Gene bank math

Two decades ago, CIMMYT scientist Jose Crossa began to apply sophisticated statistics and population genetics to the management of seed collections in gene banks. Now, the methodologies developed at CIMMYT for the maintenance and classification of genetic resources are used throughout the world.

Germplasm banks (also called gene banks) often seem like museums or bank vaults, keeping precious treasures locked away for the next generation. Banks like the Wellhausen-Anderson Plant Genetic Resources Center at CIMMYT are certainly built to withstand disasters such as earthquakes, hurricanes, and power failures. But there’s much more to it than security. Germplasm banks are really more like a zoo, and a batch of stored seeds is more like a cage full of monkeys than the Mona Lisa. Zoo animals require constant maintenance. Old animals die and have to be replaced, often by careful breeding to ensure that the genetic diversity of a captive population is maintained.

The scientists who manage collections of genetic resources face similar challenges. Even at the very low temperatures and humidity levels used in germplasm banks, seeds can’t be stored indefinitely. Over time, they lose their ability to germinate: how fast this happens depends on species and storage conditions, but all the seeds in a collection will eventually become useless for breeders or farmers. Before this happens they have to be planted, grown to maturity, and a new generation of seeds harvested for return to the bank. CIMMYT’s seeds are monitored every 5-10 years, and scheduled for regeneration when their viability drops below 80%.

This process is relatively straightforward in self-pollinating crops like wheat, where the offspring are genetic copies of the parent. But it is much trickier in cross-pollinated crops like maize, where the offspring have a jumbled-up mixture of the parents’ genes. A single maize ‘accession’—a batch of seeds from a single variety or race of maize—contains seeds with quite different combinations of genes. When these genes are recombined in the next generation, the risk is that some, rarer, genes will be lost.

Twenty years ago, Crossa had recently arrived at CIMMYT as a biostatistician, when colleague Suketoshi Taba, now Head of the Maize Germplasm Collection, approached him with a problem. He wanted to know the best way to regenerate accessions to retain a high level of genetic diversity, including how to work out how many seeds to plant, and how to manage the pollination process. “I had no idea,” says Crossa, “but I started looking into it. I really wanted to work on genetic resources because I knew how valuable they were. I’d been working in the US, and everyone spoke about how unique and important CIMMYT’s collection was.” And so began a long and fruitful collaboration.

Crossa realized that ideas from population genetics held the key, but these were not being applied to genetic resources. The crucial concept was effective population size (EPS), a measure of the number of parents that contribute to the next generation. A larger sample is likely to contain more of the population’s genetic diversity, so the progeny are likely to represent the original population better. Therefore the number of seeds planted for regeneration should be as large as possible—but in reality this is limited by the capacity and funding available.

However, the effective population size of the parent sample can also be maximized by carefully controlling the regeneration process so that each parent contributes equally to the progeny. The plants are not allowed to cross-pollinate freely. Instead, crossing is done by hand, with each plant being used to pollinate one other, ensuring that the male reproductive cells or “gametes” (i.e. pollen) from each plant are represented equally. So that the contributions of female gametes are also equal, a fixed number of seeds is taken from each plant, rather than simply harvesting all the seeds.

The models developed by Crossa and his colleagues allow scientists to make informed trade-offs between genetic diversity and regeneration costs. For example, to have a high probability of retaining a gene variation that occurs in 3% of the population, making it fairly rare, an EPS of 200 is needed. Using systematic regeneration to maximize EPS, this means planting around 250 seeds, to allow for some regeneration failure.

“I think the work has made a difference,” says Crossa. “People used to use much smaller samples, but now they are more aware of the genetic erosion caused by not using appropriate sample sizes, and the need to control male and female gametes.” The methodologies developed at CIMMYT have shown scientists around the world how genetic diversity can be managed successfully, and are used to ensure the preservation of many national and international collections. The team’s models have been extended to species with any degree of self- and cross-fertilization—even wheat, since in reality accessions are never completely homogeneous.

Crossa continues to apply the tools of statistics and population genetics to the field of genetic resources. His team has done a great deal of work on core collections, small subsets of accessions that represent as much as possible of collections’ overall genetic diversity. In the case of maize, they have grouped farmer landraces into racial groups, and generated core collections for each one. These allow researchers to study a few tens of accessions rather than trying to select from the thousands available. The team has developed ways to combine a large amount of data in order to select the most varied subsets, including data from molecular markers and data on physical traits, both quantitative and qualitative. This way of organizing and combining many types of data is now being applied as a valuable tool for selection for plant breeding. Crossa has also worked on the challenges faced by researchers out in the field collecting samples for germplasm banks, developing methodologies to efficiently capture the genetic diversity of farmers’ crops.

“When Taba first asked me about seed regeneration I knew nothing about it, and there wasn’t much work in the area, so the challenge really appealed to me,” says Crossa. “Twenty years later I’m still happy I can make my contribution to preserving genetic diversity. There aren’t many people working in this field—because, although each gene bank is extremely important, numerically there really aren’t very many—so every advance we make has a big impact.”

For more information: Jose Crossa, Head, Biometrics and Statistics (j.crossa@cgiar.org)