher Scientific) for 1 minute and discarded. The chloroform was allowed to dry in a fume hood for ≈5 d. In preparation for GCMS, wax extracts were dissolved in 500 μL HPLC-grade chloroform, 600 μL acetonitrile (Fisher Scientific), and 210 μL HPLC/GC-grade N,O bis(trimethylsilyl)trifluoro-acetamide (Sigma-Aldrich), and incubated for 30 min at 80 °C. Identification and quantification of waxes used GCMS conditions as described previously (Damon et al., 2014).
Peak areas of individual waxes were adjusted using areas of the docosane peaks in each run and therefore represent amounts of individual waxes on a gram fresh weight basis. Adjusted peak areas from the two samples from the same leaf were averaged and used for all analyses. Analyses of variance and Spearman correlations for amounts of individual waxes were calculated using RStudio (R Foundation for Statistical Computing, Vienna, Austria). Visually scored foliar phenotypes and amounts of individual waxes were analyzed using a binary model and composite interval mapping (CIM), respectively, with the R/qtl package (Broman and Sen, 2009; Broman et al., 2003). Haley Knott regression with 10 marker covariates within a window size of 10 cM was used to determine significant associations (Haley and Knott, 1992), and a 95% significance logarithm of odds (LOD) threshold was used after 1000 permutations. Multiple QTL were detected using the stepwiseqtl function, dropping one marker at a time, with parameters of 0.05 significance level and 1.5-LOD score. Effects of the candidate QTL were refined using makeqtl, refineqtl, and fitqtl, and the estimated effects provided the percentage of the phenotypic variation explained by the QTL. Epistatic interactions among loci were evaluated using scannonetwo, sim.geno, and the penalized LOD score.
GCMS analyses of the parental inbreds revealed that the three most predominant waxes were the ketone H16 and fatty alcohols Oct1 and Tri1 (Table 1), which is in agreement with Damon et al., (2014). Glossy inbreds B9885 and B9867 had significantly lower amounts of H16 than semiglossy B5351 and waxy B8667, and B5351 had significantly less H16 than waxy B8667 (Table 1). Amounts of Oct1 and Tri1 were not significantly different among these inbreds (Table 1). The Spearman correlation between the two fatty alcohols (Oct1 and Tri1) was highly significant at 0.92 (P < 0.001), and correlations between amounts of H16 and the two fatty alcohols (Oct1 and Tri1) were also highly significant (P < 0.001) at 0.77 and 0.78, respectively.
The cross of glossy B9885 by waxy B8667 produced hybrids with waxy foliage, and self-pollination of six hybrids yielded F2 families segregating for foliar glossiness. Visual scoring of glossy vs. waxy foliage (Fig. 1) yielded 20 glossy and 191 waxy F2 progenies for segregations combined across families after confirming homogeneity of errors, which did not fit the 1:3 ratio for a single recessive locus (P < 0.001). Jones et al. (1944) also reported too few glossy progenies in F2 families from crosses of glossy plants from ‘White Persian’ with waxy plants; however, Molenaar (1984) reported acceptable fits to a single recessive locus in crosses between B9885 and waxy inbred MSU611. Segregations for foliar glossiness in one F2 family from B9885 by B8667 fit the 1:3 ratio (P = 0.355), and genotyping revealed 499 SNPs segregating in this family. One SNP was discarded because there were no maternal homozygotes and segregations did not fit the expected 1:3 ratio for a dominant locus (P < 0.001). All the remaining 498 SNPs showed goodness-of-fits at P > 0.002 and were used for genetic mapping. A total of 490 SNPs segregated into eight linkage groups at LOD 4.0, with a total map length of 8.98 Morgans (Supplemental Table 1). Eight SNPs remained unlinked and were not used for quantitative analyses.
Citation: Journal of the American Society for Horticultural Science J. Amer. Soc. Hort. Sci. 145, 1; 10.21273/JASHS04840-19
Mapping of glossy vs. waxy foliage as a binary trait in the F2 family from B9885 by B8667 revealed a major QTL on chromosome 8 at 46.2 cM explaining 65% of the phenotypic variation (Table 2). A significant QTL was revealed by CIM controlling amounts of H16 on chromosome 8 at 41 cM (LOD 15.8) and explained 46% of the phenotypic variation (Table 2). The chromosome region from the waxy parent B8667 had the additive effect of increasing adjusted peak areas for H16 by 0.57, with a dominance effect of 0.33. The most significant SNP on chromosome 8 associated with amounts of H16 was isotig19082_1721, which mapped very closely to visually scored glossy vs. waxy phenotypes (Table 2). Although amounts of Oct1 and Tri1 were not significantly different between B9885 and B8667 (Table 1), CIM revealed a QTL on chromosome 8 at 46.2 cM (LODs 3.7 and 5.5 for Oct1 and Tri1, respectively) associated with amounts of these fatty alcohols and mapping to the same location on chromosome 8 affecting amounts of H16 (Table 2). The QTL explained 17% of phenotypic variation for amounts of Oct1 (Table 3), and the chromosome region from B8667 showed an additive effect of 0.36 and a dominance effect of 0.31 (Table 2). For Tri1, this same QTL explained 24% of the phenotypic variation, with the region from B8667 increasing amounts of Tri1 with an additive effect of 0.30 and a dominance effect of 0.19 (Table 2). These results demonstrate that a region on chromosome 8 from B9885 is associated with the visually glossy phenotype and reduced amounts of the ketone H16 and fatty alcohols Oct1 and Tri1.
The cross of glossy B9897 by semiglossy B5351 produced waxy hybrids that were self-pollinated. Visually scored F2 progenies into two phenotypic classes (9 glossy vs. 81 nonglossy) did not fit models for one or two recessive loci (P < 0.001) as a result of a paucity of glossy plants. This observation is consistent with F2 segregations reported by Jones et al. (1944) and from our B9885-by-B8667 families, in which too few glossy plants appeared. Genotyping of F2 progenies from B9897 × B5351 resulted in 142 segregating SNPs, and mapping yielded eight linkage groups at LOD 8.0 with a total map length of 10.4 Morgans ((Supplemental Table 2).
Significant QTL were revealed by CIM controlling amounts of H16 on chromosome 8 at 53 cM (LOD 13.3) and chromosome 5 at 38 cM (LOD 7.6), explaining 21% and 11% of the phenotypic variation, respectively (Table 3). The two QTL model accounted for 38% of the phenotypic variation at LOD 18.1. The region on chromosome 5 from B5351 showed an additive effect reducing H16 amounts, in agreement with Damon and Havey (2014). The region on chromosome 8 from the semiglossy parent (B5351) showed an additive effect increasing the amount of H16 (Table 3). Interactions between the QTL on chromosomes 5 and 8 were not significant. Amounts of H16 were affected by an epistatic interaction between regions on chromosomes 1 and 8 (Fig. 2). Significantly greater amounts of H16 were conditioned by homozygous regions on chromosome 1 from semiglossy B5351 and on chromosome 8 from glossy B9885. Chromosome 1 had no effect on amounts of H16 when the region from B9885 was present, regardless of the genotype in chromosome 8 (Fig. 2). There was no interaction between the QTL on chromosomes 1 and 5 (P = 0.773). The three QTL model (chromosomes 1, 5, and 8) for amounts of H16 explained 54% of the phenotypic variation at LOD 16.6.
Citation: Journal of the American Society for Horticultural Science J. Amer. Soc. Hort. Sci. 145, 1; 10.21273/JASHS04840-19
A region on chromosome 1 at 0 cM was detected by CIM affecting amounts of Oct1 and Tri1, explaining 37% of the phenotypic variation at LOD 9.7 for Oct1 and 46% of the phenotypic variation at LOD 13.7 for Tri1 (Table 3). The region from semiglossy B5351 reduced amounts of Oct1 and Tri1. No significant QTL affecting amounts of Oct1 or Tri1 were detected on chromosome 8 as observed in the B9885-by-B8667 family.
We studied the genetic basis of the glossy phenotype of onion using two inbred lines (B9885 and B9897) derived from ‘White Persian’ onion (Goldman et al., 2001; Molenaar, 1984). Although both B9885 and B9897 have glossy foliage, there were differences for major QTL affecting amounts of epicuticular waxes. In the segregating family from B9885 crossed with B8667, a region on chromosome 8 was significantly associated with the visual glossiness of the foliage and amounts of the ketone H16 and fatty alcohols Oct1 and Tri1 (Table 2). In the B9897 × B5351 family, the same region on chromosome 8 significantly affected amounts of H16, but not Oct1 and Tri1 (Table 3). The region on chromosome 8 in the B9897 × B5351 family interacted with one on chromosome 1 to increase the numbers of waxy progenies (Fig. 2). This interaction likely contributed to observations by Jones et al. (1944) and our research of too few glossy progenies in F2 families segregating for the glossy phenotype from ‘White Persian’. The region on chromosome 1 from glossy B9897 may act as a dominant repressor of H16 accumulation. In the segregating family from B9897 × B5351, we identified a significant QTL on chromosome 5 affecting amounts of H16 (Table 2), which is in agreement with the work by Damon and Havey (2014), who detected this same region in the cross of waxy DehyA × B5351. The QTL on chromosome 5 showed no significant interaction with chromosome 1 for amounts of H16.
In the glossy B9885-by-waxy B8667 family, the region on chromosome 8 was significantly associated with amounts of Oct1 and Tri1 (Table 2). However, in the family from glossy B9897 by semiglossy B5351, amounts of Oct1 and Tri1 were significantly associated with a region on chromosome 1 (Table 3) and no effect was detected on chromosome 8. In previous research (Damon and Havey, 2014), we detected a region on chromosome 2 from semiglossy B5351 reducing amounts of Oct1 and Tri1. Because chromosomes 2 and 8 had no significant effect on amounts of Oct1 and Tri1 in the B9897 × B5351 family, it may be there was no variation between B9897 and B5351 for these QTL.
Jones et al. (1944) studied the genetics of the glossy phenotype from onion cultivars Australian Brown and White Persian. A glossy plant from ‘Australian Brown’ was self-pollinated for two generations and progenies were crossed with waxy plants from ‘Australian Brown’ to produce an F2 family. Visually scored foliar phenotypes indicated that a single locus (gl) conditioned glossy foliage (Jones et al., 1944). Segregating families from crosses of glossy plants from ‘White Persian’ with waxy plants resulted in a paucity of glossy progenies. Jones and Mann (1963) reported that crosses between glossy plants from ‘White Persian’ and ‘Australian Brown’ produced waxy hybrids, and therefore different loci must condition the glossy phenotype in these two populations. In our current research, genetic mapping of the glossy phenotype from ‘White Persian’ revealed a region on chromosome 8 associated with the visual foliar phenotype and amounts of H16. Because this region was detected in two segregating families, we propose the name glwp for a recessive locus on chromosome 8 from ‘White Persian’ (wp) that reduces the accumulation of epicuticular waxes on onion foliage.
Our research demonstrates that major QTL on chromosomes 5 and 8 affect amounts of H16, the main epicuticular wax on onion foliage (Tables 2 and 3) (Damon and Havey, 2014). Amounts of the two most predominant fatty alcohols (Oct1 and Tri1) are highly correlated and controlled by QTL on chromosomes 1 (Table 3), 2 (Damon and Havey, 2014), and 8 (Table 2). Lower amounts of H16 relative to the other epicuticular waxes are significantly associated with reduced damage by onion thrips (Damon et al., 2014; Munaiz et al., 2019). The glossy phenotype of onion has been reported as susceptible to powdery mildew (Mohan and Molenaar, 2005), which limits its usefulness for commercial production. However, selection of onion for lower amounts of H16 conditioned by QTL chromosomes 5 or 8 and greater amounts of the other wax components, such as fatty alcohols on chromosomes 1 and 2, may produce onion populations that suffer less damage by onion thrips and accumulate enough total epicuticular wax to be commercially acceptable.
Waxy (left) vs. glossy (right) foliage on F2 progenies from the cross of B9885 × B8667 of onion.
Epistatic interaction between genotypes on chromosome 1 (isotig31382_1702) vs. chromosome 8 (isotig20266_1040) affecting amounts of hentriacontanone-16 in the F2 family from B9897 × B5351 of onion. A indicates allele from the female parent (B9897) and B is the allele from the male parent (B5351).
Contributor Notes
We gratefully acknowledge the support of U.S. Department of Agriculture (USDA) -National Institute of Food and Agriculture grants 2008-51180-04875 from the Specialty Crops Research Initiative and 2016-6013-24590 from the Organic Research and Extension Initiative.
Names are necessary to report factually on available data; however, the USDA neither guarantees nor warrants the standard of the product, and the use of the name by the USDA implies no approval of the product to the exclusion of others that may also be suitable.
M.J.H. is the corresponding author. E-mail: michael.havey@usda.gov.
Waxy (left) vs. glossy (right) foliage on F2 progenies from the cross of B9885 × B8667 of onion.
Epistatic interaction between genotypes on chromosome 1 (isotig31382_1702) vs. chromosome 8 (isotig20266_1040) affecting amounts of hentriacontanone-16 in the F2 family from B9897 × B5351 of onion. A indicates allele from the female parent (B9897) and B is the allele from the male parent (B5351).
