Plant materials.

Elite taxa were collected from botanical gardens, arboreta, and nurseries (Table 1) as either containerized plants or cuttings. Many taxa were labeled with both cultivar and trademark names. However, for H. syriacus and many ornamental taxa, usually one name becomes common in the nursery trade as the “market name.” For simplicity, only market names (cultivar or trademark) will be used hereafter.

Table 1.

Hibiscus syriacus germplasm source and internal accessions.

View Table

Crosses.

Crosses were made during summer in a glasshouse kept free of pollinators with day/night set temperatures of 25/20 °C and a 16-h photoperiod. Flowers were open for 2 days before stigmas reflexed in an effort to self-pollinate. Therefore, flowers were pollinated in the morning of their first flowering and stigmas were thoroughly covered with a dense layer of pollen. Fresh pollen was collected on the day of pollination. Because of the spatial separation of pollen and stigma, and because of pollen morphology, emasculation was unnecessary. Pollen of H. syriacus is large (108 to 169 μm), which prevents it from becoming airborne (Bae et al., 2015). Pollen grains also produce numerous, sticky spines from their exine that cause the pollen to clump (Bae et al., 2015). Therefore, for pollination, clumps of pollen were placed on stigmas with forceps. Forceps were sterilized in 70% ethanol between pollinated flowers. When flowers were abundant, pollination was performed directly using the monadelphous stamen of the male parent.

Developed seeds were removed and direct sown into 1.3-L containers filled with peat-based growing medium (Metro-Mix; Sun Gro Horticulture, Agawam, MA) in seed lots of ≤30 seeds per container. Seedlings were transplanted into 2.5-L containers filled with douglas fir (Pseudotsuga menziesii)–based potting substrate during summer and grown under conditions described previously. Because of space limitations, plants were grown and evaluated in several glasshouses, two polyhouses, and two field locations. Because of variation in growth rates and flowering, plants were evaluated over multiple years (2013–16). Floral traits were assumed to be highly heritable and no blocking was done to determine variation in response to environmental effects or years.

Eyespot.

Because of previous reports of disomic inheritance in Hibiscus and a report that spotless is often inherited as a simple Mendelian recessive trait, self-pollination and reciprocal testcrosses were conducted using three spotless taxa: ‘Buddha Belly’, ‘Diana’, and White Chiffon^®. Crosses were conducted on dozens of taxa with eyespots, and seedlings were grown and evaluated over multiple years according to the methods listed previously. Self-pollination of ‘Buddha Belly’, ‘Diana’, and White Chiffon^® were expected to produce homozygous recessive flowers without eyespots. Crosses of spotless taxa with taxa that produce eyespots were expected to have a 1:0 segregation of eyespot:spotless for homozygous dominant parents or 1:1 for heterozygous parents. For cultivar groups designated as putative heterozygotes for the eyespot trait, self-pollination and cross-pollination was performed with an expected 3:1 segregation of eyespot:spotless.

Flower color.

Color measurements were collected for each taxa and hybrid using a portable colorimeter (BC-10; Konica Minolta, Tokyo, Japan) that reports in the CIE L*a*b* scale. The CIE L*a*b* scale is a Cartesian coordinate system that represents lightness (L*) with values from 0 (black) to 100 (white), red/green color opposition (a*) with values from −100 (green) to 100 (red), and yellow/blue color opposition (b*) with values from −100 (blue) to 100 (yellow) (García-Marino et al., 2012). One to two petals were randomly collected from each plant and measured during peak bloom in late summer. Petal color was measured on fresh flowers only on their first day because of changes in petal color during senescence (Kim et al., 1989b). Measurements were taken in the center of the petal body between the eyespot and distal end of the petal. Before measurement, the colorimeter was calibrated using a pure white calibration plate (Konica Minolta) and petals were measured on the calibration plate to standardize the background and remove any variation from petal translucence. Individual values for L*, a*, and b* were recorded and averaged together to estimate the color for each individual taxon.

Statistical analyses.

Segregation of flower color was investigated by χ² goodness-of-fit tests to compare expected ratios for F₁, F₂, and backcross seedlings to observed ratios. For single-gene traits that segregated in simple amphidiploid ratios (1:0, 0:1, 3:1, and so forth), diploid (rather than tetraploid) genotypes are reported for simplicity. Only tetraploid taxa were used in the current study, with ploidy confirmed in a previous study (Lattier et al., 2019). Cross combinations that deviated from predicted segregation ratios were included at the end of each χ² table. It is unlikely that these crosses represent a multigene model or polysomic inheritance at the same loci being investigated, but rather were errors in labeling or data collection. Further work will be needed to clarify these discrepancies.

Eyespot.

A total of 246 seedlings were evaluated from self-pollination and intercrosses among taxa without eyespots. None produced floral pigment, yielding a 0:1 (eyespot:spotless) segregation (Table 2). The only self-compatible spotless taxon was White Chiffon^®. All 74 S₁ seedlings of White Chiffon^® were spotless. Based on these results, the three spotless taxa were used as testers in recessive testcrosses with a wide range of cultivars exhibiting eyespots (Fig. 1). Reciprocal combinations with testers that yielded seedlings segregating 1:0 (eyespot:spotless) were grouped together as a homozygous dominant group. This group included ‘Aphrodite’, Bali™, Fiji™, Lavender Chiffon™, Lil’ Kim™, ‘Lucy’, Pink Chiffon^®, and Strawberry Smoothie™ (Fig. 1). Reciprocal combinations with testers that yielded seedlings with a segregation of 1:1 (eyespot:spotless) were grouped together as heterozygotes. This group included blue-flowering taxa (‘Blue Bird’, Blue Chiffon™, Blue Satin^®), as well as ‘Minerva’, ‘Red Heart’, and ‘Woodbridge’ (Fig. 1). The reciprocal cross between Blue Satin^® and ‘Buddha Belly’ produced the most seedlings (168) for observation, resulting in a segregation of 1:1 (eyespot:spotless) (χ² = 0.21, P = 0.64) (Table 2). Only one cross between Blue Satin^® and ‘Diana’ yielded segregation ratios that diverged from the expected 1:1, possibly a result of accidental self-pollination or mislabeling (Table 2).

Table 2.

Segregation of spotless phenotype in F₁ and S₁ seedlings of Hibiscus syriacus.

View Table

View Full Size

Testcrosses predict Hibiscus syriacus genotypes for the recessive spotless gene. Self-pollination of White Chiffon^® and reciprocal crosses among ‘Diana’, ‘Buddha Belly’, and White Chiffon^® resulted in only spotless phenotypes, confirming the group as homozygous recessive. Reciprocal crosses between the spotless taxa (‘Diana’, ‘Buddha Belly’, and White Chiffon^®) and taxa exhibiting eyespots revealed two groups. Putative homozygous dominant group produced a 1:0 ratio of eyespot:spotless seedlings, as well as self-pollination and cross-pollination within the group. Putative heterozygous group produced a 1:1 ratio of eyespot:spotless when crossed with the homozygous recessive group. Self-pollination and cross-pollination within the putative heterozygous group revealed 3:1 ratio of eyespot:spotless seedlings. Taxa from the heterozygous group were then used to confirm homozygous dominant genotypes of two taxa (China Chiffon™ and ‘Blushing Bride’) that were not crossed with the recessive group. Self-pollination of another taxa (Hawaii™) were used to confirm its genotype as heterozygous.

Citation: Journal of the American Society for Horticultural Science J. Amer. Soc. Hort. Sci. 145, 4; 10.21273/JASHS04824-19

• Download Figure
• Download figure as PowerPoint slide

Within the homozygous dominant group, self-pollination and intercrosses were used to further confirm these taxa were not heterozygous. The resulting 276 seedlings from these crosses exhibited a 1:0 (eyespot:spotless) segregation (Table 2). In an additional step, intercrosses were performed between these taxa and heterozygotes, yielding a segregation of 1:0 (eyespot:spotless) in the 478 seedlings evaluated. Only 11 spotless seedlings were recorded in these crosses, with six observed in combinations between homozygous dominant ‘Aphrodite’ and heterozygotes ‘Woodbridge’, Blue Satin^®, and ‘Blue Bird’ (Table 2). Four spotless seedlings were recorded from the cross ‘Lucy’ × ‘Red Heart’ (Table 2). One spotless seedling was recorded from the cross ‘Blue Bird’ × Lil’ Kim™ (Table 2). These spotless seedlings could be the result of accidental self-pollination, chance mutations, or the result of mislabeling. Two additional taxa were added to the homozygous dominant group based on crosses with heterozygotes (Fig. 1). China Chiffon™ and ‘Blushing Bride’ were categorized as homozygous dominant after evaluating a combined 114 seedlings from testcrosses with heterozygotes, yielding a segregation of 1:0 (eyespot:spotless) (Table 2).

Within the heterozygous group, self-pollination and intercrosses were used to further confirm these taxa. All seedlings from self-pollination yielded a segregation of 3:1 (eyespot:spotless) with ‘Woodbridge’ proving the most self-fertile with 66 seedlings observed (χ² = 0.31, P = 0.58) (Table 2). Of the intercrosses, reciprocal combinations between Blue Chiffon™ and ‘Red Heart’ yielded the most seedlings for observation (108) and a segregation of 3:1 (eyespot:spotless) (χ² = 0.79, P = 0.37) (Table 2). Only one taxon, Hawaii™, was not initially included in the recessive testcrosses but was categorized as a putative heterozygote based on self-pollination (Fig. 1). Self-pollination of Hawaii™ revealed a segregation of 3:1 (eyespot:spotless) (χ² = 0.07, P = 0.80) (Table 2); however, only five plants flowered and further confirmation of Hawaii™ as a heterozygote will be necessary.

Based on these results, we propose that the presence of an eyespot is controlled by a single recessive allele (named spotless) with eyespot exhibiting complete dominance in heterozygous taxa. No flowers were observed to be spotless and also have a colorful petal body. Therefore, the spotless gene is likely upstream in the flavonoid biosynthetic pathway resulting in elimination of flower color. Although H. syriacus is reported to be a tetraploid, the spotless phenotype segregates according to simple Mendelian diploid inheritance patterns. Therefore, for simplicity, we propose the diploid genotype notation of ss to represent the spotless phenotype and (SS, Ss) to represent the genotypes for eyespot (Table 3). All blue-flowering taxa, including Blue Satin^®, ‘Blue Bird’, and Blue Chiffon™, were found to be heterozygotes and carry the spotless allele (Table 3). In contrast, most taxa categorized as homozygous dominant exhibit white to blush pink petal bodies (Fig. 1; Table 3). Considering the flavonol pathway from Petrussa et al. (2013), perhaps having only one copy of the eyespot gene in the cyanidin pathway allows more precursors for the pelargonidin, delphinidin, petunidin, and malvidin pathways, resulting in deeper blue or pink/lavender flowers, as observed in the blue taxa and the other heterozygote, ‘Woodbridge’. However, one white flower was discovered that also carried the spotless allele, ‘Red Heart’, whereas colorful taxa, such as ‘Aphrodite’, Lavender Chiffon™, and ‘Lucy’, proved to not carry spotless (Fig. 1; Table 3). Clearly, other genes and gene interactions are involved in determining the specific hue of the petal body.

Table 3.

Flower phenotypes (color and eyespot) and putative genotypes (spotless and geisha) for cultivars of Hibiscus syriacus.

View Table

Flower color.

To determine segregation of color in the petal body, crosses that failed to yield spotless seedlings (SS × SS, SS × Ss, and SS × ss) were first observed for additional segregation patterns. By sorting the seedlings on color depth (CIE L*), segregation patterns began to emerge between plants that produced full-color pigment in the petal body (“colorful” hereafter) and plants that produced little to none (Fig. 2). Elite taxa that exhibited flowers with a red eyespot and white, bicolor, or blush pink petal bodies (“geisha” hereafter) were observed to produce only seedlings with geisha phenotypes resulting from intercrosses and self-pollination. Therefore, this group was used as testers to explore the possibility of a recessive allele that downregulates pigment production in the petal body (Fig. 3). For these segregation tests, all seedlings exhibiting the spotless phenotype were removed from the analyses. Only six combinations yielded unclear segregation patterns, likely because of low numbers of seedlings, accidental self-pollination, and mislabeling (Table 4).

View Full Size

Hibiscus syriacus seedlings from the cross Pink Chiffon^® × White Chiffon^® segregating for dark pink seedlings and blush to white seedlings. No spotless phenotypes recovered. Seedlings sorted from lowest to highest CIE L* value. Color bar represents the flower color for each individual seedling based on recorded CIEL*a*b* values using a colorimeter.

Citation: Journal of the American Society for Horticultural Science J. Amer. Soc. Hort. Sci. 145, 4; 10.21273/JASHS04824-19

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Testcrosses used to predict Hibiscus syriacus genotypes for the recessive geisha gene. Self-pollination and reciprocal crosses among Bali™, ‘Blushing Bride’, ‘Diana’, Fiji™, ‘Helene’, Lil’ Kim™, Pink Chiffon^®, and ‘Red Heart’ resulted in only geisha phenotypes, confirming this group as homozygous recessive. Reciprocal crosses were performed between geisha taxa and taxa exhibiting full color in the petal body. Two groups were revealed based on their segregation ratios. Taxa exhibiting segregation of 1:0 (colorful:geisha) were classified as homozygous dominant and taxa exhibiting segregation of 1:1 (colorful:geisha) were classified as heterozygous. Strawberry Smoothie™, not included in self-pollination or intercrosses, was considered homozygous recessive based on its 1:1 segregation with the heterozygous group. Self-pollination and intercrosses within the homozygous dominant group and heterozygous group further confirmed their genotypes.

Citation: Journal of the American Society for Horticultural Science J. Amer. Soc. Hort. Sci. 145, 4; 10.21273/JASHS04824-19

• Download Figure
• Download figure as PowerPoint slide

Table 4.

Segregation of geisha phenotype in F₁ and S₁ seedlings of Hibiscus syriacus.

View Table

A total of 253 seedlings were evaluated from self-pollination and intercrosses among the homozygous recessive taxa for the geisha phenotype (Table 4). No seedlings resulting from the self-pollination and intercrosses produced fully colorful petals, resulting in a segregation of 0:1 (colorful:geisha) (Table 4). Initially, all white-flowered spotless taxa were included in the putative recessive group, but White Chiffon^® and ‘Buddha Belly’ were found to produce colorful seedlings when crossed with other taxa in the putative recessive group (Table 4). This result is likely because of recessive epistasis of the spotless gene over the genes that control petal body color. The only spotless taxon that segregated 0:1 (colorful:geisha) in recessive testcrosses was ‘Diana’. Therefore, the final recessive testcross group was determined to include Bali™, ‘Blushing Bride’, ‘Diana’, Fiji™, ‘Helene’, Lil’ Kim™, Pink Chiffon^®, and ‘Red Heart’ (Fig. 3). An additional taxon, Strawberry Smoothie™, was added to the recessive group based on testcrosses with taxa in the heterozygous group (Fig. 3). Another likely member of the recessive group, China Chiffon™, was not included because of a lack of appropriate crosses and low numbers of seedlings (Table 4). Further efforts will be necessary to confirm China Chiffon™ as homozygous recessive.

These taxa were used in as testers in recessive testcrosses with a wide range of cultivars exhibiting colorful and spotless phenotypes (Fig. 3). Reciprocal combinations with testers that yielded seedlings exhibiting a 1:0 segregation (colorful:geisha) were classified as homozygous dominant (Fig. 3). These taxa included all blue-flowered taxa (‘Blue Bird’, Blue Chiffon™, Blue Satin^®, and Hawaii™), one spotless taxon (‘Buddha Belly’), and one pink-flowered taxon (‘Lucy’) (Fig. 3).

Reciprocal combinations with testers that yielded seedlings exhibiting a 1:1 segregation (colorful:geisha) were classified as heterozygotes (Fig. 3). These taxa included pink-flowered taxa (‘Aphrodite’ and ‘Woodbridge’), pink-lavender taxa (‘Minerva’ and Lavender Chiffon™), and one spotless taxon (White Chiffon^®). Of a total of 397 total seedlings evaluated, the most prolific crosses with 1:1 segregation (colorful:geisha) were between ‘Aphrodite’ and ‘Diana’, yielding 81 seedlings (χ² = 0.11, P = 0.74) and the cross Pink Chiffon^® × White Chiffon^®, yielding 78 seedlings (χ² = 1.28, P = 0.26) (Table 4). Although White Chiffon^® exhibits the spotless phenotype, these segregation tests confirm that it breeds like a heterozygote for the geisha trait once recessive epistasis of spotless is released. To further confirm the heterozygous group, self-pollination and intercrosses were performed within the group, yielding a total of 167 seedlings for evaluation (Fig. 3). Self-pollination of two taxa yielded the most seedlings among all combinations. Self-pollination of ‘Woodbridge’ yielded 47 seedlings and a segregation of 3:1 (colorful:geisha) (χ² = 0.01, P = 0.93). Self-pollination of ‘Aphrodite’ yielded 44 seedlings and a segregation of 3:1 (colorful:geisha) (χ² = 1.94, P = 0.16) (Table 4).

Among the intercrosses in the heterozygous group, the cross Lavender Chiffon™ × White Chiffon^® was most prolific, yielding 31 seedlings that segregated 3:1 (colorful:geisha). In addition, Strawberry Smoothie™ was later confirmed by reciprocal crosses with Lavender Chiffon™ (Fig. 3), yielding 33 seedlings that segregated 1:1 (colorful:geisha).

Recessive testcrosses between the geisha group and the homozygous dominant group yielded a segregation of 1:0 (colorful:geisha) for all cross combinations (Table 4). Of 1083 seedlings evaluated, only five seedlings were classified in the geisha phenotype group (Table 4). These seedlings were likely spotless flowers, rather than white flowers with eyespots, and mistakenly classified in the geisha phenotype category. The most prolific testcross was the reciprocal cross between Blue Chiffon™ and ‘Red Heart’, yielding 84 colorful seedlings and no geisha phenotypes (Table 4). Surprisingly, the cross Lil’ Kim™ and ‘Buddha Belly’ resulted in a segregation of 1:0 (colorful:geisha), suggesting that although ‘Buddha Belly’ exhibits the spotless phenotype, it breeds like a homozygous dominant once recessive epistasis of spotless is released. However, only seven seedlings were observed, and further testcrosses will be needed to confirm this result.

To further confirm the homozygous dominant group, self-pollination and intercrosses were performed (Fig. 3), resulting in 167 seedlings segregating 1:0 (colorful:geisha) (Table 4). Only one seedling among this group was classified as a geisha phenotype, but was likely a spotless seedling incorrectly categorized as geisha. In another attempt to confirm the homozygous dominant group, testcrosses were made to the heterozygous group, yielding 173 seedlings that segregated 1:0 (colorful:spotless). Only two seedlings were categorized as geisha, but were likely miscategorized spotless seedlings.

Based on these combined results, we propose that the geisha phenotype (white to blush pink phenotype with an eyespot) is controlled by a single recessive allele (named geisha) with the colorful phenotype exhibiting incomplete dominance in the heterozygous taxa. Nearly all taxa in the homozygous dominant group were blue, with exception of ‘Lucy’ (a deep pink) and ‘Buddha Belly’ (a spotless white) (Fig. 3; Table 3). Nearly all taxa in the heterozygous group were pink to lavender, with the exception of White Chiffon^® (a spotless white) (Fig. 3; Table 3). The true color (underlying genes for color in the petal body) of ‘Buddha Belly’ and White Chiffon^® are likely masked by an epistatic interaction with the spotless allele. Although H. syriacus is reported to be a tetraploid, the geisha phenotype was observed to segregate according to simple Mendelian diploid inheritance patterns. Therefore, for simplicity, we propose the diploid genotype notation of gg to represent the geisha phenotype, and (GG, Gg) to represent the colorful phenotype (Table 3). However, all petal body phenotypes are under the control of the recessive epistatic spotless gene.

One possible explanation for the segregation patterns of spotless and geisha phenotypes emerges when considering the flavonoid biosynthetic pathway (Petrussa et al., 2013) (Fig. 4). If the gene conferring the spotless phenotype lies upstream of the three pathways involved in flower color expression in H. syriacus, then cyanidins responsible for the red eyespot, peonidins and pelargonidins responsible for salmon-pink and blush pink color, and delphinidins, petuniadins, and malvidins responsible for deep pink, lavender, and blue flowers will not be expressed (Fig. 4). In addition, if the gene conferring the geisha phenotype inhibits or downregulates the delphinidin pathway, responsible for dark pink, lavender, and blue pigments, then plants that produce only cyanidins, peonidins, and pelargonidins may make up the geisha phenotype (Fig. 4). Plants that produce only cyanidins, isolated to the eyespot region, would likely be white with a red eyespot. Plants that produce trace levels of cyanidin, peonidin, and pelargonidin in the petal body may be responsible for the white/pink bicolor and blush-pink flowers seen in some of the geisha phenotypes (Fig. 4).

View Full Size

Proposed Hibiscus syriacus flower phenotypes and genotypes, and proposed gene pathway arranged on a simplified flavonoid biosynthetic pathway from Petrussa et al. (2013). Flowers with no pigment production controlled by a recessive gene upstream in the flavonoid biosynthetic pathway were called spotless, resulting in pure, white flowers. Flowers with at least one dominant allele for spotless result in flowers with eyespots. Of the flowers with eyespots, color segregation is controlled by another recessive mutation called geisha that disrupts the delphinidin biosynthetic pathway. Flowers homozygous recessive for geisha result in cyanidin, peonidin, and pelargonidin-type flowers. Because red is not expressed in the petal body, cyanidin-rich flowers express as white with an eyespot. Flowers with more peonidin and pelargonidin pigments express as blush pink.

Citation: Journal of the American Society for Horticultural Science J. Amer. Soc. Hort. Sci. 145, 4; 10.21273/JASHS04824-19

• Download Figure
• Download figure as PowerPoint slide

A previous study on extracted anthocyanins in H. syriacus lends evidence to this theory. Kim et al. (1989a) analyzed pigments from H. syriacus flowers exhibiting eyespots, with petal bodies including white (geisha), blush (geisha), dark pink, lavender, and blue. Total anthocyanins were significantly reduced in the white and blush flowers, with most pigments in the petal body made up of cyanidin and pelargonidin (Kim et al., 1989a). Dark pink flowers (putative heterozygotes) produced the most total anthocyanins with a significant percentage of anthocyanins from all five categories and the lowest percentage from pelargonidins (Kim et al., 1989a). Most pigments produced in lavender flowers were from the delphinidin pathway: delphinidin, petunidin, and malvidin (Kim et al., 1989a). The vast majority of pigments in blue flowers were from the malvidin group, with lower percentages of delphinidin and petunidin compared with lavender flowers (Kim et al., 1989a).

Additional segregation tests.

To further confirm color segregation, self-pollination and backcrosses were made to develop F₂ and backcross populations in a range of F₁ seedlings. However, growth was stunted in many of the F₂ and backcross seedlings, likely because of inbreeding depression. Therefore, data were collected only on two crosses that provided enough F₂ and backcross seedlings to perform segregation analysis on the spotless and geisha phenotypes. Spotless phenotypes were removed from the segregation tests on the geisha phenotype.

In the first cross, Lil’ Kim™ (SSgg) × Blue Chiffon™ (SsGG), the F₁ population made up of 28 seedlings segregated 1:0 for both eyespot (eyespot:spotless) and for color (colorful:geisha) (Table 5). Based on the parent genotypes, F₁ seedling genotypes likely segregated 1:1 for (SSGg:SsGg). Because of time constraints and sizes of each F₁ seedling, plants were grouped together into genotype families based on the spotless phenotypes in their F₂ (S₁) seedlings after self-pollination. Therefore, the first F₂ family from self-pollination of F₁ (SSGg) family segregated 1:0 (eyespot:spotless) and 3:1 (colorful:geisha) (χ² = 1.02, P = 0.31). The second F₂ family from self-pollination of F₁ (SsGg) family resulted in a segregation of 3:1 (eyespot:spotless) (χ² = 2.23, P = 0.14) and a segregation of 3:1 (colorful:geisha) (χ² = 3.21, P = 0.07) (Table 5).

Table 5.

Segregation of spotless and geisha phenotypes in F₁, F₂, and backcrosses from the cross Hibiscus syriacus Lil’ Kim™ × H. syriacus Blue Chiffon™.

View Table

Next, backcrosses of the F₂ families to both parents were attempted. Backcross of the F₁ (SSGg) family to Lil’ Kim™ (SSgg) resulted in a segregation of 1:0 (eyespot:spotless) and a segregation of 1:1 (colorful:geisha) (χ² = 1.82, P = 0.18) (Table 5). Backcross of the F₁ (SsGg) family to Lil’ Kim™ (SSgg) resulted in a segregation of 1:0 (eyespot:spotless) and a segregation of 1:1 (colorful:geisha) (χ² = 0.03, P = 0.87) (Table 5). Backcrosses to Blue Chiffon™ (SsGG) were successful only with the F₁ (SsGg) family, yielding a segregation of 3:1 (eyespot:spotless) and segregation of 1:0 (colorful:geisha) (Table 5). Only 3 of the 90 seedlings evaluated were scored as geisha, but were likely spotless seedlings that were miscategorized during data collection (Table 5).

In the second cross, Fiji™ (SSgg) × White Chiffon^® (ssGg), the F₁ population made up of 61 seedlings segregated 1:0 (eyespot:spotless) and 1:1 (colorful:geisha) (χ² = 1.98, P = 0.16) (Table 6). Individual plants were grouped together into genotype families based on their petal body phenotype: F₁ (Ssgg) geisha family and F₁ (SsGg) colorful family (Table 6). Self-pollination produced few seedlings for segregation analysis (Table 6); however, the F₂ family from self-pollination of the F₁ (Ssgg) family resulted in a segregation of 3:1 (eyespot:spotless) (χ² = 0.67, P = 0.41) and segregation of 0:1 (colorful:geisha) (Table 6). The F₂ family from self-pollination of the F₁ (SsGg) family resulted in a segregation of 3:1 (eyespot:spotless) (χ² = 0.10, P = 0.76) and a segregation of 3:1 (colorful:geisha) (χ² = 0.76, P = 0.38). Next, backcrosses of the F₂ families to both parents were attempted. Backcross to Fiji™ failed but was successful to White Chiffon^® (Table 6). Backcross of the F₁ (Ssgg) family to White Chiffon^® (ssGg) resulted in a segregation of 1:1 (eyespot:spotless) (χ² = 0.07, P = 0.79) and a segregation of 1:1 (colorful:geisha) (χ² = 0.00, P = 1.00) (Table 6). Backcross of the F₁ (SsGg) to White Chiffon^® (ssGg) resulted in few seedlings for analyses and further segregation tests will be necessary.

Table 6.

Segregation of spotless and geisha phenotypes in F₁, F₂, and backcrosse from reciprocal crosses of Hibiscus syriacus Fiji™ and H. syriacus White Chiffon^®.

View Table

In the current study, most crosses among heterozygous taxa (SsGg) yielded too few seedlings for segregation analysis of recessive epistasis of spotless over colorful and geisha phenotypes. However, self-pollination of ‘Woodbridge’ yielded 66 seedlings with significant recessive epistatic segregation of 9:3:4 (colorful:geisha:spotless) (χ² = 0.51, P = 0.77) (Table 7). In addition, self-pollination of the heterozygous F₁ family (SsGg) produced from the cross Lil’ Kim™ × Blue Chiffon™ yielded 79 seedlings with recessive epistatic segregation of 9:3:4 (colorful:geisha:spotless) (χ² = 0.77, P = 0.06) (Table 7).

Table 7.

Segregation test for recessive epistasis of spotless over geisha in heterozygote (SsGg) self-pollination of Hibiscus syriacus.

View Table

Results from crosses with blue-flowered taxa revealed an interesting segregation pattern. Hybrid seedlings exhibiting blue flowers were recovered only from crosses among blue-flowered taxa. All other hybrid combinations with blue flowers resulted in pink to lavender flowers or spotless flowers in the F₁ seedlings. In addition, in the anthocyanin study by Kim et al. (1989a), blue flowers expressed most of the pigments from the malvidin group of anthocyanins, with minimal amounts of delphinidin and petunidin. Therefore, another recessive gene downstream in the delphinidin biosynthesis pathway may effect the transition to pink or lavender flowers (petunidin) leaving only blue (malvidin) pigments in the homozygous recessive taxa. CIE L* a* b* values for all self-pollinated and intercrossed taxa in the blue-flowered group were investigated, and sorted for their color components. Based on this set of blue seedlings, we determined that a true blue flower exhibits a CIE L* < 65, CIE a* < 18.3, and CIE b* < −18.3. This score was used to bin blue and pink-lavender flowers in F₁, F₂, and backcross segregation tests for a recessive blue allele in the cross Lil’ Kim™ × Blue Chiffon™ (Table 8). Spotless phenotypes were removed from the segregation analyses.

Table 8.

Segregation of blue flowers in in F₁, F₂, and backcrosses in Hibiscus syriacus.

View Table

A total of 52 (S₁) seedlings resulting from self-pollination of ‘Blue Bird’, Blue Chiffon™, and Blue Satin^® yielded a segregation of 0:1 (pink/lavender:blue) (Table 8). In addition, a total of 19 F₁ seedlings from the cross Blue Satin^® × ‘Blue Bird’ resulted in a segregation of 0:1 (pink/lavender: blue) (Table 8). For the cross Lil’ Kim™ × Blue Chiffon™, we observed a segregation of 1:0 (pink/lavender:blue) in the F₁ population (Table 8). Self-pollination of the F₁ family produced an F₂ population that segregated almost entirely pink/lavender after 93 observed seedlings, deviating significantly from the expected 3:1 segregation (Table 8). Backcrosses to Blue Chiffon™ yielded slightly more blue seedlings (8 blue of the 87 observed), but seedlings deviated significantly from the expected 1:1 segregation (Table 8). When CIEL*a*b* estimates from the backcross seedlings were compared with the blue seedlings resulting from blue-flowered self-pollination and blue-flowered intercrosses, the wide segregation of blue to pink/lavender color in the backcross seedlings becomes obvious. This could be because delphinidin (the precursor to petunidin and malvidin) is a stable pigment and the interplay among these three types of anthocyanins makes it difficult to delineate a true blue pigment. In addition, the blue trait in H. syriacus could simply be controlled by multiple genes at different loci.

Previous work has shown that metal ions and pH changes have contributed to blue pigmentation in other flowers (Griesbach, 1996). Blue color in Tulipa gesneriana was caused by a 25 times higher Fe³⁺ content in the vacuole of blue cells (Shoji et al., 2007). Metal ions Fe³⁺ and Mg²⁺ were found to contribute to the blue pigmentation in Centaurea cyanus (Kondo et al., 1994, 1998; Shiono et al., 2005; Takeda et al., 2005) and Meconopsis grandis (Yoshida et al., 2006). Changes in pH alone have been responsible for blue pigmentation in Ipomoea tricolor (Yoshida et al., 1995, 2005) and Petunia hybrida (Griesbach, 1996). Often, the inheritance of pH and metal ions is more complicated than the simple Mendelian inheritance patterns of anthocyanins (Griesbach, 1996), which could explain our inability to determine segregation in blue-flowered H. syriacus.

Other phenotypic observations.

Within the geisha phenotype group, one taxa, Fiji™, was found to exhibit bicolor petals, with pink or sometimes red pigment (reminiscent of the eyespot) developing on the abaxial side of mature buds. This unusual and attractive phenotype was heritable in F₁ crosses with Fiji™; the bicolor phenotype was most easily identified in mature, expanding buds. A similar taxon, ‘Elegantissimus’ (also known as ‘Lady Stanley’), was used to investigate anthocyanin composition in petal tissues in a previous study (Kim et al., 1989a). Kim et al. (1989a) found that most of the pigment in the petal body was composed of pelargonidins, with a lower, but significant, percentage of cyanidins present. However, in the current study, the expression of this trait was variable, with full-sib seedlings exhibiting a color range from conspicuous bicolor buds to blush-pink buds.

Although most plants exhibited stable flower phenotypes, bud sport mutations are inevitable when observing thousands of branches upon thousands of hybrid seedlings. However, we observed only two interesting sport branches throughout our study. The first was a pink-flowered hybrid that produced a geisha branch (white petal body with an eyespot). The second was a branch sport on an F₁ hybrid of ‘Minerva’ × Blue Satin^® that produced blush pigment in the petal body, but lacked an eyespot. The latter sport branch was propagated, and stability of this trait will be evaluated over subsequent years. If pigment production in the petal body can be bred in the absence of an eyespot, this discovery could lead to novel flower forms and new cultivars.

The current study represents a significant achievement in understanding segregation of floral traits of althea, including simply inherited traits such as eyespot and petal body color. Cultivars included in this study have been classified by phenotype and genotype for the spotless and geisha traits (Table 3). This information will aid future breeders and further research into the heritability of traits in H. syriacus. Quantitative improvement on color depth may be possible by eliminating the geisha allele from a breeding population and making selections in each generation for low CIE L* values. In addition, spotless alleles would have to be removed from the breeding population so as not to waste time and space, especially in the blue-flowered taxa. In the current study, all blue-flowered taxa carried the spotless allele, and it remains unclear if there are any true-breeding (SSGG) blue taxa available for breeders of H. syriacus. Further work must be done to establish if blue flowers are inherited through Mendelian segregation of anthocyanin genes, or if changes in metal ions and pH contribute to the perceived blue color. In addition, future breeding efforts will focus on crossing blue-flowered seedlings to recessive testers (‘Diana’, White Chiffon^®, ‘Buddha Belly’) to develop a true-breeding blue H. syriacus.