G. Namkoong
G. NAMKOONG is a geneticist with the U.S. Forest Service, Southeastern Forest Experiment Station, Genetics Department, North Carolina State University, Raleigh.
In our rapidly advancing field of forest genetics we have answered a lot of questions in recent years, but the answers have raised new and more difficult questions. We have learned enough techniques to be certain that we can develop new breeds. Now we need to look at what we have learned and decide how to organize breeding and gene management systems that will be the most beneficial to the future of mankind.
We have come to realize that the original gene pools are being physically lost. We have also begun to realize that unimproved gene pools will become economically useless as high yield breeds become incorporated into regular forestry practice. But with the power at hand to control the evolution of economically useful breeds, many questions need to be considered: in what directions should we develop future breeds? Should the nations or companies now bearing the cost of breeding programmes unilaterally direct the evolution of future tree breeds? How can we blend national and international objectives? How should we balance present with future obligations to people and forests?
The blending of national with international objectives is particularly difficult, and the complexity is increased by confusion about our exact obligations to future generations. It is clear that the design of a breeding system will be profoundly influenced by the inclusion of international objectives and future needs. It is not clear, however, how much weight the breeders of one nation should give to the needs of people in other places and times. According to Rawls' (1971) theory of justice, we should construct a system such that if a reasonable person is detached from any one particular place and time, he would judge that the costs and benefits of that system are fairly distributed. Then, the principle of equality dictates that we consider other people's needs as equal to our own and that we consider any time in the future to be as important as the present. If we accept something close to this idea, we must still ask who is going to pay the bills for other people and other times. Though all should bear some costs to reap some benefits, the poorest nations obviously cannot afford to match the richest nations in gene management. To support the principle of equality, the powerful nations and agencies will have to breed for other people's needs in addition to their own. And some method will have to be found to pay the bills. I believe that this can be done on utilitarian grounds.
A requirement that gene conservation and breeding efforts be efficient transcends geopolitical and time boundaries. We must also be constantly aware of technological and practical constraints as we consider long-term and often idealistic goals. In this report I will, therefore, first review our present technological status and gene management systems. I will, however, ignore some substantial current problems in production techniques and assume that sufficient seed or clonal propagates for reforestation can be produced from managed breeds.
The array of activities discussed in previous sessions is ample evidence of the importance, capacity, and economic feasibility of forest genetics technology. The purpose of this session is to discuss the ways in which our technology can be translated into breeding programmes under biological, economic, political, and personnel constraints. I am confident that improved breeds of trees can be produced under these constraints, but not without problems. I will first give some examples of excellent programmes that are under way.
POLLINATING SPRUCE IN CANADA searching for the best combinations
The comprehensive Douglas-fir programme of the British Columbia Forest Service has all the ingredients of a complete, self-contained breeding system. Mr. Heaman (1977) has gathered both the data and the plant materials necessary for immediate gains and for future shifts as needs for change become apparent. The main materials are selected from the best available provenances and crossed in small, disconnected, blocked, partial diallel patterns which now include 350 parents. As they are found, additional materials are tested for inclusion. Samples from wider provenances are now being tested, along with inter-provenance hybrids. A large initial base population has been established and organized so that single or multiple breeds can be developed. The utility of breeding for specific combining abilities and for creating synthetic varieties with high specific or general combining ability will be determined, and the data and material for creating these breeds will be available. Since the tests are conducted on a range of sites, it will be possible to determine the value of breeding for specific sites and using genotype × environment interactions. Depending upon test results, the breeding population can either be combined or split into multiple breeds. Finally, seeds will be produced from orchards containing the best available subset of materials at a given time. The present materials in these orchards were selected for general adaptability, but future selection for orchards may well be on a different basis.
This programme emphasizes the maintenance of options for future breeding. Such heavy investment in the future implies a faith that reasonable genetic gains can continue to be achieved in successive generations of seed orchards. Critical factors in justifying the investment include forest ownership patterns and public attitudes toward forests in British Columbia. Almost all of the forest land there is owned by the Provincial Government, and the people are conscious of both their forest heritage and their economic dependence on forests.
The Pinus radiata programme in New Zealand evolved under somewhat similar constraints and therefore has similar objectives. Burdon, Shelbourne, and Wilcox (1977) examined a wide range of alternative mating designs for various purposes including estimates of variances and combining abilities, development of breeding populations, and production of seed.
They observe that no single design is best for all purposes and no single purpose need be served by only one design. The actual mating design used in each generation is also partly determined by the time and costs of obtaining various types of breeding material. In their programme, the design used has changed from the first to the second generations. Their initial breeding material included about 600 seed parents. Their second generation material is obtained from their wind-pollinated offspring. For producing the third generation, crosses are currently being made in disconnected half diallels of five trees each.
Dr. Lindgren's (1977) work on mating designs suggests that reasonably good progeny tests can be made with a limited number of trees in any of several designs including common testers, partial diallels, poly crosses and open pollinations in seed orchards. Designs therefore can be chosen primarily for their usefulness in breed production or variance estimating.
In developing a broad breeding programme for Eucalyptus in Papua New Guinea, John Davidson (1977) was faced with different constraints. As a result, the breeding programme there is quite different. Short timber rotations and the achievement of sexual maturity at a low age for Eucalyptus permit full breeding generations to be cycled in five years. A large array of studies on genetic variation and breeding techniques therefore have already been completed in natural stands and in provenance and other plantings. Heritabilities for growth and wood quality traits have been estimated; and populations, families, and individuals promising substantial gain have already been selected and tested. Cutting and grafted seed orchards and seedling seed orchards are all being used, and their relative efficiencies are being compared.
Current seed production comes from 28-50 parents, but a new breeding programme involving 150 parents is being developed. It will include family and individual selection and single pair matings. In addition, provenance collections and other assorted materials will serve as a gene reservoir. Thus, in this fast-developing programme the value of improved varieties has already been demonstrated, provision has been made to satisfy immediate needs for improved seeds, and a population has been provided for future evolution.
The financing of this programme is largely external. Supporters of the programme may view some of the benefits as altruistic or political and some as indirect. The indirect benefits of establishing this programme would include gaining of information and experience that might prove useful in other programmes or with other species elsewhere. The programme may also be viewed as an attempt by one community to improve the wellbeing of a more inclusive community with other people and times.
Success in any breeding programme depends largely upon having the right data when you need it. Data problems exist... and they are more than a minor hindrance.
The benefits of international cooperation in breeding programmes are sometimes both immediate and direct. Burley and Nikles (1977) describe the organization of an international breeding programme for tropical pines. A network of breeding populations has been established in which increases in yield are being obtained at various locations. The size of the genotype × environment interaction is the key factor in determining whether to develop special locally adapted breeds or more generally adapted ones.
There is ample evidence that immediate gains can be achieved with only preliminary data and simple breeding techniques from breeding populations of various sizes. We also know that, by increasing population size and complexity, provision can be made for future uses. With some further increase in complexity, a programme can also become international in scope. Since various forms of international cooperation already exist, it seems clear that there are advantages to this approach.
Success in any breeding programme depends largely upon having the right data when you need it. As the complexity of the programme increases, so does the complexity of the data collection, handling, and storage. Data problems exist, and they are more than a minor hindrance.
Data recording and storage for future use can be a weak link in the breeding process, particularly when information is needed about juvenile-mature correlations. While the inevitable missing data and irregular format problems can be handled by subroutines, the necessary analyses may be too difficult for the typical tree breeder. Furthermore, since forestry experiments are costly and time consuming, all forestry data are expensive. Hence, we must make best use of the data we have through analytical procedures of maximal efficiency. New research on estimation procedures such as quadratic estimators would be important for forestry experiments.
In multiple-trait selection, data problems are especially critical. Franklin and Stonecypher (1977) show that use of variances and covariances in index selection is important for maximizing gain. But they also show that estimation errors are often damagingly high. One dilemma is that trait correlations which affect gain cannot be ignored but that these correlations can only be approximated with uncertain error.
Despite these difficulties, Burdon, Shelbourne, and Wilcox (1977) feel that in Pinus radiata, long term pedigreed breeding populations promise enough genetic gains to warrant the effort involved. Selection at young ages is also considered to be a worthwhile attempt for accelerating the genetic gain rate. They propose a hierarchy of populations - one for seed production, a larger one for breeding, and a still larger one as a gene pool. Advances in the breeding population are achieved through recurrent selection, and nonrecurring bonus gains are obtained by secondary selection in the seed production population. This programme is similar to that proposed by Namkoong et al., (1971) with the addition of an unimproved gene pool population. They also suggest that the breeding population be split into sublines which may undergo inbreeding but can be intercrossed for seed production. This proposal is similar to that of Baker and Curnow (1969), except that the latter authors recommend selective elimination of some sublines.
If it is clear that we can change breeding populations and their derived seed populations in ways that appear useful, it is also clear that to achieve these gains requires important choices. The breeder must compare costs with benefits and, perhaps more importantly, he must limit population size. In a sense, when a breeder decides not to improve a particular species he chooses a gene pool, a population size, and a set of selection coefficients - he accepts the values that accidentally occur in nature. When he decides to breed, he must choose a reduced size for the population and gene pool; these reductions are achieved by selecting for some specific combination of traits. He might allow the desire for immediate gains to cause him to reduce the gene pool so much that the remainder is undesirable in some way. Intensive selection in a single population down to less than eight parents maximizes the selection differential and expected gain for this generation at a large expense in genetic variations for gains in future generations. Less intensive selection, with larger population sizes, reduces present gains in seed populations but leaves more opportunities for future gains. The breeder can make such choices less painful through careful selection of breeding designs, but difficult choices will always remain.
The hierarchial breeding system recommended by Burdon, Shelbourne, and Wilcox (1977) is an attractive way to resolve the conflict between present and future gains. In this system the gain achieved in the seed population does not affect the long-term gain potential of the breeding population, which is less intensively selected. Furthermore, advances in the breeding population are backed up by a reservoir in an unselected gene pool population. If the system has a weakness it is the present and continuing cost of maintaining this unimproved gene pool population which may never be used and can be used only with great difficulty. The system's large breeding population implies relatively slow present gains. It is felt that the reductions in present gains will be offset by increases in future gains. Such a choice seems justified in many cases.
We and our inheritors all inhabit one planet with a common array of species to develop. The need for joist programmes on gene-pool evolution is obvious.
An alternative system of breeding multiple populations may also prove feasible. Dividing the breeding population into subunits can achieve more gain than selection with comparable parable numbers in a single breeding population (Baker and Curnow, 1969). When the future is uncertain, selecting in several subunits for different trait combinations or site conditions can give greater expected gain than single-population selection (Namkoong, 1976). Deployment of multiple populations in several directions around combinations of high expected value can give high immediate gains, guard against selection in an improper direction, and provide for future population flexibility and variation. This strategy is similar to those adopted by experimental designers who seek to place sample points near critical or extreme points. Samples are deployed to assure that the critical points are well encircled. In designing the future evolution of forest tree species, we should consider such multiple deployment of small breeding subpopulations.
If we must strike a balance between present and future gains, we must also strike one between national and international objectives. Nations bear unequal burdens and reap unequal benefits, and regardless of national aspirations, present and future generations may unequally share the costs and profits. Selection for a forest management system profitable to one agency may not be best for another. Just as we must accept reduced immediate progress to ensure future gains, we must make some sacrifices to accommodate conflicting aims. Hierarchical breeding systems have the simplest organization for such accommodations, but multiple breeding subunits are also a possibility. The multiple approach promises more security of gains but is harder to manage. The optimal system will undoubtedly vary by species or species group.
Since we and our inheritors all inhabit one planet with a common array of species to develop for greater or lesser use, the need for joint breed management programmes seems obvious. If we can afford joint programmes on gene pool conservation, we can surely afford joint programmes on gene pool evolution to manage them for diverse goals. The work of the Commonwealth Forestry Institute and the many bilateral efforts in breed development should be continued and supported. To assure international cooperation in complex gene management systems, FAO and IUFRO might assume at least planning responsibility. In such systems, intercrossing and overall population deployment could be managed internationally, and individual nations and agencies could develop special programmes to satisfy immediate needs from the hierarchical sub-breeds or from semi-independent multiple breeds. Short-term breeding goals could then be accommodated in long-term programmes with international objectives. The experience of IUFRO in cooperative work on species breeding and gene conservation might be extended to a more general programme under FAO auspices.
References are at the back of the magazine, following the last article.
