A high level of genetic variability for growth habit types is observed in tetraploid, cultivated coleus (Solenostemon scutellarioides (L.) Codd). Very few cultivars with trailing habit exist, and those that are available generally have dark green or purple foliage color. Coleus with trailing growth habit is desirable as it increases its utility for use in hanging baskets, mixed containers, and as ground cover. There is a lack of published information on the genetic mechanism controlling growth habit and the development of new trailing types with orange colors. Two commercial cultivars, “Red Trailing Queen” (RQ) with trailing habit and “Sedona” (S) with upright stature, were selfed and crossed (RQ × S) to produce self and F1 populations. F2 populations were produced by selfing plants in the F1 population. For each population analyzed, growth habit was rated on a visual 1–5 phenotypic scale, where 1 = upright, 2 = semi-upright, 3 = prostrate, 4 = semitrailing, and 5 = trailing. Genotypes were assigned to each phenotype, assuming that upright was dominant to trailing. In this study, growth habit was observed to be controlled by a single gene (U) with additive effects, with upright growth habit designated with a UUUU genotype and trailing growth habit designated with a uuuu genotype. In addition, foliage color was rated on a visual 1–5 phenotypic scale, and purple foliage color was found to be dominant to yellow–orange color. Several new coleus selections with trailing growth habit and orange foliage color were successfully developed.
Solenostemon scutellarioides (L.) Codd (syns Coleus blumei Benth; Coleus scutellarioides (L.) Benth), commonly known as coleus, is a member of the Lamiaceae family and is valued for its vibrant and colorful foliage (Pedley R and Pedley K 1974). More than 300 species and several hundred cultivars exist for the genus Solenostemon, and commercial cultivars are characterized by their leaf colors, leaf shape, growth habit, and flowering characteristics (Lebowitz 1985). Most brightly colored coleus varieties commonly found on the market have vigorous upright growth habits and are commonly used in the landscape as tall bedding plants. Few trailing habit varieties exist on the market, and those available have foliage that is generally dark green or purple. Trailing growth habit in coleus is desirable as it increases its utility as an ornamental plant for use in mixed containers, hanging baskets, and as ground cover in the landscape. Varieties with trailing habits are characterized by having an extensive lateral branching pattern with a reduction in apical dominance and a larger quantity of axillary meristems (Davies 1995). The manipulation of plant architecture has been a common focus of many breeding programs during the improvement of many crop species (Mirzaie-Nodoushan et al. 1999).
Genes controlling growth habit vary among plant species because variations in plant architecture reflect diverse growth and reproductive strategies (Goncharov 1999). In narrow-leafed lupin, Lupinus angustifolia (L.), restricted branching (lack of branching) is a simply inherited recessive trait (Adhikari et al. 2001). In red clover, Trifolium pratense (L.), a dominance × dominance epistasis controls prostrateness, with prostrate being partially dominant to erect (Mirzaie-Nodoushan et al. 1999). Although hundreds of different types of coleus exist with extensive variation in growth habit (Boye and Rife 1938), information on the genetics for growth habit is limiting because most of the genetic studies in coleus have focused on leaf color (Boye and Rife 1938; Rife 1939; Lebowitz 1985) and leaf shape (Hakeem and Rife 1966; Rife 1972). Types of growth habit observed in coleus are upright and trailing habits, with varying levels for each habit type.
Contrasting views have been proposed as to the origins of modern trailing types because most were the result of interspecific hybridizations (Pedley R and Pedley K 1974; Lebowitz 1985). According to Bailey Hortorium (Bailey 1976), current trailing varieties were developed from interspecific crosses between S. scutellarioides (L.) and Solenostemon pumilus Blanco. The majority of cytological studies on S. scutellarioides (L.) report a 2n = 48 chromosome number (Reddy 1952; Morton 1962; Hakeem and Rife 1966). Multivalents appear during meiosis in cultivars with 2n = 48 chromosomes and indicate polyploidy. Because many genetic experiments have produced simple diploid ratios (Rife 1948; Reddy 1952), the species is thought to be an allotetraploid (Lebowitz 1985).
The overall objective of this study was to provide further information on the genetics of trailing growth habit in coleus. Understanding the inheritance of trailing habit will provide useful information for the development of new coleus cultivars with trailing growth habit combined with new foliage color. Little information is available regarding the genetics of trailing growth habit in coleus. Furthermore, no known study to date has been published on the development of new trailing coleus cultivars with orange foliage colors through the method of controlled hand pollinations. Rife (1948) reported the interaction of 6 multiple alleles affecting the amount and distribution of anthocyanin in coleus foliage. These alleles he designated as the purple series, in which their interactions resulted in 9 phenotypic variations with purple color being dominant (Rife 1948). Purple color observed in “Red Trailing Queen” (RQ) may be hypothesized as being dominant to the orange color of “Sedona” (S). The report by Rife (1948) is used as the basis to classify the range of foliage colors observed in this study into several phenotypic classes. Through this research, new relevant information on the trailing variety “RQ” is provided. The specific objectives of this study were to 1) use RQ and S as breeding parents to produce hybrids, 2) determine the genetics for trailing habit and isolate hybrids with trailing habit, 3) introgress yellow/orange color into trailing habit types, and 4) develop new coleus varieties with trailing habit and orange foliage colors.
Materials and Methods
The commercial cultivars RQ and S (obtained from Proven Winners, Sycamore, IL) were used as parental lines for the production of F1 hybrids in the breeding program. RQ has dark purple leaves, prostrate growth habit, and lateral branches that extend more than 1 m in length. S has an upright growth habit and bright yellow/orange foliage. Both cultivars flower profusely, and flowering induction occurs year-round, facilitating hand pollinations in the greenhouse. Both RQ and S were selfed, and approximately 100 progeny each were grown and evaluated for growth habit as described later.
Analysis for Ploidy Level Differences among Parental Genotypes and Hybrids
Because of concerns in older literature related to ploidy levels and whether RQ and S were cross-compatible, flow cytometry and comparative cell size measurements were used to compare ploidy levels for RQ, S, and several F1 plants. For flow cytometry analysis, fresh apical meristematic tissues (ca., 20 mg) were chopped and nuclei extracted in 1 ml lysis buffer and 4′,6-diamidino-2-phenylindole (Kapuscinski 1995). Large tissue debris was removed by filtration (50-μm filter) and the extract subsequently used for analysis, with a minimum of 2500 nuclei counted per extraction solution. Samples were analyzed with a PA1 Ploidy Analyzer (Partec, Münster, Germany) and DNA quantified using a DPAC software provided by the same company. No internal standard was used, and no genome size is reported for the coleus used in this study. A 1-parameter histogram was obtained and ploidy levels were compared among each variety and a couple of F1 plants.
Comparative cell size analysis was performed by measuring the length and width of fresh pollen grains and stomatal guard cells and the length of glandular trichomes. Stomatal guard cells and glandular trichomes were obtained from epidermal peels as previously described (Reddy 1952). Measurements (in microns) were taken at a magnification of 4× for glandular trichomes and at 40× for pollen grains and stomatal guard cells using a compound light microscope (Nikon Labophot-2). Data shown are the means and standard deviations of 25 pollen grains and stomatal guard cells and 20 glandular trichomes for RQ, S, and 5 F1 plants.
All pollinations (selfs and reciprocal crosses) were performed in the morning (8–11 AM) in a greenhouse under ambient light and standard temperature conditions (23 ± 3 °C). Plants used for pollinations were not fertilized as it was the most effective means to generate seeds; plants grown under lush conditions did not produce as many seeds as those grown under a somewhat stressed condition (Nguyen, personal observation). Seeds were collected from pollinated flowers approximately 3 weeks later when the seeds and calyx (seed capsule) were dark brown in color. Cross-pollinations were performed by emasculating flowers (preanthesis) in the morning with tweezers. To prevent pollen contaminations, the tweezers were dipped in 95% ethyl alcohol after all emasculations. Pollinations were made on a daily basis until approximately 100 seeds were collected from each cross. Seeds were collected from selfing RQ and S and the cross RQ(♀) × S(♂), whereas no seeds were collected from the S(♀) × RQ(♂) cross. The F2 and F3 populations were generated by selfing selected F1 and F2 plants, respectively.
Growth of Seedlings
Seeds collected from self and cross-pollinations were germinated in a mist house with misting set at 20 s every 30 min. They were removed after 90% of seeds had germinated (10–14 days), moved to a polycarbonate shade house, and afterward transplanted into a 72-cell-pack tray filled with soilless media (Fafard 2-mix, Apopka, FL). After 3 additional weeks, they were transplanted to 15.2-cm pots and irrigated as needed with liquid fertilizer at 150 ppm nitrogen (15-5-15, N-P2O5-K2O—Scotts, Apopka, FL) at every other irrigation. The plants were grown for an additional 6 weeks, at which point growth habit and foliage color were evaluated.
Rating Scale for Growth Habit and Foliage Color Brightness
Growth habit was rated on a visual 1–5 rating scale, where 1 = upright, 2 = semi-upright, 3 = prostrate, 4 = semitrailing, and 5 = trailing (Figure 1). The 1–5 rating scale was used because this scale has been used for other plants and all 5 growth habit types have been observed in coleus (Pedley R and Pedley K 1974). Additionally, a 5 category scale is appropriate for single gene tetrasomic ratios.
Although foliage color is a more subjective trait, with several genes or epigenetic factors likely contributing to pattern and color, we attempted to rate foliage color in order to establish cursory data on the factors that contribute to this important trait. Based on Rife's (1948) report that allelic interactions resulted in phenotypic color variations, foliage color in this study was divided into 5 phenotypic classes for simplicity. Foliage color was rated on a visual 1–5 rating scale, where 1 = dark purple (color as dark as RQ), 2 = purple (color slightly lighter than RQ), 3 = light reddish/purple (color between RQ and S), 4 = dark yellow/orange (color slightly darker than S), and 5 = orange (color as bright as S).
Rating Scale for Genotypic and Chi-square Analysis
A plant observed as having an upright growth habit (rating of 1) was designated as quadraplex genotype (UUUU), assuming that upright growth habit was dominant to trailing. Based on the 1–5 scale, 1 = upright (UUUU), 2 = semi-upright (UUUu), 3 = prostrate (UUuu), 4 = semitrailing (Uuuu), and 5 = trailing (uuuu). A chi-square (χ2) goodness-of-fit analysis based on a single gene tetrasomic locus with complete additive gene action and chromosomal segregation as illustrated by Burnham (1962) was used.
Chi-Square Test for Homogeneity
After the chi-square goodness-of-fit test was applied to the RQ and S selfed and F1 populations (as described above), the genotypes of RQ and S were determined. To further support that the phenotypes matched the given genotypes, 3 plants from the F1 (or F2) population from each genotypic group (UUUU, UUUu, UUuu, Uuuu, and uuuu) were selfed and progeny were analyzed for plant habit and color. A chi-square test for homogeneity based on an r × c table, as previously described by Cochran (1954), was performed.
Introgression of Color Using the Selfing Method
Thirty F1 plants characterized with semitrailing habit were selected for use as lines to introgress new foliage color. The F1 plants were selfed over a 1-month period and seeds collected to generate an F2 population. F2 plants exhibiting both trailing habit and nonpurple foliage color were selected and self-pollinated to produce F3 progeny.
Determination of Ploidy Level Differences
Solenostemon scutellarioides is assumed to be a tetraploid based on reports obtained in the literature (Rife 1948; Reddy 1952). In this study, no internal standard was used and no genome size was reported; thus, only relative DNA and ploidy levels were compared among the plants tested. The relative DNA content for RQ and S were not different, using the appropriate gain setting on the ploidy analyzer. RQ and S mean values were 112 and 110, respectively, and the F1 plant DNA content ranged from 91 to 118. These results from flow cytometric analysis showed that no differences in ploidy levels existed for RQ, S, and the F1 plants tested (data not shown).
The applicability of pollen, stomatal guard cell, and glandular trichome length as indirect methods to further support the similar ploidy levels found for RQ, S, and several F1 plants was investigated. Comparative cell size analysis has been shown to differentiate ploidy levels between species, where increases in ploidy levels correlate with increases in cell size (Reddy 1952; Aryavand et al. 2003). There was an overlap in the measurements between RQ (38 μm), S (39 μm), and the 5 F1s (range of 37–41 μm) for pollen grain length (data not shown). In addition, similar observations were found for stomatal guard length, with measurements of 30 μm (RQ), 32 μm (S), and a range of 30–34 μm (F1s). Additionally, an overlap was observed for the length of glandular trichomes when RQ, S, and the 5 F1s were compared (data not shown). These results further support the flow cytometry results showing that RQ and S have the same ploidy. Further segregation analysis suggested that both are tetraploids (discussed later).
Differences in seed set and fertility were observed when self and reciprocal cross-pollinations were performed on RQ and S. Very low self seed set was observed in RQ, whereas self seed set was greater than 95% in S. Of a maximum of 4 seeds that can be produced in a coleus flower, an average of one seed was normally produced by RQ, whereas 3–4 seeds were usually produced from S (personal observation). Many species have low seed set, and pollen quality and/or pollen sterility due to nonfunctional pollen have been reported as factors affecting seed set for some plant species (McDade and Davidar 1984; Byers 1995). In order to determine if the low seed set observed in RQ was due to poor pollen quality, quality was estimated using acetocarmine staining. Less than 15% of RQ pollen was stainable, whereas 95% of S pollen was stainable (data not shown).
Growth Habit Segregation
Photographs were taken of the 5 categories of growth habit forms (Figure 1) observed in RQ and S selfed populations and in the F1 population. All 5 different growth habit types were observed in the RQ selfed population (Figure 1). Of 96 selfed progeny evaluated, the observed ratios reported were 1:19:53:21:2 for category 1, 2, 3, 4, and 5, respectively. The occurrence of progeny from the RQ parent which were more and less trailing than RQ suggested that RQ was not homozygous for trailing habit. In the S selfed population, 3 different growth habit types were observed (Figure 1). Of 100 progenies evaluated, the observed ratios reported were 20:57:23 for category 1, 2, and 3, respectively. The 20 progeny in category 1 were taller and had poorer branching than S, suggesting that S was not homozygous for the upright trait as previously assumed. Four growth habit types were observed in the F1 (RQ × S cross) (Figure 1). Of the 250 F1 plants evaluated, the observed ratios were 22:108:94:26 for category 1, 2, 3, and 4, respectively. None of the F1 plants had the trailing growth habit (category 5) observed in some RQ selfed progeny, whereas those that were upright were similar to the plants in category 1 in the S selfed population. The distribution for growth habit types for RQ selfed, S selfed, and F1 population is shown in Figure 2.
Determination of Parental Genotypes
A chi-square test for goodness of fit (χ2) was used to test hypotheses of possible genotypes for RQ and S. For RQ, all types of growth habits were observed in the selfed population. Because both the quadraplex (UUUU) and nulliplex (uuuu) forms were present, the genotype proposed for RQ was the duplex form (UUuu). A χ2 value of 2.02 (P = 0.73) was calculated for RQ selfed progeny (Table 1). In regards to S, because only growth habits in category 1, 2, and 3 were observed in the selfed population, the triplex form (UUUu) was proposed. A χ2 value of 2.14 (P = 0.34) was calculated for S selfed progeny (Table 1). If the above genotypes were correct for RQ (UUuu) and S (UUUu), the F1 population should theoretically segregate in a 1:5:5:1 ratio. A χ2 value of 2.51 (P = 0.47) was calculated for this segregation ratio for the F1 population (Table 1). Results from the chi square indicated a goodness of fit for the observed ratios in the RQ selfed, S selfed, and F1 populations and thus supported the proposed genotypes for RQ (UUuu) and S (UUUu).
|RQ × S||250||22||108||94||26||0||1:5:5:1||2.51||3||0.47|
|RQ × S||250||22||108||94||26||0||1:5:5:1||2.51||3||0.47|
Growth habit was classified into 5 phenotypic/genotypic categories: category 1 = upright (UUUU genotype), category 2 = semi-upright (UUUu genotype), category 3 = prostrate (UUuu genotype), category 4 = semitrailing (Uuuu genotype), and category 5 = trailing (uuuu genotype). The χ2 values of 2.02 (P = 0.73) for RQ and 2.14 (P = 0.34) for S supported the genotypes proposed for RQ (UUuu) and S (UUUu), respectively. A χ2 value of 2.51 (P = 0.47) in the cross RQ × S further supported the genotypes proposed for RQ and S. df, degrees of freedom.
To further support that the phenotypes matched their assigned genotypes (e.g., a plant characterized as having a semi-upright growth habit [category 2] had the triplex [UUUu] genotype), plants from each of the 5 phenotypic plant growth habit categories were selected. Three F1 plants from category 2, 3, and 4 were selfed and segregation for growth habit in the F2 population was tested using the chi square. Because no nulliplex F1 progeny were obtained and the quadraplex F1s were lost (not due to lethality in the homozygous state but were prematurely discarded), plants of phenotypes in category 1 and 2 from the F2 generation were selfed and the F3 population was analyzed.
Chi-square tests to validate that a given F1 or F2 phenotype corresponded to the postulated genotypes are shown in Table 2. For the category 1 phenotype, the F3 populations produced from each of the 3 F2 plants all had upright growth habits (Table 2). For the category 2 phenotype, the F2 population for each of the 3 plants segregated according to the expected ratios with χ2 values of 1.84 (P = 0.34), 0.34 (P = 0.18), and 0.92 (P = 0.63) (Table 2). For the category 3 phenotype, each of the 3 plants segregated according to the expected ratios with χ2 values of 4.04 (P = 0.40), 6.41 (P = 0.17), and 4.93 (P = 0.29) (Table 2). Because selfed progeny from each of the 3 plants in the first 3 categories segregated according to the expected ratios, it was not surprising that the selfed progeny in the category 4 phenotype did the same with χ2 values of 1.22 (P = 0.57), 0.52 (P = 0.77), and 1.69 (P = 0.43) obtained (Table 2). For the category 5 phenotype, all plants segregated for trailing growth habit. Data from the chi-square test for homogeneity are listed as the total χ2 and P values in Table 2. The finding of homogeneity among families with each category further supported the results from the individual chi-square tests. The data for growth habit had a goodness of fit to a tetraploid model, and genetics for growth habit in coleus appear to be controlled by a single tetrasomic locus gene with completely additive effects. We propose the gene symbol U/u for upright growth habit with the quadraplex (UUUU) being fully upright.
Three plants from each of the 5 phenotypic categories: 1 = upright (UUUU), 2 = semi-upright (UUUu), 3 = prostrate (UUuu), 4 = semitrailing (Uuuu), and 5 = trailing (uuuu) were selfed. Values for the χ2 test are listed as individual χ2 with the P values for each of the 3 plants analyzed for each category. Values for the χ2 test for homogeneity are listed as the total χ2 and the P values for each category. N/A, not applicable.
Mean rating values were calculated for foliage color in the RQ selfed, S selfed, F1, and the F2/F3 populations (Figure 3). A mean value of 4.1 was observed for the S selfed population, whereas a mean value of 1 was observed in the RQ selfed population. Progeny produced from selfing RQ were all dark purple and green in color, with the 2 colors varying slightly in degree of intensity. For the S selfed population, a range of new colors were observed, with colors ranging from yellow, orange, red, green, and mottled yellow/orange. In the F1 population, some combinations of foliage colors were observed which were not present in either the RQ or S selfed population. Foliage colors ranged from dark purple (similar to RQ), purple with large green margins, mottled light purple/green, mottled brown/green, and medium red. The introgression of new foliage color was observed in the F1 population because some individuals had light green color in their foliage, which was not observed in the RQ selfed population. Many F1 plants had foliage colors that were darker than S but still lighter than that of RQ. In addition, purple color from RQ appeared to be dominant and the orange color from S to be recessive because all the F1 plants had some form of purple in their foliage. When the means for color were compared, color improved slightly in the F1 population. The mean rating value for color of 1.3 was slightly higher than that of the RQ selfed population (value of 1) but was still much lower than that of the S selfed population (value of 4.1). In the F2/F3 populations produced from selfing, the 9 F1 and 6 F2 plants for genotypic analysis; the mean rating value for color was 2.7, although this value is still lower than that of the S selfed population.
Introgression of Color
Our initial plan was to use a backcross approach to introgress new color traits from S into F1 plants with semitrailing (category 4) growth habits. However, F2 progeny from some F1 plants selfed for the genotypic analysis described above showed yellow/orange color as seen in the parent S. Subsequently, 30 F1 plants with category 4 growth habit and color ratings of 2 or higher were selfed to produce an F2 population. Foliage color observed in this F2 population ranged from purple (as seen in RQ), green, mottled pink with green margins, mottled green/red, to orange color (as seen in S).
A difference in ploidy level did not appear to be the factor contributing to the low selfed seed set in RQ and the low cross seed set found in S(♀) × RQ(♂) but not in RQ(♀) × S(♂). In a study conducted on Eupatorium species (Eupatorium resinosum and Eupatorium perfoliatum) by Byers (1995), seed set (number of seeds and heads in each cross) was used as a measure of compatibility, with a cross-considered compatible if there was at least 1 seed produced per head. Ten plants were used in a complete diallel crossing within each population (Byers 1995). Low seed set for both populations was suggested to be due to poor pollen quality. Poor RQ pollen quality appears to have been a factor contributing to its low selfed seed set and unsuccessful seed set when RQ was used as the pollen donor in the cross S(♀) × RQ(♂). Decreased pollen quality has been observed in other species such as Paspalum maritimum (Adamowski et al. 2000), Clarkia exilis (Vasek 1961), and Pfaffia tuberosa (Taschetto and Pagliarini 2004). In these other species, low frequencies of irregular meiotic behaviors resulting in pollen mother cells (PMCs) with univalents, trivalents, and tetravalents were found to be due to precocious chromosome segregation leading to micronuclei formation. The micronuclei remained in the tetrad stage and resulted in pollen with variable fertility and sterility. In RQ PMCs, preliminary microsporogenesis analysis showed the occurrence of micronuclei with some PMCs containing tetrads with microspores of different sizes (Nguyen 2007). Variations in pollen stainability were also observed in other cultivars of coleus (personal observation). Therefore, it can be speculated that poor pollen quality in RQ may be due to irregular meiotic behavior; however, further cytogenetic studies are needed to accurately assess poor pollen quality in RQ.
The genetics of trailing habit in coleus appear to be controlled by a single tetrasomic gene with fully additive gene action. Complete dominance was not observed for the gene controlling trailing habit in coleus because F1 plants with genotypes UUUU, UUUu, UUuu, and Uuuu all had different growth habits. Therefore, additive gene action is proposed, with the effect of each additional allele U (for upright growth habit) being equal. One shortcoming that needs to be noted is the limited genetic base used to test the hypothesis of disomic tetraploid inheritance. As only 2 genotypes (RQ and S) were used, the conclusions drawn in this study are desirable for preliminary studies; therefore, more genotypes are needed to accurately test the hypothesis reported in this study.
The genetics for foliage color was not the main focus of this work, and we realize that quantification of a wide range of colors and patterns on a 5-point scale and using only 2 parental genotypes is somewhat simplistic; however, interesting observations were made. Rife (1948) stated that 6 multiple alleles, designated as the purple series, affected the amount and distribution of anthocyanin. Their interactions resulted in 9 phenotypic variations ranging from uneven purple and brown to green, with purple color being dominant (Rife 1948). In support of his results, purple color from RQ was observed in this study to be dominant to the orange color from S as all the F1 plants had some purple color in their foliage. However, there does appear to be some interaction with the genes controlling yellow/orange color in S because foliage color was improved in the F1 population. Although all F1 plants were still relatively dark in color with remnants of purple foliage color, and no plants had orange foliage color similar to S, some F2 plants with trailing habit and orange colors were identified. The goal of developing a plant with a more trailing growth habit than RQ and orange foliage color similar to S was achieved: an F2 seedling (H05-66-5) produced by selfing the F1 plant H05-66 possessed these 2 desirable characteristics. Although complete growth habit data were not collected on these F3 progeny, it appears to be a simplex (Uuuu) because all F3 plants had category 3, 4, or 5 growth habits. Thus, in 3 generations, trailing habit and orange foliage color were combined in one individual (Figure 4).
There is little foliage color diversity available in trailing coleus cultivars. In our study, we determined that trailing growth habit is qualitatively inherited and controlled by a single gene with additive effects. We propose the gene symbol U/u for upright growth habit with the quadraplex (UUUU) being fully upright. Through our breeding program at the University of Florida, we have successfully developed several dozen new coleus selections with trailing growth habit and orange foliage color. These selections are currently being evaluated for landscape performance.
Florida Nursery Growers and Landscape Association and the Florida Agricultural Experiment Station.
The authors gratefully acknowledge Cristina Tan and Becky Hamilton for assistance with all greenhouse work and Dr. Kevin Folta for review of the manuscript.