Plant materials.

During a resource survey in Kunming area, the research group collected a total of 27 Z. hybrida cultivars (the main color groups in Z. hybrida were covered by all selected cultivars) mainly derived from the greenhouse of Kunming Trifecta orchid Nursery Co., Ltd. (lat. 24°33′N, long. 102°41′E), in May 2016. The cultivation conditions were as follows: 20 °C with a 12 h photoperiod provided by incandescent lights and humidity of 30% (constant temperature and humidity). Plants showing the same growth pattern were selected, and fully opened spathes were collected for flower color determination, slice preparation, and pigment analysis.

Color analysis.

The colors of the fresh spathe of the 27 variants were first compared with the Royal Horticultural Society Color Chart, and the surface color of the spathe was analyzed with a Chroma Meter CR-400 [Konica Minolta (China) Investment Ltd., Shanghai, China] at C/2. Three spathe samples of each cultivar were taken from different plants. The Z. hybrida spathe was placed on clean white paper and the light source was aligned to the middle part of the spathe for measurement. Finally, using PASW Statistics 18 software, hierarchical clustering analysis was performed using phenotype values of spathe flower colors measured by colorimeter. The cluster analysis method using the furthest neighbor, coupled with CIElab color space data (Voss, 1992) was used to characterize spathe color.

Extraction of anthocyanins and TF.

Zantedeschia hybrida spathes were quickly ground into fine powder in liquid nitrogen. After the liquid nitrogen was evaporated, 250 mg of spathe powder was added to 1 mL of anthocyanin extraction solution [V (methanol) : V (water) : V (formic acid) : V (trifluoroacetic acid) = 70:27:2:1]. Total flavones and flavonols were extracted from 100 mg of fresh sample with 1 mL of methanol as the extraction solution using the same processing method. Both extracts were incubated at 4 °C followed by a 24 h extraction, with vortexing every 12 h (Hashimoto et al., 2000; Jia et al., 2008; Yoshitama, 1981; Zhang et al., 2009). The extracts were filtered through a 0.2 μm filter to obtain the samples to be tested.

Measurement of TA and TF and their structural identification.

Total anthocyanin was measured using HPLC (Li et al., 2008). TA was measured using an Agilent HPLC equipped with a P680 pump, UltiMate 3000 autosampler, DAD-100 ultraviolet-visible (ultraviolet-vis) detector, TCC-100 column oven and Agilent ZORBAX SB-Aq (4.6 mm × 250 mm) column. For the HPLC measurements, the mobile phase consisted of solvents A and B (solvent A : water : formic acid : trifluoroacetic acid = 97.9:2:0.1; solvent B : water : acetonitrile : formic acid : trifluoroacetic acid = 62.9:35:2:0.1). The mobile phase was subjected to ultrasonic degassing and 0.2 μm ultrafiltration before injection. The injection volume was 10 μL, the column temperature was 25 °C, and the flow rate was 0.8 mL·min⁻¹. The elution gradient profile of mobile phase B was 30% to 53% for 0–20 min, 53% to 53% for 20–40 min, 53% to 30% for 40–45 min, and 30% to 30% for 45–50 min. The peak areas of petal anthocyanin absorbance at a wavelength of 530 nm were determined (Jia et al., 2008).

For the HPLC-MS, HPLC-ESI-MSⁿ was used to analyze the structure of anthocyanins in typical samples using the Agilent 6540Q-TOT liquid chromatography–-mass spectrometry system. Liquid chromatography (LC) was performed based on the aforementioned conditions and procedures. Mass spectrometry (MS) conditions included ion trap analyzer scanning from m/z 100 to 1600; positive ionization mode with capillary voltage of 3500 V; nebulizer pressure of 200 kPa; capillary exit of 100 V; dry temperature of 350 °C; and drying gas (N₂) flow rate of 6.0 L·min⁻¹ (Sun et al., 2010; Yamagishi et al., 2014). The mass spectra were analyzed using MZmine software (Okinawa Institute of Science and Technology, Okinawa, Japan).

TF was measured using an UPLC-Q-TRAP (Ultimate 3000-API 3200 Q TRAP) system (SCIEX, Redwood, CA) equipped with an MSLab HP-C18 (150 mm × 4.6 mm, 5 μm) column. For the UPLC measurements, the mobile phase consisted of solvent A (water) and solvent B (acetonitrile), an injection volume of 10 μL, column temperature of 50 °C, and flow rate of 1 mL·min⁻¹.

Petal TA and TF were calculated in μg·g⁻¹ fresh petal using a semiquantitative method (compared with standards) (Wang et al., 2001). The content of each compound with standards was calculated using the external standard method (TA: y = 0.00001x − 0.000, R² = 0.999; TF: y = 0.00005x − 0.09369, R² = 0.998).

The anthocyanin quantitative standard was Cy, and the quantitative and qualitative standards for flavones and flavonols were naringenin, apigenin, luteolin, dihydromyricetin, dihydrokaempferol, dihydroquercetin, myricetin, kaempferol, and quercetin. The standards were purchased from Shanghai Yuanye Bio-Technology Co., Ltd. (Shanghai, China).

Extraction and measurement of carotenoids.

Zantedeschia hybrida spathes (100 mg) were ground in liquid nitrogen and added to 5 mL of extraction solution [V (petroleum ether) : V (acetone) = 1:1] with thorough vortexing, followed by an 8 h overnight extraction in the dark (wrapped in aluminum foil) at 4 °C. The extract was then filtered and the filtrate was subjected to ultraviolet–visible spectroscopy at a wavelength range of 400–500 nm using a double-beam ultraviolet–visible spectrophotometer (TU-1901; Beijing Persee General Instrument Co., Ltd., Beijing, China) (Zhao et al., 2004).

Observation of the spathe cross section of Z. hybrida.

The center regions of Z. hybrida where spathes maintain even coloration were used for the preparation of plant slices (Rasika et al., 2003). The center cross section of the spathes was manually cut using a shaver blade and placed on a slide for observation using a light microscope (Digital Sight DS-Fi2; Nikon Corporation, Tokyo Japan); photographs were taken using an attached camera.

Analysis of the spathe color phenotypes of different Z. hybrida cultivars.

The CIE L*, redness (a*), and yellowness (b*) values of Z. hybrida spathes with different colors were measured using colorimetry and the data were subjected to cluster analysis. The flower colors were divided into six groups based on the transition line graphs plotted for black, amaranth, red, yellow, pink, and orange groups (Table 1).

Table 1.Spathe colors and color parameters of 27 Z. hybrida cultivars.

View Fullscreen

The spathe colors of the 27 Z. hybrida cultivars were mainly distributed in Quadrants I, II, and IV on a* and b* hue coordinates, and no spathe color was distributed in Quadrant III of the blue area (Fig. 1A). The yellow group was distributed in Quadrant I, orange and red groups were distributed in Quadrant II, black and amaranth groups were mainly distributed in Quadrant IV, and the pink group was distributed in Quadrants I, II, and IV. The a* values (redness) ranged from −7.25 (‘Blanc’) to 45.46 (‘Figo’); the b* values (“yellowness”) ranged from −4.56 (‘Promise’) to 79.59 (‘Solo’); the h_ab (“hue angle”) ranged from −19° (‘Picasso’) to 110° (‘Blanc’).

View Figure

• Download as Powerpoint
• Download Figure

The three-dimensional coordinate distribution of the L*, a*, and b* values of the flower colors of each cultivar is shown in Fig. 1B. The L* values of black, amaranth, and red groups were lower than those of the pink, yellow, and orange groups with a concentrated distribution. Among them, the black group had the lowest L* value, whereas the pink and yellow groups had the highest L* values. The L* values ranged from 19.20 (‘Mirages’) to 83.80 (‘Solo’); the C values (chroma) ranged from 1.53 (‘Cantor’) to 80.02 (‘Solo’).

Different color groups of Z. hybrida have different relationships between L* and C (Fig. 2). There was a significant positive relationship between C values and L* values in the orange, yellow, and black groups (r² = 0.552, r² = 0.845, and r² = 0.546, respectively), where the L* increased linearly with the increase in C. On the contrary, there was a significant negative correlation between the C value and the L* values in the red group (r² = 0.904), where the L* decreased linearly with the increase in C. No significant relationship was found for C values and L* values in the amaranth and pink groups (r² = 0.0123, and r² = 0.228, respectively) (Supplemental Fig. 1).

View Figure

• Download as Powerpoint
• Download Figure

Analysis of pigment composition and content in Z. hybrida spathe.

Spathe anthocyanins in Z. hybrida were identified using HPLC-PAD to obtain the HPLC chromatogram of different color groups. The type of anthocyanin can be determined depending on the retention time of the anthocyanin peak (Yasuda, 1989a, 1989b). Two types of anthocyanins were detected from the results, namely A1 and A2, with retention times of 26 and 28 min, respectively. No anthocyanin absorption peak was detected in the yellow group or for ‘Ventura’ in the pink group (Fig. 3).

View Figure

• Download as Powerpoint
• Download Figure

The mass spectra of anthocyanins showed that their molecular ions often carry a hydrogen ion and are present in the form of [M + H]⁺. Anthocyanins were mainly classified based on the mass-to-charge ratio of glycoside ions [Y₀]⁺ in their mass spectra (Beatriz et al., 2009; Cuyckens and Claeys, 2004; Stobiecki, 2000). In this study, we performed an MS analysis on these two types of anthocyanins (Supplemental Fig. 2). The molecular ions at m/z 595 [(M + H)⁺] were obtained from A1, which produced the characteristic fragment ions of Cy at m/z 287 [(Y₀)⁺] and other fragment ions at m/z 449. The molecular ions at m/z 595 [(M + H)⁺] were obtained from A2, which produced the characteristic fragment ions of pelargonidin (Pg) at m/z 271 [(Y₀)⁺] and of other fragment ions at m/z 433. Hence, it was inferred that A1 was Cy and that A2 was Pg. The common component in the spathe of all color groups, except for the yellow group, was A1 (Cy). It was also the main anthocyanin component in Z. hybrida spathes. By contrast, A2 (Pg) was mainly found in certain cultivars of red and orange groups (‘Murano’, ‘Trinity’, ‘Durance’, and ‘Lido’).

The black group had a significantly higher spathe TA than the other color groups (Table 2; Fig. 4), followed by the amaranth, red, orange, and pink groups, and the TA content in each flower color group was highly significantly different (P < 0.01) (Table 3). ‘Murano’ had the highest proportion of Pg (38.5%) among four cultivars containing both Cy and Pg. All 27 cultivars were rich in TF with a copigmentation index (CI) that ranged from 8.12 to 989.06 (Table 2). The TF content in each flower color group was significantly different (P < 0.05) (Table 4).

Table 2.Pigment constitutions and relative quantity of anthocyanins in 27 Z. hybrida cultivars.

View Fullscreen

View Figure

• Download as Powerpoint
• Download Figure

Table 3.The analysis of variance analysis of the content of total anthocyanin among six flower color group of Z. hybrida.

View Fullscreen

Table 4.The analysis of variance analysis of the content of TF among six flower color group of Z. hybrida.

View Fullscreen

UPLC-Q-TRAP-MS was used to analyze the composition of flavones and flavonols in the spathes of two cultivars (‘Romance’ and ‘Black Girl’). Ten peaks from primary MS data were found to be associated with flavones and flavonols using comparison of standards. Comparison of the total ion chromatogram of spathe samples and standard total ion chromatogram (Supplemental Fig. 3) indicated that the samples from the two cultivars contained naringenin, luteolin, myricetin, dihydromyricetin, dihydrokaempferol, dihydroquercetin, kaempferol, quercetin, catechins, and epicatechin. Quantitative analysis using standards was performed (Table 5).

Table 5.The composition and content of flavones and flavonols in ‘Black girl’ and ‘Romance’.

View Fullscreen

Measurement of the spathe carotenoid extract using ultraviolet–visible spectrometry indicated that none of the 27 Z. hybrida cultivars showed the characteristic absorption peak at wavelengths of 400–500 nm, indicating the absence of carotenoids in the Z. hybrida spathe.

Analysis of microstructure and pigment distribution in the Z. hybrida spathe.

Analysis of the cross sections of the spathes of different Z. hybrida color groups showed that the spathe had flattened upper and lower epidermal cells, and that pigments were distributed in both upper and lower epidermal cells and mesophyll cells. In addition, spathes of different color groups had different numbers of stained cell layers, and spathes with a thicker stained cell layer had a deeper color (Fig. 5). Pigments of the black group were richly distributed in epidermal and mesophyll cells, and in the two to three layers of mesophyll cells close to both upper and lower epidermis. In the amaranth group, the pigments were distributed in the epidermal cells and the two layers of mesophyll cells; the lower epidermal cells and mesophyll cells close to the lower epidermis presented reduced accumulation compared with the upper epidermal cells. In the red and orange groups, pigments were distributed in the epidermal cells and a single layer of mesophyll cells. In the pink group, pigments were present only in the epidermal cells. Anthocyanins were not found in the yellow group or in ‘Ventura’ of the pink group. Thus, spathe coloration in Z. hybrida was closely associated with pigment distribution in the spathe.

View Figure

• Download as Powerpoint
• Download Figure

Correlation among spathe color parameters in Z. hybrida.

This study performed a hierarchical analysis of flower color phenotypes L*, a*, and b*, and established a measurement system for defining and classifying the flower color of Z. hybrida. The 27 collected cultivars were classified into six major color systems. When drawing the various color systems of Z. hybrida on the two-dimensional hue a, b coordinate scatter plot, we found that the yellow-color group was mainly distributed at 90 degrees around the coordinates, whereas the red-, black-, pink-, and amaranth-color groups were distributed at zero degrees around the coordinates, and the orange-color group was distributed at the second quadrant of the coordinates. This is generally consistent with the definition and distribution of colors in the CIElab color space (Voss, 1992), proving that the definition and classification of Z. hybrida flower colors in this study are scientific and reasonable.

Different Z. hybrida color groups had different relationships between L* and C. L* values and C values of orange and yellow groups had a significant positive relationship, whereas the red group had a significant negative relationship. The linear regressions in the black, amaranth, and pink groups were nonsignificant. In previous studies, some color groups of Chrysanthemum morifolium cultivars (Hong et al., 2012) and some color groups of Consolida ajacis cultivars (Hashimoto et al., 2000) showed a significant negative relationship with the L* and C values. The differences in the correlation between L* and C values may be due to differences in anthocyanin compositions and distribution, as well as different petal cross-sectional structures in different cultivars.

Relationship between the main pigments and phenotypes in Z. hybrida.

The flower color is an important ornamental trait of plants (Nielsen et al., 2003). This study identified spathe anthocyanins in 27 Z. hybrida cultivars and found that the yellow group and ‘Ventura’ of the white group did not contain anthocyanins. Cultivars of all other color groups contained anthocyanins with Cy as the main component, whereas Pg only accumulated in four cultivars. Cy produced varying degrees of amaranth coloration of the spathe, similar results have been obtained in previous studies (He et al., 2011), whereas Pg produced varying degrees of vermilion color. According to a study of Papaver nudicaule, Pg is also an important coloring material that makes the petals red and orange (Bettina et al., 2016). The flavonoid composition of typical Araceae plants is relatively simple (Williams et al., 1981); for the Z. hybrida cultivars tested in this study, there was only one anthocyanin in 18 cultivars, and only two anthocyanins were found in the four cultivars. Anthocyanin was not detected in other cultivars, and carotenoids were not detected in any cultivars. Therefore, we believe that TF was the main pigment type producing yellow and orange colors. However, Lewis et al. (2003) found accumulation of carotenoids in cultivars of yellow and orange groups, possibly as the result of differences among cultivars. Flavones and flavonols are important copigments that exert their copigmentation effects when the CI exceeds 5 (Asen et al., 1971). In this study, the CI values of all tested cultivars were greater than 5, indicating that flavones and flavonols play a relatively greater role in the coloration of Z. hybrida.

Previous studies have demonstrated that petal anthocyanins affect light absorption (Hughes et al., 2006; Merzlyak et al., 2008). The study of the relationship between the spathe phenotype and TA in Z. hybrida revealed a significant negative relationship between TA and L* value (r² = 0.615) (Fig. 6), i.e., large amounts of TA decreased the L* values. The pink (r² = 0.849), red (r² = 0.667), and orange (r² =0.511) (Supplemental Fig. 4) color groups showed significant positive relationship between a* value and TA, demonstrating that TA content in these color groups is closely associated with redness. The b* value and TF value and the orange-color group showed a significant positive relationship (r² = 0.766) (Supplemental Fig. 4), suggesting that the formation of the orange-color group resulted from the combined effects of TA and TF. In previous studies on Z. hybrida (Lewis et al., 2003), the test materials were relatively uniform (orange, red, and pink color groups), and the L* values were distributed from 28.4 to 89.5, the h values were distributed from 35° to 359°, and the TA concentration was ≈0–24 μg·g⁻¹. In the test materials in this study, the L* values were distributed around 19.2–83.8, h values were ≈0.2° to 110°, and TA content was ≈0–1179.8 μg·g⁻¹, which basically covered all existing major Z. hybrida color groups and cultivars and meant that we systematically analyzed the relationship between the type and content of anthocyanins and flower color phenotypes.

View Figure

• Download as Powerpoint
• Download Figure

Black color in flowers is a highly attractive trait in the floricultural industry. Flower color shows species specificity and black flowers in different species have different coloring mechanisms. Dahlia flowers are presented as black generally because of low L* and low C (Deguchi et al., 2013, 2015). The black cultivars collected in our study also have low L* and C values. The petals of black Dahlia cultivars accumulated large amounts of anthocyanins (Deguchi et al., 2013). Similarly, the black cultivars in our study also showed the largest accumulation of anthocyanin content (1179.8 μg·g⁻¹). In addition, TA content in the spathes of 27 cultivars showed a significant negative relationship with L* values, i.e., when L* values were reduced to present a black phenotype, TA was an important factor. The main reason for black Alcea rosea ‘Nigra’ presenting as a black flower is the copigmentation of flavonols and flavones and accumulation of large amounts of anthocyanins (Hosaka et al., 2012). In the six black cultivars collected in this study, in addition to extensive accumulation of anthocyanins, flavonols and flavones accumulated in large amounts at the same time, and copigmentation may thus play some role in black coloration.

To further analyze the causes of the formation of the black group, this study compared the intermediate metabolites of flavonoids in ‘Black Girl’ of the black group and ‘Romance’ of the pink group, and characterized the metabolic pathway of anthocyanins in both cultivars (Fig. 7A and B). The amount of intermediate metabolites in both cultivars varied greatly (Table 5), with significantly higher amounts in ‘Black Girl’ than in ‘Romance’. In addition, ‘Black Girl’ cultivars did not accumulate flavones, possibly because of the inhibition of FNS expression in ‘Black Girl’ that leads to inhibition of flavone synthesis as a result of competition with anthocyanins synthesis, thereby resulting in high accumulation of anthocyanins. This result was similar to that previously reported in Dahlia (Deguchi et al., 2013). Dihydromyricetin and myricetin were detected in both cultivars, which suggests that F3′5′H may exist in Z. hybrida. However, these plants were incapable of accumulating delphinin to produce a blue coloration. This result requires further study.

View Figure

• Download as Powerpoint
• Download Figure

Relationship between colorant distribution in the Z. hybrida spathe and phenotype.

When light shines on the petals to penetrate the pigment layer, a higher number of pigment layers lead to higher light absorption and the proportion of incident light entering the cells increases, resulting in a deeper flower color (Gorton and Vogelmann, 1996; Noda et al., 1994; Rasika et al., 2003). It was found in this study that colorants were distributed in the upper and lower epidermal cells, as well as in multiple layers of mesophyll cells near the epidermal cells (palisade and spongy tissues) of the Z. hybrida spathe. In addition, cultivars of different color groups had different numbers of pigmented cell layers; a spathe with a thicker pigmented layer had a deeper color and the black group had the thickest pigmented cell layer. This is similar to the effects of pigment distribution in grape hyacinth on coloring (Qi et al., 2013). Cross-sectional structural observation indicated that the flavones and flavonols that cause yellow flower coloration are mainly distributed in the mesophyll cells of spathes, which is a different cell layer from where anthocyanins are distributed. This resulted in a background effect, particularly in orange coloration of spathes.

The shape of the petal epidermal cells affects the optical properties of the anthocyanin they contain, thereby affecting the appearance of the petals. Most angiosperms have conical petal upper epidermal cells that enhance the proportion of incident light entering the epidermal cells and the light absorption by pigments, thereby increasing the color intensity of the flower (Baumann et al., 2007). However, analysis of the microstructure of the spathes of the 27 cultivars examined in this study indicated that their epidermal cells did not have a conical shape, but rather a flattened shape, which resulted in greater light reflection that reduced the color intensity of the flowers (Gorton and Vogelmann, 1996). Therefore, the shape of the spathe epidermal cells was unrelated to the formation of the black Z. hybrida group. Conical epidermal cells can enhance color saturation (Baumann et al., 2007), whereas the flattened spathe epidermal cells of Z. hybrida reduce color saturation, which may be the main reason for the absence of correlation among C value, TA, and TF in this study.

In conclusion, the coloration of the Z. hybrida spathe is closely associated with its pigment components and contents as well as the cross-sectional structure of the spathe. The formation of the Z. hybrida spathe in the black group was closely associated with high TA and accumulation of anthocyanins in multiple cell layers, and was not correlated with the shape of epidermal cells. Analysis of the relationship between the coloration of Z. hybrida and its pigment compositions, contents, and distribution is important for resolving the coloration mechanism of the Z. hybrida spathe and for molecular breeding to improve the color of the flowers.