It is generally agreed that pearl millet was domesticated in Africa, probably on the southern edge of the Sahara, west of the Nile, some 3,000 to 5,000 years ago and subsequently spread to southern Asia (Harlan 1975; Brunken et al. 1977). The gene flow from probable progenitor wild species [spp. monodii (syn. P. violaceum Maire)] still occurs in West Africa, where weedy segregates (spp. stenostachyum Klotzsch) are common in cultivated varieties. The frequency of occurrence of these segregates declines through eastern to southern Africa, and they are absent in Asia.
Three germplasm pools are recognized in respect of cultivated pearl millet (Harlan and de Wet 1971). The primary pool contains all cultivated, weedy, and wild diploid (2n = 14) pearl millets which are freely cross-fertile. The secondary pool is solely P. purpureum (Shum.) (2n = 28), elephant or Napier grass, a vigorous perennial species also from Africa. The cross between pearl millet and elephant grass is easily made (but is sterile unless the chromosome number is artificially doubled) and is widely used as a forage propagated by cuttings. Hanna (1990, 1991b) has demonstrated that part of the purpureum genome can be usefully transferred to pearl millet. The tertiary pool contains numerous more distantly related Pennisetum species of various ploidy levels which do not naturally interbreed with the primary pool, but can potentially be accessed through various wide crosses (Dujardin and Hanna 1989, 1990).
The cultivated gene pool in pearl millet contains a truly enormous range of genetic variability with no incompatibility and few linkage problems. Many important physiological and morphological traits essential for breeding a crop suitable for combine harvesting are readily available. These are reduced plant height and early maturity--independent of photoperiodic control, synchrony of tiller flowering, angle of tillering, stem and peduncle thickness, peduncle length, grain color (white or cream simply dominant over gray), and mesocotyl length. Ample variation exists for head size (length ' diameter) and grain number and size though, as expected, correlations between yield components are large and negative.
In areas where frost-kill can occur before harvest, standability is crucial. Both peduncle and stem lodging resistance are evident in inbred stocks, but most material is susceptible and this is one trait which will require much breeding attention. Differences in tolerance to some herbicides (propachlor and atrazine) have been noted and incorporated in selection and breeding strategies.
Pearl millet is a naturally cross-pollinating species, which is achieved through protogyny, since all the sessile flowers on each head are perfect (i.e. both male and female fertile). On any one head, all flowers first exert stigmas over a 1 to 3 day period progressing from the mid-top to the bottom of the head. Anthesis occurs one to as many as 4 days later, in the same sequence from the same flowers, and sometimes, later from the pedicellate flowers. Thus, there is a period for each head, when flowers can only be fertilized by external pollen which is freely wind-born. Stigmas wither about 8 h after pollination. Self-pollination can occur when stigma emergence on later flowering tillers overlaps with the anthesis of earlier heads on the same plant. In random-mating situations (as in landrace cultivar populations or breeder created populations--synthetics or composites), the amount of self-pollination (considerations of common parentage and effective population size apart) is influenced by the degree of tillering, relative size and flowering relationships of tillers, and whether all or only primary tillers are harvested. As a generality, about 20% selfing is normal (Burton 1974; Chirwa 1991).
Selfed seed in pearl millet can be produced simply by placing a bag over a head prior to stigma emergence. If the stigmas are not short lived, 100% selfed seed set will then occur. Similarly, 100% hybrid seed can be made by pollinating a previously bagged head once at full protogyny prior to anther emergence. The breeding opportunities in pearl millet can be illustrated by the following: each of 3 heads on one plant in a population (tillering can be promoted by planting at reduced density) can be used for different objectives--one can be selfed, one crossed (full-sib, testcross, topcross) and one left to random-mate. Seed from each head will be sufficient to plant 20 plots each of 7.5 m2.
At least four cytoplasmic-genic systems causing male sterility (CMS) are available in pearl millet (Kumar and Andrews 1984; Hanna 1989). The first and currently most widely used source (now termed A1) was discovered by Burton (1958) in Tifton, Georgia, and released as Tift 23A in 1965 (Burton 1965). All forage hybrids in the United States are made with this CMS system. Its most extensive use, however, has been in grain hybrids in India, where an average of over 2 million ha have been grown annually over the last 23 years.
New lines in which male sterility is stable in all environments are difficult to breed in the A1 system and have never been obtained with A2 and A3. The Am source (Hanna 1989) using monodii cytoplasm is more stable and appears easier to breed, but restorers are scarce and hybrids have not been widely tested. An additional source of CMS has been reported by Marchais and Pernes (1985) but its relationship with others has not been established.
Pearl millet resembles maize in many respects in regard to gene action in performance traits. In general, additive effects are larger than non-additive effects, which can however be significant. Inbreeding depression is large--some 30% from one generation of selfing in populations (Khadr and El-Rouby 1978; Rai et al. 1984); however, vigorous inbred lines yielding 60 to 70% of open-pollinated cultivars of comparable maturity can be selected. As a generality, good hybrids will yield 20 to 30% more than the best open-pollinated cultivars of comparable maturity (Table 1) (Andrews and Rajewski 1990, 1991; Dave 1986).
With the correct selection of parent lines in regard to phenotype and relative maturity, hybrids can also be made in pearl millet by utilizing the natural period of protogyny. This method allows quicker hybrid development, greatly increases the range of possible parent combinations, and avoids diseases which are associated, particularly in Africa, with the use of CMS seed parents. These pro-hybrids, as they are termed, appear to have the most utility for developing countries where existing or reselected leading open-pollinated cultivars could be directly used as male parents for topcross hybrids.
Heterotic effects in pearl millet are large and most completely expressed in single crosses, though yields from topcross hybrids are similar in all but the highest yielding situations. Topcross hybrids have several advantages including stability and durability of performance and ease of production (Andrews 1986).
Why breeding grain hybrids in pearl millet did not immediately commence in the United States in the 1960s to parallel hybrid development in grain sorghum is not clear. Contributing factors may have been that grain production in sorghum was already established using dwarf and semi-dwarf inbred varieties that mostly had sufficient stalk strength to be combine harvested; and there was a relatively much larger germplasm base of adapted sorghum stocks from which to breed whereas the initial stocks from the Tifton program were phenotypes primarily intended for use in forage production. Also in the 1950s and 1960s, the relative nutritional advantage of pearl millet grain compared to sorghum was not widely appreciated.
In 1969, Kansas State University began a grain breeding program in pearl millet at Manhattan, Kansas, which grew partly out of the USDA/OAU joint Cereals Research Project 26 in Africa, which supported genetic research in pearl millet at Serere, Uganda. The millet breeding program at the Fort Hays Experiment Station started in 1971. Early sources of germplasm for the Hays program came from both East and West Africa; India; Tifton, Georgia; and the USDA Plant Introduction Station pearl millet germplasm collection, Experiment, Georgia. While the Tift A1 cytoplasm has been the basis for the development of seed parents at Kansas State University, another accession (PI 185642) from the Ghana/Togo landrace called Iniati/Koupela, has been a parent of fundamental importance in transmitting the character associations of large (12 to 16 g/1000), round, slate-gray or yellow grain, relatively large head width/length ratio, good combining ability, and earliness uninfluenced by photoperiod response. Dwarf derivatives of another Togo type cultivar (Serere 3A) have contributed early maturity, large seed size, and high grain yield potential to numerous imported accessions, inbred lines, and populations used as sources of pollen parents of experimental hybrids. The breeding value of the Iniati/Koupela germplasm was independently recognized in breeding programs in India, East and now West and Southern Africa. Seed parents from Hays lines have been released via ICRISAT in India, and are used extensively in hybrid production in northwest Indian states.
Work at Hays, Kansas, now supported by INTSORMIL, is focused on improving stand establishment, fertility restoration, and lodging resistance--characteristics necessary for mechanized production of hybrid millet. Large seed size and ability to emerge from deep (7.5 to 10 cm) field plantings have been selected at Hays to overcome establishment difficulties (Stegmeier 1990). These materials emerge from normal planting depths up to one day earlier than unselected lines, which is advantageous when weather conditions cause either crusting or rapid drying of seedbeds.
Fertility restoration of the A1 cytoplasm has been difficult to stabilize within the variable environment of the central Great Plains, but inbred lines have been identified that have consistently produced fertile hybrids in 20 or more tests during the past five years.
Severe stalk lodging and breaking of stem internodes occurs within all germplasm sources, lines, and hybrids selected for improved grain yield. Two sources of improved stalk quality have been found that reduce the incidence of lodging and are being incorporated into elite inbred lines.
Grain yield levels of up to 5.3 t/ha have been recorded (Christensen et al. 1984). Grain yield comparisons of sorghum and pearl millet hybrids of similar maturities (W.D. Stegmeier unpubl. data) indicate millet yields are to 60 to 90% as large as sorghum when grown on silty clay loam soils, 85 to 100% on silt loams, and will often exceed the yield of sorghum on sandy soils. On sandy soils in southcentral Kansas, Stegmeier (1990) reported pearl millet hybrids producing up to 76% more grain than the commercial sorghum hybrid check yield of 2.4 t/ha.
Research on grain production started as an adjunct to the on-going pearl millet forage breeding and wide crossing program at Tifton, Georgia, in the early 1980s. Dominant resistance to pearl millet rust (Puccinia substriata Ell. & Barth. var. indica Ramachar & Cumm) and blast (Piricularia setariae Nisikado) has been incorporated into A1 seed parents (Hanna 1991a), while pollen parents have been obtained from crossing the doubled (6x) pearl millet x elephant grass cross back to pearl millet (Hanna 1991b). A hybrid (Tift 90DAE x 8677), with these parents has been released under an exclusive license and is being grown on a pilot scale of a few hundred hectares in 1991 on the sandy soils of Georgia and South Carolina.
The breeding program for grain pearl millet commenced at the University of Nebraska-Lincoln and High Plains Agricultural Station at Sidney in 1984 with the support of INTSORMIL. Germplasm introduced earlier had been random-mated into a population early enough to mature in western Nebraska. Breeding material was extensively introduced from India and Africa. Both population and pedigree breeding are being used to produce adapted inbreds for use as hybrid parents, in synthetics, and to make new populations. New seed parents have been produced in A1 cytoplasm with improved seed set and lodging resistance. Seed and pollen parents are also being produced with the Am (monodii) cytoplasm.
The possibility of producing hybrids by using the species natural protogyny, which would increase potential hybrid combinations and greatly reduce hybrid development time, is currently being investigated (Andrews 1990). Tests with mechanical mixtures of "seed parent" lines and pro-hybrid seed have been conducted to estimate the effect on hybrid performance of any self-pollination that might occur in the pro-hybrid seed parent during hybrid seed production. Provided the hybrid has a dominant phenotype, no significant loss in hybrid yields was found in three different hybrids when 20% inbred seed of the female parent was added (Andrews 1990). Actual losses were from 4 to 6% (for detailed results see Andrews et al. 1993 in this vol.). Much less than 20% selfing would be expected in a well managed seed plot. Protogynous hybrids, therefore, seem feasible to produce and may be particularly useful in African situations.
Pearl millet regional grain yield trials testing initial experimental hybrids and other entries from ARS/USDA, Tifton, Georgia; Kansas State University, Hays; and University of Nebraska, Lincoln; have been grown cooperatively at 5 locations in the United States since 1988 (Fig. 1). The 1990 results shown in Table 1 are typical. Across locations the best pearl millet hybrids averaged 85% of the grain yield of the sorghum hybrid checks. Only where the season was short as in North Dakota, and in double cropping after wheat in Indiana (sorghum failed to mature) did millet yields exceed sorghum. Considerations of maturity, height, lodging, leaf disease occurrence, and relative genotype yields suggest that there are at least two contrasting adaptation areas within the region in which the tests have been conducted. These are the Midwest High Plains and the Southeast. Cultivars from the Southeast are too late maturing in the Midwest (will mature in Kansas but require more moisture), tend to be tall, and have little resistance to lodging following frost. Conversely, Midwest cultivars are too early in the South, and have little resistance to leaf diseases.
Recent chick feeding experiments, Sullivan et al. (1990) (Table 2) and Hancock et al. (1990) (Table 3) show that weight gains and feed/gain ratios obtained in pearl millet based diets are equal to that of maize and some sorghums. Smith et al. (1989) similarly report that pearl millet can replace maize in chick diets without affecting weight gain or feed efficiency. Both the gross energy and MEn values of pearl millet tend to be higher then those of maize and many have been previously underestimated by 20% (Fancher et al. 1987). Tribble et al. (1986) reported that they were also able to substitute pearl millet for sorghum in sorghum based diets for growing pigs without affecting performance. Calder (1955, 1961) had previously concluded that pearl millet was suitable for pig feeding.
Studies on the comparative value of pearl millet with sorghum or corn for cattle are few. When millet and sorghum grain were compared in high-silage growing rations for steers adjusted to equal protein intake, the results suggested millet protein had a high biological value as the addition of Rumensin to the rations gave millet grain a 10% advantage over sorghum grain (Brethour 1982) (Table 4). With finishing steers, Brethour and Stegmeier (1984) comparing rations where 25% of the sorghum component was replaced with pearl millet, reported that average daily gains were 1.40 and 1.20 kg, and feed/gain ratios were 7.53 and 8.03, respectively, for millet based versus sorghum based diets. Estimated net energy value of pearl millet was 4% higher than for sorghum. In both experiments, the amounts of soybean meal and/or urea needed for iso-N rations were less when pearl millet was used.
In a metabolism trial with steers, Hill and Hanna (1990) compared a diet with 79% pearl millet (PM) to diets of 76% sorghum + 2.8% soybean meal (GS) with a control (C) of 73% maize + 6% soybean meal. Ether extract and crude protein digestibilities were higher for C and PM than GS while retained N was similar for all. In an accompanying growth trial with yearling heifers, diet C gave a higher daily gain than PM, but feed:gain ratios were similar for all diets (8.5, 9.1, and 8.2 kg feed/kg gain-1, respectively, for PM, GS, and C).
In general, feeding test results support data from biochemical analyses which indicate that pearl millet is similar to maize and superior to sorghum as a feed grain. A number of factors are thought to be responsible. Pearl millet grain generally has a higher crude protein level by 1 to 2 percentage points relative to sorghum grown with similar cultural practices. Pearl millet is still deficient in essential amino acids, but averages 35% more lysine than sorghum (Rooney and McDonough 1987). Pearl millet grain has 5 to 6% oil and a lower proportion of the less digestible cross-linked prolamins (Jambunathan and Subramanian 1988). These differences can be partly attributed to the different structure of the kernel. The proportion of germ in pearl millet grain (17%) is about double that of sorghum, while the endosperm accounts for 75% as against 82% in sorghum (Table 5). Amounts of bran are similar.
Major recessive genes that strongly influence grain protein lysine levels, as discovered in sorghum and maize, have not been found in pearl millet, despite an extensive survey of the world collection. However, selection for grain protein level in pearl millet has resulted in inbreds where crude protein levels (and consequently higher levels of lysine per sample) are 4 to 6 percentage points higher than normal, without affecting endosperm development (Singh et al. 1987). Hybrids made between these high protein inbreds and normal parents gave normal yield levels but with some elevation in grain protein, indicating partial dominance for the expression of grain protein content (ICRISAT 1984). It would appear possible to breed for moderately higher protein grain content levels (and higher lysine/sample) in pearl millet without the use of a high lysine gene that adversely affects endosperm development.
More agronomic research is needed now that new hybrids are available, particularly on seedling establishment and control of grassy weeds. Preliminary observations at UN-L indicate that the choice of hybrid phenotype (medium maturity, 120 to 130 cm height, elongated, and closed canopy), planting date (delay sufficient to allow germination and removal of some grass seed), and row spacing (mechanical cultivation of wide rows vs non-cultivated narrow rows) are important in reducing competition effects of foxtail and fall panicum grasses. Pearl millet seed protectants and safeners are not available for use with the amide family of herbicides (metolachlor and alachlor). We have made good progress at the University of Nebraska, Lincoln on selecting for propachlor tolerance and the tolerance is not limited to one source of germplasm. Selection for large seed size and long mesocotyl at Kansas State University, Hays has identified genotypes with better seedling establishment and early growth which thus can be planted a little deeper into assured moisture.
Apart from the rust and leaf blast in the South, no major diseases have so far been identified on pearl millet in the Midwest. Bacterial leaf spotting caused by (Psuedomonas syringae pv. syringae van Hall) (Odrody and Vidaver 1980) has occasionally occurred in July in pearl millet forage crops, but subsides later in the season. Increased stalk lodging can occur in high nitrogen soils (>112 kg N/ha) with some hybrids. As with sorghum, some peduncle attack of the European corn borer [Ostrinia nubilalis (Hübner)] can occur and cinch bugs [Blissus leucopterus leucopterus (Say)] can spread to pearl millet from adjacent wheat to increase head and stalk lodging or kill plants during the growing season. Pearl millet has two distinct advantages over sorghum or proso--its seed will not over-winter in moist soil, and there are no wild relatives in the United States to which it will naturally outcross, so it would not become a weed in subsequent crops.
Following the demonstration of the yield potential of early maturing hybrids in India, breeding commenced in the early 1970s on grain pearl millet in the United States at Kansas State University, Hays and was joined in the 1980s by USDA/ARS Tifton and the University of Nebraska-Lincoln and Sidney. Fully dwarf experimental hybrids which can be grown like sorghum have been produced and tested regionally since 1988, giving yields averaging 2.3 to 3.8 t/ha. Highest yields on a field basis (5.3 t/ha) were recorded in Kansas. Two adaptation zones, the Southeast and Midwest High Plains are evident from these tests.
Feeding tests on cattle, swine, and particularly chickens have shown pearl millet is at least equivalent to maize and often superior to sorghum in feed rations, generally because of high energy and grain protein levels.
Initial breeding efforts and utilization tests have given encouraging results. Further cultivar improvements can be expected. Opportunities for production will be dependent on a number of factors including marketing possibilities, which may first be by specific contracts. Clearly, more agronomic research is needed on determining optimum cultivation practices, but pearl millet crops can presently be grown well with existing row crop equipment and practices. Potential production areas are those where pearl millet will have a relative advantage over other summer cereals, such as in the Southeast coastal sands, in the drier or short-season parts of the Midwest High Plains, and possibly in double-cropping after wheat in the central Midwest.
|Mean grain yield (kg/ha)|
|68A x MLSz||2300||3790||3120||4130||4440||3670|
|68A x EDSz||2980||3780||3070||3610||4360||3670|
|90PV0046 x 0049y||2640||3060||1920||2670||3190||1510|
|90PV0003 x 0005y||4790||2570||2460||4010||3480||790|
|90PV0016 x 0017y||3690||2980||2320||2900||2560||2730|
|90PV0016 x 0015y||3220||3330||2840||4050||3770||2190|
|H23DA1E x 77x||2950||1930||1260||3720||2510||810|
|RR23DAE x 77x||4950||2320||1180||4020||2590||1380|
|1163 x 86-7907x||2670||3960||3130||3330||3670||3810|
|2068 x 87-8025x||4610||2930||1990||3830||3030||2220|
|DK 39 sorghumx||5650||---||4860||6090||4170||1210|
|F 2233 sorghumx||5840||---||3660||5660||5220||2720|
|Grainy||Added fat (%)x||Weight gain (g) d 1 to 42||Gain:feed d 1 to 42|
|Crop||Crude protein |
d 7 to 21 (g)
|Avg daily ration (kg)|
|Rolled pearl |
|Premix||Air dry |
|Avg. daily |
45.4 kg gain
|Grain fraction||% of grain||Protein||% of grain||Protein|
Fig. 1. Dwarf pearl millet grain hybrid (protogyny type), 1990 Pearl Millet Regional test, Hays, Kansas (2-row plots).