Plant material.

An initial selection of 15 C. moschata and 15 C. pepo cultigens (cultivars, heirlooms, and PIs) was made based on subjective descriptions of flesh color with samples representing a range of whites, yellows, and oranges. Cultivars and heirlooms were purchased from commercial seed producers, whereas PIs were obtained from the USDA-ARS North Central (Ames, IA) and Southern (Athens, GA) Regional Plant Introduction Stations.

Field trials.

A preliminary field study was conducted 6 Apr. to 15 June 2007 at the Plant Science Research and Education Center (PSREC) located in Citra, FL, to evaluate and select from the 15 C. moschata and 15 C. pepo cultigens those that would provide a range of white, yellow, and orange flesh color based on colorimetric analysis using the method described subsequently. A randomized complete block design with two replications per cultigen, eight plants per plot, was used. Temperatures recorded by the Florida Automated Weather Network (FAWN) recorded 22 °C average, 38 °C maximum, and 5 °C minimum over the course of the growing season.

From the preliminary study, six C. moschata and five C. pepo cultigens were selected for both colorimetric and HPLC analysis after plantings at two locations, PSREC and the Gulf Coast Research and Education Center (GCREC), Wimauma, FL, during 11 Mar. to 27 June 2008. Temperatures recorded by FAWN were 22 °C average, 33 °C maximum, and 4 °C minimum and 24 °C average, 38 °C maximum, and 22 °C minimum for PSREC and GCREC, respectively. A randomized complete block design with two replications per cultigen, eight plants per plot, was used at each location. Recommended conventional cultural practices and fertility rates for Florida squash were followed for both 2007 and 2008 (Olson and Simonne, 2007).

Colorimetric and high-performance liquid chromatography analysis.

A total of 220 mature fruit (11 cultigens × five fruit × two replications per location × two locations) was harvested for colorimetric and HPLC analysis. Fruit maturity was based on common commercial practices for squash type. Color was recorded using a Minolta CR-400 Colorimeter (Minolta Camera Co., Ltd., Ramsey, NJ) tristimulus color analyzer equipped with an 8-mm diameter measuring area and diffuse illumination of a 2° standard observer. The L* coordinate indicates darkness or lightness of color and ranges from black (0) to white (100). Coordinates, a* and b*, indicate color directions: +a* is the red direction, –a* is the green direction, +b* is the yellow direction, and –b* is the blue direction. Chroma is the saturation or vividness of color. As chromaticity increases, a color becomes more intense; as it decreases, a color becomes more dull. Hue angle is the basic unit of color and can be interpreted, for example, as 0° = red and 90° = yellow. Both chroma and hue are derived from a* and b* using the following equations: metric chroma: C* = √ (a*)² + (b*)² and metric hue angle: h = tan⁻¹ (b*/a*) (degrees).

Slicing each fruit transversely, L*a*b* color space measurements from the edible flesh (mesocarp) of each fruit were recorded within 5 min to avoid discoloration. Our preliminary study revealed replicate measurements of the mesocarp of each fruit per cultigen not significantly (P ≥ 0.05) different; therefore, avoiding the seed cavity and surrounding tissue (≈10 mm), three random measurements per fruit were recorded. A total of 660 (11 cultigens × five fruit × three colorimeter measurements per fruit × two replications per location × two locations) colorimetric values were recorded.

The separation and quantification of carotenoids were accomplished by HPLC. A total of 22 samples (11 cultigens per location) were prepared for HPLC analysis directly following colorimetric measurements. From each of five fruit harvested per replication, a 10.0 ± 0.1 g FW cubed flesh sample was cut and combined to make a 100.0 g ± 0.5 g sample per cultigen per location. Samples were vacuum-sealed in plastic storage bags within 50 min of initial fruit slicing, wrapped with aluminum foil, labeled, and held at –20 °C. Samples were then sent frozen to Craft Technologies, Inc. (Wilson, NC) for saponification and HPLC processing. Carotenoids measured included lutein, zeaxanthin, cis-lutein-zeaxanthin, α-carotene, β-carotene, cis-β-carotene, and total carotenoids. All fruit was processed with the colorimeter and prepared for HPLC within 3 d of field harvest and underwent HPLC analysis within 4 weeks of harvest.

Data analysis.

Color space values and carotenoid content of the Cucurbita cultigens, across locations, were subjected to analysis of variance by the GLM procedure of SAS (Statistical Analysis System Version 9.2; SAS Institute, Cary, NC). Cultigens, locations, replications, fruit, and replicate measurements within fruit were considered as random effects. Least significant differences among cultigens were determined at a 5% significance level. Cultigen means across locations were calculated for each trait. Components of variance were estimated by using the VARCOMP procedure of SAS. Broadsense heritabilities, based on cultigen means, were calculated for each trait according to Fehr (1987). Spearman's coefficient (r_s) of rank correlations (Steel et al., 1997) were calculated to test differences in rank order among the cultigens between the two locations. Pearson correlation coefficients (r) and linear regression (R²) between color space values and carotenoid content were calculated from the means of the cultigens across locations using the CORR and REG procedure of SAS. Scatterplots of the data indicated that all relationships between color space values and carotenoid content were linear.

Colorimetric evaluation.

Significant (P ≤ 0.0001) differences were observed among cultigens for flesh color represented by the five color space values: L*, a*, b*, chroma, and hue. This demonstrates that genetic variation for flesh color is present among the cultigens tested. Locations were not significantly (P ≥ 0.60) different for each color space value, but there was a significant (P < 0.0001) cultigen-by-location interaction. Based on Spearman's rank correlation tests, no significant (r_s ≥ 0.89) change in rank order of the cultigens between locations was noted. This indicates that the flesh color of each cultigen was consistent despite different environments. Replicate measurements within fruit of each cultigen were not significantly (P ≥ 0.18) different, but significant (P ≤ 0.02) differences were observed for the color space values among fruit within cultigens. Variance component analysis, however, revealed fruit within cultigen estimates to be less than 1.5% of the phenotypic variance for each value. With such a small effect, little would be gained by evaluating more fruit per cultigen. Broadsense heritability (H²) estimates of the five color space values (Table 1) ranged from 0.81 to 0.93 indicating that genetic improvement and effectiveness of selection for color would be moderate to high.

Table 1.

Means and broadsense heritability (H²) estimates of color space values, L*, a*, b*, chroma, and hue^z measured in fruit flesh of 11 Cucurbita cultigens.

View Table

Subjective observations of flesh color were consistent with the color space value hue in that lower hue angles corresponded to more orange–red flesh and higher hue angles corresponded to more yellow flesh (Table 1). ‘Butterbush’ (C. moschata), with a mean hue angle of 77.5°, represented orange–red flesh color in this study and was significantly (P ≤ 0.05) different from all other cultigens evaluated. ‘PI 314806’ (C. pepo) reflected a whitish yellow flesh color with a mean hue angle of 102.9°. Differences in hue of the cultigens tested may be the result of the various types and ratios of pigments present.

Mean color space values a* (+a* red direction; –a* green direction) ranged from –4.9 for ‘Tennessee Sweet Potato’ (C. moschata) and ‘Fordhook Acorn’ (C. pepo) to 14.8 for ‘Butterbush’ (C. moschata) (Table 1). The lowest mean color space value, b* (+b* yellow direction; –b* blue direction) was 9.8 found within PI 314806 (C. pepo) with the highest, 71.4, found within ‘Waltham Butternut’ (C. moschata). Cultigens with hue angles 90° or greater had yellow flesh, mean a* values ranging from 0.2 to –4.9, and mean b* values ranging from 9.8 to 47.8. Cultigens with hue angles 90° or less had yellow–orange flesh, mean a* values ranging from 6.6 to 14.8, and mean b* values ranging from 47.3 to 71.4.

A range of L*, lightness (83.8) to darkness (70.5) of color, and chroma, dullness (10.0) to vividness (72.2) of color, was present within the cultigens evaluated (Table 1) and may reflect different concentrations of pigments. Three yellow–orange cultigens with hue values not significantly (P ≥ 0.05) different from each other included ‘Waltham Butternut’ (C. moschata), ‘Sucrine DuBerry’ (C. moschata), and ‘Table Gold Acorn’ (C. pepo). The flesh color of ‘Waltham Butternut’ is lighter (mean L* = 73.4) and more vivid (mean chroma = 72.2) compared with ‘Sucrine DuBerry's’ flesh color, which is darker (mean L* = 70.5) and duller (mean chroma = 47.8). The yellow–orange flesh and rind color of ‘Table Gold Acorn’ is the result of the introgression of the B (Bicolor) gene. Typically, acorn-type squash have light yellow flesh color at maturity as reflected by ‘Table King Bush’ (mean L* = 83.8; mean hue angle = 93.3), ‘Thelma Sander's Sweet Potato’ (mean L* = 81.6; mean hue angle = 93.8), and ‘Fordhook Acorn’ (mean L* = 83.4; mean hue angle = 97.7). Several genes that affect rind as well as the flesh color of squash have been described (Paris and Nelson Brown, 2005; Tadmor et al., 2005) including D (dark), l-1 (light coloration–1), l-2 (light coloration–2), and B (bicolor). Tadmor et al. (2005) studied the effects of these particular genes in different genetic combinations within NILs of C. pepo and reported that when the dominant allele of B interacts with the dominant allele of L-2, an intense orange rind and flesh color will develop. The flesh color of ‘Table Gold Acorn’ is lighter (mean L* = 76.5) than both ‘Waltham Butternut’ and ‘Sucrine DuBerry’ and falls in between these two cultigens for color vividness (mean chroma = 65.3).

Carotenoid content.

Among cultigens tested, significant (P ≤ 0.03) differences were observed for α-carotene, β-carotene, cis-β-carotene, and total carotenoids suggesting genetic variation in carotenoid accumulation is present. Statistically, no significant differences were observed for lutein (P = 0.08), zeaxanthin (P = 0.07), or cis-lutein-zeaxanthin (P = 0.06), but this may be a reflection of restricted sampling for analysis. In this study, locations were not significantly (P > 0.05) different for each carotenoid measured. Broadsense heritability (H²) estimates of the carotenoids measured ranged from 0.37 to 0.85 (Table 2) with genetic improvement and effectiveness of selection being most effective for α-carotene (H² = 0.85) and β-carotene (H² = 0.74).

Table 2.

Means and broadsense heritability (H²) estimates of carotenoids (μg·g⁻¹ fresh weight) measured in 11 Cucurbita cultigens.

View Table

On average, orange–red and yellow–orange flesh colored cultigens contained 21.5 μg·g⁻¹ FW total carotenoids, whereas yellow flesh colored cultigens contained 2.4 μg·g⁻¹ FW total carotenoids (Table 2). This represented a ninefold increase in total carotenoids provided within orange–red and yellow–orange flesh-colored cultigens versus yellow flesh-colored cultigens. The most abundant carotenoids within the orange–red and yellow–orange flesh-colored cultigens were lutein (average, 9.8 μg·g⁻¹ FW), α-carotene (average, 4.5 μg·g⁻¹ FW), and β-carotene (average, 5.4 μg·g⁻¹ FW). The most abundant carotenoid within the yellow flesh colored cultigens was lutein (average, 1.7 μg·g⁻¹ FW). Zeaxanthin (range, 0.1–0.2 μg·g⁻¹ FW), cis-lutein-zeaxanthin (range, 0.1–1.2 μg·g⁻¹ FW), and cis-β-carotene (range, 0.1–1.7 μg·g⁻¹ FW) were measured but not found within all cultigens evaluated.

Tadmor et al. (2005) evaluated the carotenoid content of NILs of C. pepo differing in fruit pigmentation loci B (bicolor), D (dark), l-1 (light coloration–1), and l-2 (light coloration–2). In their study, genetic backgrounds that lacked either the dominant D or dominant L-2 alleles were yellow flesh-colored. On average, these NILs contained 1.2 μg·g⁻¹ FW lutein, 0.6 μg·g⁻¹ FW α-carotene and β-carotene, and 1.8 μg·g⁻¹ FW total carotenoids. Within our yellow flesh-colored C. pepo cultigens, similar carotenoid concentrations were obtained with an average lutein concentration of 1.8 μg·g⁻¹ FW, an average α-carotene and β-carotene concentration of 0.6 μg·g⁻¹ FW, and an average total carotenoid concentration of 2.7 μg·g⁻¹ FW (Table 2). In our study, no significant (P ≥ 0.05) differences among the yellow flesh-colored C. pepo cultigens for any of the carotenoids were identified.

In this study, we evaluated ‘Table Gold Acorn’, a yellow–orange flesh-colored C. pepo cultigen. The yellow–orange flesh color of this cultigen is the result of the introgression of the B pigmentation allele. In our study, this cultigen had at least a threefold increase in total carotenoid content compared with other C. pepo cultigens (Table 2). In the Tadmor et al. (2005) study, the effect of B on carotenoid content within their C. pepo NILs was dependent on the genetic combinations of pigmentation loci D, l-1, and l-2. In genetic backgrounds with either dominant D or dominant L-1, the B/B genotype had an approximate twofold increase in total carotenoid content. In NILs with B/B and L-2/L-2 genotypes, total carotenoid content increased 10-fold. The genetic makeup of ‘Table Gold Acorn’ will need to be further investigated to determine its allelic state at the pigmentation loci D, l-1, and l-2 and the cause of its increased carotenoid content.

The C. moschata cultigen ‘Butterbush’ contained the highest total carotenoid content (42.3 μg·g⁻¹ FW) and significantly (P ≤ 0.05) more α-carotene (14.9 μg·g⁻¹ FW) and β-carotene (15.3 μg·g⁻¹ FW) compared with other C. moschata cultigens evaluated in this study (Table 2). This suggests genetic variability is present within C. moschata that leads to elevated carotenoid content and concentrations. Additional evidence for variability in carotenoid content and concentration within C. moschata is provided by Murkovic et al. (2002). The C. moschata cultigens in their study averaged 86.7 μg·g⁻¹ FW total carotenoids (range, 41.6–143.0 μg·g⁻¹ FW), 28.2 μg·g⁻¹ FW α-carotene (range, 9.8–59.0 μg·g⁻¹ FW), and 51.0 μg·g⁻¹ FW β-carotene (range, 31.0–70.0 μg·g⁻¹ FW). The carotenoid averages and their ranges were higher than those found in our study. It would be interesting to investigate the genetic basis that leads to elevated carotenoids within C. moschata and to compare this with the C. pepo findings of Tadmor et al. (2005).

Correlations between color space values and carotenoids.

Pearson correlations coefficients (r) between color space values, L*, a*, b*, chroma, and hue, with lutein, α-carotene, β-carotene, and total carotenoids, the four most prominent carotenoid levels measured in this study, were calculated (Table 3). The color value L* (lightness or darkness) correlated (P < 0.03) negatively with lutein (r = –0.68) and total carotenoids (r = –0.66). A negative correlation between L* and certain carotenoids would be expected because any increase in pigment would increase the darkness and thereby decrease L*. The strength of our two linear relationships, however, would be considered weak (r ≤ –0.70). The color value a* (color direction in red or green) was strongly (r ≥ 0.85) correlated with total carotenoids (r = 0.91) followed by moderate (0.75 ≤ r ≤ 0.84) to weak correlations with lutein (r = 0.84), β-carotene (r = 0.77), and α-carotene (r = 0.70). The color value b* (color direction in yellow or blue) and chroma (saturation or vividness of color) correlated strongly with lutein (r = 0.87 for both) and moderately with total carotenoids (r = 0.75 and r = 0.76). Hue (tint of color, an angular measure) was weakly correlated with α-carotene (r = –0.62) and β-carotene (r = –0.69) but moderately correlated with lutein (r = –0.80) and total carotenoids (r = –0.83). The negative correlation observed between hue and the carotenoids measured in this study suggests that as hue angles decrease, carotenoid concentrations would increase.

Table 3.

Pearson correlation coefficients (r) between color space values (L*, a*, b*, chroma, and hue) and carotenoids calculated from the means of 11 Cucurbita cultigens.

View Table

Studies relating colorimetric values with total carotenoids and/or β-carotene in winter-type squash have previously been reported. Francis (1962) evaluated cultigens of C. maxima and C. moschata and identified moderate to strong correlations between L* (r = –0.78), a* (r = 0.83), b* (r = 0.79), chroma (r = 0.93), and hue (r = –0.96) with total carotenoids. In a study evaluating C. maxima germplasm, Seroczynska et al. (2006) reported poor to fair correlations between L* (r = –0.53), a* (r = 0.77), b* (r = 0.76), and chroma (r = 0.77) with total carotenoids and poor to fair correlations between L* (r = –0.54), a* (r = 0.74), b* (r = 0.66), and chroma (r = 0.67) with β-carotene. Comparing our findings with those of Francis (1962) and Serocyznska et al. (2006) revealed similar correlations and strengths, except our correlation between a* (r = 0.91) and total carotenoids was stronger.

In other vegetable crops, studies have reported correlations between colorimetric values and carotenoids. For example, in tomato, both D'Souza et al. (1992) and Arias et al. (2000) found L* (r = –0.91 and r = –0.92) and a* (r = 0.87 for both) to be strongly correlated with lycopene content. In sweetpotato, Simonne et al. (1993) reported a strong positive correlation between hue and β-carotene (r = 0.99), whereas Ameny and Wilson (1997) reported a moderate correlation between b* and β-carotene (r = 0.74). All four reports viewed colorimetric analysis as an appropriate estimator of carotenoid concentrations. A report by Reeves (1987), however, cautioned readers about making valid assessments of carotenoid concentrations by tristimulus colorimetry. He stated that although correlations may show statistical significance, the variation explained by the correlation may not be adequate enough to be of practical use. In his evaluation of pepper color and carotenoid data, the best parameter to predict total carotenoid content was a negative correlation (r = –0.713) with L* from pureed peppers. In his view, a mathematical equation based on this correlation would, at best, explain only 51% (R²) of the observed variation between L* and total pigment content.

Considering the recommendation by Reeves (1987), the strongest linear relationship, in our study, was found between a* and total carotenoids (r = 0.91; R² = 0.83). The R² value indicates that 83% of the variation in total carotenoids can be accounted for by the change in a*. The color values b* and chroma were also strongly correlated with lutein (r = 0.87; R² = 0.76 for both). The R² values indicate that 76% of the variation in lutein can be accounted for by the change in b* or chroma. Although regression equations based on these correlations may account for only 76% to 83% of the variation for lutein and total carotenoids, respectively, they may still be useful for estimating these concentrations. Regression equations for the prediction of lutein and total carotenoids based on a*, b*, and chroma, based on our findings, are provided in Table 4.

Table 4.

Best fit (r ≥ 0.85) regression equations for pairs of related carotenoids with color space values.

View Table