D.P. Fowler
D.P. Fowler is Research Scientist with the Maritime Forest Research Centre, Canadian Forestry Service, Fredericton, New Brunswick, Canada.
Forest tree breeding was conceived in the latter part of the 19th century, born in the early 1900s, and developed into adolescence in the 1950s. It is now, as we enter the final quarter of the twentieth century, approaching maturity. The first World Consultation on Forest Tree Breeding, held in Sweden, could be characterized as having been hopeful; the second, held in the United States, was full of promises and predictions; now, at the Third Consultation, we can look back and see that the hopes, the promises, and the predictions of the earlier Consultations are being fulfilled. The number of active tree breeders has increased from a few score in the early 1950s to several hundred (Toda, 1974), and tree breeding has become a science. Genetically improved seeds and seedlings now provide a small, but significant, contribution to several afforestation programmes throughout the world.
Although severall breeding programmes are reaching maturity and in some instances work on second generation programmes has begun (Weir and Zobel, 1975), there are many more that are still in their infancy and some that are yet in their conceptual phase. Most of the relatively advanced breeding programmes deal with coniferous species, especially Pinus, and it is from these advanced programmes that much of our present information is derived. Only recently has a significant portion of our resources been directed to improvement of non-coniferous species. Angiosperms, especially the tropical ones, are now beginning to receive attention.
THREE-YEAR-OLD CANDIDATE FOR SECOND-GENERATION SELECTION creating an elite
The experience and information gained from the more advanced breeding programmes will undoubtedly prove useful for the development of new programmes with "genetically unknown" species Just how much of the information from programmes with coniferous species will be directly applicable to other genera and species is a rather moot question.
An understanding of reproductive biology and natural breeding systems is important for the development of breeding strategies and for determining the actual genetic gain attained in improvement programmes. (1977) emphasizes the importance of obtaining this information at an early stage in the development of improvement programmes. It is not enough to accept and utilize the strategies of other programmes for the improvement of new, genetically unknown species. For example, a strategy that works well for a generally outbreeding, highly variable species would be quite inappropriate for a species containing little variation or for a species in which selfing is an important means of reproduction.
Conifers. Over the past decades much has been learned about the reproductive biology and breeding systems of several important coniferous species. The reproductive biology of conifers is reasonably simple. They are essentially diploid, monoecious, and wind-pollinated, and have evolved only a few simple, but effective, mechanisms to control inbreeding (Koski, 1973). These mechanisms include spatial separation of male and female strobili and, in a few species, phenological differences in receptivity and pollen shedding. Self-incompatibility in conifers is dependent on the frequency of recessive lethal (or semi-lethal) genes within the individual tree genome and not upon highly sophisticated sterility gene systems (Hadders and Koski, 1975) which have evolved in many non-coniferous plants. Estimates of natural selfing for different conifers are surprisingly similar considering the large number of species and the varied conditions under which they grow. Although most coniferous species exhibit considerable tree-to-tree variation in self-compatibility, natural selfing is generally less than 10% under stand or plantation conditions.
The level of inbreeding (selfing, plus crossing of related trees) that occurs under natural conditions is still not well understood. Intuitively, it is widely accepted that within stands relationships do exist (Ledig, 1974), but it has only been recently that attempts have been made to quantify these relationships (Sakai et al., 1970; Sakai and Miyazaki, 1972; Tigerstedt, 1973; Coles and Fowler, 1976). Snyder (pers. comm., 1973) found that crossing neighbouring trees (less than 300 m apart) resulted in 2-14% reduction in seed and seedling characters. Coles and Fowler (1976) reported relationship coefficients of r = .30 and .23 for two stands of Picea glauca. Morgenstern (1972) reported relationship coefficients of r = .16 and .06 for Picea mariana growing in southern and northern Ontario, respectively.
The relationship among trees within populations has important implications for choosing the most effective selection methods (Ledig, 1974) and for determining actual genetic gain attained from a phenotypic selection programme. Several papers presented during this session of the Consultation have indicated strategy changes in which selection intensity will be reduced in the "wild" so that larger numbers of trees can be included in the programmer (Kellison and Dinus, 1977; Pederick and Griffin, 1977). Under these conditions, especially for species of limited natural range, an understanding of within-population relationships increases in importance.
A portion of the gain attained from clonal seed orchards may well be attributed to reduced inbreeding in orchards. The substantial gain in yield from unrogued seed orchards of loblolly pine over that of commercial checks (Kellison and Dinus, 1977) may include a fortuitous gain derived from reduced inbreeding compared to the level of inbreeding in natural stands, while the smaller gain reported for plantation origin Monterey pine in Australia (Pederick and Griffin, 1977) would not. One might argue that a gain is a gain, and in a sense this is true. On the other hand, if relationship coefficients had been known and taken into account, an even more effective breeding strategy might have been followed.
Non - coniferous species. With the exception of a few well-studied genera, e.g., Populus, Tectona, and Eucalyptus, information on the reproductive biology and breeding systems of angiosperms is considerably more limited than it is for conifers. What is known indicates that both these factors will be complex. Improvement of angiosperms will confront breeders with problems unknown by conifer breeders. Angiosperm breeders must contend with problems involving ploidy (Kim et al., 1977; Dancik, 1977), complex pollination mechanisms, highly evolved and more variable systems to control inbreeding (Tufuor, 1977), and quite possibly very different problems involving population structure.
The importance of basic information for the development of appropriate breeding strategies is well illustrated by Eldridge (1977), Wyk (1977), and Venkatesh and Vakshasya (1977) for Eucalyptus; Badran and El-Lakany (1977) for Casuarina; Wang (1977) for Pinus ponderosa; Toda (1977) for Larix leptolepis; and Tufuor (1977) for Terminalia ivorensis. It is imperative that a reasonable portion of the resources available to any new breeding programme involving a genetically unknown species be devoted to providing basic information on reproductive biology and breeding systems. This information will be essential for the development of optimal breeding strategies.
Following species trials and provenance trials most breeding programmes have concentrated on mass selection of phenotypically superior trees on the assumption that these trees are "above average" genetically. Working criteria for selection have been developed and are being utilized in many programmes, but these criteria differ greatly by species and by the objectives of specific programmes. In the more advanced breeding programmes, information is now available on the genetic control of most important attributes and it has been possible to modify selection criteria in some instances (Fletcher and Faulkner, 1972). Growth attributes continue to be important selection criteria, but emphasis has shifted in several programmes to selection for the more highly heritable features such as branch and stem form (Pederick and Griffin, 1977) and for disease resistance (Kellison and Dinus, 1977). Greater reliance is being placed on progeny testing for the improvement of growth and yield.
Comparison tree selection as developed by the Zobel school or some modification of this system is most widely used. However, base-line selection (Ledig, 1974) is gaining in popularity where the desired attributes are known to be under reasonable genetic control or where strong intra-population relationships are known or suspected. For several species, the rigorous selection criteria used to attain a large selection differential have been softened so that more good trees (rather than a few super trees) can be utilized. This shift to larger numbers has developed because of the relatively low heritability of growth attributes for "wild" selection (Fletcher and Faulkner, 1972), the need to maintain a broad genetic base to make possible a reasonable level improvement in the second and subsequent generation of breeding (Pederick and Griffin, 1977), and to allow for selection for attributes such as disease resistance (Kellison and Dinus, 1977; Palmberg, 1977) which may not have had high priority in the original selection criteria.
IRANIAN FORESTRY STUDENTS IN TRAINING stressing the practical
Progeny testing is an integral part of most improvement programmes. The tests are used as a basis for assessing breeding values and roguing clonal seed orchards, for the development of seedling seed orchards and as a means to provide materials for subsequent improvement efforts using inter- and intra-family selection. These tests can take many forms depending on species, improvement objectives, strategies, and are discussed in detail by Nanson (1974) and Wright (1976). Burdon and Shelbourne (1974) discuss the use of vegetative propagules for obtaining genetic information.
Clonal selection and testing have in the past been restricted to genera such as Populus and Salix which root readily from stem cuttings and exhibit a minimum of topophytic effects. Recent advances in vegetative propagation methods, including techniques for eliminating or reducing topophytic effects, improve the possibilities of clonal propagation as a mass production method (Libby, 1974). The possibility that tissue culture will provide even better means of mass producing clones is high (Brown, 1976; Durzan and Campbell, 1974). Several large-scale programmes involving "difficult-to-root". Species are presently under way, e.g., Picea abies (Kleinschmit, 1974; Lepistö, 1974), Pinus radiata (Shelbourne and Thulin, 1974).
Breeding strategies have been reduated or modified on the strength of information obtained from progeny tests. For example:
A
- Dominance variance is considerably greater than previously anticipated for the southern pines (Dorman and Zobel, 1973; Dorman and Squillace, 1974) and for Pinus radiata (Pederick and Griffin, 1977). Different breeding strategies such as the use of bi-clonal orchards or controlled pollinations will be required to capitalize on this variation.
B - Genotype × environment interactions, although often evident (Toda, 1977; Wang, 1977), do not pose insurmountable problems (Shelbourne, 1972). Most improvement programmes breed for widely adapted types rather than for types best adapted to specific environments (Kellison and Dinus, 1977).
C - Correlations between traits have generally been found to be favourable or not strongly expressed d (Barnes, 1977; Kellison and Dinus, 1977). An exception to this is a positive relationship between growth and gummosis in Acacia mearnsii (Nixon, 1977). The multi-trait criteria for selection used in most improvement programmes are satisfactory.
D - Trees with exceptionally good general combining ability have been identified in some programmes (Weir and Zobel, 1975) and support the need to utilize a pedigree breeding scheme if inbreeding is to be maintained at low levels in future generations.
E - Juvenile-mature correlations have generally been reasonably strong and provide encouragement for the development of early selection procedures.
Although most current planting programmes are still based on seed obtained from wild unselected populations, seed orchards are beginning to make a contribution to afforestation efforts with several species. Faulkner (1975) provides a good up-to-date review of all aspects and problems of seed orchards, including design, establishment, management, and protection. Seed orchards have been established in more than 40 countries for the production of seeds of approximately 50 different species (Feilberg and Segaard, 1975). First generation orchards associated with some of the more advanced breeding programmes are producing significant quantities of seed. For example, Pinus radiata orchards in Australasia provide 15-60% of seed requirements (Pederick and Griffin, 1977), and in the United States all P. elliottii and P. taeda seed will be available from orchards by 1990 (Kellison and Dinus, 1977). Dvorak (1977) reports that all Pinus caribaea seeds used in Fiji will come from orchards by 1986.
Grafted clonal orchards are by far the most widely used at the present time. Stock-scion graft incompatibility has proved to be a serious, but not insurmountable, problem with several species (Hong, 1975). Several techniques, including the use of rooted cuttings rather than grafts and selection of compatible stock have been used to successfully combat incompatibility. Seedling seed orchards which combine progeny testing and seed production are being more widely used, especially for precocious species or for species that are difficult to propagate vegetatively. Wright (1976) provides a detailed discussion of the advantages and disadvantages of seedling and clonal orchards.
A reasonable portion of the resources available to a breeding programme involving a genetically unknown species must be devoted to providing basic information on reproductive biology and breeding systems, essential for developing optimal breeding strategies.
Most seed orchards have been designed to accommodate wind-pollinated, relatively self-incompatible coniferous species. The suitability of these types of orchards for insect-pollinated or highly self-fertile species is questionable. Wyk (1977) suggests the planting of ramets of Eucalyptus grandis in hedge rows so that their branches will intermingle, or planting 23 ramets of different clones in the same planting site, as a more appropriate means of seed production for this species.
The conventional multi-clonal seed orchard is designed primarily to capture additive genetic variance. It has become evident that non-additive variance can also be important in some species (Pederick and Griffin, 1977; Kellison and Dinus, 1977) and that unique orchard designs will be required if non-additive variance is to be utilized, i.e., bi-clonal orchards, non-random distribution of clones in orchards or selective mass pollination (Pederick and Griffin, 1977). Selective mass pollination, especially during the early production years, may be required to combat background pollen or to reduce selfing, even in conventional orchards (Denison and Franklin, 1975; Furukoshi, 1977).
During the past several years methods for predicting genetic gain in tree species have developed a high degree of sophistication (Shelbourne, 1969, 1971; Namkoong, 1972; Wright, 1976) and estimates of gains for various selection and breeding strategies abound in the current literature. It is beyond the scope of this report to attempt more than a cursory review of the subject. It is enough to state that estimates have been obtained for many traits for most important species and that in most cases these estimates indicate that economically worthwhile gains are attainable. In fact in a number of improvement programmes, offspring from first-generation seed orchards have been tested against "commercial" checks to determine realized gains.
Rogued, first-generation orchards of Pinus taeda and P. elliottii have produced stock which, in operational plantings, is 10% to 20% greater in volume than commercial checks at six years (Weir and Zobel, 1975). Improved stock from first-generation orchards from five improvement programmes in Australia and New Zealand is - 10% to 53 % (X = 8-10%) greater in volume than commercial checks (Pederick and Griffin, 1977; Eldridge et al., 1977). Barnes (1977) found that polycross progenies of P. patula in Rhodesia produced volume increases of 17% at 5 years and 37% at 8 years over commercial checks and reported that comparable gains were obtained from P. elliottii and P. taeda in South Africa.
Of interest is the fact that all the above programmes were based on multi-trait selection schemes and that comparable gains in other traits, except for disease resistance (Kellison and Dinus, 1977), were also attained. The results from these tests are very encouraging in that the predicted gains are not out of line with realized gains.
A TEN-YEAR-OLD Pinus taeda ORCHARD IN THE UNITED STATES quality also requires quantity
Strategies for the development of long-term genetic improvement programmes have been discussed in detail by Burdon and Shelbourne (1971), Libby (1973), Namkoong (1974), and Weir and Zobel (1975). Although the proposed strategies differ quite markedly, they all recognize the need to separate the short-term seed production function from the long-term goal of developing and maintaining broad-based genetic populations for future advances in tree improvement. The proposed strategies also allow for the infusion of new genetic material into the breeding populations to broaden the genetic base, and thereby reduce inbreeding. Controlled crossing and complete pedigree records are essential if the greatest gains are to be achieved and inbreeding is to be avoided in advance generations (Zobel et al., 1972; Weir and Zobel, 1975).
It is recognized that a unified effort between countries or regions, working in the improvement of the same species, and the sharing of breeding materials offer many long-term advantages to a number of breeding programmes (Pederick and Griffin, 1977; Barnes, 1977; Muniswami, 1977; Dyson, 1977).
Although species hybridization has proved to be a viable improvement strategy for some genera, it is by no means the tree improvement panacea it was once considered to be. Brown (1972) provides a detailed review of the role of inter-specific hybridization in forest tree breeding. The role of species and provenance hybridization is also dealt with in detail in the proceedings of a symposium on that subject (Fowler and Yeatman, Eds., 1973). Brown (1972) concludes that hybridization is rarely a useful tool in tree improvement, which agrees with the general conclusion of the symposium.
The reasons that hybridization has failed to fulfil its promised goal are many - the most important being that there are usually better or more economical means of achieving the same goals. Heterosis or hybrid vigour in species and provenance hybrids is the exception rather than the rule (heterosis as used here is synonymous to Dobzhansky's (1964) "luxuriance" and does not imply greater fitness in the genetic sense). Genetic differentiation between species as well as between widely separated populations of the same species has usually arisen by evolutionary adaptation to local conditions (Falconer, 1975). The crossing of these species or populations disrupts the highly evolved, but different, adaptive systems and often results in hybrids that are less fit than either parent species. This reduced fitness is often expressed in reduced seed set, increased frequency of abnormal or inferior seedlings, and in reduced vigour.
A further point that often militates against hybridization as a viable alternative to other improvement approaches is the greater complexity of hybrid breeding strategies (Shelbourne, 1969) and the lower predictability of results as a result of a paucity of tested genetic theory relating to species hybrids (Stuber, 1970; Lester, 1973). The existence of strong maternal effects, as illustrated by the differences between reciprocal crosses of Eucalyptus camaldulensis and E. tereticornis (Venkatesh and Sharma, 1977), may be more pronounced in species hybrids.
Hybridization does create new gene combinations, some of which may be superior to those available within either parent species. Where these can be captured and economically utilized, hybridization offers a viable alternative to other improvement schemes. Vegetative propagation is one means of capturing exceptional gene combinations and it is in those tree genera which can be easily propagated vegetatively that species hybridization has had, and will have, its greatest impact. Species hybridization has proved to be a very successful approach to the improvement of Populus and Salix species (Zsuffa 1973). Both these genera contain many "crossable" species, most of which can be propagated vegetatively. Inter-specific hybridization does appear to hold considerable promise for Eucalyptus which has generally been considered to be "difficult-to-root." The successful rejuvenation of Eucalyptus hybrids forms strong sprouts and the direct use of these on a large scale in the Congo (Chaperon, 1977a, 1977b; Chaperon and Guillet, 1977) is most encouraging.
In recent years there has been a renewed interest in the use of vegetative propagation as a mass production technique with "difficult-to-root " species. If the problems of tree ageing can be overcome, as suggested by Libby (1974) and Kleinschmit (1977) and if more effective means of mass propagation, e.g. tissue culture, can be developed, hybridization can play an important role in the improvement of several other genera.
Well-documented examples of heterosis include:
Larix leptolepis
× L. decidua (several, see Brown, 1972)
Pinus elliottii × P. caribaea (Nikles et al., 1977)
Plantanus orientalis × P. occidentalis (Santamour, 1970)
Populus hybrids (Zsuffa, 1973)
Salix hybrids (Barrett and Rial Alberti, 1972)
Pinus nigra × P. densiflora (Wright et al., 1969; Vidakovic, 1974)
Eucalyptus hybrids (Chaperon, 1977)
Species hybridization has proved to be a useful tool for the combining of specific attributes such as disease resistance or cold hardiness with growth attributes of another species. Examples where hybridization is being used for the following purposes are:
RESISTANCE TO DISEASE
Castania dentata
× C. molissima (Diller and Clapper, 1969)
Pinus strobus × P. griffithii (Zsuffa, 1975)
Ulmus pumila × U. japonica (Lester, 1973)
Larix leptolepis × L. decidua (Brown, 1972)
INCREASED COLD HARDINESS
Fraxinus excelsior
× F. americana (Rohmeder, 1964)
Pinus rigida × P. taeda (Hyun, 1969)
Pinus attenuata × P. radiata (Griffin and Conkle, 1967)
Species hybridization has also been used successfully to determine phylogenetic relationships within tree genera (Critchfield, 1967; Wright, 1955).
Despite the successful use of species hybridization to bring together specific high-value attributes and some modest success in producing heterotic progenies, it must be concluded that intra-specific methods are more promising for the genetic improvement of most trees.
References are at the back of the magazine following the last article.
