'.~ SHORTCOMMUNICATION PlêlntGenetic ResourcesNewsletter, 1999, No. 117:55-56

The core collection of maize germplasm of BrazilANVRAVE,

R. V.Tabare Abadie1, José R. Magalhaes2, Sidney N. Parentoni2,Celia Cordeiro2 and Ramiro V. de Andrade21 Facultad de Agronomía, Garzón 780, Montevideo, Uruguay. Email: tabadie@biagro.edu.uy (corresponding author).

Centro Nacional de Recursos Genéticos e Biotecnología-EMBRAPA, SAIN-70.770-900, Brasilia, DF, Brazil3 Centro Nacional de Pesquisa de Milho e Sorgo-EMBRAPA, C. Postal 151-35701-970, Sete Lagoas, MG, Brazil

SummaryThe core collection o, maizegermplasm o,BrazilA core collection of the Brazilian MaizeGermplasm Collection was establishedusing a two-Ievel sampling strategy.Three hundred accessions were selected,representing the genetic variability of thebase collection.

Keywords: Brazil, core collection,genetic variability, maize, Zea mays

RésuméLa collection noyaubrésilienne de germoplasmede maisUne collection noyau de Ia Collectionbrésilienne de germoplasme de mais aété établie en utilisant une stratégied' échantillonage à deux niveaux. Troiscents accessions ont été sélectionnées, quireprésentent Ia variabilité génétique deIa collection de base. .

Introduction.'-.../ A core collection consists of a set of accessions selected to

represent the genetic diversity of a base collection (Frankeland Brown 1984; Hamon et aI. 1995). This strategy wasintroduced with the intention of minimizing the cost ofgenetic conservation, while ensuring maximum geneticvariation. It also allows a rapid evaluation of germplasm,and a better access to the base collection. The BrazilianMaize Germplasm Collection consists of 2263 accessions,and is currently conserved at the Centro Nacional deRecursos Genéticos (CENARGEN-EMBRAP A, Brasilia,DF), and at the Centro Nacional de Pesquisa de Milho eSorgo (CNPMS-EMBRAPA, Sete Lagoas, MG). Most ofthese accessions have adequate passport data. Our objec-tive was to develop a Core Collection for the BrazilianMaize Germplasm Collection, in order to facilitate the useof maize genetic resources by the breeding programmes.This research was developed by an interdisciplinary groupof curators, breeders and statisticians.

Materiais and methodsThe development of a Core Collection is basically a samplingexercise. The two main problerns to solve are the size of thesample, and the sampling strategy to be used. In our case, 300

accessions was considered an adequate sample size, which canbe handled easily and with relatively low cost by the curators.In addition, it is 13% of the base collection, which is slightlyhigher than the 10% guideline recommended by Brown (1989).A two-level stratified sampling strategy was used. At

the first level, the accessions were classified into four strataaccording to the origin of the germplasm. These four stratawere: (a) landraces (1554 accessíons), (b) landrace-derivedcomposites (199 accessions), (c) improved materials (222accessions), and (d) introductions (288 accessions). Strata(b) Oandrace-derived composites), was not represented inthe core, because they are composites derived from themixture of landraces of strata (a), and so their inclusionwould have rendered an undesired redundancy. Each ofthe other three strata was represented in the Core Collec-tion following approximately a proportional approach (a:78%, b: 12%, c: 10%).A second level of classification was used within each of

the three main strata. The landraces were classified in 27groups based on ecogeographical origin and grain type(Table I), as suggested by Abadie et ai. (1997). Theecogeographical regions used in this case were the sameused by Cordeiro et ai. (1995) for the classification of Bra-zilian cassava accessions. Some accessions could not be

classified by ecogeographical origin be-cause of the lack of passport informa-tion. Nevertheless, observation of thegrains conserved in the bank allowedtheir classification by grain type. A loga-rithmic strategy was used to assign therepresentation of each of these groups inthe Core Collection. The logarithmicsampling strategy within these groupsavoided an over-representation of thedent grain types, which are higher innumber in the Base Collection, but showless variability than the other grain types(Abadie et aI. 1997). Within each group,the accessions were selected by the cura-tors, taking into account their experience

Table 1. Distribution of Brazilian landraces based on grain type andecogeographical region of origin, in the Base (BC) and the CoreCollections (CC)

Ecogeographic Graintype

region pop dent tlint

BC CC BC CC BC CCSouth 29 10 279 17 23 9Cerrados 26 10 321 19 77 13Cerrados North 12 8 110 14 9 7Amazonia 35 12 121 14 94 15Caatinga 17 8 169 16 38 11Agreste Litoral 1 1 62 12 14 8non-class 4 O 10 O 5 OTolallandracesin lhe core= 235.

11"1<1

tloury & otherBC CC5

506

191O7

512581OO

ResumenCo/ección básica degermoplasma deI maíz deIBrasilSe ha establecido una colección básica degermoplasma deI maíz brasileüo uti-lizando una estrategia de muestreo endos niveles. Se seleccionaron trescientasaccesiones como representativas de Iavariabilidad genética de Ia colecciónbásica.

"56 Plant Genetic Resources Newsletter, 1999, No. 117 .

and knowledge of the crop when possible, or otherwise atrandom. Of the original 27 groups of the landrace strata,only 24 were represented in the Core Collection (Table 1).The accessions not classified by ecogeographical originwere not included in the Core Collection because theywere felt to be represented by other accessions of thesame grain type in each of the ecogeographical origina.The improved materiais were classified as: pops, non-

pops from CNPMS, and non-pops from other breedingprogrammes. This classification was suggested from a Princi-pal Components Analysis done on the data of Feldman andSilva (1984), following the methodology used by Abadie et al.(1997) in the study of the landraces (Fig. 1). The introductionswere classified based on their origin as: tropical or tempera te,and within each of these groups on the four grain types (pop,flint, floury, dent), as suggested by the experience of the cura-tors and breeders. Differing from the approach used in thelandraces strata, the representation of each of the groups ofthese two strata to the Core Collection was proportional totheir size. This approach was selected because both curatorsand breeders considered that it represented the existing vari-ability. The members of the Core Collection belonging to eachof these groups were selected by the breeders, who tried tomaintain maximurn genetic variation and to include thoseaccessions representative of the main genetic pools used inthe breeding programmes.

Results and discussionA Core Collection of 300 accessions was obtained. Thecomposition of the Core Collection is shown in Table 2. Thedevelopment of this Maize Com Collection has resulted ina new classification of the Brazilian Maize Collection. This

PC39%

", ~.

,....

..... " ;,

,.

~

3

21

O

-1

-2-3

• CNPMSo Other+ Pop

-3

~3

21

O-1

-2-3

PC212% PC160%

Fig. 1. Spatial distribution of the improved accessions for the firstthree PrincipalComponents.

Table. 2. Accessions corresponding to each 01thestrata and groups within strata 01 the Core Collection

CoreCollectionstrata No.accessions

Landrace pop 48Landrace dent 90Landrace flint 63Landrace floury 34Improvedmaterial- pop 5Improvedmaterial- CNMS 15Improvedmaterial- other 15Introducedmaterial- tropical 20Introducedmaterial- temperate 10

classification was based on experimental results (Abadie etal. 1997), and on the knowledge and experience of thecurators and breeders. In addition, the classification wasdeveloped using available passport and characterizationdata, without the need of additional expensive experimen-tal work. The appeal of the new classification is both itssimplicity and its intuitively sound biological and practi-cal basis. We expect that this will interest breeders to usethe Brazilian Maize Germplasm in their programmes.The Core Collection itself is a subset representa tive of

the genetic variability of the base collection for which alarger amount of seed will be kept readily available for useby the breeders. This subset will be a logical and efficientstarting point for projects involving screening of the BaseCollection in the search for desirable traits.Additional characterization and evaluation of the Core

Collection will allow for refinements in the classification ofthe Base Collection, increasing the knowledge about thegenetic variation contained in it, and adding to the strate-gic value to the whole collection. Lastly, the Core Collec-tion can be the basis for further studies about the phyloge-netic origin of the maize crop in the region.

AcknowledgementTo Dr Emesto Paterniani for his valuable comments andsuggestions during the development of this research.

ReferencesAbadie, T., J.R. Magalhaes, C. Cordeiro, S. Parentoni and R. de

Andrade. 1997. Obtençao e tratamiento analítico de dados paraorganizar Coleçao Nuclear de Milho. EMBRAPA, ComunicadoTécnico NQ20, Novembro/97:1-8.

Brown, A.H.D. 1989. The case for core collections. Pp, 136-156 in TheUse of Plant Genetic Resources (A.H.D. Brown, O.H. Frankel, D.R.Marshall and J.R. Williams, eds.). Cambridge University Press,Cambridge, UK.

Cordeiro, C.M.T., E.A.V. Morales, P. Ferreira, D.MS. Rocha, I.R.S.Costa, A.c.c. Valois and S. Silva. 1995. Towards a Brazilian corecollection of cassava. Pp. 155-167 in Core Collections of PlantGenetic Resources (T. Hodgkin, A.H.D. Brown, T.J.L van Hintumand E.A.V. Morales, eds.). [ohn Wiley and Sons, New York.

Feldman, R. and J. Silva. 1984. Ca::;;talogo de germoplasma de milho,Zea mays L EMBRAPA CENARGEN, Brasilia DF.

Frankel, O.H. and A.H.D. Brown. 1984.Current plant genetic resources- a critical appraisal. Pp. 1-11 in Genetics: New Frontiers, Vol IV.Oxford and IBH Publishing Company, New Délhi, India.

Hamon, S., S. Dussert, M. Noirot, F. Anthony and T. Hodgkin. 1995.Core Collections - accomplishments and challenges, Plant Breed.Abst. 65(8):1125-1133.