Barnyard millet globalcore collection evaluation in the submontane Himalayan region of India using multivariate analysis

Slej Sood*,Rjesh K.KhuleArun Kumr R.Pwn K.AgrwlHriD.Updhyy

aIndian Council of Agricultural Research—Vivekananda Institute of Hill Agriculture,Almora,Uttarakhand 263601,India

bInternational Crop Research Institute for the Semi-Arid Tropics(ICRISAT),Patancheru,502 324 Telangana,India

Barnyard millet globalcore collection evaluation in the submontane Himalayan region of India using multivariate analysis

Salej Sooda,*,Rajesh K.Khulbea,Arun Kumar R.a,Pawan K.Agrawala,HariD.Upadhyayab

aIndian Council of Agricultural Research—Vivekananda Institute of Hill Agriculture,Almora,Uttarakhand 263601,India

bInternational Crop Research Institute for the Semi-Arid Tropics(ICRISAT),Patancheru,502 324 Telangana,India

A R T I C L E I N F O

Article history:

Accepted 6 August 2015

Available online 15 August 2015

Agro-morphological variation Barnyard millet core germplasm Cluster analysis Echinochloa spp Principal component analysis

Barnyard millet(Echinochloa spp.)is one of the most underresearched crops with respect to characterization of genetic resources and genetic enhancement.A total of 95 germplasm lines representing global collection were evaluated in two rainy seasons at Almora, Uttarakhand,India for qualitative and quantitative traits and the data were subjected to multivariate analysis.High variation was observed for days to maturity,five-ear grain weight,and yield components.The first three principal component axes explained 73%of the total multivariate variation.Three major groups were detected by projection of the accessions on the first two principalcomponents.The separation of accessions was based mainly on trait morphology.Almost all Indian and origin-unknown accessions grouped together to form an Echinochloa frumentacea group.Japanese accessions grouped together except for a few outliers to form an Echinochloa esculenta group.The third group contained accessions from Russia,Japan,Cameroon,and Egypt.They formed a separate group on the scatterplot and represented accessions with lower values for all traits except basal tiller number.The interrelationships between the traits indicated that accessions with tall plants,long and broad leaves,longer inflorescences,and greater numbers of racemes should be given priority as donors or parents in varietal development initiatives.Cluster analysis identified two main clusters based on agro-morphological characters.

?2015 Crop Science Society of China and Institute of Crop Science,CAAS.Production and hosting by Elsevier B.V.This is an open access article under the CC BY-NC-ND license

(http://creativecommons.org/licenses/by-nc-nd/4.0/).

1.Introduction

Barnyard millet(Echinochloa spp.)is one of the oldest domesticated millets in the semiarid tropics of Asia and Africa.Two main species,Echinochloa esculenta(A.Braun)H.Scholz;syn. Echinochloa utilis Ohwi et Yabuno(Japanese barnyard millet) and Echinochloa frumentacea Link;syn.Echinochloa colona var. frumentacea(Link)Ridl.(Indian barnyard millet)are cultivated and grown as cereals.It is a staple cereal in areas where climatic and edaphic conditions are unsuitable for rice cultivation[1].In India,barnyard millet is grown in the Himalayan region from the north to the Deccan plateau in the south.It is generally cultivated in hill slopes and undulating fields of hilly,tribal,or marginal areas,wherefew options exist for crop diversification.In addition to the two domesticated species,the genus includes about 20-30 annualand perennialwild species distributed worldwide[2,3], many of which can grow in wet or well-watered situations and compete successfully with rice.

Barnyard millet has a wide adaptation capacity and can grow up to an altitude of 2000 m above mean sea levelduring summer season[4].Diversity in barnyard millet has fast eroded,owing to a considerable reduction in acreage and changing socio-cultural and economic dimensions of the farming community in India[5].Many efforts have been made to preserve the crop diversity ex situ,but information about on-farm and in situ conservation of all small millets is scarce[6].

The classification of the genus Echinochloa on the basis of inflorescence morphology into two species,four subspecies,and eight races is simple and reliable and helps to elucidate not only the patterns of variation but also the paths of evolutionary history.However,it is difficult to categorize the variation in germplasm collections for economic purposes[7].Multivariate methods are useful for characterization,evaluation,and classification of plant genetic resources when a large number of accessions are to be assessed for severalcharacters ofagronomic and physiological importance[8].The utility of multivariate methods for handling morphological variation in germplasm collections has been demonstrated in many crop plants(finger millet[9];sorghum[10];barnyard millet[4]).The information generated can be useful for identifying groups of accessions that have desirable characters for crossing,planning efficient germplasm collecting expeditions,establishing core collections, revealing the patterns ofvariation in germplasm collections,and investigating aspects of crop evolution[8,11-15].

The present study describes the characterization of a recently developed barnyard millet global core germplasm collection[16]and identification of groups of accessions with similar quantitative characters,using a range of multivariate statistical methods,for genotypes grown in the submontane Himalayan region where this crop occupies a special place as food and fodder[17].

2.Materials and methods

2.1.Plant materials

The 95 germplasm accessions used include 89 accessions from the barnyard millet core collection developed by ICRISAT and six check genotypes[16].The accessions with their source countries are presented in Table S1.

The crop was raised from July to November,2011 and 2012 at the experimental farm of the ICAR-Vivekananda Institute of Hill Agriculture(79°39′E latitude and 25°35′N longitude, 1250 m above sea level).During 2011,a single row of each accession was planted in an augmented design,whereas in2012 two rows of each accession were planted in an alpha lattice design with two replications.Five blocks constituted one replication with 19 accessions in each block.The row length was 3 m with a row-to-row spacing of 22.5 cm.Thinning was applied within a month after sowing to maintain a plant-to-plant spacing of 7.5 cm within rows.

Table 1–Qualitative trait analyses in a barnyard millet core germplasm collection based on two years of data.

Fertilizer was applied at 40:20:0(N:P:K)kg ha-1,where the entire amount of phosphorus and half of the nitrogen was applied as a basal dose during field preparation.The remaining half of the nitrogen was applied as top dressing 45 days after sowing and after the second weeding.Manual weeding was performed twice during the crop season,20 and 40 days after sowing.

2.2.Data recording

Data were recorded for 7 qualitative and 16 quantitative traits, following the descriptors of barnyard millet[18].For every accession in a plot,five individualplants were used for recording the data,exceptfor days to flowering and days to maturity,which were recorded on a plot basis.

2.3.Statisticalanalysis

The agro-morphologicaldata of each year separately as wellas the pooled data were analyzed.The data could not be subjected to a combined analysis of variance,owing to the unavailability of replicated data in the first year.The adjusted mean values of the first and second years were used for further analysis.The adjusted mean values of the two years were pooled and mean values were computed for combined analysis.Statistical analyses were performed using JMP 2009(JMP,Version 9.0.0.SAS Institute Inc.,Cary,NC)and SPSS(Statistical Package for Social Science,SPSS Inc.,Chicago,IL).For multifactorial comparison, principal component analysis(PCA)was used to display the correlations between the various morphological and related parameters and their relationship with different barnyard millet genotypes.Two-way cluster analysis was performed with JMP 2009.One accession(IEc 566)did not set seeds and was not included in the analysis.

3.Results and discussion

The qualitative trait data showed that the predominant growth habit in barnyard millet is erect with green plant pigmentation. Culm branching was low in most ofthe accessions belonging to the E.frumentacea group.Medium and high culm branching was observed in eight accessions of Japanese and Russian origin.Among inflorescence traits,a pyramidalopen shape with straight lower racemes was abundant.The five accessions having branching in the lower raceme were all of Japanese origin(Table 1).The qualitative trait data showed marked differences between Japanese accessions in comparison to other accessions,indicating them to be of a group of different origin:E.esculenta.

Morphological characterization is important for identification of accessions with desirable traits intended to be employed directly as cultivars or as trait donors for use in crop improvementprograms[19].We founda wide range ofvariation in agronomic performance among the accessions evaluated. The mean,range,genotypic variance,error variance,heritability(bs-broad-sense)and coefficient of variation(CV)of quantitative traits are presented in Table 2.In the analysis of quantitative traits,the coefficient of variation varied from 0.79%for days to maturity to 36.43 for basal tiller number.The h2bsestimates ranged from 70.14 for inflorescence length to 99.87 for days to maturity.Allofthe studied traits exhibited high heritability.Sonnad et al.[20]also observed high heritability for allquantitative traits in finger millet.

The highest five-ear grain weight was found in check variety PRJ 1(24.25 g)followed by all accessions of Indian origin up to the yield level of 17.54 g.The accessions from Japan were low-yielding,except for PRJ 1,IEc 552(17.52 g),and IEc 530 (16.71 g).We observed a range of 58-91 for days to maturity, 1.14-9.28 for basal tiller number,2.57-9.17 mm for culm thickness,79.68-156.85 cm for plant height,4.19-9.16 for number of nodes,12.28-31.26 cm for flag leaf length, 1.38-3.02 cm for flag leaf width,6.86-14.42 cm for flag leaf sheath length,8.09-29.84 cm for peduncle length,from-0.47 to 19.0 cmfor panicle exertion,12.12-24.02 cmfor inflorescencelength,2.36-5.79 cm for inflorescence width,8.73-49.74 for raceme number,1.78-6.16 cm for lower raceme length,and 4.20-24.25 g for five-ear grain weight in the global core germplasm collection(Table 2).The normal maturity duration of released varieties of barnyard millet in this ecology is 80-90 days.While,we observed several accessions with less than 64 days to maturity(Table 5),which can be used in the breeding program for earliness in crop maturity.Most of these early accessions were of Japanese and Russian origin.We observed high variation for grain yield,also reported previously [21-23].The availability of genetic diversity in the core germplasm collection provides an opportunity to select best genotypes for different environments.

Table 2–Variance components of a barnyard millet core germplasm collection.

3.1.Principalcomponent analysis

Interrelationships among the different parameters were evaluated by principal component analysis(PCA).The first three PCA components provided a reasonable summary of the data and explained 73%of the total variation,and subsequent components contributed 5%or less(Table 3).The first principal component(PC1)was the most important and explained 45%of the total variation.PC1 was attributed to days to flowering,days to maturity,culm thickness,plant height,number of nodes,flag leaf length,flag leaf width, inflorescence length,raceme number,and five-ear grain weight for largest positive loadings.Basal tiller number,flag leaf sheath length,peduncle length,panicle exsertion,and lower raceme length had largest negative loadings.As a result,the first PC differentiated the accessions mainly by the contribution of high values for culm thickness,number of nodes,and raceme number.The second PC explained an additional 15.75%of the total variation and was attributed to positive loadings of plant height,flag leaf length,flag leaf sheath length,peduncle length,panicle exsertion,inflorescence length inflorescence width,and lower raceme length. The third PC,which explained 11.7%of the total variation, differentiated the accessions by higher numbers of basal tillers and narrow flag leaves and low levels of grain yield. Three fourths of the parameters occupied the right side of the biplot and one fourth were observed in the upper left side (Fig.1).

Three major groups were detected on the basis of projection of the accessions on the first two principal components.The majority of Indian and origin-unknown accessions were on the right-hand side in the biplot,whereas most of the Japanese accessions,namely,IEc 455,IEc 487,IEc 435,IEc 448,IEc 449,IEc 452,IEc 471,IEc 516,IEc 423,IEc 498,IEc 552,IEc 404,IEc 521,IEc 530,and PRJ1 were on the top left.Six accessions from Japan(IEc 436,IEc 519,IEc 517,IEc 511,IEc 537,and IEc 561),three from Russia(IEc 330,IEc 331,and IEc 338),and one each from Cameroon(IEc 624),Egypt(IEc 353),and the Syrian Arab Republic(IEc 346)constituted the third cluster(Fig.2).

Principalcomponentanalysis revealed that days to flowering, days to maturity,culmthickness,plant height,number of nodes, flag leaf length,flag leaf width,inflorescence length,raceme number,and five-ear grain weight contributed most to genetic diversity.A scatterplot of PC1 and PC2 showed overlapping of accessions of Indian origin with accessions of unknown origin along with one accession from Pakistan.A possible explanationfor this overlap may be that all accessions of unknown origin either originated in the Indian subcontinent or have similar morphologies.Given thatthe separation ofaccessions was based mainly on agro-morphological traits,all the Indian accessions along with accessions ofunknown origin and one accession each of Pakistan and Malawi grouped together.These accessions possibly belong to the E.frumentacea group.The Japanese accessions formed a second group,indicating them to be of the E.esculenta group.The third group contained a mix of accessions from Russia,Japan,Cameroon,and Egypt.The reason for the clear separation of Indian and Japanese accessions was their trait morphology and growth habitat.The accessions of Indian and unknown origin in the first group were characterized by thicker culms,taller plants,greater number ofnodes,longer flag leaves,longer inflorescence,greater number ofracemes,and late maturity.The Japanese accessions in the second group,in contrast,typically had longer flag leaf sheath,longer peduncles, high panicle exsertion,and early maturity.The third group comprised accessions with lower values for alltraits except basal tiller number.This finding matches previous observations that the races of E.colona do not correspond to geographic,ecological, or ethnological divisions,but are instead based on morphology [24].E.frumentacea accessions are generally of longer growth duration and are adapted to both temperate as well as tropical climatic conditions,whereas E.esculenta accessions are adapted to temperate ecology and perform poorly in tropical conditions. Our results ofthree different groups in the barnyard millet global core collection are in agreementwith results of Wallace et al.[25] who also obtained three groups using multidimensionalscaling and PCA in SNP data generated by genotyping by sequencing in the same core germplasm.The accessions in the third group were speculated to be results of seed contamination[25],but hybrids between these two species are known to be sterile[26]. We accordingly propose that the third group consists of accessions belonging to wild,weedy species of Echinochloa,given their higher basal tiller numbers,weak culms,low yield, and values for yield component traits.

Table 3–Principal component analysis based on morphological and agronomic traits of 94 barnyard millet accessions,and significant loadings(in bold)of the first three principalcomponents from mean data of two years.

Fig.1–Loading plot of PC1–PC2 for 94 barnyard millet genotypes.DTFF,days to 50%flowering;DTM,days to maturity;BTN, basaltiller number;CT,culm thickness;PH,plant height;NN,number of nodes;FLL,flag leaflength;FLW,flag leafwidth;FLSL, flag leaf sheath length;PL,peduncle length;PE,panicle exsertion;IL,inflorescence length;IW,inflorescence width;RN,raceme number;LRL,lower raceme length;FEGW,five-ear grain weight.

3.2.Relationship between traits

The correlation coefficients between traits are presented in Table 4.Character associations may be used to identify a few traits that are less relevant and could be of low priority in germplasm evaluation[19].The evaluation data provides a valuable opportunity for assessing relationships among traits to test the similarity between different groups[27].This practice simplifies work and saves resources.Association studies among different traits are important for barnyard millet breeders in effective selection of desirable genotypes. Of the 120 character associations estimated,five associations, namely peduncle length with panicle exsertion(0.925),days to maturity with days to flowering(0.852),number of nodes with raceme number(0.801),days to maturity with number of nodes(0.796)and culm thickness with raceme number(0.790) had high estimates,indicating that in future characterization of barnyard germplasm,tedious observations such as of panicle exsertion,number of nodes,and culm thickness may be avoided.Upadhyaya et al.[19]also observed a strong association of peduncle length with panicle exsertion in finger millet and emphasized recording observations of the easy trait,peduncle length,and not panicle exsertion.Gupta et al.[4]found a positive association of grain yield with raceme number and flag leaf width.In finger millet too,a positive association of finger number per ear with grain weight per ear was observed[20].These associations suggest that raceme number and flag leaf width will be effective selection indices for grain yield.In addition,culm thickness and flag leaf width were highly positively correlated with five-ear grain weight, revealing the roles of biomass and photosynthesis,respectively,in sink development.

Fig.2–Score plot of PC1–PC2 for 94 barnyard millet genotypes.

?

Fig.3–Two way hierarchical clustering of 94 barnyard millet core accessions.

3.3.Cluster analysis

Two-way cluster analysis separated the accessions as well as traits into two major groups(Fig.3).Group A contained 43 accessions,of which eight(IEc 786,IEc 701,IEc 706,IEc 788,IEc 731,IEc 722,IEc 747,and IEc 751)were of unknown origin,one each were from Japan(IEc 402)and Malawi(IEc 348)and the rest were Indian.This group was further subdivided into two groups.Group B contained 51 accessions of diverse origins. This group was subdivided into three subgroups(B1,B2,and B3)representing 27,10,and 14 accessions,respectively. Subgroup B1 contained one accession each from Pakistan (IEc 661),Japan(IEc 459),Cameroon(IEc 624),Egypt(IEc 353) and Syrian Arab Republic(IEc 346)and four accessions(IEc 264,IEc 725,IEc 758,IEc 699)of unknown origin,and the rest were Indian.Subgroups B2 and B3 contained all Japanese accessions except for three lines from Russia.

Two-way cluster analysis broadly separated the accessions based on trait variation.The first group A contained accessions with late maturity,high culm thickness,more nodes, taller plants,larger and broader flag leaves,larger inflorescences,more racemes,and high five-ear grain weight.This group contained all Indian and origin-unknown accessions, clearly indicating that Indian and origin-unknown accessions can be used as donors for these traits.The accessions in group B had more basal tillers,longer flag leaf sheaths,longer peduncles,high panicle exsertion,wider inflorescences,and longer lowest racemes.The subgrouping in cluster B also showed separation of accessions based on geographical origin.All the accessions from Japan and Russia were grouped together in subgroups B2 and B3,whereas B1 contained accessions of Indian,origin-unknown,Syrian Arab Republic, Egypt and Cameroon origin.This diversity may be due to migration of material from one region to another,and some ecological conditions could also influence the traits.The check genotypes VL 207(E.frumentacea)and PRJ-1(E.esculenta) occupied A1 and B3 clusters,respectively,clearly indicating the separation of the two different species of Echinochloa by hierarchical clustering.The results are in agreement with Wallace et al.[25]where the separation of two species was based on SNP data and the intermediate accessions were reported to be seed mixtures.But,in our opinion,the intermediate(based on trait values)accessions among both species are potential candidates for exploiting trait variation for genetic improvement of the crop.

Overall,there was a high level of genetic diversity of morphological and agronomic characters in the barnyard millet core collection.Gowda et al.[28]also observed high diversity in smallmillet germplasmcollections of ICRISATand identified trait-specific diverse lines in foxtail and finger millet.We too identified some promising trait donors (Table 5)which could be efficiently used in breeding programs for the improvement of this orphan crop.The PCA and cluster analyses provided a simplified classification ofbarnyard millet core accessions for use in breeding.Categorizing germplasm accessions into morphologically similar and presumably genetically similar groups is useful for selecting parents for crossing[15].Crossing accessions belonging to different clusters would maximize opportunities for transgressive segregation because of the higher probability that unrelated genotypes will contribute unique desirable alleles at multiple loci[8,9,29,30].Thus,the grouping of accessions by multivariate methods in the present study will be of practical value to barnyard millet breeders in allowing them to choose elite accessions from different clusters as parental lines for crossing programs.

Acknowledgments

We thank Dr.M.V.C.Gowda,Project Coordinator,AICSMIP,UAS, Bangalore,India for initiating and facilitating the movement of barnyard millet core germplasmfrom ICRISAT.Authors are alsograteful to Prof.James Nelson from Kansas State University, United States for English editing of the manuscript.

Table 5–Promising trait donors identified from mean data of barnyard millet core germplasm evaluation.

Supplementary material

Supplementary material to this article can be found online at http://dx.doi.org/10.1016/j.cj.2015.07.005.

R E F E R E N C E S

[1]T.Yabuno,Japanese Barnyard Millet(Echinochloa utilis, Poaceae)in Japan,Econ.Bot.41(1987)484-493.

[2]W.D.Clayton,S.A.Renvoise,Genera Graminium,in:T.A. Cope(Ed.),Grasses of the World,Royal Botanic Garden,Kew, UK 2009,pp.280-281.

[3]K.W.Hilu,Evidence from RAPD markers in the evolution of Echinochloa millets(Poaceae),Plant Syst.Evol.189(1994)247-257.

[4]A.Gupta,V.Mahajan,M.Kumar,H.S.Gupta,Biodiversity in the barnyard millet(Echinochloa frumentacea Link,Poaceae) germplasmin India,Genet.Resour.Crop.Evol.56(2009)883-889.

[5]R.K.Maikhuri,K.S.Rao,R.S.Semwal,Changing scenario of Himalayan agro-ecosystem:loss of agro-biodiversity an indicator of environment change in Central Himalaya,India, Environmentalist 21(2001)23-29.

[6]S.Padulosi,M.Bhag,S.Bala Ravi,J.Gowda,K.T.K.Gowda,G. Shanthakumar,N.Yenagi,M.Dutta,Food security and climate change:role of plant genetic resources of minor millets,Indian,J.Plant Genet.Resour.22(2009)1-16.

[7]K.E.Prasada Rao,M.H.Mengesha,V.G.Reddy,International use of a sorghum germplasm collection,in:A.H.D.Brown, D.R.Marshall,O.H.Frankel,J.T.Williams(Eds.),The Use of Plant Genetic Resources,Cambridge University Press, Cambridge 1989,pp.49-67.

[8]J.P.Peeters,J.A.Martinelli,Hierarchical cluster analysis as a tool to manage variation in germplasm collections,Theor. Appl.Genet.78(1989)42-48.

[9]S.H.Hussaini,M.M.Goodman,D.H.Timothy,Multivariate analysis and geographical distribution of the world collections of finger millet,Crop Sci.17(1977)257-263.

[10]A.Ayana,E.Bekele,Multivariate analysis of morphological variation in sorghum(Sorghum bicolor(L.)Moench) germplasm from Ethiopia and Eritrea,Genet.Resour.Crop. Evol.46(1999)273-284.

[11]A.Camussi,E.Ottaviano,T.Calinski,Z.Kaczmarek,Genetic distances based on quantitative traits,Genetics 111(1985) 945-962.

[12]N.M.Cowen,K.J.Frey,Relationships between three measures of genetic distance and breeding methods in oat(Avena sativa L.),Genome 29(1987)97-106.

[13]J.S.Brown,Principal component and cluster analyses of cotton cultivar variability across the U.S.Cotton Belt,Crop Sci.31(1991)915-922.

[14]M.C.Perry,M.S.McIntosh,Geographical patterns of variation in the USDA soybean germplasm collection:I.morphological traits,Crop Sci.31(1991)1350-1355.

[15]E.Souza,M.E.Sorrells,Relationships among 70 American oat germplasm:I.Cluster analysis using quantitative characters, Crop Sci.31(1991)599-605.

[16]H.D.Upadhyaya,S.L.Dwivedi,S.K.Singh,S.Singh,M. Vetriventhan,S.Sharma,Forming core collections in Barnyard,Kodo,and Little Millets using morpho-agronomic descriptors,Crop Sci.54(2014)2673-2682.

[17]S.Sood,R.K.Khulbe,A.Gupta,P.K.Agrawal,H.D.Upadhyaya, J.C.Bhatt,Barnyard millet-a potential food and feed crop of future,Plant Breed.134(2015)135-147.

[18]IPGRI(The International Board for Plant Genetic Resources),Echinochloa millet descriptors,Rome,http:// www.bioversityinternational.org/uploads/tx_news/ Echinochloa_millet_descriptors_394.pdf 1983(accessed April 3rd,2015.).

[19]H.D.Upadhyaya,C.L.L.Gowda,R.P.S.Pundir,V.G.Reddy,S. Singh,Development of core subset of finger millet germplasm using geographical origin and data on 14 quantitative traits,Genet.Resour.Crop.Evol.53(2006) 679-685.

[20]S.K.Sonnad,G.Santhakumar,P.M.Salimath,Genetic variability and character association studies in white ragi (Eleusine coracana Gaertn.),Karnataka,J.Agric.Sci.21(2008) 572-575.

[21]B.B.Bandyopadhyay,Genotypic differences in relation to climatic adaptation of two cultivated barnyard millet species at Garhwal Hills,Indian,J.Genet.Plant Breed.59(1999) 105-108.

[22]H.Mehta,P.C.Tyagi,K.P.Mohapatra,Genetic diversity in Barnyard millet(Echinochloa frumentacea Roxb.),Indian,J. Genet.Plant Breed.65(2005)293-295.

[23]V.Joshi,Assessment of Genetic Variability and identification of genotypes for different traits in Barnyard millet (Echinochola spp.),Int.J.Agric.Food Sci.Technol.4(2013) 65-67.

[24]K.E.Prasada Rao,J.M.J.De Wet,V.Gopal Reddy,M.H. Mengesha,Diversity in the smallmillets collection at ICRISAT,in:K.W.Riley,S.C.Gupta,A.Seetharam,J.N. Mushonga(Eds.),Advances in Small Millets,Oxford and IBH Publishing Co.Pvt.Ltd.,New Delhi,India 1993,pp.331-346.

[25]J.G.Wallace,H.D.Upadhyaya,M.Vetriventhan,E.S.Buckler, C.Tom Hash,P.Ramu,The genetic makeup of a global barnyard millet germplasm collection,Plant Genome 8(2014) http://dx.doi.org/10.3835/plantgenome 2014.10.0067.

[26]S.Sood,R.K.Khulbe,N.Saini,A.Gupta,P.K.Agrawal, Interspecific hybrid between Echinochloa esculenta(Japanese barnyard millet)and E.frumentacea(Indian barnyard millet)-a new avenue for genetic enhancement of barnyard millet,Electron.J.Plant Breed.5(2014)248-253.

[27]D.Z.Skinner,G.R.Bauchan,G.Auricht,S.Hughes,A method for the efficient management and utilization of large germplasm collection,Crop Sci.39(1999)1237-1242.

[28]C.L.L.Gowda,H.D.Upadhyaya,V.G.Reddy,S.Singh, Diversity in small millets germplasm and enhancing its use in crop improvement,http://www.intlcss.org/files/icss/ congress-proceedings/2008-papers/cs2-s1/cs2-s1-o2-laxmipathi-c-l-gowda.pdf 2008(accessed April 3rd,2015).

[29]S.C.Beer,J.Goffreda,T.D.Phillips,J.P.Murphy,M.E.Sorrells, Assessment of genetic variation in Avena sterilis using morphological traits,isozymes,and RFLPs,Crop Sci.33(1993) 1386-1393.

[30]K.S.Kanwal,R.M.Singh,J.Singh,R.B.Singh,Divergent gene pools in rice improvement,Theor.Appl.Genet.65(1983) 263-267.

16 April 2015

in revised form22 July 2015

.Tel.:+91 9411706285;fax:+91 5962241250.

E-mail address:salej1plp@gmail.com(S.Sood).

Peer review under responsibility of Crop Science Society of China and Institute of Crop Science,CAAS.