In some crosses, some pollen germinates and some does not

The inhibition of pollen germination is therefore the physiological basis of incompatibility in sweet potato. A series of multiple alleles controls this incompatibility and these alleles act sporophytically to determine the pollen phenotype. Since the sweet potato is a hexaploid, it is assumed that the incompatibility loci have been duplicated, or even triplicated, during its evolution (Martin, 1982).

Sterility in sweet potato results from a series of processes that are easy to detect (Table 10.1). Several days after anthesis, the ovaries of pollinated flowers begin to develop. During the process, all ovules grow together up to the third day. One or two fertilized ovules increase rapidly, whereas the growth of the others is arrested. As the good seeds continue to grow, the aborted ovules are compressed on one side of the capsule and die. When the fruit dries, the mature seeds shrink to approximately 3 mm in diameter, about half their size when green and mature. The aborted ovules dry to appear as scales of approximately 0.5–1.5 mm long. Large seeds germinate more rapidly than smaller seeds. Small seeds, however, can represent up to 50% of the total number of seeds obtained, depending on the genotypes involved.

It is likely that sterility is caused by high polyploidy. It can be caused by accidents occurring in meiosis, resulting in defects and recombination that lead to an unbalanced gene distribution, producing embryos with a defective combination of genes that cannot function properly. It is possible that open-pollination practised over several generations could favour fertile plants that tend to flower freely. Martin and Jones (1971, 1986) have shown that flowering increased up to 300% in only six generations of open-pollinated crosses. It is unlikely that selection pressures increase sterility or

self-incompatibility.

Table 10.1. Detectable failures in the reproductive processes of sweet potato.

Failures Method of detection Estimated damage

Male gametes are abnormal Pollen production failure Rare Female gametes are abnormal Microscope examination Very rare Some pollen does not germinate Pollen counts on stigma Very common Stigma does not stimulate Pollen fails to germinate Rare but known

germination

Pollen tubes do not pass from Microscope observation Common, 30–95%

stigma to style of studied cases

Pollen tubes do not pass Microscope observation Common, 10–20%

through entire style

Pollen tubes do not result Microscope observation, Common, 10–60%

in fertilization seed and scale counts

Ovules develop as scales Observations and counts Very common, 75%

Seeds do not develop normal Observations and counts Very common,

endosperms 25–50%

Seeds do not germinate Observations and counts Common, 10–25%

Seedlings are weak and die Observations and counts Uncommon, 2–3%

Seedlings do not grow into Observations and counts Uncommon, 1%

mature plants

Mature plants do not flower Observations Rare Source: Martin (1982).

CROSSING TECHNIQUES AND TRUE SEED PRODUCTION

Most sweet potato genotypes flower naturally within the tropics and it is not necessary to use grafting, girdling or day-length control, which breeders in temperate countries have to use to induce flowering during long days (Edmond and Ammerman, 1971). Parent clones, planted in crossing blocks isolated from other flowering sweet potatoes, are open-pollinated by naturally occurring insects. Sweet potato flowers best during short days and, in tropical countries, the cool season is the best period for producing seeds. In the southern hemisphere, for example, the crossing block is planted during the first 2 weeks of April. Flowering begins 6 weeks later and continues for 3–4 months. Seeds are harvested from June to September. In Taiwan, in the northern hemisphere, the best season for pollination is from the beginning of November to the middle of December, when the average daily temperature is between 20°C and 25°C, with a maximum seed set occurring when the mean daily temperature is about 23.9°C (Wang, 1982).

The vine cuttings of the parent clones are planted at 1 ⫻ 1 m, with two cuttings per planting position. Usually, ten plants of each genotype are enough, although more may be needed for genotypes with poor flowering. Usually, the climbing vines of four plants are mixed together on a pyramid-like system with four 2 m-high stakes tied together. Eventually, wires connect the pyramids to allow trellising of the vines, which promotes flowering. Staking facilitates hand pollination and insect pollination, but such plants can be damaged easily by strong winds. It is not recommended that crossing blocks be fertilized with nitrogen as this promotes lush and leafy vines without flowering. Various insects and diseases can reduce flowering, especially scab (E. batatas) in the wet tropics.

When hand pollinating, it is necessary to ensure that the flowers are well protected from pollination by insects. The buds and flowers due to open the following morning are prevented from opening by clipping the tip of the corolla. The best time to clip flowers is during the afternoon or the evening of the day before hand pollination. The flower from the male parent is carried to the female parent and the clip is removed gently without destroying the corolla. The petals are then spread and the anthers of the male parent are rubbed gently over the stigma of the female parent. In order to prevent pollen contamination, the corolla of the pollinated flower is tied together so that insects cannot reach the stigma. Another technique uses 2 cm-long pieces of drinking straw, which are pushed on the unopened corollas 1 day before anthesis. After hand pollination, the corolla is rolled again and pushed into the straw again.

For genetic studies, the female parent flowers are emasculated by hand to eliminate all possibility of pollen pollution. Success rates depend on the weather and the health of the plants used, but approximately 50% of the pollinated flowers produce two seeds. It is possible, however, to improve this rate with better nursery techniques. For example, in Japan, 12 bi-parental crosses were made

and grafting on I. nil was conducted in order to promote flowering. Hand pollination was done inside a glasshouse, with great care being taken to isolate the plants from pollinating insects. From 14,256 hand pollinations, 8910 seeds were finally harvested, representing a success rate of 62.5%, which is considered as very high for such an operation (Mok et al., 1998).

Seeds mature between 4 and 6 weeks after pollination. Each capsule is harvested by hand when it is fully brown and the pedicel has dried. It is necessary to collect them in the crossing block every morning as mature capsules fall off easily or dehisce and release their seeds while opening. Seeds are then extracted and those that are lightweight, deformed or with insect or fungus damage are discarded. An easy way of selecting the healthy seeds is to put them in a container with water and to eliminate those that float. Once properly dried, they can be stored and remain viable for up to 20 years if the storage conditions are well controlled (18°C, 50% relative humidity) and for at least 5 years in a simple desiccator with silica gel lodged in a refrigerator.

Since sweet potato seeds have a very hard coat, they germinate slowly and irregularly. The most practical way of scarifying substantial volumes of seeds is to soak them in concentrated sulphuric acid (98% H₂SO₄) in a glass beaker for 40 min. The seeds are then rinsed under running water for 5–10 min. This technique gives about 95% germination success (Wilson et al., 1989). It is also possible simply to soak the seeds overnight in water and this improves germination, although it will be irregular, occurring over several weeks.

Immediately after scarification or soaking, the seeds are placed individually in Jiffy® pots or equivalent and germination occurs readily. Seedlings are planted at 0.5 ⫻ 0.5 m and are ready to harvest at 10 weeks. Cuttings are then replanted for the first ‘hill trial’.

SELECTION METHODS AND PROGRAMMES

Recurrent selection appears to be suitable for sweet potato breeding as it permits minor and recessive genes to be expressed and selected with a progressive increase in population. With this type of mass selection, the capture of additive effects is straightforward, consisting of the selection of a number of genotypes for one or more desirable traits and their hybridization in a polycross block by honeybees.

Numerous seedlings are screened for desirable traits and the best are used, with or without the best parents, in a new polycross block for a second cycle of natural hybridization. This technique results in the rapid accumulation of suitable genes.

A study conducted with DNA markers has confirmed the usefulness of a polycross breeding strategy, in spite of frequent cross-incompatibility. Moreover, the high level of genetic variation in polycross breeding lines can certainly assist in selecting elite material (Hwang et al., 2002) (Fig. 10.1).

However, this simple selection method has to be complemented with efficient screening techniques and for many traits; the identification and the

measure-Fig. 10.1. Morphological variation found in polycross breeding lines (in Vanuatu).

ment of a particular trait is often the weakest operation (Martin, 1988). In practice, breeding lines are evaluated in a series of trials conducted in research stations and in farmers’ fields. Undesirable genotypes are discarded as early as possible and the selection process concentrates on eliminating the poorest genotypes, rather than on selecting the best. It takes approximately 2 years from the crossing block until the first harvest of the IT-1 (Table 10.2). Consequently, if a crossing block is planted once a year, there are two recurrent-selection populations to be managed at the same time.

Superior genotypes usually have an efficient photosynthetic surface and produce high yields of high quality in a relatively short time. The factors determining the photosynthetic surface are the length of the stem and the number of leaves per unit length of stem. Simple visual tools are developed to screen thousands of genotypes efficiently for such traits. Some bushy types, however, can develop with short stems an ample photosynthetic surface over a small area. Selection has to be cautious and to proceed step by step, combining important characters. The short shelf life of the sweet potato is being improved and a simple technique for assessing rates of deterioration is the measurement of weight loss during the first week of storage, a good indicator of the subsequent rates of deterioration (Rees et al., 1998).

In Japan, Shochu alcohol production is one of the largest markets for processed sweet potatoes. ‘Joy White’, a new cultivar released by the Kyushu National Agricultural Experiment Station in 1994, has a starch content of approximately 25% DM and its roots have low sugar content (approximately 2.7% DM). Shochu alcohol obtained from ‘Joy White’ is fruitier and has a lighter taste than the usual sweet potato alcohol.

The colorant industry is looking for natural red colour because consumers prefer natural food ingredients to artificial ones. New cultivars with high anthocyanin content and high yields are being developed. ‘Ayamurasaki’ for example, has elongated, fusiform and uniform good shaped roots with dark-purple skin and a deep dark-purple flesh. Anthocyanins extracted from its roots are used for confectionery and in various foods as a colorant. A dry powder and a paste made from this variety are also used for breads, noodles and snack foods (Yamakawa, 1998; Yoshinaga, 1998). It is now possible to predict anthocyanin content and composition by appreciation of the paste colour without going through expensive HPLC analyses (Yoshinaga et al., 1999). Anthocyanins, responsible for the bright purplish colour, are synthesized throughout the stage of storage root development, but not in a steady manner. It is possible to screen numerous genotypes but only after the 9th week of growth, when the percentage of peonidin, which is an index for anthocyanin composition in purple-fleshed sweet potato, becomes constant and when the storage roots reach a diameter over 20 mm (Yoshiniga et al., 1999).

Major compounds can be quantified routinely using near infrared reflectance spectroscopy (NIRS) and this technique seems promising for breeders (Lu and Sh, 1990). On the other hand, for ␤-carotene,

spectro-Table 10.2. Characters evaluated in each trial of a sweet potato breeding programme.

No. of genotypes in trial Planting pattern Characters evaluated

2000 seedlings in seedling 50 ⳯ 50 cm, 1 seedling Leaf scab score at harvest, vine length nursery (SN) per genotype and thickness, storage root skin colour

and flesh colour.

100 genotypes in hill trial 1 ⳯ 1 m, 3 plants per Leaf scab score, early maturity, little-leaf (HT) genotype (2 cuttings score, virus score, vine length and

per mound) thickness, storage root skin and flesh colour, yield (low, medium, high) and specific gravity measured in the field.

100 genotypes in preliminary 1 ⳯ 1 m, 6 plants per Leaf scab score, early maturity, little-leaf trial (PT) genotype (2 cuttings per score, virus score, vine length and

mound) thickness, storage root skin and flesh colour, root shape, skin smoothness, skin cracking, number of storage roots per plant and individual size, yield (low, medium, high) and specific gravity measured in the field.

25 genotypes in intermediate 1 ⳯ 1 m, 10 plants per Leaf scab score, early maturity, little-leaf trials IT-1 and 13 genotypes genotype (2 cuttings score, virus score, vine length and in IT-2 per mound) thickness, storage root skin and flesh

colour, root shape, skin smoothness, skin cracking, number of storage roots per plant and individual size, marketable weight, edible weight per plant and specific gravity measured in the field.

7 genotypes in advanced 1 ⳯ 1 m, 16 plants per Same as IT plus tuber dry weight and trials AD-1 and genotype, 4 datum eating quality.

5 genotypes in AD-2 plants (2 cuttings per mound)

2 genotypes in on-farm trials Planting patterns Average leaf scab score over the season, determined by farmers. virus score, marketable weight and Number of replications tuber numbers and eating quality, as and number of plants per judged by farmers and farmers’ choice replication determined of which clone will be replanted.

by availability of cuttings.

Trial is best located in the middle of the farmer’s field.

Source: Wilson et al. (1989).

photometry over estimates the HPLC values for ␤-carotene content because of the presence of minor carotenoids. It is thought, however, that screening large numbers can be done using a cost efficient spectrophotometer and that the expensive HPLC is necessary only for the accurate quantification of ␤-carotene (Kimura et al., 2007).

The flavour of sweet potato is of utmost importance to consumers but it is difficult to measure accurately, and this is an obstacle to reliable selection. If flavour could be measured analytically, then the number of genotypes that could be screened accurately would be increased. A study conducted in Georgia, USA, indicates that it is possible to distinguish differences in aroma between genotypes using gas chromatography analysis. Compounds such as sugars and organic acids can also be quantified. The advantages are an accurate parent line and progeny selection (Kays et al., 1998). It is now important to evaluate the practicalities and economics of establishing such a system on a routine basis for breeders.

In Guadeloupe, French West Indies, consumers prefer the non-sweet varieties of sweet potato and studies were conducted to obtain genotypes with low sugar content, high yield and high DM content. Eighteen introduced and local cultivars were characterized using HPLC and results showed that sucrose was the major sugar in all clones, with glucose and fructose contents being higher than maltose or raffinose. These less sweet clones are now being crossed in order to lower the sugar content of local varieties (Mathurin et al., 1998).

In the USA, breeders are also attempting to reduce sugar levels. The cultivar ‘GA90-16’, an open-pollinated seedling selection derived from a polycross nursery, has been released to farmers because of its lower levels of odour-active volatiles and endogenous sugars. In baking trials, it has a bland flavour compared to the intensely flavoured cv. ‘Jewel’. However, when prepared as French fries, ‘GA90-16’ absorbs less fat than ‘Jewel’. These are yellowish, dry tasting and devoid of the typical sweet potato flavour, while

‘Jewel’ French fries are, because of their high sugar content and very dark, oily and soft texture (Kays et al., 2001).

HERITABILITY OF MAJOR TRAITS

It has been determined that flesh colour, flesh oxidation, percentage dry weight, percentage crude starch, percentage crude protein, skin colour, resistance to root-knot nematode and vine length all have high heritabilities (Table 10.3).

In Bogor, Indonesia, heritability of storage roots is about 61.2% for the family and 58.6% for individual genotype response, agreeing with estimates in temperate countries, but it is not certain that this would be the case for other traits. For DM content, heritability is high enough to make rapid progress with phenotypic variation (Mok et al., 1998).

Qualitative characters are distinguished easily one from the other (for example, storage root skin colour) and generally are controlled by only one or two sets of genes. Quantitative characters such as the shape and yield of the storage roots are indistinct and continuously grade into each other, involving many sets of genes. White flesh colour seems to be dominant over orange flesh.

For such characters, progenies often resemble the parents and genetic improvement through recurrent selection is rapid. As for other root crop species, the total carotene content appears to be controlled by several genes, probably six that are additive.

Starch, along with carotene and protein, are the three essential characters in sweet potato breeding programmes for human consumption. The broad sense heritability of starch digestibility is very high and its improvement is theoretically feasible, but it necessitates facilities for routine analysis and screening of numerous genotypes. The costs are so high that unless a simple screening tool is available, it does not seem practical to retain this trait in a conventional breeding programme (Zhang and Li, 2004). Apparently, the starch content of sweet potatoes is determined mainly by the additive effect of polygenes and, therefore, the accumulation of genes controlling high starch content is recommended. A few genes with simple dominance seem to control the inheritance of fibre size, while the total fibre content is controlled by several genes, suggesting that low-fibre improvement is quite feasible. Unfortunately, carotene content seems to be correlated negatively with DM, while starch content and DM are correlated positively: both are associated with eating

Table 10.3. Narrow sense heritability estimates for sweet potato.

Trait Heritability estimates (%)

Root weight 25, 41, 44

Growth cracking 37, 51

Flesh colour 53, 66

Flesh oxidation 64

Dry matter 65

Fibre 47

Skin colour 81

Sprouting 37, 39

Vine length 60

Leaf type 59

Flowers/inflorescence 50

Fusarium wilt resistance 50, 86, 89

Nematode egg mass index 57, 69, 75

Insect complex resistance 45

Flea beetle resistance 40

Weevil resistance 84

Source: Jones et al. (1976), Jones (1986) and Martin (1988).

quality. A genotype with only high carotene content probably would have a low DM and high moisture and this would be unpalatable for most consumers throughout the tropics. A starch content of about 20% and a DM above 30%

should be considered as criteria more important than either protein or carotene content (Wang, 1982).

Selection for high DM content is very effective because there is tremendous variation for this trait in the germplasm, ranging from only 14% to more than 44%, and because the heritability of DM has been estimated at 75–88% (Zhang and Li, 2004).

The Japanese approach for accumulating the genes for starch content is via inbreeding by selfing and sib cross. The development of inbred lines with high DM content and subsequently crossing them with elite cultivars is an attractive approach. The first step is to develop inbred lines that are derived

The Japanese approach for accumulating the genes for starch content is via inbreeding by selfing and sib cross. The development of inbred lines with high DM content and subsequently crossing them with elite cultivars is an attractive approach. The first step is to develop inbred lines that are derived