Progressive Tree Improvement Program In Coastal Douglas-fir Roy R. Silen and Joseph G. Wheat ABSTRACT-Since it began in 1966, the "progressive" system of tree improvement has been applied on 18 million acres of the Douglas-fir (Pseudotsuga menziesii [Mirb.] Franco) region. It departs markedly from earlier practice, needing no grafted seed orchard. Its flexibility, low cost, minimal needs for skilled personnel, and large genetic base attracted landowners who were reluctant to invest in grafted orchards, which were then beset with many problems. The early adoption of a second phase, involving crossing of all parent trees and establishment of seedling seed orchards, has accelerated the region toward widespread use of second-generation seed about 15 years earlier than origi­ nally anticipated. Other than an initial research note (Sileo 1966), there has been no formal documentation and little pub­ licity covering the most widely used tree improvement program in the coastal Douglas-fir region. The pro­ gram was proposed by the senior author in 1966 and was accepted that year by the Industrial Forestry As­ sociation. Under the Association's sponsorship, most of the private forest owners in Oregon and Washington were gradually included. The U.S. Forest Service, the Bureau of Land Management, and the state of Oregon have incorporated its basic features into their pro­ grams. Today it is applied over nearly 18 million acres. This paper documents its features and development. A Typical Program The name "progressive" arises from incremental genetic gains with each succeeding seed crop as infor­ mation about parent trees improves. Commercial seed is initially collected directly from parent trees instead of from a grafted orchard. Main basis of genetic gain is from family selection based on a wind-pollinated prog­ eny test of all parent trees. Seed from a best parental fraction, usually the top one-fourth, is used for plant­ ing through the first 15 years. As test results improve, so does the choice of parents. A tandem part of the program quickly starts a second generation, in which progeny resulting from crossing the parents are used for a seedling seed orchard; the aim is to furnish seed with a second increment of improvement after 15 years. The wind-pollinated progeny test for evaluating the first-generation parents is also used to estimate performance of crosses for their combined contribu­ tion to the second-generation seed. For the third gen­ eration, the best individuals of best faQlilies of the wind-pollinated progenies are to be crossed to estab­ lish an orchard and field test. The three phases of a typical program can be 78/JoURNAL OF FORESTRY/February 1979 illustrated by an actual example. The first installation was near Vernonia, Oregon, in 1966 on 100,000 acres of the Crown Zellerbach Corporation; trees were 30 years old. In this example, Phase I describes activities associated with the parent trees and their seed. Phases II and III, subsequently described, refer in turn to succeeding generations. The landowner has the option of stopping in Phase I with an anticipated genetic gain in the 10-percent range. Alternatively, he can go into subsequent phases, each anticipated to provide addi­ tional gains of about 10 percent. The steps by year were: Phase 1-Seed from Parents Year 1966 Breeding activities Selection of 300 parent trees, collection of wind-pollinated seed for progeny test Spring 1%9 Outplanting of progeny of all parents on 9 test sites (300 families x 12 seedlings per family on each site) Winter 1971, Measurement of prog1976, 1981, eny at 9 test sites 1986 Seed production activities Collection of commer­ cial seed from same 300 parents First commercial planting of seedlings from selected parents Three genotypic up­ gradings of seed col­ lected from parent trees on the basis of test results Phase IT-Seed from Progeny Orchards (Note that activities overlapped in time with Phase I) Spring 1971 Single-pair mating of None original 300 parents (150+ crosses) Establishment of sec­ Spring 1974 None ond generation full-sib seed orchard from 1971 matings Winter 1977- Progressive roguing of None 79 poorest ~ of individu­ als in each family of second-generation or­ chard Winter 1980 Roguing poorest None (planned) ~ of orchard families Fail 1988 Final roguing to Initial collection of (planned) best 3 !.6 of families commercial seed from second-generation or­ chard Phase III-Seed Orchards from Progeny Crosses Phase III would establish a seedling orchard from crosses among progeny of the wind-pollinated test, with roguing planned over a 15-year period as in second-generation or­ chards. crossing of None Spring 1989 Field (planned) best performing trees in progeny test for third-generation or­ chard Spring 1990 Establishing field Establishment of third­ (planned) test of above crosses generation full-sib or­ chard 1990-2025 Measurements and As testing progresses, (planned) roguing as in second phase third-generation generation seed into use In more recent examples, the time frame has been compressed by using year-old, container-grown seed­ lings for progeny evaluations and starting the single­ pair matings earlier. If these alternatives had been used at Vernonia, the field crossing for the third­ generation seed orchard might have been possible by I982 instead of I989. Departures The program was developed because the only other system at the time was failing for Douglas-fir (Silen and Copes I972). Six innovations, designed specifi­ cally for conditions of the Douglas-fir region, were included.· I. A very localized breeding unit (generally under 150,000 acres). Unique features of the mountainous Douglas-fir region include unusually high growth rates of its species, despite a droughty summer climate and substantial mile-by-mile site variation. Such a combi­ nation places a major constraint on tree improvement. Douglas-fir, the omnipresent species, forms tall, uni­ form stands from sea level nearly to timberline. Asso­ ciated species, however, show that such uniformity is accomplished by unusual genetic diversity. For exam­ ple, Douglas-frr competes at sea level with Sitka spruce and at high elevation with mountain hemlock; but neither of these wide-ranging species could be in­ terchanged. Douglas-fir from sea level or high eleva­ tions also cannot be interchanged. In fact, there is evi­ dence now that genetic adaptation to local conditions in Douglas-fir is so precise (Campbell 1979) that each local population probably is genetically different in some respects from all others. Hence, to take advan­ tage of the balance of adaptive traits naturally evolved in the local population, genetic improvement in growth had best be limited at first to small breeding zones. These zones were initially defined as ecologically simi­ lar units under I50,000 acres in size and with a change in elevation of less than I ,000 feet. ll. A much larger parent tree base (usually three trees per 1,000 acres) than usual. The quantity of seed needed for sustained commercial planting required a large parental population. A large base also lessens the possibility that any strain of a native pest can over­ come a large fraction of our plantations. Of equal im­ portance, it minimizes inbreeding and allows con­ tinued selection in subsequent generations without re­ duction to a few vulnerable genotypes. Ill. Reliance on family selection for growth, as evaluated primarily by genetic testing rather than phenotypic selection. For some traits less affected by site variation, like resistance, straightness, or form, ample gains are expected from phenotypic selection. Selection for growth, however, placed major reliance on family performance in field tests rather than on phenotypic selection. There has been no discouragement to any phenotypic selection level the owner wished. Several landowners have used very intensive methods. Most have preferred moderate intensities and confined selection to the network of roads sampling their own­ ership. Extreme microsite differences, mountainous topog­ raphy, differences in within-stand tree age, competi­ tive and browsing differences at early ages, and ten­ dency for individuals to be uniquely adapted make use of comparison-tree or other selection methods very difficult in the Northwest. This difficulty is illustrated by the outcome of recent progeny-test data from two landowners' selections for volume production. Each used IOO groups of IO trees whose only qualification was production of \4 bushel of cones. Out of each IO trees, one was randomly chosen. In addition, the best and second-best of the remaining trees were chosen. Average diameters of the 8- and 10-year progeny in inches for the three groups were: Landowner I: 8-year progeny Landowner II: 10-year progeny Best Second best Random 0.823±0.007 0.826±0.006 0.826±0.006 1.405±0.012 1.398±0.010 1.402±0.010 Although greater benefits of selection may appear when the progeny attain parent age, at present each group is yielding about equal numbers of superior fami­ lies. IV. Use of wind-pollinated seed for testing (rather than seed offull-sib families). Use of wind-pollinated seed for progeny testing was still questionable in I966. The equivalency with an average of many outcrosses was, however, soon demonstrated. We have several studies in which height growth of wind-pollinated fami­ lies was highly correlated with the average height of a number of outcrosses made with the same set of parent trees. Many screening programs of parent trees throughout the world now employ wind-pollinated seed. V. Planting of progeny tests immediately to test each parent. Advantages of early test establishment are obvious in maximizing gain per unit of time. As tests on so large a scale were rare in the mid-I960s, guidance was sought from geneticists throughout the world. We were advised to apportion tests involving more than I00 parents into sets of 50 parents or less to be tested separately. Eight to twelve test sites were recommended to properly sample a breeding zone, the number also influenced by the difficulty of finding uni­ form sites more extensive than IO acres in mountain­ ous forestlands. A typical test of 300 parents would require I20 seedlings to test each parent. Use of 10 sites with three replications, each containing four ran­ domly distributed seedlings of the same family, is the most common pattern. Parents are ranked initially ac­ cording to performance over all sites. The accuracy of the ranking at any one site is of less concern than February 1979/JouRNAL oF FoRESTRvn9 adequate sampling for performance over the breeding unit. Subsequently, those that perform well for certain elevations or aspects, as determined from regression analysis, are used for special planting situations. VI. Immediate procurement of seed from phenotyp­ ically selected parents in the forest to meet needs of commercial planting (rather than seed from commer­ cial collections or later from grafted orchards). The need for commercial seed can be met from the outset from parent trees in the forest. Selection of the top one-fourth progresses from phenotypic to genotypic as the test proceeds. With equal selection intensity, the remaining distinc­ tion between parent trees in the forest and grafts in a seed orchard is the genetic disadvantage of knowing the contribution of only one parent. Knowing only one parent appears to have the potential effect of cutting usable additive variation in half. In early days of tree improvement, this disadvantage loomed so large as to cut out such wind-pollinated programs from further consideration. By 1966, with a failing Douglas-ftr seed orchard program, we were practically forced to give such programs another look. Once we did, we found at least seven attractions or simpliftcations of using wind-pollinated seed directly from forest trees: I. Seed of best wind-pollinated families col­ lected directly from the trees can provide genetic gains in the 10-percent range. 2. Such programs are quickest and least ex­ pensive to initiate and complete. 3. Such programs provide largest base popula­ tion per unit of cost. 4. Seed is immediately available in commercial quantity and roughly comparable in quality to seed from orchards. 5. No seed-orchard specialists are initially re­ quired. 6. Because capital outlays occur primarily by year 3, forestry staffs are quickly freed to con­ centrate on activities of the next generation of tree improvement. 7. Early crossing of all parents by single-pair matings provides seedlings for early establish­ ment of second-generation seed orchards, thus avoiding a 15-year delay in production of second-generation seed. (Single-pair matings be­ came part of the program in 1971. The rationale for early crossing of parent trees is discussed later.) Most of these attractions are obvious, or have be­ come widely accepted, but perhaps point four needs clarification. Advantage of genetic contribution from both parents is not immediate in orchard seed. The first orchard crops often receive their pollen mainly from trees surrounding the orchard, whereas all seed from forest parents are pollinated from locally adapted trees. Also, the early mix of seed from grafted or­ chards is mostly from a few productive and phenologi­ cally synchronous clones. The mix from forest trees contains reasonably uniform amounts of seed from Table 1. Cooperative tree improvement programs, Industrial Forestry Association and Pacific Northwest Forest and Range Experiment Station. Test established Programs 1 Species2 Vernonia Molalla Burnt Woods Umpqua Dallas Forks Snow Peak Tillamook Gold Beach OF OF OF OF OF OF OF WH OF Coquille Mapleton Cowlitz OF OF OF Cowlitz Snoqualmie Snoqualmie Skagit Roseburg NF OF NF OF OF Medford OF Oakridge Totals OF Cooperators OS, LF, CZ, IP CZ, PP, OS,LF OS, SF BL, IP, OS WI, BC, HF, IP A, CZ, MR TS, WI, CT, OS FS,OS,PP,BL FS, A, BL, CT, GP, SL, C BL, M, IP, CC FS, CT, 01 FS, BN, MP, CT, BC,M FS, BN, MP FS, BN, SA FS, BN, SA FS, SP, GP FS, BL, RL, CT, LF, SS, CD, OS FS, BL, MD, BC, JC, LF, OS, CM FS,PT 1Abbreviations: Program size M acres 326 103 69 582 141 72 172 130 Breeding units Year started Parent trees Sites Seedlings Area Acres 306 19 8 20 11 Number 91,800 40,716 15,456 218,752 48,840 27,520 46,560 47,520 928 570 450 33 13 0 121,368 57,680 0 219 102 0 1976 1976 1976 1976 1976 1,770 360 540 460 980 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 18 1978 5,330 0 0 0 914 124 20 2 1978 1978 4,164 495 0 0 6,205 74 0 0 173 0 0 1,578 1 2 1 5 2 1 2 1 1967 1968 1968 1969 1971 1971 1971 1974 900 375 161 1,709 441 215 388 270 299 214 345 3 2 2 1975 1975 1976 596 125 183 132 346 5 1 2 2 2 1,332 20,506 12 9 8 40 716,212 94 36 467 112 64 107 71 A =Agnew Timber Products; BC = Boise Cascade Corp.; BL = Bureau of Land Management; BN = Burlington Northern; C = Crook Estate; CC = Coos County; CD = C&D Lumber Company; CM = Cotton and Miller; CT = Champion Timberlands; CZ =Crown Zellerbach Corp.; Dl =Davidson Industries; FS =U.S. Forest Service; GP =Georgia-Pacific Corp.; HF =Hampton Tree Farms; IP = International Paper Co.; JC =Josephine Co.; LF =Longview Fibre Co.; M = Menasha; MD = Medford Corporation; MP = Murray Pacific; MR = Merrill & Ring; OS = Oregon State Forestry Dept.; PP = Publishers' Paper Co.; PT =Pope and Talbot; R = ITT-Rayonier; RL = Roseburg Lumber; SF= Starker Forests; SL = Simonson Lumber; SP = Scott Paper; SR = St. Regis; SS = Sun Studs, Inc.; TS = Timber Service; WI = Willamette Industries. 2DF = Douglas-fir, WH = western hemlock, NF = noble fir. SO/JouRNAL OF FoRESTRY/February 1979 Figure 1. Interior view of the special containerized greenhouses erected by the Industrial Forestry Associa­ tion to grow progeny for testing of parent trees. Over each parent. The advantage of the contribution of both parents to seed in Douglas-fir orchards does not occur much before 11 to 15 years. By this time, both pro­ grams should begin phasing into use of second­ generation seed. Thus, both kinds of seed have advan­ tages and both have shortcomings. A Regional Program Of the nearly 18 million acres in the program, about two-thirds are on large blocks of single ownerships, primarily U.S. Forest Service lands. About 6.2 million acres, however, are represented by intermingled own­ erships covered by cooperative programs. Early events shaped a pattern that was to become an or­ ganized regional activity. The first was the establishment of cooperatives (ta­ ble 1 ).. In 1966 the Crown Zellerbach program for northwest Oregon was viewed with high interest by neighboring landowners. Before the excellent 1968 seed year, the state of Oregon and Longview Fibre Company had decided to participate in similar pro­ grams. At the urging of the authors of this paper, a single plan for three ownerships was accepted. In northwest Oregon each organization owned about 100,000 acres of intermingled forestlands. By pooling plans, each contributed 300 parent trees to form a 900-parent base population. A fourth cooperator, In­ ternational Paper Company with 30,000 acres, joined too late to contribute parent trees. Each of the four developed three sites for a test of the 900 parents. 1,500 parental lots have been grown yearly since /972 and distributed to 173 outplanting sites. Under no circumstances could any of the owners have then carried such a comprehensive program alone. Yet costs were comparable since each owner was selecting the same number of trees and testing with the same number of progeny as before. The main difference with a 900-parent base was the assurance that not only would the landowner use the best of known parentage on his own lands, but also share in the best on his neighbors' land. Thus, at its inception the tree improvement cooperative became a hallmark of the progressive program. In the decade which fol­ lowed, 16 additional cooperatives were formed, each independent yet integrally meshed by common land ownerships and by uniform technical guidance from the two authors. A second early development was a regional center to grow progeny for testing. The first test seedlings were produced in nurseries. With the need to grow more than 5,000 parent-tree seed lots collected from the 1971 crop, the Industrial Forestry Association constructed greenhouses at NisquaUy, Washington, with annual capacity to grow 1,500 progeny lots in containers (fig. 1). The third event was the coordination of program guidance. For a typical program, the two present auTHE AUTHORs-Roy R. Silen is principal plant geneticist, USDA Forest Service, Pacific Northwest Forest and Range Experiment Station, Forestry Sciences Laboratory, Corvallis, Oregon. Joseph G. Wheat is director, Tree Improvement Labo­ ratory, Industrial Forestry Association, Olympia, Washington. February 1979/JoURNAL oF FoRESTRY/81 thors jointly propose a basic plan that spells out the cooperators' acreages, breeding units, division of par­ ent tree numbers, and test sites. The exact design of the test usually is an activity of the Pacific Northwest Forest and Range Experiment Station; advances in methodology are incorporated into each succeeding program. The Industrial Forestry Association provides coordination and guidance, and grows the test seedlings. Scientists of the Pacific Northwest Sta­ tion's genetics project have assured that best research information is applied. A considerable portion of the project's research arises out of field problems encoun­ tered by cooperators. The fourth development has been an informal re­ gional organization. As individual cooperatives began to meet regularly, we saved travel and time by setting aside certain weeks early in the year for-general meet­ ings. Problems of each cooperative are aired before all members, and newcomers receive advice and help. Unsolved problems are directed toward researchers who also attend. Outdoor meetings are held in early summer with visits to progeny outplanting sites or seed orchards. The result of these meetings has beena high level of uniform regional practice in planning, test seedling production, testing, measurement, and data processing. Despite its scope, complexity, and con­ tinuity, the organization has no formal name. A fifth development, the second-generation seed or­ chard, is discussed immediately below. The Second-Phase Program By 1971, the Vernonia Cooperative had reached a maintenance stage. The cooperators asked, "What next?" The outcome ofthis question was a simple plan that will produce next-generation improved seed 15 years earlier than originally programmed. A bumper cone crop was anticipated for 1971, though the financial climate was bleak. Only the most austere mating design for crossing of parents could have any hope of acceptance. The plan proposed by the senior author was to employ single-pair matings (Libby 1968, 1969}--e.g., parent AxB, Cx D, Ex F, etc., giving half as many crosses as the base parentage-and then plant the progeny as a seedling seed orchard. When the existing progeny test reaches 15 years, data cari presumably be used with confi­ dence. The seed orchard will then be rogued of all crosses except those representing parents in the top three-sixteenths of field performance, made up of crosses representing the best one-fourth of females and males crossed with the best one-half of the other sex. These winning families will be allowed to interpol­ linate for production of next generation seed. The novel feature was that the crosses were made in 1971 82/JoURNAL OF FORESTRY/February 1979 at random, before we had any information on parental rankings, so that both the test and the orchard could simultaneously mature. The members of the Vernonia Cooperative pro­ duced 499 crosses in 1971, the seedlings of which have been established in two seed orchards in the Willamette Valley (table 1). The initial planting, now 7 years old, is already producing some seed. Second­ generation seed production is expected to phase into the program before 1988. Molalla, Burnt Woods, and Umpqua cooperatives have since started second­ generation orchards on the same sites following the same plan. The crosses can be rogued accurately on the basis of field performance of progeny tests, but within-family roguing may not be as exact because this orchard is located outside the breeding zone. Two unusual developments have occurred. One is use of consulting forestry firms to make the pollina­ tions. Presently several such firms constitute an un­ usual regional asset, providing economical and readily available climbing and crossing skills. The other development has been an expanded role for the state of Oregon in providing a second­ generation seed orchard service for cooperatives. Large landowners, like Crown Zellerbach, Interna­ tional Paper Company, and the Bureau of Land Man­ agement, are setting up their own orchards. The state of Oregon participates directly in eight cooperatives. In order to provide the state's needs for second­ generation seed, a 400-acre seed orchard site was pro­ cured in the Willamette Valley near St. Paul (fig. 2). As development began, it became efficient to simply enlarge the acreage for each cooperative in order to supply seed for the other, usually smaller, landowners on a formal cost-sharing arrangement. Thus, orchards for part or all of the landowners in the Vernonia, Molalla, Burnt Woods, and Umpqua cooperatives have been developed on this site. Each cooperative breeding unit has agreed that ownerships too small to be members will be provided seed surplus to the mem­ bers' seed needs at cost. This second phase of the progressive program, like the first phase, is designed toward the minimal effort that will produce a satisfactory product. Like the ac­ cepted shortcomings of wind-pollinated seed in the first phase, the crosses that will furnish the seed for the second generation will not necessarily be those that would be made with more complete data on parentage performance in 1985. The best of present random mat­ ings will simply provide the best available seed for a decade until seed from better matings is produced. Even now landowners are beginning to supplement the basic plan with additional cross pollinations to assure progressive improvement. Looking Ahead The first grafted seed orchards of Douglas-fir were installed in 1957. By 1960 there were 16 in the region, totaling 145 acres. Thus, there was a decade of dis­ couraging experience with the more common method oftree improvement by 1966. Had plus-tree selection and grafted seed orchards been as inexpensive and convenient to apply as with most pine species, there would have been little reason for posing an alternative. Application of grafted orchards to Douglas-fir was neither, and the progressive approach was a welcome alternative. Several major landowners were deeply committed to grafted orchards before 1966, and have continued these programs. Most of these have also added progressive programs to expand base popula­ tions. For the uncommitted landowners, however, there had been reluctance to take on the heavy capital expense, and the known problems of dying grafts, pol­ len contamination, slow and unequal seed production, and costly specialized personnel. It was only after 1966, when there was an alterna­ tive, that most landowners committed themselves to tree improvement. The pivotal consideration for land­ owners was that they could carry such a program through the first phase with existing forestry staffs on low budgets and a limited commitment. Now, a decade later, the region can look back on an exciting experience. Many foresters and workers who were abruptly given the tasks of choosing hundreds of parent trees, establishing nurseries, and measuring complex test plantations felt this was their first inten­ sive forestry experience. Few had previously faced the challenge of bringing every planted test-seedling through alive. By early 1979, almost-% million tagged Figure 2. The second-generation seed orchard site main­ tained by the state of Oregon in the Willamette Valley. Oldest full-sib progeny are 6 years from seed. Orchards seedlings from 5,957 parent-trees were being tested on 173 sites, with this number certain to double when the region has another bumper seed crop. The huge ge­ netic base and small breeding zones are gradually being recognized as safeguards against possible loss of adaptiveness. The establishment phase is now almost completed as the last major ownerships west of the Cascade Range are forming into cooperatives, and with programs under way for two other major species: western hemlock and noble fir. It is now clear that initial concerns over whether to use grafted seed orchards or the ''progressive'' system were unjustified. This is because second-generation programs have turned out to be very similar, regard­ less of initial programs. In all programs, the best par­ ents or progeny in the testing are crossed and tested, and the progeny used in another seed orchard. In all programs, the sooner the crossing can be properly done, the sooner the initial program can be phased out in favor of seed populations from crosses of known performance. On a large scale, the progressive pro­ gram is reaching this time goal more quickly than most foresters expected. • Literature Cited CAMPBELL, R. K. 1979. Genecology of Douglas-fir in an Oregon Cascades watershed. Ecology (in process). LIBBY. W. J. 1968. Mating designs for second-generation selection in forest trees (abstr.). P. II in Proc. West. For. Genet. Assoc. Meet., Corvallis, Ore. LIBBY, W. J. 1969. Seedling versus vegetative orchards. Pap. to FAO-N.C. State For. Tree Improv. Train. Cent., Sch. of For. Resour., N.C. State Univ ., Raleigh, p. 306-316. StLEN, R. R. 1966. A simple, progressive tree improvement program for Douglas-fir. USDA For. Serv. Res. Note PNW-45, 13 p. SILEN, R. R., and D. L. COPES. 1972 Douglas-fir seed orchard-a progress report. J. For. 70:145-147. for partial or complete fulfillment of seed needs of three cooperatives are established. February 1979/JouRNAL oF FoRESTRY/83