Seed Development and Seed Fill in Hull-less Seeded Cultigens of Pumpkin (Cucurbita pepo L.)

Kelly J. Vining and J. Brent Loy

Department of Plant Biology, University of New Hampshire, Durham, NH 03824

Additional index words. edible seeds, seedcoat, embryo, embryogenesis

Abstract. In Summer 1997 a field study was conducted on seed development in two hull-less seeded pumpkin cultigens and their F1. Data were obtained on dry biomass accumulation in developing seeds and partitioning of biomass among seed organs (embryo, endosperm and seedcoat). Developmental changes in fruit size and skin color patterns were observed and related to stages of seed maturity. Microscopic examination showed that most embryos were not fully expanded until 30 to 35 days postanthesis (PA). Endosperm was detectable between 10 and 35 days PA. Seedcoat biomass peaked at 20 days PA, comprising 95% to 96% of total seed dry weight at that time. The subsequent decline in seedcoat biomass could account for 20% to 24% of the increases in embryo biomass (seed fill). Losses in fruit biomass (fruit ¥ percent dry matter) between 30 to 60 days PA accounted for another 41% to 55% of the increases in seed fill among the three genotypes. In fruit harvested 10 to 5 days prematurely and stored for 10 days in a greenhouse, total seed dry biomass increased significantly in all cultigens.


Pumpkin seeds are an excellent high- energy food source, containing 32% to 40%protein and 40% to 50% oil by weight. At trade shows and demonstrations, we have found their rich nutty flavor to be extremely appealing to consumers. In the ever-widening search for alternative foods for developing nations with burgeoning populations, such a concentrated source of protein and oil may prove to be useful.

Cucurbita pepo varieties are not widely grown for their seeds. In North America decorticated pumpkin seeds can sometimes be found alongside sunflower seeds and peanuts, especially in natural food stores. These so-called pepitas are usually imported from Mexico or China. In Eastern Europe hull-less or naked-seed varieties of C. pepo are grown for pharmaceutical use and for extraction of vegetable oil; whereas, most of the pumpkin seeds consumed as a snack food have the normal hulled condition or intact seedcoat.

A breeding program to develop high seed-yielding cultivars of hull-less seeded pumpkins has been in progress at the University of New Hampshire since 1981. Several breeding lines have been developed which are being used to produce high yielding F1 hybrids that are currently being

tested. More recent efforts (Carle and Loy, 1994, 1995) have focused on increasing seed size, an important trait for consumer appeal. Some of the new hybrids being tested exhibit large seed size (200 to 260 mg each) in addition to high yield potential.

A major obstacle to the commercial production of hull-less pumpkin seeds is the paucity of basic information regarding seed development and maturation in these cultigens. Our earlier efforts focused on seedcoat composition and development (Stuart and Loy, 1983; 1988), but we lacked information on stages of embryo development and accumulation of storage reserves (seed fill) in embryos and on the contribution of seedcoats and fruit mesocarp to seed fill. Further, it was not known whether larger-seeded cultigens required longer maturation periods for seed fill than smaller-seeded cultigens, nor had the relationship between perceived stages fruit maturity and stages of seed maturity been established.

The objectives of the current study were to determine the time course of embryogenesis, fruit maturation and seed fill in hull-less seeded cultigens, and compare these parameters in three genotypes differing in seed size. As part of this study,

Cucurbitaceae '98

we sought to determine the extent of seed fill occurring in fruit harvested prematurely from plants and stored.

Materials and methods

A field study was conducted in the summer and early Fall 1997 on Paxton loam at the Woodman Research Farm, Durham, N.H. Three hull-less seeded genotypes were used: PI 285611S3 (NH285), with fruit size 3 to 5 kg and seed size 210 to 250 mg; NH29-13, a highly inbred line, with fruit 0.5 to 1.0 kg and seed 140 to 170 mg; NH1003, the F1 hybrid of the above lines, having 2 to 4 kg fruit and 180 to 220 mg seed.

Seeds were sown on 5 June using a randomized complete-block design with 6 blocks, three genotypes, and 36 data plants per genotype subplot. Rows were 1.8 m apart and within row spacing was 0.6 m for NH29-13, 0.9 m for NH1003 and 1.2 m for NH285. Hand pollinations were made between 22 July and 7 Aug, with 90% of pollinations made between 29 July and 1 Aug. Fruit were harvested at 3- and 5-day intervals for thirteen samplings dates between 10 and 65 days postanthesis (PA). Fruit were stored at 4 °C for 2 to 10 days before analysis, at which time fruit were weighed and seed samples removed. For fruit harvested from 10 to 35 days PA, 30-seed samples were used; for 40- to 60-day PA fruit, 20-seed samples were used. Ten-seed samples were used for 65-day PA fruit. Fresh and dry weights were obtained for whole seeds, seedcoats, embryos and endosperm. For dry weights, seed organs were dried in an oven at 55 to 60 °C for a minimum of 24

h. Percent dry matter of the mesocarp was determined from core samples taken from the midportion of each fruit with a 1.5-cm corkborer, midway between the stem and blossom ends of the fruit.

To determine the extent of seed fill during storage of immature fruit, fruit were harvested and stored in a greenhouse (25 to 30 °C day and 18 to 21 °C night) for 10 days before seed removal, weighing and drying. Fruit were harvested and initial seed weights obtained at 5-day intervals between 35 to 50 days PA. Seed fill (embryo weight) in stored fruit was compared to that of fruit left intact on plants. Because we had anticipated differences in fruit maturity among genotypes, fruit for storage were not obtained for all genotypes at each sampling period.


Fruit development. Maximum fruit size was attained in all three genotypes by 20 days PA and near maximum size by 15 days PA in NH285 and NH29-13 (Fig. 1). Percent dry matter (%DM) peaked at 30 days PA and then declined in all three genotypes (Fig. 2). Fruit of NH29-13 had the highest %DM; %DM was lower in fruit of NH285 and NH1003 throughout development. Because of the decrease in percent dry matter between 30 to 60 days PA and relatively constant fruit size during that period, total biomass per fruit declined by 50, 80, and 95 g, respectively, in NH29-13, NH1003 and NH285.

Fruit of NH29-13 began to show orange skin color as early as 25 days PA and reached full

Table 1. Dry biomass accumulation in embryos of seeds from fruit harvested prematurely [35, 40, 45, and 50 days postanthesis (PA)] and stored in a greenhouse for 10 days compared to seeds from fruit allowed to develop intact on plants for 10 days.

Harvest intervals (days PA)

Genotypes Treatment 35 to 45 40 to 50 45 to 55 50 to 60

10-day dry biomass accumulation, mg/embryoz

NH29-13 Intact fruit 44.1 41.9

Stored fruit 38.5y 33.2

NH1003 Intact fruit 71.5 40.8 44.6

Stored fruit 59.2 34.0y 18.6x

NH285 Intact fruit 53.2 52.8

Stored fruit 55.6 54.6

zValues are means of six replications.

yStored fruit harvested 1 day later (36 vs 35 days PA) than control fruit.

xStored fruit harvested 2 days later (47 vs. 45 days PA) than control fruit.

Cucurbitaceae '98

orange rind color at 35 days PA. Fruit of NH1003 began to change color between 31 to 35 days PA, but were not 100% orange-yellow until nearly 55 days PA. Fruit of NH285 were presumed to be mature when showing yellow-orange and green striping. At 60 to 65 days PA only unshaded sides of the fruit exhibited the orange and green striping, with the rest of the fruit showing light and dark green striping.

Seed development. Total seed biomass increased progressively in all genotypes between 10 and 60 days PA (Fig. 3), with the exception of a dip at 50 days PA in NH1003 and NH285, and at 55 days in NH29-13. At 60 days PA NH285 had the largest seeds, followed by the F1 hybrid (NH1003), and then NH29-13. Larger seed biomass in NH285 as compared to NH29-13 and NH1003 was evident early in development. The seedcoat comprised most of the dry matter of seeds between 10 (99.8%) and 25 days (89% to 94%) PA (Fig. 4 a­c). Seedcoat dry matter increased until 20 to 25 days PA, and then steadily declined until 50 to 55 days PA, leveling out at 15 to 20 mg per seed. Peak seedcoat dry weight was significantly higher in NH285 than either the hybrid (NH1003) or NH29-13. Even though seeds of NH1003 are larger than those of NH29-13, seedcoat biomass was often greater in NH29-13 between 25 to 60 days PA, reflecting its partial hull-less nature at seed maturity.

Endosperm was detectable in some seeds of all genotypes between 10 and 35 days PA, but size and consistency was highly variable within seeds of all genotypes. At its maximum size, at 20 days PA, endosperm represented only 2% to 6% of seed biomass, but 10% to 17% of total seed fresh weight (data not shown). By 35 days PA traces of endosperm were present in only a small percentage of seeds.

Embryos were not large enough to efficiently dissect until 20 days PA. There was considerable variation in embryo size between 20 to 31 days PA, both within and among fruit. Embryo biomass comprised only 1% to 6% of the total dry matter at 25 days PA, but increased nearly linearly in all genotypes from 25 days until 55 to 60 days PA. Rates of seed fill as represented by changes in embryo biomass were highest in NH285 and NH1003.

Cucurbitaceae '98

Seed fill in stored fruit. Seed fill, calculated as embryo biomass, increased in 10-day stored fruit in all cultigens and at all sampling times (Table 1). There were, however, genotypic differences in the extent of seed fill in stored compared to intact fruit. In NH285, seed fill in stored fruit was comparable to that of fruit left on the plant (intact fruit) at both sampling periods (40 to 50 and 50 to 60 days PA). In the inbred line NH29-13, seed fill in stored fruit was 12.7% (35 to 45 days PA) to 20.5% (40 to 50 days PA) lower that in intact fruit. In the hybrid NH1003, seed fill in stored fruit varied from 17.7% to 58.3% lower than that of intact fruit. There was no consistent relationship between the extent of seed fill in stored compared to intact fruit and the time of harvest.


Because the developing seed becomes the dominant sink for assimilates in most crop species after anthesis (Thorne, 1985), knowledge of assimilate partitioning within fruit and seed tissues is vital for assessing yield potential. Such information is well-established in several major dicotyledonous agronomic crops such as pea (Flinn and Pate, 1968), safflower (Leininger and Urie, 1964), and soybean (Fehr et al., 1971; Smith and Nelson, 1985), but is almost totally lacking in cucurbits.

The seedcoat is the primary sink early in seed development in C. pepo and source of assimilates for later rapid embryo enlargement. In a previous study of seedcoat development (Stuart and Loy, 1988), seedcoat biomass peaked at 20 days PA, and then declined until seed maturity. The decline coincided with decreases in several structural (pectins, hemicelluloses) and nonstructural (starches, fats, sugars, amino acids) components of the seedcoat. In our study, seedcoat biomass also peaked at 20 days PA, and the decline in seedcoat biomass coincided precisely with the rapid growth phase of embryos (Fig. 2 a­c). Seedcoat reserves, interpreted as seedcoat biomass loss between 20 and 60 days PA, amounted to between 32 to 48 mg/seed (depending on genotype), comprising between 20% to 24% of the total increase in embryo biomass between 20 and 60 days PA. This is a significant contribution to seed fill, and is equal to the contribution of both the seedcoat and pod reported in

Cucurbitaceae '98

field pea (Flinn and Pate, 1968).

Endosperm may be an important source of nutrition for early embryo development, but comprises a relatively small proportion of seed biomass in C. pepo. It may also be a primary tissue through which symplastic transport of assimilates from the seedcoat to the embryo occurs via the chalazal haustorium (Chopra, 1958), until the embryo enlarges within the seedcoat. However, the endosperm represents only 2% to 3% of seed biomass at 30 days PA when most embryos are fully enlarged, and thus contributes little to subsequent seed fill.

In the present study, we had anticipated that NH285, because of its huge seed and late development of fruit color, would exhibit a much longer period of seed fill as compared to the other two genotypes. This proved not to be the case in the 1997 growing season, with seeds of all three genotypes reaching peak biomass at 60 days PA (Fig. 1). The F1 hybrid, NH1003, appeared to be slightly earlier, with seeds reaching near peak biomass at 55 days PA. We had limited fruit for sampling beyond 60 days PA and the data on fruit and seed size was therefore highly variable. Therefore, we cannot unequivocally state that maximum seed fill was reached at 60 days PA.

Maximum attainment of fruit and seedcoat size occurred simultaneously between 15 and 20 days PA, suggesting that regulation of expansion of the two organs is coordinated. Fruit expansion must be coincident with seedcoat expansion to accommodate growth of seedcoat tissues within the fruit locules. Seed size is largely determined by seedcoat size, and expansion of seedcoats are in turn delimited by size of fruit. For example, we have found in our large-seeded breeding populations that fruit size must be 1.4 kg or greater to obtain average seed sizes >200 mg with good seed set. In small-fruited genotypes having the genetic potential for large seeds, we commonly see one of two conditions: 1) extensive seed abortion occurs and the remaining seeds are large or 2) few seeds are aborted, but seeds are only moderately large and have a distorted, compressed appearance, suggesting a physical restriction of seed development.

Because maximum fruit fresh weight occurred at or before 20 days PA and %DM peaked at 30 days PA, then the maximum amount of biomass and presumably assimilates available for export from fruit mesocarp tissue to developing seeds occurred at 30 days PA. The gradual decline in total fruit biomass (%DM ¥ fresh weight) between 30 and 60 days PA coincided well with the accelerated increases in seed fill in all three genotypes. The estimated loss of fruit mesocarp biomass could account for between 41% and 55% of the increases in embryo (primarily cotyledonary) biomass among the three genotypes during the same period of development.

We were unable to control powdery mildew infection effectively during the last 3 weeks of growth, thus a deficiency of photosynthates for allocation to both fruit and seed development was likely, accounting for the decline in %DM of fruit at the expense of increases in seed biomass. This interpretation agrees with an earlier study of Culpepper and Moon (1945) who found that total solids content of two cultivars of C. pepo pumpkin peaked at 30 days PA, but remained relatively constant until 60 days PA. Nonetheless, our results demonstrate that C. pepo fruit function as an important storage organ that can shunt assimilates to developing seed, especially under conditions in which assimilate export from leaves are limiting, as in unhealthy plants, or halted, as in fruit harvested prematurely. Together, the fruit mesocarp and seedcoat contributed as much as 55% to 70% of total required assimilates for seed fill between 30 and 60 days PA.

An important finding was the lack of a consistent association of perceived fruit ripeness with seed maturation. This was especially evident in NH29-13 which began to show orange fruit color as early as 25 days PA and reached full orange fruit rind color at 35 days PA, 25 days ahead of peak seed biomass. Before this study, we had routinely harvested fruit of this inbred line at 40 to 45 days PA, assuming that seed maturation was complete. In NH285, it was difficult to ascertain seed maturity from changes in fruit color because the changes were quite subtle and occurred late in development. In the F1 hybrid, NH1003, fruit rind color

Cucurbitaceae '98

began to change at 31 days PA, at which time seed fill was only 31% complete. However, fruit were generally 100 % yellow-orange at 55 days PA, corresponding well with near peak seed fill. As much as 30% of the seed fill occurred between 50 to 60 days PA in two of the three genotypes studied. For maximizing seed yields, seed producers should be aware of the long seed maturation period in some genotypes of C. pepo and adjust harvest periods and fruit storage accordingly.

Literature cited

Carle, R.B. and J.B. Loy. 1994. Heritability of seed size in hull-less seeded strains of Cucurbita pepo L. Cucurbit Genet. Coop. Rpt. 17:125­127.

Carle, R.B. and J.B. Loy. 1995. Heritability of seed size and its association with fruit size in a hull-less seeded population of Cucurbita pepo L., p. 221­223. In: G.E. Lester and J.R. Dunlap (eds.). Cucurbitaceae '94: Evaluation and enhancement of cucurbit germplasm. Gateway Printing and Office Supply, Inc., Edinburg, Texas.

Chopra, R.N. and Agrawal, S.. 1958. Some further observations on the endosperm haustoria in the

Cucurbitaceae. Phytomorphology 8:194­201.

Culpepper, C.W. and H.H. Moon. 1945. Differences in the composition of the fruits of Cucurbita varieties at different ages in relation to culinary use. J. Agr. Res. 71(3):111­136.

Fehr, W.R., C.E. Caviness, D.T. Burmood, and J.S. Pennington. 1971. Stage of development descriptions for soybeans, Glycine max L. Merrill. Crop Sci. 11:929­931.

Flinn, A.M. and J.S. Pate. 1968. Biochemical and physiological changes during maturation of fruit of the field pea (Pisum arvense L.). Ann. Bot. 32:479­495.

Leininger , L.N. and A. Lee Urie. 1964. Development of safflower seed from flowering to maturity. Crop. Sci. 4:83­87.

Smith, J.R. and R.L. Nelson. 1986. Relationship between seed-filling period and yield among soybean breeding lines. Crop Sci. 26:469­472.

Stuart, S.G. and J.B. Loy. 1983. Comparison of testa development in normal and hull-less seeded strains of Cucurbita pepo. L. Bot. Gaz. 144(4):491­500.

Stuart, S.G. and J.B. Loy. 1988. Changes in testa composition during seed development in Cucurbita pepo L. Plant Physiol. (Life Sci. Adv.) 7:191­195.

Thorne, J.H. 1985. Phloem unloading of C and N assimilates in developing seeds. Annu. Rev. Plant Physiol. 36:317­343.

Cucurbitaceae '98