Horticultural Traits and Breeding Objectives

by Todd C. Wehner
Department of Horticultural Science
North Carolina State University
Raleigh, NC 27695-7609


Vine length of watermelon varies from dwarf to long. For example, ‘Charleston Gray’ and ‘Jubilee’, large-fruited varieties, have vines up to 30 feet long. Short or medium length vines are well suited to varieties with small or medium sized fruit. For example, ‘Sugar Baby’, ‘New Hampshire Midget’, and ‘Petite Sweet’ are short vined, and ‘Crimson Sweet’ has intermediate vine length.

Dwarf mutants have been discovered in watermelon. Two genes cause dwarfing when they are in homozygous recessive condition: dw-1 and dw-2. ‘Kengarden’ has the genotype dw-1 dw-1. Another gene mutant (Japanese Dwarf, dw-2 dw-2) has increased branching from the crown. Dwarf plants having both sets of genes (dw-1 dw-1 and dw-2 dw-2) have hypocotyls 50% the length of normal vining plants, so can be selected in the seedling stage (Table 3.2).

Sex Expression

Most modern varieties are monoecious, and that appears to be the prefered type of sex expression for commercial seed production of inbred lines and hybrid varieties. Andromonecy (aa) is recessive to monoecy.

Most varieties have a ratio of 7 staminate to 1 perfect or pistillate flower. There are some varieties with a ratio of 4 staminate to 1 pistillate flower. It may be possible to breed for gynoecious sex expression by selecting for increased proportion of pistillate nodes in a segregating population. There is no advantage to andromonoecious sex expression, since the perfect flowers must be pollinated by bees in order to set fruit. Thus, they are no more likely to set without bees or to be self-pollinated, than monoecious varieties.

Male sterility is useful for the production of hybrid seeds without the requirement for expensive hand pollination. The glabrous male sterile (gms) mutant provides male sterility, but the plants are less vigorous, have poor seed set, and are susceptible to cucumber beetles because they lack hairs. A second male sterile mutant, the Chinese male sterile (cms), has been more useful for hybrid production (Fig. 3.3).

Fruit can be set parthenocarpically. Although there are no gene mutants that make plants parthenocarpic, fruit set may be achieved without pollination by applying growth regulators to the plants. Thus, commercial production of seedless watermelon may be possible in areas where bees have been excluded by applying growth regulators at a particular growth stage to diploid pistillate flowers that would otherwise produce seeded fruit.


Yield varies among watermelon accessions and current varieties. Growers want high weight per acre of marketable size fruit, with a low percentage of culls. The yield goal expressed by many growers is at least one load (45,000 lb.) per acre. Most watermelon breeders are selecting for yield in their programs, but it is not clear whether significant progress has been achieved.

In the production of triploid hybrids, up to one third of the field must be planted to a diploid seeded variety. Therefore, higher yield of seedless watermelon per acre could be obtained by using a more efficient pollenizer that would allow more than two thirds of the field to be planted to the triploid variety. Alternatively, parthenocarpic fruit set (genetic or hormone-induced) to stimulate fruit set would permit the entire field to be planted to the triploid variety.


Early maturity is desirable because prices for watermelon usually are best at the beginning of the local season. However, late maturity is associated with varieties that have large fruit size and high yield. Thus, it may be necessary to sacrifice some earliness to obtain high yield or large fruit. Time from pollination to fruit harvest ranges from 26 days for early maturing, small-fruited varieties such as ‘Petite Sweet’ to 45 days for large-fruited varieties such as ‘Super Sweet’.

The selection process for early maturity should involve both days from seeding or transplanting to first fruit set, and days from first fruit set to fruit harvest. Days to fruit harvest should be based on fruit having fully developed sugars as verified by a hand-held refractometer or by taste evaluation.

Fruit Size, Shape, and Rind Pattern

Fruit size. Fruit size is an important consideration in a breeding program since there are different market requirements for particular groups of shippers and consumers. The general categories are: micro (<3 lb.), mini (3-8 lb.), icebox (9-13 lb.), small, sometimes called pee-wee (14-18 lb.), medium (19-24 lb.), large (25-32 lb.), and giant (>32 lb.). Fruit size is inherited in polygenic fashion, with an estimated 25 genes involved. Shippers in the United States work with particular weight categories, such as 19-24 lb. for seeded and 14-18 lb. for seedless.

Old varieties tend to have larger fruit size than current varieties, because one of the things growers were interested in was winning competitions for fruit weight. Competitions are still being held to grow the largest fruit, but commercial production concentrates on high quality. Another reason for larger fruit in the past is that they are more efficient for hand harvest and shipping; large fruit handled individually permit more weight to be moved per unit. Also, there was demand for large fruit to be sold or served by the slice for restaurants and cafeterias. Today, most supermarkets request fruit of seeded watermelons that weigh 19-24 lb.

Small- or medium-fruited types were the result of adapting watermelon to the northern areas of the United States. Varieties developed for the northern United States were bred from early maturing Asian varieties brought from Japan and Russia. A.F.Yeager produced the early varieties ‘White Mountain’ and ‘New Hampshire Midget’ from sources, which have 2 to 4 lb. fruit with a 65-day maturity. The early variety ‘Petite Sweet’ has 5 to10 lb. fruit.

‘Sugar Baby’, a small-fruited variety popular in some parts of the world, was selected in Oklahoma by M. Hardin in 1956. Even though mini and icebox varieties with 4 to 11 lb. fruit have been developed to fit easily in a small refrigerator, most of the demand in the marketplace for small fruit has been met using sections cut from a large fruit. A large watermelon fruit cut into quarters has the same weight as an icebox melon, but it has a different shape, and consumers can see what they are buying.

Fruit shape. Fruit shape is also an important part of market type. The general categories are round, oval, blocky, or elongate. There is one gene involved in round vs. elongate, with the F1 being intermediate (blocky). In some cases, fruit shape is related to cotyledon shape at the seedling stage. Plants with elongate fruit have elongate cotyledons, and plants with round fruit have round cotyledons. However, others have concluded that selection for fruit shape at the seedling stage is ineffective. Among old varieties with elongate-shaped fruit, there was greater susceptibility to production of gourd-neck or bottle-neck fruit, which are culls. Old varieties with round fruit were more susceptible to hollowheart. Thus, some of the first hybrids were made between elongate and round inbreds to reduce the incidence of these defects. Recently, genetic resistance to those defects has been incorporated into new varieties, and has made fruit shape less important to consider.

Rind pattern. The third area of importance in market type is rind pattern, which can be gray, striped, or solid. Stripes on the rind can be narrow, medium, or wide where the stripes are the dark green areas. The striped pattern can be on light green or medium green background. Solid rind color can be light or dark green. Solid dark green is dominant to gray rind pattern. Solid dark green is dominant to striped, and striped is dominant to solid light green rind pattern. However, the striped pattern can be seen on a solid dark green fruit after the color has been bleached by the sun.

In addition to the common rind patterns, there is furrowed vs. smooth rind, controlled by the recessive gene, f (Table 3.2). Most current varieties have smooth rind. Another interesting mutant is golden rind, which is controlled by the recessive gene, go. Its usefulness as an indicator of fruit ripeness is limited because the change in fruit color at fruit maturity is accompanied by chlorosis of the leaves. Furthermore, it does not appear to be a reliable indicator of ripeness, and may be disadvantageous for yield, especially if the grower is using a multiple harvest system.

Summary. We propose that watermelon varieties be categorized by fruit size, shape, and rind pattern as follows: Fruit size – micro, mini, icebox, small, medium, or giant. Fruit shape – round, oval, blocky, or elongate. Rind pattern – gray, solid light, solid medium, solid dark, or narrow, medium, or wide striped on a light green or medium green background (Table 3.1, Fig. 3.2). Using these categories, we would classify ‘Allsweet’ as large, elongate, with wide stripes on a light green background. ‘Crimson Sweet’ would be classified as medium size, round, with medium stripes on a light green background. ‘Charleston Gray’ would be large, elongate, with gray rind.

External Fruit Quality

Rind durability is important on varieties that are to be shipped to market. On large-fruited varieties, the rind should be thick and tough; whereas on small-fruited varieties, the rind should be thin and tough. Rind thickness should be a small percentage of flesh diameter to keep it in a balanced proportion for best appearance. Large-fruited varieties look better with a thicker rind, and need the extra protection for postharvest handling and shipping. The rind can be tough and hard as in ‘Peacock’ or tough and soft as in ‘Calhoun Gray’. Brittle rind as in ‘New Hampshire Midget’ is not useful for varieties that are to be shipped to market.

Rind flexibility can be tested by cutting a 1/16 to 1/8 inch x 3 inch piece of rind from a fruit and bending the rind into an arc. If the rind bends into a tight arc, it is flexible and tough. If it breaks early in the attempt, it is tender and explosive.

Rind toughness can be measured by driving a spring-loaded punch into the rind. A tough rind would require more force to punch through, whereas a tender or brittle rind requires less force. Watermelon breeders often use faster methods to test for rind toughness, however. One method is to drop the fruit onto the ground from a particular height (for example, knee height) to see whether it breaks open or not. The drop height would depend on the soil type of the field being used. Another method is the “thumb” test, where the breeder presses on the rind at a particular location on each fruit. If the rind breaks when only a small amount of force is applied, then it has a tender rind; otherwise it should be resistant to shipping damage.

Internal Fruit Quality

Flesh color is one of the primary traits consumers look for in a watermelon fruit. Color can be scarlet red, coral red, orange, canary yellow, salmon yellow (golden), or white. Coral red (YY) is dominant to orange (y-o y-o), which is dominant to salmon yellow (yy). Canary yellow (CC) is dominant to non-canary yellow (cc), and epistatic to (overcomes) the y locus for red-orange-salmon yellow. Coral red is recessive to the white flesh color, which is found in citron (Table 3.2). Scarlet red color from ‘Peacock’ has been used to develop many new varieties because of its attractive color. The inheritance of scarlet red flesh color is due to the single dominant gene, Scr. Varieties with scarlet red flesh include ‘Dixielee’, ‘AU-Sweet Scarlet’, ‘Red-N-Sweet’, and ‘Sangria’.

Sugar content is a major component of flavor. Breeders select for high sugar content as indicated by taste and refractometer readings. Refractometer readings are easily made in the field using a handheld unit, and provide data on percentage of soluble solids (°Brix). These translate to sugar content, which should be a minimum of 10%. Newer varieties have Brix as high as 14%. Some varieties have higher levels of fructose, which tastes sweeter than sucrose. The difference in taste is not measured by a refractometer.

Selection should be made for good watermelon flavor, independent of sweetness (sugar content). Flavor should include freedom from bitterness, which is controlled by a single dominant gene, and may be introduced in crosses with C. colocynthis accessions. Another component is caramel flavor as in ‘Sugar Baby’ fruit, which some taste testers find unpleasant. The flavor is sometimes associated with dark red flesh color. Its inheritance is not known, but caramel flavor does respond to selection. Thus, breeders should select lines with mild (not bitter) taste, high sugar content (°Brix), freedom from caramel flavor, and excellent “watermelon” taste. It is important that varieties with excellent taste be included as checks in all selection blocks to provide a comparison for the plant breeder. Examples of varieties with good quality that are commonly used include ‘Allsweet’, ‘Crimson Sweet’, and ‘Sweet Princess’.

Flesh texture is an important part of internal quality. Watermelon fruit can have flesh that is soft or firm, and fibrous or non-fibrous. The objectives for plant breeders should be to develop varieties with flesh that is firm and non-fibrous. The genes controlling those traits are not known, but they are heritable.

Seeds and Seedlessness

Seed color can be white, tan, brown, black, red, green, or mottled. White seed color usually is not preferred since it suggests that the fruit is immature, and can make it difficult to distinguish mature from immature seeds. On the other hand, white seeds may be a useful objective for the development of near-seedless varieties that have few, small, and inconspicuous seeds. Black seed color is attractive with red or canary yellow flesh color. Black, brown, or tan seeds look good with orange flesh color.

Seed size should be large for confectionery (edible seeded) type, and small or medium sized for the standard (edible flesh) type (Fig. 3.4). A new seed size mutant discovered recently is called tomato seed. The seed size is about half that of the small watermelon seed size, and is controlled by a single recessive gene, tss.

Seed number should be high for the confectionery type, but should be low or medium for the edible flesh type. Seed number should be lower in small-fruited varieties so that the seeds will not appear to include more than the usual percentage of the fruit volume. Seed number should be high enough to make seed production economical, but low enough to make the flesh easy to eat.

In theory, seedless triploid hybrids should provide higher yield than diploid hybrids because no energy is used in seed production. However, in practice this may not be the case. Fruit production in triploids is limited by the availability of viable pollen to induce fruit set.

During the development of tetraploid inbreds, seed yield is often low in early generations, so selection for fertility is essential. Some tetraploids are more fertile than others, and should be selected to keep seed costs low for triploid hybrid production, since the hybrid seeds are produced on the tetraploid parent line.

Triploid hybrids are generally seedless, but occasionally hard seed coats form in the fruit The presence of objectionable seed coats is affected by environment, but can also be selected against in the development of the inbred parents of the hybrid. Inbred parents that do not develop objectionable seed coats in the fruit in different production environments should be selected for triploid hybrids.