Treasures in the
Attic: Finding the
Diversity Stored in the Maize
Genebank
CIMMYT and its partners have increased their efforts in "prebreeding"--accessing and refining raw diversity to make it breeder-friendly.
When CIMMYT's predecessor organization began work in the 1940s on improved maize for developing countries, its first step was to gather seed of diverse landraces from fields, markets, and farm households throughout Latin America. This seed was classified by race, the ecology where it was best adapted (that is, the lowland tropics or midaltitude, subtropical, or highland areas), grain type, and color. Each class was later used to form a genetic pool to which appropriate material from other sources—say, US or developing country breeding programs—was added.
Breeders from CIMMYT and partner organizations have drawn on these pools to develop hundreds of productive maize cultivars sown in the tropics and subtropics. "The pools are the foundation of our entire breeding program," says Suketoshi Taba, head of maize genetic resources and prebreeding at CIMMYT. "They link the enormous diversity of genebank seed collections to improved varieties, which return this diversity to farmers' fields in a more productive form."
Reinventing Gene Pools
Taba and his team have lately renovated the pools. They enrich them continually with genetic diversity from varied sourcesthe genebank, partners' breeding stocks, and collections from farmers, to name a few. "We see the gene pools as evolutionary maize populations for the future, a merging point for many useful maize genotypes," says Taba. "In the pools, potentially useful diversity gets refined and made available for advanced breeding."
The genebank is a valuable source of useful traits for pools, but with 23,000 or more registered seed collections—called accessions—it is also akin to grandma's attic: you need to look through many boxes to find its treasures. Taba's group has employed sophisticated statistical analysis and models to distill useful, accessible subsets from bank contents and farmers' seed. These "core subsets" are carefully chosen to embody most of a specific race's diversity and to feature useful traits, such as high yield or disease resistance. Core subsets are virtual groupings, rather than actual collections of seed. They are linked both to agronomic data and to records on the original accessions, so users can locate specific maize types or traits and, ultimately, the seed itself.
CIMMYT maize pools are genetically diverse, but their components still fall into discrete groups that researchers are trying to break up and mix together. The clustering shown here for genotypes in Pool 25--al tropical, late-maturing, yellow maize of flint kernel type--was done using data for important agronomic traits. The key gives the number of genotypes in each cluster. |
Taba and his team use the subsets as "samplers" of diversity, crossing them with elite inbred lines to identify genotypes that possess useful landrace genes without typical landrace weaknesses. The best products are added to the pools.
Breaking Ties that Bind
The group also works with the 32 pools to enhance desirable characteristics and weed out unwanted ones. "Yield, for instance, is not a dominant trait. It results from many recessive alleles—forms of the same gene—working together," Taba explains. "You're trying to gather the best alleles for each of maybe 30 or 50 traits, so that their small effects accumulate."
Because of the way genome segments are broken up and recombined in reproduction, genes that are nearer to each other on a chromosome are more likely to be passed on as a single block to succeeding generations; they are said to be "linked." As a result, in diversity's banquet, desirable qualities are often served along with unwelcome side dishes of inferior traits.
"Pools are composed of different race accessions, each with characteristic linkages—that's what makes them races," says Duncan Kirubi, CIMMYT adjunct scientist who has worked in prebreeding. "We try to break the normal linkages and create new ones that render useful traits more accessible to breeders." He and Taba apply statistical analyses that allow clear visualization of pool components (see figure). Those least alike genetically can be crossed to endow pools with new combinations that contain higher fractions of favorable traits. The researchers also break up close-knit subgroups to remix pool contents, and they are beginning to use DNA fingerprinting to assess and monitor diversity in pools. Finally, they have classified the pools into heterotic groups, which are pairings that can be used to develop productive hybrids.
Breeding Maize with Farmers
Taba and his associates are perfecting a method that combines in situ conservation and farmer participatory breeding of maize landraces, while enriching and taking advantage of gene pools. (In situ conservation of cultivated crop species, such as maize, is the conservation of genetic resources in farmers' fields rather than in genebanks.) According to Matthew Krakowsky, a postdoctoral fellow at CIMMYT, the first step is cataloguing the genotypes grown in a center of diversity for a particular landrace. "We analyze what farmers have, pinpoint the genotypes they want to improve, and cross them with our improved materials to enhance them," he says.
For example, in 1997 Taba and researchers from INIFAP, Mexico's national agricultural research program, began work with farmers in the Central Valleys of Oaxaca, where varieties bred by researchers have had little impact. They focused on improving Bolita, a drought-tolerant landrace that farmers especially appreciate for its tortilla-making quality. Initial efforts resulted in refined versions of key Bolita types, and farmers throughout the Central Valleys are purchasing Bolita seed.
The researchers will now take a selection of the best genotypes from the area and from Bolita core subsets developed with farmers, cross them with plants from improved pools, and cross the resulting progeny again with the original landrace samples. The first cross with pools will contribute improved traits; the final backcrossing to the landrace ensures conservation of the original landrace type—that is, the grain quality and appearance that farmers like. "This approach also gives us access to valuable traits from the landrace," says Krakowsky. According to Taba, similar methods may be perfected and extended to many landraces grown in Latin America.
 |
For more information: Suketoshi Taba (s.taba@cgiar.org) Matthew Krakowsky (m.krakowsky@cgiar.org) |
August, 2004