Evolution of finger millet: evidence from random amplified polymorphic DNA

Genome Downloaded from www.nrcresearchpress.com by MICHIGAN STATE UNIV on 01/26/17 For personal use only.

Khidir W. Hilu

Abstract: Finger millet (Eleusine coracana ssp. coracana) is an annual tetraploid member of a predominantly African genus. The crop is believed to have been domesticated from the tetraploid E. coracana ssp. africana. Cytogenetic and isozyme data point to the allopolyploid nature of the species and molecular information has shown E. indica to be one of the genomic donors. A recent isozyme study questioned the proposed phylogenetic relationship between finger millet and its direct ancestor subspecies africana. An approach using random amplified polymorphic DNA (RAPD) was employed in this study to examine genetic diversity and to evaluate hypotheses concerning the evolution of domesticated and wild annual species of Eleusine. Unlike previous molecular approaches, the RAPD study revealed genetic diversity in the crop. The pattern of genetic variation was loosely correlated to geographic distribution. The allotetraploid nature of the crop was confirmed and molecular markers that can possibly identify the other genomic donor were proposed. Genotypes of subspecies africana did not group closely with those of the crop but showed higher affinities to E. indica, reflecting the pattern of similarity revealed by the isozyme study. The multiple origin of subspecies africana could explain the discrepancy between the isozyme-RAPD evidence and previous information. The RAPD study showed the close genetic affinity of E. tristachya to the E. coracana - E. indica group and underscored the distinctness of E. multiflora. Key words: finger millet, Eleusine, RAPD, evolution, germplasm. RCsumC : Le millet indien (Eleusine coracana ssp. coracana) est une annuelle tktraploi'de qui fait partie d'un genre prCsent surtout en Afrique. L'espece cultivCe rCsulterait de la domestication de l'espece tktraploi'de E. coracana ssp. africana. Les donnCes cytogCnCtiques et isoenzymatiques indiquent une nature allopolyploi'de pour ces espkces, et des informations molCculaires identifient E. indica comme un des parents. Une Ctude rCcente a l'aide d'isoenzymes a remis en question la relation phylogCnCtique proposCe entre le millet indien et son ancgtre directe, la sous-espece africana. Une approche employant des ADNs polymorphes amplifiCs au hasard (RAPDs) a permis ici d'Ctudier la diversit6 gCnCtique et d'kvaluer diverses hypotheses sur 1'Cvolution des especes annuelles domestiques et sauvages d'Eleusine. Contrairement aux approches molCculaires antkrieures, 1'Ctude a l'aide de RAPDs a rCvClC une diversit6 gCnCtique a I'intCrieur de l'espece cultivke. Le motif de variation gCnCtique Ctait faiblement corrClC a la distribution gkographique. L'Ctat allotCtraploi'de de l'espece cultivCe a CtC confirm6 et des marqueurs molCculaires permettant potentiellement d'identifier l'autre parent sont suggCrCs. Les gCnotypes de la sous-espece africana n'Ctaient pas tres proches de ceux de l'espece cultivCe mais montraient de plus grandes affinitCs avec E. indica tel que l'avait montrC 1'Ctude isoenzymatique. L'origine multiple de la sous-espece africana pourrait expliquer la divergence entre les donnCes isoenzymatiques/RAPDs et les donnCes antkrieures. L'Ctude RAPD a montrC la proximitt5 de E. tristachya avec le groupe E. coracana - E. indica et souligne la nature distincte de E. multiflora. Mots cle's : millet indien, Eleusine, RAPD, Cvolution, germoplasme. [Traduit par la RCdaction]

introduction

I

Corresponding Editor: J.P. Gustafson. Received July 5, 1994. Accepted October 14, 1994.

Khidir W. Hilu. Department of Biology, Virginia Polytechnic Institute and State University, Blacksburg, Virginia 24061, U.S.A.

Genome, 38: 232-238 (1995). Printed in Canada 1 Imprime au Canada

Finger millet (Eleusine coracana (L.) Gaertn. ssp. coracana) is one of the oldest domesticated millets in Africa south of the Saharan desert (Hilu et al. 1979). The crop is valued for its drought tolerance, good yield, and superior nutritional value (Barbeau and Hilu 1993). These agricultural aspects

Table 1. Eleusine species and accessions used in the RAPD study and the origin of the seed collections.

Genome Downloaded from www.nrcresearchpress.com by MICHIGAN STATE UNIV on 01/26/17 For personal use only.

Genotype number

Species

Subspecies

coracana coracana coracana coracana coracana coracana coracana coracana coracana coracana indica indica indica indica indica coracana coracana coracana coracana tristachya tristachya multiflora

coracana coranaca coracana coracana coracana coracana coracana coracana coracana coracana

africana africana africana africana

Accession number

Origin

4264, deWet & Harlan 4007, deWet & Harlan 3310, deWet & Harlan 3968, deWet & Harlan PI 321 126, U.S.D.A. PI 321 130, U.S.D.A. KH 264 K. Hilu PI 225569, U.S.D.A. PI 318897, U.S.D.A. KH 267, F. Attere PI 217609, U.S.D.A. 37 11, deWet & Harlan 17 14, deWet & Harlan KH 202, K. Hilu PI 321 126, U.S.D.A. PI 226270, U.S.D.A. KH 268, K. Hilu KH 225, K. Hilu KH 254, K. Hilu PI 477078, U.S.D.A. PI 309992, U.S.D.A. 5 168, deWet & Harlan

India, Sikkim, Koda India, Madras India, Andhra Pradesh, Sharipiralla India, Himachal Pradesh, Mandal Uganda, Mugusu Uganda, Fort Portal Kenya, Homa Bay Zimbabwe, Salisbury Ethiopia, Adiugn Tanzania India, Nilgiri Hills, Coonor Nigeria, Baissa Nigeria, Zowkwa Kenya, Kangoru Uganda, Mugusu Zimbabwe, Salisbury Kenya, Mathana Valley Kenya, Handi Hill Kenya, Kisii Uruguay, Cerrekargo Brazil, Porto Alegre Ethiopia, Alemeya

Note: The genotype number corresponds to the numbers in Figs. 2 and 3. "Genotypes used in the first set of species (see Materials and methods). ' ~ e n o t ~ ~used e s in the second set of species (see Materials and methods).

make finger millet a valuable crop in the semi-arid regions of Africa. Finger millet is also widely cultivated throughout India (Hilu and deWet 1976b). The extensive cultivation of finger millet in both Africa and India has raised some questions regarding the area of origin and the evolution of the crop. Finger millet is an annual tetraploid with a basic chromosome number, x = 9 (Chennaveeraiah and Hiremath 1974). It is a member of a predominantly African genus, Eleusine, which includes nine species that are diploid or polyploid and are of an annual or perennial habit. Eleusine coracana is comprised of the two subspecies coracana and africana (Kennedy-O'Byrne) Hilu and deWet [syn. E. africana Kennedy-O'Byrne]. Various hypotheses concerning the origin and evolution of finger millet have been proposed. It has been hypothesized that finger millet was domesticated from Eleusine indica in India, others suggested an African domestication from E. coracana ssp. africana, and still others believed in independent domestication from E. indica and subspecies africana in India and Africa, respectively (see Hilu and deWet 1976b). Eleusine indica is a diploid, while subspecies africana is a tetraploid; both are based on x = 9. Morphological, biochemical, molecular, and cytogenetic evidence point to subspecies africana as the possible direct progenitor of finger millet (reviewed in Hilu and Johnson 1992). Chromosome pairing and isozyme segregation patterns point to the allopolyploid nature of E. coracana (Hiremath and Chennaveeraiah 1982; Werth et al. 1993,

1994). Information from restriction fragment length polymorphism (RFLP) of the chloroplast DNA (cpDNA), ribosomal DNA (rDNA), chromosome pairing, and isozyme segregation point to the diploid E. indica as one of the genomic donors of wild and domesticated forms of E. coracana (Hilu 1988; Hilu and Johnson 1992; Hiremath and Salimath 1992; Werth et al. 1994). The second genomic donor is yet to be identified. A recent study of isozyme variation in Eleusine species (Werth et al. 1994) showed the two subspecies of E. coracana to have some distinct allelic compositions, raising the question of whether E. coracana ssp. africana is truly the direct ancestor of the crop. This hypothesis has to be tested with new approaches. An alternative hypothesis to the isozyme-based one is that E. coracana ssp. africana is highly variable and that the populations and (or) genotypes examined in the isozyme study are not necessarily the ones from which the crop was domesticated. In this study, an approach using random amplified polymorphic D N A (RAPD) is employed to examine genetic diversity in domesticated and annual wild species of Eleusine and the evolution of finger millet. Previous studies showed the perennial species to be unlikely candidates as genomic donors to the tetraploid annual crop (Hilu and Johnsom 1992; Werth et al. 1994). The RAPD approach was chosen for its ability to resolve genetic variation at the population and species levels and to study the origin of polyploids (Devos and Gale 1992; He et al. 1992; Joshi and Nguyen 1993; Hilu 1994; Howell 1994). RFLP studies of the cpDNA and the

Genome, Vol. 38, 1995

intergenic spacer region (IGS) of the ribosomal genes in E. coracana did not resolve useful variation in the crop, nor between it and subspecies africana and E. indica (Hilu 1988; Hilu and Johnson 1992). Although this lack of variation was not particularly surprising for the highly conserved chloroplast genome, it was unexpected for the IGS region.

Genome Downloaded from www.nrcresearchpress.com by MICHIGAN STATE UNIV on 01/26/17 For personal use only.

Material and methods Two sets of species were examined; one set emphasized finger millet, its direct ancestor subspecies africana, and the proposed genomic donor E. indica, while the other centered on the annual species E. coracana, E. indica, E. tristachya, and E. multiflora. Representative genotypes of the second set of species were chosen to represent the variation in RAPD observed in the first set and the geographic distribution of the species. The perennial species were excluded, because of the annual habit of the crop and the lack of evidence for their close genetic affinity to finger millet. The accessions used in each experiment and the sources of the seed material are cited in Table 1. The finger millet accessions were selected to represent the geographic distribution of the crop, while the representation of the wild species was limited by the available seed collections. Seed collections of wild species of Eleusine, like those of most millets, are unfortunately very rare. Seeds were grown in the greenhouse and leaf material was harvested from 3to 4-week-old seedlings. The leaves were either used directly for DNA isolation or were frozen in liquid nitrogen and stored at - 80°C.

Template DNA preparation and PCR amplification Total cellular DNA was isolated from leaf material of individual plants following Hilu's (1994) modification of the procedure of Saghai-Maroof et al. (1984). DNA concentration in the samples was determined with a TKO 100 Mini-Fluorometer (Hoefer Scientific Instruments). Samples for the PCR amplification were diluted to approximately 5 ng/pL with deionized distilled water. Ten-base oligonucleotide primers (Operon Technologies primers OPA 1, 2, 3, 4, 6, 7, 8, 9, 10, 11, 13, and 18) were used for RAPD PCR amplification following the procedure of Williams et al. (1990), with some modifications. Each reaction mixture (25 pL) contained 2.5 pL of 1OX PCR reaction buffer (Promega), 2.0 pL of 25 mM magnesium chloride, 0.25 pL of each of the 10 mM deoxynucleotide triphosphates dATP, dCTP, dGTP, and dTTP, 0.047 pg of a single PCR primer, 0.9 U of Taq DNA polymerase (Promega), and approximately 15 ng of genomic DNA template. The mixture was overlaid with two drops of mineral oil. Amplification was carried out in a Perkin Elmer Cetus thermal cycler programmed for one initial denaturation cycle at 95°C for 5 min followed by 75 cycles of 94°C for 10 s, 36°C for 10 s, and 72°C for 2 min. On completion of the cycles the samples were refrigerated at 4°C before analysis. This experimental protocol has been successfully used in our RAPD studies of grasses and millets and peanut (Hilu 1994; M'Ribu and Hilu 1994; K.W. Hilu and H.T. Stalker, unpublished data). As controls, some of the experiments were repeated twice in order to check reproducibility. In addition, a control sample was

included that contained all the components of the PCR reaction mixture except for the template DNA, which was replaced by water. The effect of template DNA concentration on amplification was examined and the optimal concentration was selected. PCR amplification products were analyzed by electrophoresis on a gel containing 0.7% agarose and 1.0% Synergel (Diversified Biotech). The whole reaction sample (25 p L ) was loaded on the gel and run in TAE (Tris-acetate-EDTA) buffer at 120 V for approximately 5 h. A one-lulobase DNA ladder marker (Bethesda Research Laboratories) was used as a molecular standard. DNA was stained with ethidium bromide and photographed under uv light.

Data analysis Amplified DNA fragments were scored for the presence or absence of shared RAPD fragments. DNA fragments that were present at lower intensities were scored as present, while those that were barely detectable were scored as absent. Only RAPD fragments that were variable among accessions were included in the data analysis. Missing value scores were used in cases where amplification of DNA samples was not adequate for scoring. Three matrices of row data were analyzed: two represent the two sets of species detailed above, while the third matrix included the RAPD fragments from the accessions of domesticated finger millet only (Table I). The latter matrix was extracted from the first matrix by eliminating the data of the wild species in order to assess the grouping of the finger millet accessions alone. To generate matrices of similarity, the rectangular row data matrices were subjected to the Dice algorithm (Dice 1945), which is equal to equation 21 of Nei and Li (1979). Both algorithms calculate coefficients of similarities based on shared presence of attributes and exclude shared absence as a criterion of similarity. The resultant similarity matrices were utilized to group genotypes via the unweighted pair-group method (UPGMA). The cophenetic coefficients for the clusters were computed and the correlations between these coefficients and the similarity matrices were determined by using normalized Mantel statistic z. The matrices of similarity were also used in a principal coordinates analysis (PCO) to resolve patterns of variation among the genotypes. P C 0 has advantages over the principal component analysis for qualitative data (i.e., scored as present or absent) and also, when the original matrix contains missing values (Sneath and Sokal 1973). The minimum spanning tree (MST) was also computed from the Dice similarity matrices and was projected onto the plot of the first three factors of the PCO. The MST links nearest neighbour genotypes. The NTSYS-pccomputer program (Rohlf 1993) was used for all the data analysis.

Results and discussion Among the various amplified DNA fragments, 103 were found to be informative (present in more than one accession but not in all) for the E. coracana - africana - indica group and 140 for the annual species of Eleusine. Although DNA bands shared by the Eleusine species were observed

Genome Downloaded from www.nrcresearchpress.com by MICHIGAN STATE UNIV on 01/26/17 For personal use only.

Fig. 1. RAPD patterns in species of Eleusine. A. Band patterns generated by primer OPA 2 for E. coracana ssp. coracana (finger millet), lanes 1-9; E. indica, lanes 10-1 1 ; and E. coracana ssp. africana, lanes 12-15. DNA in lane 2 was not well amplified. The leftmost lane contains molecular weight markers identified in base pairs. B. RAPD pattern of three genotypes representing finger millet and E. indica amplified by primers OPA 8, 10, 11, 12, 13, and 18 to show bands unique to finger millet (marked by an arrow) that might be used as markers to identify the missing genome. The first two lanes on the left in each set belong to finger millet, while the third lane represents E. indica. The leftmost lane contains molecular weight markers identified in base pairs.

Fig. 2. Grouping of genotypes of Eleusine species based on RAPD data using UPGMA. A. Genotypes of E. coracana ssp. coracana (cor, finger millet) and ssp. africana (afr) and of E. indica (ind) of the first set of species described in the Materials and methods section. The numbers correspond to the genotype numbers in Table 1. The abbreviation beside the finger millet genotype represents the locality of the genotype: AP, Andhra Pradesh, India; ETH, Ethiopia; HP, Himachal Pradesh, India; KEN, Kenya; MAD, Madras, India; SIK, Sikkim, India; TAN, Tanzania; and UGN, Uganda. B. Representative genotypes of E. coracana ssp. coracana (cor), ssp. africana (afr), E. indica (ind), E. tristachya (tris), and E. multiflora (multf) of the second set of species described in the Materials and methods section. The number beside the abbreviation corresponds to the genotype number in Table 1. 0.32 6

0.48

0.64

0.80

0.96 cor 1 - S I ~ cor 3-AP cor 4-HP cor 7-KEN cor 5-UGA cor 9-ETH cor 8-ZIM cor 6-UGA

A

1nd3 2-Had cor 10-TAN

tor

I I 1I

(Fig. l a ) , variation within and among species in RAPD fragments was evident for all species examined (Fig. 2a). The variation was higher in the wild species than in the domesticated finger millet, as is evident from the phenogram (Fig. 2a) and the P C 0 (Fig. 3). The overall degrees of correlation in the phenogram were higher for the wild species than for the crop, and the genotypes of the wild species show a higher range of variation on the three P C 0 axes. This finding is in agreement with the RFLP study of the ribosomal DNA spacer region (Hilu and Johnson 1992). However, in contrast with both ribosomal DNA and isozymes, where little or no variability in finger millet could be resolved, the RAPD study revealed some molecular heterogeneity in the crop. Therefore, the RAPD method may represent a good approach for revealing genetic diversity

I

afr 2 afr 1

in highly self-fertile species, such as those of Eleusine. This is particularly true for crop plants where domestication from small subsets of the wild species populations represent a bottleneck pattern of evolution. An additional bottleneck may be imposed by the dispersal of a few accessions (via trading) to new geographic regions, with subsequent inbreeding and human selection for particular superior cultivars further augmenting the loss of genetic diversity. No differences were observed in the grouping of finger millet genotypes, when analyzed separately or included with the genotypes of subspecies africana and E. indica. Therefore, to avoid duplication, only the phenogram based on all three taxa will be used in the discussion. The 10 genotypes of finger millet appeared in a single welldefined cluster that included genotype ind3 of E. indica (Fig. 2a). This grouping was reflected in the P C 0 plot, where all the genotypes were linked to each other by the MST before joining a genotype of E. indica (Fig. 3). Interestingly, the ind3 genotype is the only E. indica that carries the 105 P G M l allele characteristic of subspecies africana (Werth et al. 1994). The RAPD-based affinities among these genotypes in the phenogram varied from 0.92

Genome, Vol. 38, 1995

Genome Downloaded from www.nrcresearchpress.com by MICHIGAN STATE UNIV on 01/26/17 For personal use only.

Fig. 3. A plot showing RAPD-based associations computed from the PC0 and MST of the genotypes of E. coracana ssp. coracana (cor, finger millet) and ssp. africana (afr) and of E. indica (ind). The number beside the abbreviation corresponds to the genotype number in Table I used in the first set of species.

to 0.65. Among the Indian collections, the genotypes from Sikkim appeared distinct. The Sikkim group of finger millet is also geographically and morphologically distinct, being grown in the very northeastern region of India (Hilu and deWet 1 9 7 6 ~ )The . finger millet cultivars from southern India are morphologically indistinguishable from the lowland East African types of finger millet. Thus, the RAPDbased grouping of the Madras genotype with the one from Tanzania is not surprising. The two genotypes from Andhra Pradesh and Himachal Pradesh were genetically the most similar of the finger millet genotypes, in spite of their origins from different geographic regions in India. The East African genotypes, with the exception of the one from Tanzania, appeared in a group that included two of the Indian genotypes (Fig. 2a). The two genotypes from Uganda appeared to be quite distinct. The RAPD data did not provide a clear-cut separation between the African and Indian cultivars of finger millet. Such segregation of cultivars from isolated geographic areas has been reported in RAPD studies of Panicum, Echinochloa, and Paspalum millets (Hilu 1994; M'Ribu and Hilu 1994; K. M'Ribu and K.W. Hilu, in review). The RAPD results are in line with the morphological studies of Hilu and deWet ( 1 9 7 6 ~ ) and Hussaini et al. (1977). The lack of well-defined genetic and morphological differentiation resulting from geographic isolation could be the result of recent and sequential introductions of finger millet to India from Africa, which are estimated to have occurred around the second century B.C. (Hussaini et al. 1977; Hilu et al. 1979). The representative genotypes of E. c o r a c a n a ssp. africana appeared in a group that included the Kenyan genotype of E. indica (Figs. 2a and 3). The genotype of subspecies africana from Zimbabwe grouped last and at a much lower similarity value (Fig. 2a; genotype afrl). This pattern of genetic association was supported by the P C 0 and MST analyses (Fig. 3), including the genetic affinity of the genotype from Zimbabwe to one of the

genotypes from Kenya (Fig. 3; genotype afr4). The clustering of the three genotypes of subspecies africana (Fig. 2a; afr2, afr3, and afr4) with E. indica (genotype ind4) is quite intriguing, since all came from Kenya. The close genetic association between genotypes of the same geographic region of diploid E. indica and tetraploid subspecies africana might be an indication of multiple origin for the tetraploid E. coracana ssp. africana. The isozyme study (Werth et al. 1994) showed substantial variability among the genotypes of subspecies africana, some of it resulting from gene silencing, but much of it involving functional alleles of differing electrophoretic mobility. This variation also suggests that the wild tetraploid is derived from multiple origins. The multiple origin of polyploid species has been documented in various plant groups (Werth et al. 1985; Wyatt et al. 1989; Soltis and Soltis 1991). Therefore, a better sample representation by further collecting of both taxa from a wider geographic region could test this hypothesis. The hypothesis of a multiple origin for the tetraploid E. coracana ssp. africana could find support from the way it grouped with subspecies coracana. The groups containing predominantly subspecies coracana and africana displayed an overall similarity value of 0.54. The RAPDbased genetic affinities between domesticated and proposed wild finger millet (ssp. africana) are less than one would expect from RAPD-based affinities between a crop and its direct wild ancestor (Hilu 1994; Hilu and M'Ribu 1994; K.W. Hilu and H.T. Stalker, unpublished data). The MST showed the nearest neighbors of most of the representatives of E. c o r a c a n a ssp. africana to be the genotypes of E. indica but not those of domesticated finger millet. Therefore, it is possible that the evolutionary lineages of subspecies africana that were examined are not the same as those from which the crop was domesticated. Information from chromosome pairing, flavonoid compound distribution, chloroplast genome and the rDNA spacer region sequence variation, and morphological characteristics point to subspecies africana as the direct ancestor for the crop. Both the isozyme (Werth et al. 1994) and this RAPD study (Fig. 1) have shown the presence of molecular markers that are not common to the two subspecies but that are shared between subspecies africana and E. indica. Isozymes and RAPD, in addition to their potential role as genomic markers, were capable of resolving more variability at the subspecies level in both subspecies of E. coracana than were the other approaches. Previous studies cited above have laid down the overall pattern of evolution of the crop. With the aid of isozyme and RAPD methods, we are looking at genetic affinities at a higher resolution, i.e., affinities at the population levels. Unfortunately, wild species of millet are poorly represented in the world germplasm collections. With the high degree of genetic variability observed in the wild species (Fig. 2a), it is very likely that the genotypes of subspecies africana that were examined might not represent those genotypes and populations from which the crop was domesticated. This is particularly true if subspecies africana had a multiple origin. Only a better representative collection of the wild species can provide a more accurate and detailed picture of the genetic affinities between finger millet and subspecies africana. Thus, the information from the RAPD and

Genome Downloaded from www.nrcresearchpress.com by MICHIGAN STATE UNIV on 01/26/17 For personal use only.

isozyme approaches contribute to a new phase in studying the evolution of finger millet at the population level. The RAPD analysis centering on representatives of the annual species of Eleusine showed the domesticated and wild subspecies of E. coracana segregating into two distinct groups, with subspecies africana displaying higher affinities to E. indica (Fig. 2b). These three taxa formed a welldefined cluster. The grouping of subspecies africana and E. indica was at a slightly higher similarity value than the grouping of the two to the crop (Fig. 2b). Eleusine indica is a diploid species that has been shown to represent the A genome donor of the tetraploid species E. coracana (Hilu 1988; Hilu and Johnson 1992; Hiremath and Salimath 1992). Therefore, the overall grouping of these three taxa (Fig. 2b) reflects their overall close genomic affinity. The presence of two genomes in tetraploid finger millet is supported by cytogenetic and isozyme data (Hiremath and Salimath 1992; Werth et al. 1994). Finger millet possessed DNA bands that were not found in the accessions examined for the "A" genome donor E. indica (Fig. lb). These DNA markers could possibly have been contributed by another diploid taxon that would represent the "B" genome donor, and potentially could be useful in its identification. However, the possibility that these bands represent polymorphism in E. indica should be excluded by examining a larger sample size of the species. Eleusine tristachya is the only New World species of this predominantly African genus. The species is widely distributed in South America, extending north to the southwestern parts of the United States (Hilu 1980). Eleusine tristachya is found as a weed in the northeastern parts of Africa (Phillips 1972). Chloroplast DNA, rDNA, and isozyme studies have shown this New World species to be the most similar to the E. coracana - E. indica species complex (Hilu 1988; Hilu and Johnson 1992; K.W. Hilu and J.L. Johnson, unpublished data; Werth et al. 1994). This relationship is further supported by the RAPD study (Fig. 2b). Eleusine multiflora is native to Africa. The species, however, is morphologically very distinct from the remaining species of Eleusine. The raceme-type inflorescence is quite distinct from the digitate inflorescence characteristic of the genus and the flower structure and seed surface are also atypical for the species of Eleusine (Phillips 1972). The species is considered to be intermediate between Eleusine and Achrachne in floret disarticulation, shape of lemma tip, and the persistence or shedding of the pericarp (Phillips 1972). Flavonoid compounds characteristic of this species are quite different from those of the other species in the genus (Hilu et al. 1978). Similarly, molecular data from cpDNA and rDNA point to the isolated position of the species in Eleusine (K.W. Hilu and J.L. Johnson, unpublished data, Hilu and Johnson 1992). Eleusine multiflora appeared very distinct in the RAPD-based phenogram, showing only 0.15 similarity to the other species. The RAPD data provide additional evidence for the distinctness of this species. Thus, the RAPD approach is particularly useful in providing measures of genetic diversity and for generating markers for the study of germplasm resources in crop plants at intraspecific and interspecific levels. The molecular

characters provided by the RAPD approach outlined a pattern of genetic affinities among the annual domesticated and wild species of Eleusine that is well supported by molecular data from the chloroplast and nuclear genomes, cytogenetics, flavonoid distribution, isozymes, and morphological characteristics. This impressive pattern of similarity was obtained from only two samples per taxon, but with the aid of nine primers and 140 informative bands. The RAPD approach also provided the opportunity to study crop evolution at the population level, which points to exciting potential studies of centers of diversity and origin of crop plants. DNA bands that were present which could mark individual plants, subspecies, and species represent very valuable characters for germplasm resource studies and plant breeding. It cannot be emphasized enough that additional collections of the wild species of finger millet and other millets are essential and are absolutely needed in order to assess the genetic diversity of the various gene pools and to determine the ones genetically closest to the crops. These wild species could be the source of valuable and irreplaceable genetic resources.

Acknowledgements The author thanks Charles Werth for valuable comments on a draft of this paper, the U. S. Department of Agriculture Regional Plant Introduction Station at Ames, Iowa, the International Crops Research Institute for the Semi-Arid Tropics, and J. R. Harlan and J. M. J. deWet for contributing some of the seed collections used in the study. This research was supported by grant no. DHR-5600-G00-1073-00, Program in Science and Technology Cooperation, Office of the Science Advisor, U. S. Agency for International Development.

References Barbeau, W.E., and Hilu, K.W. 1993. Protein, calcium, iron, and amino acid content of selected wild and domesticated cultivars of finger millet. Plant Foods Hum. Nutr. (Dordr.), 43: 97-104. Chennaveerdiah, M.S., and Hiremath, S.C. 1974. Genome analysis of Eleusine coracana (L.) Gaertn. Euphytica, 23: 489-495. Devos, K.M., and Gale, M.D. 1992. The use of random amplified polymorphic DNA markers in wheat. Theor. Appl. Genet. 84: 567-572. Dice, L.R. 1945. Measures of the amount of ecological association between species. Ecology, 26: 295-302. He, S., Ohm, H., and Mackenzie, S. 1992. Detection of DNA polymorphisms among wheat varieties. Theor. Appl. Genet. 84: 573-578. Hilu, K. W. 1980. Eleusine tristachya (Lam.) Lam. (Poaceae). Madrono, 27: 177- 178. Hilu, K.W. 1988. Identification of the "A" genome of finger millet using chloroplast DNA. Genetics, 118: 163-167. Hilu, K.W. 1994. Evidence from RAPD markers in the evolution of Echinochloa millets (Poaceae). Plant Syst. Evol. 189: 247-257. Hilu, K.W., and deWet, J.M.J. 1976a. Racial evolution

Genome Downloaded from www.nrcresearchpress.com by MICHIGAN STATE UNIV on 01/26/17 For personal use only.

Genome, Vol. 38, 1995

of finger millet Eleusine coracana. Am. J. Bot. 63: 131 1-1318. Hilu, K.W., and deWet, J.M.J. 1976b. Domestication of Eleusine coracana. Econ. Bot. 30: 199-208. Hilu, K.W., and Johnson, J.L. 1992. Ribosomal DNA variation in finger millet and wild species of Eleusine (Poaceae). Theor. Appl. Genet. 83: 895-902. Hilu, K.W., deWet, J.M.J., and Seigler, D. 1978. Flavonoids and the systematics of Eleusine. Biochem. Syst. Ecol. 6: 247-249. Hilu, K.W., deWet, J.M.J., and Harlan, J.R. 1979. Archaeobotany and the origin of finger millet. Am. J. Bot. 66: 330-333. Hiremath, S.C., and Chennaveeraiah, M.S. 1982. Cytogenetical studies in wild and cultivated species of Eleusine (Gramineae). Caryologia, 35: 57-69. Hiremath, S.C., and Salimath, S.S. 1992. The 'A' genome donor of Eleusine coracana (L.) Gaertn. (Gramineae). Theor. Appl. Genet. 84: 747-754. Howell, E.C., Newbury, H.J., Swennen, R.L., Withers, L.A., and Ford-Lloyd, B.V. 1994. The use of RAPD for identifying and classifying Musa germplasm. Genome, 37: 328-332. Hussaini, S.H., Goodman, M.M., and Timothy, D.H. 1977. Crop Sci. 17: 257-263. Joshi, C.P., and Nguyen, H.T. 1993. Application of the random amplified polymorphic DNA technique for the detection of polymorphism among wild and cultivated tetraploid wheats. Genome, 36: 602-609. M'Ribu, K., and Hilu, K.W. 1994. Detection of interspecific and intraspecific variation in Panicum millets through random amplified polymorphic DNA (RAPD). Theor. Appl. Genet. 88: 4 1 2-4 1 6. Nei, M., and Li, W.H. 1979. Mathematical model for studying genetic variation in terms of restrictions endonucleases. Proc. Natl. Acad. Sci. U.S.A. 76: 5269-5273.

Phillips, S.M. 1972. A survey of the genus Eleusine Gaertn.(Gramineae) in Africa. Kew Bull. 27: 25 1-270. Rohlf, F.J. 1993. NTSYS-pc. Numerical taxonomy and multivariate analysis system. Version 1.8. Exeter Publishing Ltd., Setauket, New York. Saghai-Maroof, M.A., Soliman, K.M., Jorgensen, R.A., and Allard, R.W. 1984. Ribosomal DNA spacerlength polymorphisms in barley: Mendelian inheritance, chromosomal location, and population dynamics. Proc. Natl. Acad. Sci. U.S.A. 81: 8014-80 18. Sneath, P.H.A., and Sokal, R.R. 1973. Taxonomic structure. In Numerical taxonomy. W.H. Freeman and Co., San Francisco. pp. 234-235. Soltis, P.S., and Soltis, D.E. 1991. Multiple origins of the allotetraploid Tragopogon minus (Compositae): rDNA evidence. Syst. Bot. 16: 407-413. Werth, C.R., Guttman, S.I., and Eshbaugh, W.H. 1985. Recurring origins of allopolyploid species in Asplenium. Science (Washington, D.C.), 228: 731-733. Werth, C.R., Hilu, K.W., Langner, C.A., and W.V. Baird. 1993. Duplicate gene expression for isocitrate dehydrogenase and 6-phosphogluconate dehydrogenase in diploid species of Eleusine (Gramineae). Am. J. Bot. 80: 705-710. Werth, C.R., Hilu, K. W., and Langner, C.A. 1994. Isozymes of Eleusine (Gramineae) and the origin of finger millet. Am. J. Bot. 81: 1186-1197. Williams, J.G.K., Kubelik, A.R., Rafaliski, K.J., and Tingey, S.V. 1990. DNA polymerase amplified by arbitrary primers are useful as genetic markers. Nucleic Acids Res. 18: 653 1-6535. Wyatt, R., Odrzykoski, I.J., Stoneburner, A., Bass, H.W., and Galau, G.A. 1989. Allopolyploidy in bryophytes: Multiple origins of Plagiomnium medium. Proc. Natl. Acad. Sci. U.S.A. 85: 5601-5604.