Index | Search | Home | Table of Contents

Loy, J.B. 1990. Hull-less seeded pumpkins: a new edible snackseed crop. p. 403-407. In: J. Janick and J.E. Simon (eds.), Advances in new crops. Timber Press, Portland, OR.

Hull-less Seeded Pumpkins: A New Edible Snackseed Crop

J. Brent Loy

    1. Inheritance of Seedcoat Development and Color
    2. Plant Fruiting and Growth Habit
  7. Table 1
  8. Table 2
  9. Fig. 1
  10. Fig. 2
  11. Fig. 3


Pumpkin seed (Cucurbita spp.) are consumed as a snack in many cultures throughout the world, and the seed is especially popular in Central American countries where strains have been selected primarily for edible seed use. World production of Cucurbita seed for food use is meager, however, because hand dehulling is necessary for obtaining marketable seed and there has been little effort to improve seed yields.

Tschermak-Seysenegg (1934) reported a mutant trait in the seedcoat of C. pepo L. in which the seedcoat was much reduced in thickness. The "naked seed" trait permitted more efficient extraction of vegetable oil and seed could be consumed directly without the necessity of dehulling. Genetically "hull-less" seed produce all seedcoat tissue layers, but secondary wall thickening is reduced in the outer tissues (epidermis, hypodermis, and sclerenchyma). As mature hull-less seed dry, the outer tissues collapse, producing a thin seedcoat (Stuart and Loy 1983).

Pumpkin seed are rich in both oil and protein; oil contents are generally in the 40 to 50% range (Bemis et al. 1968, Jacks et al. 1972) and protein content ranges from 30 to 40% (Jacks et al. 1972, J.B. Loy unpubl.). The amino acid content of pumpkin seed protein is not as well utilized as soybean protein (Zucker et al. 1958), being low in methionine, threonine, and tryptophan (Bressani 1963, Robinson 1975). Limited studies indicate that pumpkin seed oil is highly unsaturated and semi-drying (Lazos 1986). In the United States, Curtis (1947) recognized the potential value of hull-less seeded pumpkins for producing a high quality vegetable oil, for use as a snackseed, and for making a sandwich spread. He initiated a breeding program at the University of Connecticut to improve seed yields of pumpkin, but did not remain in Connecticut long enough to carry the project to fruition.

As a result of sampling pumpkin seeds at various trade shows, I observed that roasted seed have excellent consumer appeal. Moreover, in the U.S. alone, the potential market for new and superior products in the confectionery trade (snacks, trail mixes, granola cereals, salad bars, etc.) is huge. For example, 185,000 metric tons of sunflower were marketed in the U.S. for this purpose in 1981 (Lusus 1985). Also, use of pumpkin seed in developing countries could provide a concentrated food source, high in protein and energy value.

In 1977, I began investigations on the genetics and breeding of hull-less seeded pumpkins and soon thereafter attempts to develop high seed-yielding strains suitable for supporting a snackseed industry


Inheritance of Seedcoat Development and Color

Several inheritance studies conducted in the early 1950s (Heinisch and Ruthenberg 1950; Mudra and Neuman 1952; Schoeniger 1950, 1952, 1955; Weiling and von Becherer 1950) produced conflicting interpretations as to the mode of inheritance of the hull-less condition. More recent work by Stuart (1983) showed convincingly that a single recessive gene conditions the existence of a thin, parchment like seedcoat. At least two additional modifying genes, one of which appears to be dominant (J.B. Loy unpubl.), further reduce the degree of seedcoat development. In seed classified as completely hull-less, the outer seedcoat layers are reduced to the extent that seed take on the appearance of the inner seedcoat layer (chlorenchyma) which is normally dark green (Fig. 1, 2). I have found two color variants of chlorenchyma issue, light green, recessive to dark green, and yellow, recessive to both dark green and light green. When the outer seedcoat is partially developed (monorecessive condition) seedcoats appear grey against a green chlorenchyma background and tan against a yellow chlorenchyma background.

Plant Fruiting and Growth Habit

I am developing strains with the bush rather than vine habit of growth for several reasons. Bush plants appear to be more productive than vine plants because of a higher harvest index and adaptability to higher density plant culture (Broderick 1982). Bush plants set fruit around the crown of the plant resulting in more uniform ripening and easier harvesting.

Most hull-less seeded plant accessions currently available have moderate to large fruit size, and some of the large-fruited strains have a high seed yield per fruit. However, when seed yield is evaluated in terms of seed weight per kg of fruit, I have obtained the highest yields in small-fruited (0.5 to 1.5 kg) selections. Moreover, small-fruited lines tend to set fruit better, set fruit earlier, and set a larger number of crown fruit than lines bearing large fruit. Although large seed size is often associated with large fruit, there appears to be no genetic or physiological barriers to developing large-seeded strains with small fruit.


To evaluate components of yield, a spacing trial was conducted in 1986 with two F5 strains of pumpkin, NH14-40-6 and NH55-7-20. Plant density was varied by using gradient density plots in which row width was kept constant at 0.9 m and within row subplots were arranged in progressively decreasing within row spacing from 1.5 to 0.3 m. This produced plant populations between 7,890 to 35,880 plants/ha. Subplots consisted of 5 data and 2 guard plants. Seed yields were low in 1986, but the spacing results revealed important relationships between plant density and components of yield among bush lines. With both strains, per plant yields decreased significantly with increasing plant density (Fig. 3). On the other hand, seed yield per unit area increased with plant density (although due to extreme variability within a few plots, linear regressions were not significant at P = 0.05).

Some important components of yield and their relationship to plant density are given in Table 1. Fruit number, and to a lessor extent, mean fruit weight decreased with increasing plant density in both strains. Seed yield per fruit usually decreases with fruit size, but this relationship was not consistent in 1986. Plant spacing and fruit size had no effect upon seed size, although I have observed that seed size decreases in extremely small fruit (<0.3 kg). An important criterion for assessing the efficiency with which a plant partitions photosynthates into seed is the Seed Index (SI) or ratio of seed dry weight to total fruit dry weight. Within the strains used for the spacing trial, SI's were low and variable, and showed no relationship to plant density.


A third cycle of snackseed selections has been derived from a cross of NH14-40-6 to a seedy, partially hull-less strain, NH54-10-14. The major selection criteria have been small fruit, thin pericarp, high seed weight per kg fruit, high harvest index, vigorous growth at close spacing, good pollen production, and crown fruit set. A summary of performance of individual selections and selections within lines in F2 and F3 generations obtained in 1986 and 1987 is given in Table 2. The most noteworthy observation is that seed yield (g/kg fruit wt) and seed index have been dramatically improved as compared to the lines used in the spacing trial. Furthermore, these lines show improved germination under field conditions, more vigorous growth, and several are adaptable to high density culture. Seed yields were, in general much higher in 1987 (F3) than in 1986 (F2). However, seed indices were not higher because pericarp dry matter was also higher.

It was expected that potential seed yields of advanced lines could be evaluated in 1988. Eighteen F4 and F5 snackseed lines were grown in rows 1.8 m apart and 20 m long with 0.3 to 0.6 m between plants. However, adverse weather conditions for pollination and fruit set markedly decreased crown set and seed yields, and few, plant selections were made. Nonetheless, seed were harvested from fruit of 4 lines which looked most productive and had the least amount of green fruit by mid-September. Factoring near-ripe, immature fruit (17 to 28% of total harvest) into total yield estimates, seed yields of 1,100 to 1,500 kg/ha were obtained. Yields of this magnitude would be sufficient for supporting commercialization of hull-less pumpkin seed for a confectionery market.


Pumpkin seed with reduced seedcoat tissues, termed hull-less, can be eaten directly without dehulling, and make a delicious roasted seedsnack. Within Cucurbita pepo there is enormous genetic variation in plant growth habit, fruit type, and seed traits which can be utilized for improving both seed yield and various culinary traits of seed such as taste, size, color, and degree of fiber. Current snackseed pumpkin breeding lines developed at the New Hampshire Agricultural Experiment Station during the past 10 years show excellent potential for use in developing a snackseed industry


Table 1. Yield components for two New Hampshire breeding lines at 5 plant densities in a gradient density spacing trial, 1986.z

Plant density/m2
Yield component 0.8 1.02 1.43 2.39 2.59
Fruit (no./plant) 3.5±0.6 2.7±0.5 2.1±0.2 2.1±0.1 1.5±0.3
Mean fruit wt (kg) 1.2±0.1 1.2±0.1 1.0±0.1 0.9±0.04 0.8±0.1
Seeds (g) /fruit 17.1±1.7 18.0±2.2 21.0±4.5 14.4±0.3 14.6±1.4
Seed wt(g/100) 9.8±1.2 9.7±1.3 11.2±1.2 10.0±0.2 10.8±0.6
Seed Index (SI)y 24.7±2.3 22.7±1.4 28.2±3.9 27.1±1.9 29.4±7.1
Fruit (no./plant) 3.6±0.6 4.2±0.4 3.2±0.6 1.5±0.4 1.7±0.3
Mean fruit wt. (kg) 1.2±0.1 1.2±0.1 1.1±0.1 1.0±0.1 0.9±0.03
Seed (g) fruit 20.4±6.3 19.4±5.7 15.5±5.3 22.9±9.1 15.8±4.4
Seed wt. (g/100) 10.8±1.7 10.4±0.5 9.9±1.0 11.3±0.7 11.3±1.3
Seed index (SI) 25.3±1.0 22.6±4.0 20.4±12.3 29.2±6.5 23.3±8.1
zValues ±SD for 3 replications

ySeed index (SI) = seed dry wt/fruit dry wt x 100

Table 2. Yield components for advanced snackseed selections.

Breeding line
and generation
Mean fruit weight (kg) % Dry matter pericarp g seed/kg fruit Seed index
F2 Generation
614-6 0.91 5.8 56.7 51.7
614-100 1.15 6.2 41.0 42.0
614-213 1.12 7.0 46.4 42.0
614-226 0.95 8.1 72.7 48.0
F3 Generation
(Ave±SD, 5 selections)
0.66±0.16 7.9±3.3 65.3±11.5 49.4±4.8
(Ave±SD, 4 selections)
0.87±0.23 8.9±4.5 63.1±20.9 44.6±13.8
(Ave±SD, 2 selections)
0.61±0.06 10.1±4.5 63.4±5.2 42.8±13.4
(Ave±SD, 5 selections)
0.67±0.13 7.4±1.9 64.4±23.9 49.8±11.9

Fig. 1. Seedcoat phenotypes of air dried seed of C. pepo. Upper row = normal, hulled seed; second and third rows = partially hull-less seed; bottom row = completely hull-less seed.

Fig. 2. Seedcoat phenotypes in hull-less strains of C. pepo, illustrating differences in seed size, shape, and degree of seedcoat development in seed margins.

Fig. 3. Relationship of seed yield to plant density for two F5 breeding lines (1986). Open symbols = NH55-7-20; closed symbols = NH14-40-6. R2 for g seed yield per plant was 0.93 for NH14-40-6 and 0.89 for NH55-7-20. R2 for g seed yield per m2 was 0.82 for NH14-40-6 and 0.87 for NH55-7-20.

Last update September 4, 1997 by aw