“MI-I. .TIILI \‘I' III}? 1‘52“ MAI; I4 I. .. III; I} .V‘I I\‘ l M £3“ IVII '3 “ III I. l'; u. ' . ‘IIIIIL'I, QI H: II III “I III III II, 1.‘I- II IL“. “VIZ-.1; I' .I' I'I III I I°I|I I II I I I” III(' I I I I _I6!II III:KI.I: ’ II’I( {II I W; *6” lIIIksIIIIII IIIIIII IyII III-WEI" IIII “ A .I_i "T"; ‘IIII‘I-I‘, MINI. Il- . ’1'.) 7 H I', ‘ ‘. A IIIIII IéIIIIIH II). \'.' ”I "J". ‘I: ”HIM-l I . I ' I.’ _ I'I‘ I‘YI'I, p'W‘I .3 IIIIIZ’iI . . W I‘ I'Ig.I;.-.I, III (I I I 'I' ' l I l - I ,IIIl. lv .III',I “I .“.!,I I. . I r I ‘ y IIII I I ";v‘1r‘ll "H!“ ' ’. a "\II I . . ‘ '.' .| . I' D II I ... .VI. . ' 0| ‘ I ...... f WI: 4“?“ I I 0 {A}? LII| ., . . “I .. ' I ' ' IIIIIIIIIIIIIII‘I ' 2-. *II' I .3315 III» “I“ ~ Q I" .‘“',I'I..I: ‘ H.” . . ,tIlmI'I‘” I‘IIIII III: “.IIIIII“ “It? I'J“. “3:11 lI' I .: 'I I ‘I :12”) I (I'm IIIIII III'I'I 33 .I I". II' I II' IIIIII .. I II II... 3 - I ’ I" ~39. .' ..i.‘,IsI;s. . III! ., ~ II"r" "If”. . " “'I,'I.' I .1“ II 'I" I’I'f" .33.: II 'I-III I’IIII'II II " I ' 'I'~ I II III I I IIIIIIIII III-III I: l :l‘ ‘ I. . I“! . ' | [I .II . ‘ID 0 I | I IJ.. . I. . . I”... *1. I, .II I). :"lI. ' I. I '.II ‘. IIIISII I “I I I - ‘I I. a I I; If“ fl'II'III'I“? III,I III II 'II-IIIII I III" IIII'CII “I :’.'I'.. -.IIfII=£Ii.; '. I” . .-. III?- I II- ' I' I I " III ‘ I¢|III . | 7'. 4' I:--. . III II IIIIIII " II III . "I'IJII IIIII IIII " IIR'IIIL IIIIIII III I II? MI: .I‘ I; III. - I;I,.-.II‘; I“ . 1.153 .. Y ‘* Afichigan State University W” This is to certify that the thesis entitled EARLY RESULTS OF PROVENANCE STUDIES OF LOBLOLLY AND SLASH PINES IN BRAZIL presented by Antonio Jose de Araujo has been accepted towards fulfillment of the requirements for Ph . D . figgree in ForEStry " vim / Major professor Date August 21,1 0-7 639 .‘ OVERDUE FINES: 3-1.5 ‘ _ ’ ‘ 25¢ per day per item : (fix. j RETURNING LIBRARY mum: '34:.” 5'}! r ‘( Place in book return to ream “MI/r -, charge from circulation recor ‘, S" EARLY RESULTS OF PROVENANCE STUDIES OF LOBLOLLY AND SLASH PINES IN BRAZIL By Antonio Jose de Araujo A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Forestry 1980 ABSTRACT EARLY RESULTS OF PROVENANCE STUDIES OF LOBLOLLY AND SLASH PINES IN BRAZIL By Antonio Jose de Araujo The objectives of this study were to determine if there are: (l) genetic differences in growth traits among provenances, (2) any trends of significance for locating the most suitable provenances, and (3) any genetic-environment interaction. Additional objectives were to determine (4) future uses of the trials, (5) improvements of the experimental procedures, and (6) future direction of the experi- mentation aiming at the genetic improvement of these two pines. Thirty-five seedlots of loblolly pine and 20 seedlots of slash pine were tested in two series. Randomized and replicated trials were established in four locations. Mortality was higher in Series 1 for both species due to long-distance transportation of bare root seedlings. There were very important differences in loblolly pine growth rate due to provenance in both series. One seedlot from Livingston Parish was consistently among the leaders at all test locations. Other seedlots from Florida, Mississippi, South Carolina, and one Brazilian seedlot were also fast growing. Genotype-environment inter- action were strong in Series l and smaller in Series 2. A general Antonio Jose de Araujo trend of coastal seedlots to grow faster than interior ones was observed in both series. The most highly recommended area for seed collection for plant- ing in Brazil is the coastal region stretching 120 kilometers from the coast of the Gulf of Mexico and the Atlantic Ocean in seven southern states. The consistent superiority of Livingston Parish indicates that Brazilian growers can obtain immediate improvement by using seed from that area. There were significant differences in slash pine growth rates in both series. Differences were smaller than those observed for loblolly pine. In both series there were some best seedlots. Sizable differ- ences occurred among stands close together geographically. No clear geographical trend or pattern was found. Genotype-environment inter- actions were strong in both series. No region in southern United States can be generally recommended fbr seed collection for planting in Brazil. Only individual stands can be suggested as result of their good performance. The close correla- tion of my results and American results indicates that genetically improved seeds from the United States can be used in Brazil for both loblolly and slash pines. The studied trials should have three major uses in the future: (1) as producers of information, (2) for demonstration purposes, and (3) as breeding arboreta. A fourth possible use as seed orchards is not recommended. All Nelder-Bleasdale systematic spacing designs should be measured and analyzed in the future. They may or may not Antonio Jose de Araujo give any useful genetic information. Valuable information about spacing can be learned from these systematic grids. Future trials should follow a different approach, including hundreds of seedlots and sampling several families from each of many stands. Further improvement can be pursued by using already improved American material or superior provenances or stands in the natural range. Seeds from American seed orchards can be used to establish test plantations for future selections. Livingston Parish and sur- rounding parts of southeastern Louisiana are a natural choice for intensive sampling trials of loblolly pine. In slash pine such a region of superiority has not been detected. The sending of a Brazilian team for seed collection in southeastern Louisiana is neces- sary for an intensive sampling of stands. Future genetic research in Brazil should take advantage of some vital improvements in the experimental procedures such as small plots, a greater number of replications, and a greater number of locations. Future intensive sampling trials should be established in such a way that the test plantations can be converted into seed orchards. Dedicated to my parents, Antonio Daniel and Auracelia; my wife, Michiko; and my sons, Alexandre and Daniel, in recognition of their sacrifices, patience, and love. ii ACKNOWLEDGMENTS I express my sincere gratitude to Dr. Jonathan w. Wright, my major professor, for his wise guidance, assistance, and encouragement throughout my doctoral program. I am grateful to Drs. Victor J. Rudolph, Donald I. Dickmann, and Peter G. Murphy for their helpful comments and suggestions as members of my Guidance Committee. Appre- ciation is extended to Drs. Dale E. Linvill and Charles E. Cress, who also served on my Committee. Thanks are due to the Instituto Brasileiro de Desenvolvimento Florestal (IBDF) and Empresa Brasileira de Pesquisa Agropecuéria (EMBRAPA). The Regional Unit of Forestry Research at Colombo, PR, provided effective assistance in my research through Mr. Jarbas Y. Shimizu. Special thanks are due to the Brazilian Ministry of Education and Culture for providing me a fellowship through the Programa da Educacfio Agricola Superior (PEAS) and Coordenacao do Aperfeicoamento de Pessoal de Nivel Superior (CAPES). I am also thankful to Univer- sidade Federal do Parana for granting me the necessary educational leave. Very special acknowledgments to my wife, Michiko, for her under- standing, confidence, and very definite help during the course of my research. iii TABLE OF CONTENTS Page LIST OF TABLES ........................ vi LIST OF FIGURES ........................ ix Chapter I. INTRODUCTION ..................... l 11. OBJECTIVES ...................... 6 Objectives of the PRODEPEF Provenance Trials . . . . 7 Objectives of the Present Study ........... 8 III. MATERIALS AND METHODS ................. 9 Personnel ...................... 9 Seed Procurement .................. lO Germination Tests and Stratification ........ ll Handling in the Nursery ............... 13 Nursery Measurements and Analysis .......... l4 Outplanting Procedure ................ l5 Plantation Measurements ............... 20 Analysis ...................... 2l IV. LOBLOLLY PINE PROVENANCE TRIALS ............ 23 The Species ..................... 23 Condition of the Plantations ............ 25 Growth Rates .................... 27 Comparison With Results of Previous Studies ..... 39 Application of the Results ............. 44 V. SLASH PINE PROVENANCE TRIALS ............. 45 The Species ..................... 45 Condition of the Plantations ............ 47 Growth Rates .................... 49 Comparison With Results of Previous Studies ..... 58 Application of the Results ............. 64 iv Page VI. FUTURE USES OF THE PRODEPEF TRIALS Provenance Trials .................. 65 Nelder-Bleasdale Spacing Trials ........... 67 VII. WHERE DO NE GO FROM HERE? ............... 69 Future Improvement ................. 7O Improvements of the Experimental Procedures ..... 7l Progeny and Combined Provenance-Progeny Tests . . . . 72 VIII. SUMMARY AND CONCLUSIONS ................ 74 Loblolly Pine Growth Rate Results and Conclusions . . 76 Slash Pine Growth Rate Results and Conclusions . . . 77 Future Uses of the Trials .............. 78 Future Research ................... 79 APPENDICES .......................... 8l A. HEIGHT GROWTH DATA OF LOBLOLLY AND SLASH PINE SEEDLOTS TESTED IN SOUTHERN BRAZIL .......... 82 B. SAMPLES OF THE RESULTS OF THE STATISTICAL ANALYSIS USING MSU CDC CYBER 750 COMPUTER ........... 87 C. THE USE OF NELDER-BLEASDALE DESIGNS IN THE PRODEPEF TRIALS .................... 92 BIBLIOGRAPHY ......................... ITO Table 10. ll. 12. LIST OF TABLES Number of Seedlots of Pinus Tested in PRODEPEF Provenance Trials ................... Geographic and Climatic Site Parameters by Location . . . Location, Mortality, and Average Height of Eight Loblolly Pine Test Plantations in Southern Brazil . . . Origin and Nursery Data for Loblolly Pine, Series I, Grown in the Irati Flona's Nursery for 2l3 Days (May 7, I973) ..................... Location and Relative Growth Rate of Loblolly Pine Seedlots (Series 1) Collected From Various Parts of the Species Range and Tested in Southern Brazil Location and Height Rank of Loblolly Pine Seedlots (Series l) Tested in Southern Brazil ......... Rank Correlations (rs) of Height Growth of l2 Loblolly Pine Seedlots (Series 1) Tested at Four Locations in Southern Brazil .................. Location and Relative Growth Rate of Loblolly Pine Seedlots (Series 2) Collected From Various Parts of the Species Range and Tested in Southern Brazil Location and Height Rank of Loblolly Pine Seedlots (Series 2) Tested in Southern Brazil ......... Rank Correlations (rs) of Height Growth of 16 Loblolly Pine Seedlots (Series 2) Tested at Four Locations in Southern Brazil .................. Location, Mortality, and Average Height of Seven Slash Pine Test Plantations in Southern Brazil ....... Origin and Nursery Data for Slash Pine, Series I, Grown in the Irati Flona's Nursery for 2l3 days (May 7, l973) ..................... vi Page 12 I7 26 28 29 3O 31 32 33 34 48 50 Table Page 13. Location and Relative Growth Rate of Slash Pine Seedlots (Series I) Collected From Various Parts of the Species Range and Tested in Southern Brazil . . 52 14. Location and Height Rank of Slash Pine Seedlots (Series l) Tested in Southern Brazil ......... 53 l5. Rank Correlations (rs) of Height Growth of Nine Slash Pine Seedlots (Series I) Tested at Three Locations in Southern Brazil .................. 54 16. Location and Relative Growth Rate of Slash Pine Seedlots (Series 2) Collected From Various Parts of the Species Range and Tested in Southern Brazil . . 55 l7. Location and Height Rank of Slash Pine Seedlots (Series 2) Tested in Southern Brazil ......... 56 18. Rank Correlations (rs) of Height Growth of Eight Slash Pine Seedlots (Series 2) Tested at Four Locations in Southern Brazil .................. 57 Al. Location and Height Growth of Loblolly Pine Seedlots (Series l) Tested in Southern Brazil ......... 83 A2. Location and Height Growth of Loblolly Pine Seedlots (Series 2) Tested in Southern Brazil ......... 84 A3. Location and Height Growth of Slash Pine Seedlots (Series l) Tested in Southern Brazil ......... 85 A4. Location and Height Growth of Slash Pine Seedlots (Series 2) Tested in Southern Brazil ......... 86 Bl. Sample of an Analysis of Variance Table for Height Growth ........................ 88 82. Sample of an Analysis of Variance Table for Height Growth of the 25% Tallest Trees ............ 88 B3. Sample of an Analysis of Variance Table for Diameter at Breast Height ................... 89 B4. Sample of an Analysis of Variance Table for Forked Trees ......................... 89 35. Sample of an Analysis of Variance Table for Ant Damage . 9O vii Table 86. B7. Cl. C2. Sample of an Analysis of Variance Table for Foxtail . . . Sample of Pearson Correlation Coefficients Among Traits in One Set of Measurements .............. Parameters of Nelder-Bleasdale Design--Plan A ...... Parameters of Nelder-Bleasdale Design-~Plan B ...... viii Page 90 91 100 102 Figure Cl. C2. C3. LIST OF FIGURES Natural Range for Parana pine (Araucaria angustifolia) . Suitable Area for Loblolly and Slash Pines in Southern Brazil ........................ Location of the Provenance Trials in Southern Brazil . . Natural Range for Loblolly Pine, Pinus taeda ...... Geographic Location of Loblolly Pine Seedlots Tested in Southern Brazil and Delimitation of Regions for Seed Collection ................... Natural Range for Slash Pine, Pinus elliottii ..... Geographic Location of Slash Pine Seedlots Tested in Southern Brazil ................... The five original Nelder Designs ............ Layout of PRODEPEF's Plan A .............. Layout of PRODEPEF's Plan B .............. ix Page I6 24 37 46 59 95 101 103 CHAPTER I INTRODUCTION The southern states of Brazil were once covered by magnificent warm, temperate forests. They were cleared mainly for agriculture and pastures. Today less than lO% of the land has a forest cover. An area of about 20 million ha in the highlands between alti- tudes of 500 and 1500 meters was the habitat of a mixed forest with the native conifer Parana pine (Araucaria angustifolia) being the dominant Species, and the principal commercial tree in southern Brazil. Many valuable hardwoods such as cedro (Cedrela fissilis), imbuia (Ocotea porosa), erva-mate (Ilex $9.), and several species of Lauraceae and Mirtaceae were present. These prize hardwoods have been heavily exploited for industrial uses or simply wasted in clearing the land for agriculture. Few are planted since early attempts often met with little or no success. The silvics of these species are not known; they are slow growing and their natural regeneration has been unsuccessful. The Araucaria forest was subjected for many years to an annual cut about 10 times the estimated annual increment (Dillewijn et al., 1966). The result was depletion of the easily accessible stands. Based on 1973 remote sensing surveys, only 5.8% of the original 7.4 million ha of Araucaria forest was left in the state of Parana, and similar situations exist in the other southern states (Siqueira, l977). In the Araucaria forest, timber extraction is so radical and regeneration so poor that recovery of the stands is uncertain or very far in the future. According to Golfari (l97l, 1978), a leading expert in Brazilian forestry, the small number of seedlings and saplings in wild stands of Araucaria and also in selectively logged stands is due to compe- tition and dominance of associated hardwoods and the presence of several seed predators. Among such enemies are insects, birds, rodents, wild and domestic ungulates. Also pe0ple utilize the large starchy seeds as food. Araucaria, despite producing timber of a quality very high and unmatched by exotic species, has been planted on a limited scale. The reasons for such apparent contradiction are its slow initial growth when compared with other conifers, the need for expensive cultural operations, and its little plasticity and very high edaphic requirements. Due to soil limitation, it is paradoxically restricted to less than 20% of its own natural area of occurrence (Figure 1). Sites potentially suitable for Araucaria are usually high-priced agricultural lands. However, within and beyond that same natural area the introduced pines, loblolly (Pinus taeda) and slash (Pinus elliottii) can be very successfully planted without any limitation (Figure 2). Loblolly and slash pines have been intensively planted in the past l5 years and have found ideal ecological conditions in the southern states of Brazil. They present an average height growth _ , ”um-0's loo- an m “:1, 1:}? \‘hi m”:.3':.~m Q ..... 7 |fl""""'l|||"' . "I- .. I. ‘ li' l “ '. ‘l v )1,“ .L‘..‘ --~ i '0“ l r .n‘ '1 ’Hi ‘iE-‘z n {{{{{{ ..... J”! I'IILIW '“" 1n|"”’” ) ‘ a“... lumber-aw I Legend: MB] Araucaria Forest Grasslands Figure 1. --Natural range for Parana pine (A raucaria angustifolia). (Adapted from Golfari, 1971. ) A.wmmp ..~m um vcmwpow soc» emuawu 109 6 FL Calhoun 101 108 104 E MS Covington -- -- 109 2 FL Columbia 98 98 105 1 FL Marion 111 96 98 8 FL Flagler llO -- 104 4 LA St. Tammany 105 94 96 7 GA Dooly 93 96 98 99 PR Irati, BRAZIL 95 100 88 5 MS Harrison 96 94 88 C FL Bay 94 -- 98 3 FL Calhoun 94 78 95 Year of measurement 1977 1974 1977 Age when measured (mo.) 49 14 52 Mean height (cm) 496 49 384 53 Table l4.--Location and height rank of slash pine seedlots (Series 1) tested in southern Brazil. Seedlots are arranged in decreasing order of their overall ranking. Seedlot number and Height rank when grown at iiitfiaiighi°32t3rigin Sgfiiio Irati 3::535 A FL Levy - l l 0 LA Allen 4 3 2 6 FL Calhoun 5 2 5 E MS Covington - - 3 2 FL Columbia 6 5 4 1 FL Marion 1 7 7 8 FL Flagler 2 - 6 4 LA St. Tammany 3 8 10 7 GA Dooly ll 6 9 99 PR Irati, BRAZIL 8 4 13 5 MS Harrison 7 9 12 C FL Bay 9 - 8 3 FL Calhoun 10 - 10 11 54 Table l5.--Rank correlations (r ) of height growth of nine slash pine seedlots (Series 1) tested at three locations in southern Brazil. Rank correlation with results at Location Irati Tres Barras Capao Bonito .18 .47 Irati .55 55 Table 16.--Location and relative growth rate of slash pine seedlots (Series 2) collected from various parts of the species range and tested in southern Brazil. Seedlots are arranged in decreasing order of their overall ranking. Relative height when grown at Seedlot number and Iggtgzrigfi)igggfiigin Lat. Long. §l§;' 852120 Irati 85::25 Pelotas (°) (°) (m) (% of plantation mean) 14 FL Co1umbia 30.3 82.6 75 110 113 110 104 6 FL Calhoun 30.6 85.1 65 101 96 102 106 11 MS Harrison 30.9 89.2 45 100 103 97 105 99 PR Irati, BRAZIL -- -- 850 97 107 92 105 12 MS Harrison 30.9 89.2 -- 101 101 92 98 13 SC Hampton 32.5 81.0 -- 94 94 109 96 10 FL Calhoun 30.4 85.4 46 93 101 105 96 8 SC Berkeley 33.2 79.7 8 101 84 93 91 Year of measurement 1977 1977 1978 1977 Age when measured (mo.) 20 20 27 25 Mean height (cm) 105 109 208 200 56 Table 17.--Location and height rank of slash pine seedlots (Series 2) tested in southern Brazil. Seedlots are arranged in decreasing order of their overall ranking. Seedlot number and Height rank when grown at IgitgaaighIogftgrigin 532120 Irati 85::55 Pel°t°5 14 FL Columbia 1 1 1 4 6 FL Calhoun 2 6 4 1 11 MS Harrison 5 3 5 3 99 PR Irati, BRAZIL 6 2 7 2 12 MS Harrison 3 4 8 5 13 SC Hampton 7 7 2 6 10 FL Calhoun 8 5 3 7 8 SC Berkeley 4 8 6 8 57 Table 18.--Rank correlations (rs) of height growth of eight slash pine seedlots (Series 2) tested at four locations in southern Brazil. Rank correlation with results at Location Irati Tres Barras Pelotas Capfio Bonito .26 .OO .38 Irati .02 .52 Tres Barras ’ -.lO 58 some plantations and among the poorest in others. The most interac- tions were presented by seedlot 99-PR Irati (in both series) and by seedlot 1-FL Marion (in Series 1) and seedlot 6-FL Calhoun (in Series 2). In Series 1 a very important interaction was displayed by seedlot 1-FL Marion, which is the best at Capao Bonito and poor at other locations. No reasonable interpretation can be given for the interactions. Differences in diameter growth due to seedlot were also signifi- cant. The correlation between height and diameter was r = .90 or higher and significant at the .01% level. Combining the data and observations from both series, it was not possible to delimit a region of fastest-growing seedlots. Growth—rate differences among stands sampled within the species range of slash pine were real and significant but exhibited no distinctive geographi- cal trends such as clines or ecotypes (Figure 7). Comparison With Results of Previous Studies Previous provenance studies in slash pine were summarized by Burley (1966), Squillace (1966a), Wells (1969), and Dorman (1976). From 1937 to 1949, the U.S. Southern Forest Experiment Station established 10 provenance plantations of slash pine in the United States and contributed to seven plantings in South Africa. None sampled the whole range of the species. Those in the United States were less well designed and maintained than the South African tests (Snyder et al., 1967). Thirty-five-year results from the seven plan- tations in South Africa showed no significant differences between the 59 J x v.- ' .1 . O l I MILES 0 100 200 300 400 l_L LL11 l l l n I a I l l I I 90° 85° fit a" .8 Location and seedlot identification in Series 1. . 3 Location and seedlot identification in Series 1. A 6 Location and seedlot identification in Series 2. Figure 7.--Geographic location of slash pine seedlots tested in southern Brazil. 60 provenances at any of the trials and for any growth traits studied. In most trials highly significant correlations were found between early and late growth (Falkenhagen, 1978). Similar results were obtained in the United States (Snyder et al., 1967). In 1951, P. C. Wakeley organized the "Southwide Pine Seed Source Study," which was summarized in Chapter IV. The slash pine portion of the Southwide Study sampled six areas and plantations in 12 locali- ties. Height and survival showed significant differences but no pat- tern of variation (Wakeley, 1961; Wells, 1969). In 1957, a combined provenance-progeny test was established in southern Mississippi testing 61 seedlots. Of these, 37 were randomly selected in five native south Mississippi stands, 16 were randomly selected in a south Mississippi plantation grown from northeast Florida seed, and 8 were phenotypically superior trees selected in a Georgia planting. Morphological and juvenile characters of the proge- nies were assessed. Six characters, including height, were highly significant among progenies within stands. Of the six traits, four (including height) varied among stands within geographic source. Of a total of nine characters studied, only height varied between geoe graphic sources, and this was because of the superior growth of the progenies from Georgia which had been selected for vigor (Snyder et al., 1967). Squillace (1966b) reported on his large provenance-progeny test, which included seedlings from 54 natural stands scattered throughout the range. Most of the traits studied showed significant differences due to geographic source. In the parental data, 61 stand-to—stand variation was relatively strong in cone, seed, and needle traits. In the progeny data, stand variation was strong for germination rate, height, and diameter. There was a major general trend in Florida. Seedlings of the South Florida variety grew faster in diameter but slower in height ' than trees of the typical or North Florida variety. This clinal pattern through Florida has little practical significance for southern Brazil planting conditions, where the typical variety is utilized almost exclusively. In the northern part of the range, the pattern was largely random and Squillace could not recommend any one portion of the range as likely to produce the fastest-growing trees. Squillace's nursery work has had a follow-up (Gansel etal., 1971), in which 270 seedlots (54 sources x 5 mother trees) were planted at four locations. The results are parallel to a 1954 Georgia-Florida study, in which slash pines from 15 seed sources were planted at seven loca- tions in the Georgia-Florida area. Differences in growth rate due to seed source and genetic-environment interactions were present and highly significant in both studies. In the 1954 study, considering the average height over the plan- tations, the best sources occurred in a narrow band extending from Calhoun County northeastward to Effingham County, Georgia. In Squillace's field follow-up trial, one plantation closely agrees with these results, but another indicates that the presumed optimum zone is farther south in central Florida. A slightly different Optimum growth zone is defined by Wells (1975). It includes southeastern South Carolina, southern Georgia, 62 part of western Florida, southern Alabama, and southern Mississippi. He recommends that seed be collected from the optimum growth zone for local plantings and up to 150 miles south of this zone. The PRODEPEF trials include five seedlots from within that "optimum zone." Seedlot 6-FL Calhoun was third best in Series 1 and second best in Series 2. The four other seedlots (13-SC, 8-SC, lO-FL, and 3-FL) were among the poorest. Obviously the data do not indicate that the "optimum zone" produces the best seed for Brazil. The overall good performance of seedlot 6-FL Calhoun when com- pared with the poor performance of two other seedlots from the same county raises questions about the extent of stand-to-stand variation. Unfortunately, neither the American nor PRODEPEF trials provide an adequate measure of stand-to-stand variation over the entire range. Additional provenance research in southern United States was carried out in Florida (Goddard et al., 1962; Goddard, 1964; Goddard & Cole, 1966; Goddard & Smith, 1969) and in Texas (van Buijtenen, 1969). From all these studies we learn that geographic variation should also be recognized in clonal selection for tree seed orchards. Slash pine provenance trials have been carried out in many Southern Hemisphere countries such as Malawi (Andersen, 1967), Kenya (Burley, 1966), Zimbabwe (Prevost et al., l973a), Australia (Slee & Reilly, 1966; Burgess, 1973), New Zealand (Streets, 1962), Colombia (Perez, 1967), Uruguay (Burley, 1966), and Argentina (Barrett, 1974). Results are quite variable due to deficiencies in experimental design, quality of care after planting, and number of seedlots tested. In 63 general, differences due to seedlot have been found in most trials, and results are comparable to the PRODEPEF experiments. Results from neighboring countries such as Argentina and Uruguay as well as from Brazil (Canavera, 1971; Fonseca et al., 1978: Shimizu, 1978) are very important for comparison purposes. A good comparison is possible with a 4-year-old provenance trial established by IPEF in Lages, Santa Catarina (Fonseca et al., 1978). That trial utilizes 7 of the 13 seedlots of PRODEPEF's Series 1 experiment. The rank cor- relation between the two tests is rS = .14. This result reinforces what was said in the previous section about the large amount of genotype-environment interaction. Barrett (1974) established a good network of well-designed trials over 20 locations in Northeast Argentina, testing 38 seedlots. Significant differences due to seed source existed. Such variation was smaller considering only the typical variety. Seedlots from Florida were faster growing than more northern ones, but no clear pattern of variation existed. Among the best provenances for each site were seedlots from Florida, Louisiana, and Georgia. A discon- tinuous variation in growth rate was found between neighbor stands. Brazilian and Argentinian results were very closely related. Shimizu (1978) reported some early results of a comparison between a Brazilian seedlot and some American progenies of slash pine. The local seedlot was collected at Capao Bonito. The American progenies of high and low vigor were families selected respectively for their high and low resin yield. Height growth of the best family was 2.7 times higher than the Brazilian seedlot. In the comparison with the 64 low-vigor families the Brazilian seedlot was significantly smaller than two of the low-vigor families. CapEo Bonito's plantations were established by using seeds from the United States of unknown origin. A great quantity of seeds collected at CapEo Bonito have been used in reforestation programs. Application of the Results There were growth rate differences due to seed source in slash pine, but the variation was random. Differences among stands were real, but no definite geographical pattern existed. Because of this, no region in southern United States can be generally recommended for seed collection for planting in Brazil. Individual stands, however, can be recommended. The specific stands which originated seedlots A-FL Levy, D-LA Allen, 6-FL Calhoun, and l4-FL Columbia should be considered as seed sources as a result of being the four best seedlots overall in Series 1 and/or 2. The stand where seedlot l-FL Marion was collected can be considered as a seed source for Capao Bonito but not Irati or Tres Barras. There was enough correspondence between my results and results in the United States and other countries to indicate that genetically improved seeds from seed orchards in the United States can be used in Brazil. CHAPTER VI FUTURE USES OF THE PRODEPEF TRIALS All genetic plantations established by PRODEPEF are presently under responsibility of EMBRAPA (Empresa Brasileira de Pesquisa Agropecuaria). The creation by EMBRAPA of a well-staffed Regional Unit of Forestry Research at Colombo, Parana, when PRODEPEF ended assures that all genetic research previously established will have continuity. Provenance Trials The three major uses of the provenance trials in the future are (l) as producers of information, (2) for demonstration purposes, and (3) as breeding arboreta. A fourth possible use is as seed orchards. Production of information is the most important function of the trials. Additional measurements should be taken at intervals of approximately 3 to 5 years. Some of the questions to be addressed by them are: 1. Are early trends consistent with later performance? 2. How strong are the age-age correlations? 3. Is there any genetic variation in the height—diameter ratio among seedlots? 65 66 4. Which measurements will be most appropriate for future cor- relations when the plots have been thinned or height measure- ments have become difficult? Future assessments may include a thorough evaluation of form, quanti- fication, and timing of flowering habits, and other traits regarded as important at the time. A comprehensive wood-quality study and a biomass study will be most apprOpriate if coincident with first thin- ning. Such studies may be advantageously undertaken as research tepics of graduate theses in joint collaboration of EMBRAPA and Graduate Programs in Forestry. Use of the trials as demonstration areas is also important. Tree improvement results ought to be transmitted to the general public. Communication and acceptance of results are more likely to be effec- tive through publications and demonstration tours of the test plan- tations rather than through publications alone. Test plantations are located in easily accessible areas. All four areas are portions of "forestry districts" in the southern states. A state forestry dis- trict is an area designated by the federal government as most suit- able for tree planting due to superior economical and ecological conditions. The demonstrational value of the test plantations can be enhanced by good maintenance. Thinning is an essential part of future care. All plantations were established in good alignment, and their growth is comparable to commercial plantings. A breeding arboretum is a plantation or groups of trees used chiefly as parental stock in forest genetic experimentation. Utili- zation of the test plantations as breeding arboreta is a potential 67 and recommended use for the trials. A favorable aspect for such use is the relatively large plot size. Selection within as well as between seedlots will be possible. Maintenance of the maximum num- ber of trees of the best seedlots until maturity will also be feas- ible. Enhancement of the breeding value of the plantations can be accomplished by thinning to increase spacing and allow better crown development, and consequently heavier flowering. Good maintenance should also be given to the areas, including weed control. Conversion of the test plantations into seed orchards is not advisable. PRODEPEF test plantations are not amenable for an effec- tive conversion into seed orchards because they have large plot sizes. Conversion into seed orchards is performed by a series of thinnings. Maximum improvement results if thinning removes all trees of the poor- est seedlots and the poorest trees of the best seedlots. Removal of all trees of the poorest seedlots is feasible only in small plot experiments. In large-plot experiments as the PRODEPEF trials, thinning would leave large and undesirable gaps in the plantations. Nelder-Bleasdale Spacing Trials Several test plantations using Nelder-Bleasdale systematic spacing designs have been established by CPFRS in southern Brazil from 1973 to 1976. PRODEPEF trials were the first in the country to utilize the full layout of a systematic spacing design in a forestry experiment. A detailed discussion of the designs including sugges- tions for theiranalysis is presented in Appendix C. 68 Provenances were not replicated in Series 1, and a few repli- cations (up to four) were used in Series 2 trials. The lack of rep- lication and randomization in Series 1 and the moderate mortality with replacements by different provenances has discouraged measure- ments and analyses of the established trials. PRODEPEF systematic grids may or may not give any useful genetic information. Their main value will be the spacing data. Nelder- Bleasdale systematic grids used in PRODEPEF trials should be measured and analyzed in the future. Measurements and analyses should be done every year from age 4 to 10 when competition is expected to start for most of the spacings under test, and at intervals of 2 to 3 years afterward. There are very few spacing experiments with loblolly and slash pine in Brazil. The valuable information about spacing that can be learned from these systematic grids is by itself a very strong reason to maintain, measure, and analyze all existing plantations. CHAPTER VII WHERE DO WE GO FROM HERE? The PRODEPEF provenance trials of loblolly pine were successful in identifying a region producing seed of superior quality for Brazilian conditions (Figure 5). The same wasn't true for slash pine, for which no definite geographic pattern existed. In loblolly pine this provenance study made a real contribution to practical tree improvement in Brazil. In slash pine, however, it has not. PRODEPEF trials followed the conventional approach of prove- nance testing. In slash pine, they succeeded in selecting a few better stands among a dozen or so tested. They did not succeed, however, in delimiting any region for seed collection nor in select- ing the very best stands in the natural range. If additional prove- nance trials are to be carried out in Brazil, they should follow a different approach. They should include hundreds of seedlots, sampling several families from each of many stands. Wright (1980) advised that in species containing a limited amount of geographic variation there should be less of the conven- tional type of provenance test and more of an intensive sampling type of experimentation. Wells and Switzer (1971), following this approach, were very successful with loblolly pine and reached important conclu- sions from the practical standpoint. 69 70 Future Improvement Further improvement is, of course, desirable and needed in Brazil. This can be pursued by (a) using already improved American material or (b) superior provenances or stands in the natural range. American seed orchards do not produce enough seeds for American needs. Seeds from American seed orchards, however, can be procured in enough quan- tities to establish test plantations for future selections. The choice of a superior provenance for more intensive sampling trials is simple in the case of loblolly pine. Livingston Parish and sur- rounding.parts of southeastern Louisiana are the natural choice. In slash pine such a region of superiority has not been detected, and northern Florida may be as appr0priate as western Florida for further trials. Seed procurement for an intensive sampling of southeastern Louisiana or any other region will probably require not only mail contacts with industries and government agencies in the region but also the sending of a Brazilian team for seed collection. The extent of work to be carried out by such a team will depend upon the amount of material which could be obtained from the forestry industry and government in the area. A Brazilian team selecting in Louisiana should be able to cause at least a 1% improvement in growth rate of loblolly pine. Such improvement would be worth many thousands of dollars--much more than the cost of sending the team. 71 Improvements of the Experimental Procedures Genetic research in Brazil has followed a very traditional approach. Usually a small number of seedlots (2 to 20) have been tested using a small number of replications (2 to 5) and large plots (25 to 121 trees). Such procedures have limited the scape of the trials in terms of the material tested, and they have also decreased the precision and efficiency of the trials. Statistical precision is the ability of an experiment to differ- entiate among treatments. Statistical precision is greater in a plantation containing small plots and many blocks than in one con- taining large plots and fewer blocks. Small plots make up smaller blocks with less intrablock site variability, and therefore more reliable are the comparisons among seedlots planted in the same block. Statistical efficiency is the ability of an experiment to pro- duce the greatest amount of useful information per unit of cost, as measured by number of trees planted, measurement time, etc. It has been demonstrated that efficiency decreases with increasing plot size (Wright & Freeland, 1959; Conkle, 1963; Wright, 1963). Calculations made by these same investigators showed that'L-to lO-tree plots were statistically most efficient, giving the most information per unit of time spent on measurement and analysis. Another serious limitations of past Brazilian research has been the limited number of locations tested in a given experiment. PRODEPEF trials including four locations were the first step toward a more desirable situation. Such a number, however, is minimal and 72 should be increased for a more reliable sampling of environmental conditions. Genetic-environment interaction exists for both pines, and the only way of studying it is through multiplantation experi- ments. Future genetic research in Brazil with loblolly and slash pines should make use of small plots (1 to 10 trees), a greater number of replications (8 to 10), and a greater number of locations (15 to 30). This recommendation is especially important for the intensive sampling approach suggested for future experimentation in Brazil. Progeny and Combined Provenance-Progeny Tests The intensive sampling approach for the testing of already improved American material or superior provenances or stands in a chosen region should be performed as progeny tests or combined provenance-progeny tests. Such trials should be established in such a way that the test plantations can be converted into seed orchards, as discussed by Nanson (1972). The three most important features which will allow efficient conversion are a very large number of seedlots (100 to 300), small plot sizes (1 to 4), and a large number of replications (8 to 10). A few advantages of using this approach in Brazil are the following: 1. Low costs: It will not involve much more cost than ordi- nary provenance experiments. 2. Rapidity and ease of establishment: Seedling plantations are easy to establish because no special skills are required beyond those necessary for a commercial planting. 73 3. Broadggenetic basis: A very broad genetic basis will be assured by the diversity of provenances, stands, and individuals making up the seed orchard. The trees produced by such a seed orchard will be a genetic synthesis of individuals which are the best adapted to the planting site. 4. Reduction of inbreeding: The use of hundreds of seedlots reduces the chances of inbreeding. Modifications in design will be required to handle large progeny tests or combined provenance-progeny tests. A simplified design that can be conducted almost as simply as an ordinary provenance test was pr0posed by Wright (1978). Such experiments can yield considerable genetic improvement and be converted into seed orchards. CHAPTER VIII SUMMARY AND CONCLUSIONS Parana pine (Araucaria angustifolia) was a very important conifer in southern Brazil. The Araucaria forest has been depleted due to intensive logging, and regeneration is very poor. Araucaria has been planted but in a limited scale due to silvicultural and economic limitations. Loblolly pine (Pinus taeda) and slash pine (P. elliottii), introduced from the southern United States, can be very successfully planted. About 800,000 ha of these two species have been estab- lished. The total industrial demand for wood in Brazil requires not only more planting but better productivity. Genetic improvement based on provenance trials is one major approach to increase produc- tivity. This study is an assessment of loblolly and slash pine prove- nance trials planted in southern Brazil by the Project for Forestry Development and Research (PRODEPEF). The major objectives of this study were to determine if there are: 1. Genetic differences in growth traits among provenances, 2. Any trends of significance for locating the most suitable provenances, 3. Any genetic-environment interaction. 74 75 In addition, the following objectives were pursued: 4. To determine the advisability of future uses of the trials, 5. To determine possible improvements of the experimental procedures, 6. To determine the future direction of the experimentation aiming at the genetic improvement of loblolly and slash pines. Seeds were obtained from the U.S. Forest Service (Macon, Georgia, and Washington, D.C.). Thirty-five seedlots of loblolly pine and 20 seedlots of slash pine were tested in two series. In Series l (1973 plantations), bare root seedlings were raised in one nursery at Irati, Parana. In Series 2 (1975 plantations), container- ized seedlings were raised at the plantation sites. A nursery trial was conducted in Series 1. Plantations were established in four locations: Capao Bonito (Sao Paulo); Irati (Parana); Tres Barras (Santa Catarina), and Pelotas (Rio Grande do Sul). Plantations fol- lowed a randomized complete block design with 16- to 49-tree plots and 3 to 5 replicates. Spacing was 2 by 2 m or 2.5 by 2.5 m. Mortality was assessed, and several growth traits (height, diameter, forking, foxtail, etc.) were measured. Each set of measurement data was subjected to statistical analysis. All differences quoted as being important are statistically significant at least at the 5% level. Mortality was much lower in Series 2 for both species. The main cause of mortality in Series 1 was the long-distance transpor- tation of bare root seedlings. Loblolly pine and slash pine at Capao Bonito presented an average annual height increment of 1.4 m and 76 1.25 m, respectively--very good for Brazilian conditions. Form was generally good for both species, with few trees forked or crooked. Flowering was observed at age 4 in Capfio Bonito in 15% of the loblolly pine trees. In slash pine, 6% of trees had flowers. Nursery traits were significant for both pines. Some latitudi- nal trends were observed in loblolly pine seedlings related to height growth (r = -.66), bud set (r = .69), and Lammas growth (r = -.67). In slash pine only Lammas growth had a significant correlation with latitude (r -.78). Loblolly pine seedlots which were tallest in the nursery also tended to be tallest in the plantations. The rank correlation between nursery height and average height in all planta- tions was rS = .67. For slash pine no significant correlations were found between height in the nursery and height at the plantation sites. Correlations between first-year height and fourth—year height in the field, however, were generally strong for both species (r = .70 in the average). Loblolly Pine Growth Rate Results and Conclusions There were very important differences in growth rate due to provenance in both series. In Series 1, seedlot 6-LA Livingston was consistently among the leaders at all test locations. Seedlots B-FL Jackson and A-FL Marion were also fast growing. In Series 2, among the leaders at all locations were two seedlots from Mississippi (9-MS and 8-MS), a Brazilian seedlot from Klabin (98-PR), and a South Carolina seedlot (l7-SC). Genotype-environment interactions were strong in SGFlES‘l and smaller in Series 2. 77 A general trend of coastal seedlots to grow faster than interior ones was observed in both series. Seedlots from Maryland, Virginia, and the Piedmont from North Carolina to Alabama grew poorly from average to 50% below average. The fastest-growing seedlots were up to 2 to 2.5 times taller than northern or interior seedlots. PRODEPEF results were compared with previous studies in the United States. American and Brazilian results were very similar. Results obtained in many other countries were also comparable to those of the PRODEPEF experiments. The most highly recommended area for seed collection for plant- ing in Brazil is the coastal region stretching 120 kilometers from the coast of the Gulf of Mexico and the Atlantic Ocean in the states of Louisiana, Mississippi, Alabama, Florida, Georgia, South Carolina, and North Carolina. The consistent superiority of seedlots from Livingston Parish indicates that Brazilian growers can obtain immediate improvement by using seed from that area. The close correlation of my results and American results indicates that genetically improved seeds from the United States (recommended region) can be used in Brazil. Slash Pine Growth Rate Results and Conclusions Important differences in growth rate existed in both series. Differences were smaller than those observed for loblolly pine. In Series 1, seedlot A-FL Levy was consistently the best. Seedlots D-LA Allen and 6-FL Calhoun were also fast growing. In Series 2, seedlot l4-FL Columbia was the best. In both series it was observed that sizable differences occurred among stands close together 78 geographically. Genotype-environment interactions were strong in both series. Growth rate differences among stands sampled within the species range of slash pine were real and significant, but no clear geo- graphical trend or pattern was found. Many other experiments in the United States and other countries (especially Argentina) had similar results. No region in the southern United States can be generally recom- mended for seed collection for planting in Brazil. Individual stands which originated seedlots A-FL Levy, D-LA Allen, 6-FL Calhoun, and 14-FL Columbia can be suggested as a result of their good performance. The close correspondence of results in Brazil and in the United States indicates that genetically improved seeds from seed orchards in the United States can be used in Brazil. Future Uses of the Trials Provenance trials established by PRODEPEF should have three major uses in the future, which are (l) as producers of information, (2) for demonstration purposes, and (3) as breeding arboreta. These three uses are compatible with each other and suitable for the adopted design of the trials. A fourth possible use as seed orchards is not recommended because the test plantations have large plot sizes. Conversion into seed orchards would also conflict with the other three uses. Some plantations were established using Nelder-Bleasdale syste- matic spacing designs. They should be measured and analyzed in the 79 future. They may or may not give any useful genetic information. There are very few spacing experiments with loblolly and slash pine in Brazil. Valuable information about spacing can be learned from these systematic grids. Future Research A conventional approach of provenance testing was used in loblolly and slash pine trials. Some real contribution to practical tree improvement was achieved for loblolly pine, but not for slash pine. Future trials should follow a different approach, including hundreds of seedlots and sampling several families from each of many stands. Further improvement can be pursued by using already improved American material or superior provenances or stands in the natural range. Seeds from American seed orchards can be used to establish test plantations for future selections. The choice of a superior provenance of loblolly pine for more intensive sampling trials could be Livingston Parish and surrounding parts of southeastern Louisiana. In slash pine such a region of superiority has not been detected. The sending of a Brazilian team for seed collection in southeastern Louisiana is necessary for an intensive sampling of stands in the region. Genetic experimentation in Brazil has usually tested a small number of seedlots (2 to 20), using a small number of replications (2 to 5), and large plots (25 to 121 trees). Such procedures have limited the scope of the trials in terms of the material tested, and 80 they have decreased the precision and efficiency of the trials. Another serious limitation has been the limited number of locations tested. Future genetic research should make use of small plots (1 to 10 trees), a greater number of replications (8 to 10), and a greater number of locations (15 to 30). Future intensive sampling trials should be established in such a way that the test plantations can be converted into seed orchards. Efficient conversion can be accomplished by using a very large num- ber of seedlots (100 to 300), small plot sizes (1 to 4), and a large number of replications (8 to 10). Modifications in design which allow such large tests to be conducted almost as simply as an ordinary provenance test are indicated. APPENDICES 81 APPENDIX A HEIGHT GROWTH DATA OF LOBLOLLY AND SLASH PINE SEEDLOTS TESTED IN SOUTHERN BRAZIL 82 83 Table Al.--Location and height growth of loblolly pine seedlots (Series 1) tested in southern Brazil. Seedlots are arranged in decreasing order of their overall ranking. Height when grown at Seedlot number and transiiitingm 335% 315525 (cm) (cm) (cm) (cm) 6 AL Livingston 645 148 377 231 A FL Marion 630 -- 493 -- 8 FL Jackson -- -- 460 -- 11 NC Jones 523 143 397 193 C LA Washington 625 -- 407 -- 10 MS Stone 588 153 317 220 1 SC Berkeley 615 167 213 214 13 GA Jones 455 134 407 170 5 LA Calcasieu 590 119 323 183 0 MS Forrest 585 114 423 157 99 PR Irati, BRAZIL 600 128 370 133 2 FL Marion 705 110 337 138 12 GA Oglethorpe 460 131 327 160 4 MD -- -- 353 -- 14 GA Oglethorpe 438 92 310 175 3 MD Worcester 353 110 267 171 7 TX Nacogdoches 575 -- 313 138 84 Table A2.--Location and height growth of loblolly pine seedlots (Series 2) tested in southern Brazil. Seedlots are arranged in decreasing order of their overall ranking. Height when grown at Seedlot number and (gitgaggghlogftgrigin 332120 Irati 81::25 Pelotas (cm) (cm) (cm) (cm) 9 MS Harrison 135 111 263 240 8 MS Scott & Smith 117 121 275 -- 98 PR Klabin, BRAZIL 151 116 233 240 17 SC Charleston 136 105 248 239 18 SC Berkeley 154 90 268 235 11 NC Jones 112 116 255 222 16 MS Scott 125 87 273 232 29 TX Angelina 116 98 255 199 20 NC Robeson 121 102 225 205 30 NC Perquimans 102 104 233 194 27 SC Kershaw 107 91 228 219 13 GA Jones 115 91 230 198 22 MS Chickasaw 101 92 228 201 28 AL Talladega 88 97 200 200 24 NC Pasquotank 97 93 213 192 19 GA Greene 86 67 218 181 25 MD Worcester 56 77 190 189 21 NC Durham 76 80 195 -- 23 VA Nottoway 67 80 175 -- 26 MD Kent 64 71 175 -- 85 Table A3.--Location and height growth of slash pine seedlots (Series 1) tested in southern Brazil. Seedlots are arranged in decreasing order of their overall ranking. Seedlot number and Height when grown at state and county CBP§0 Irati Tres (or parish) of origin BOHTtO Barras (cm) (cm) (cm) A FL Levy -- 60 420 D LA Allen 510 52 420 6 FL Calhoun 500 53 400 E MS Covington -- -- 420 2 FL Columbia 488 48 403 1 FL Marion 553 47 378 B FL Flager 545 -- 398 4 LA St. Tammany 523 46 368 7 GA Dooly 460 47 378 99 PR Irati, BRAZIL 473 49 338 5 MS Harrison 475 46 338 3 FL Calhoun 465 38 363 C FL Bay 465 -- 378 86 Table A4.--Location and height growth of slash pine seedlots (Series 2) tested in southern Brazil. Seedlots are arranged in decreasing order of their overall ranking. Height when grown at Seedlot number and izitsaaz-‘shrzramm $33120 .1232. (cm) (cm) (cm) (cm) 14 FL Columbia 115 123 228 208 6 FL Calhoun 106 105 212 212 11 MS Harrison 105 112 202 209 99 PR Irati, BRAZIL 102 117 192 210 12 MS Harrison 106 110 192 195 13 SC Hampton 99 103 226 191 10 FL Calhoun 98 110 218 191 8 SC Berkeley 106 92 194 181 APPENDIX B SAMPLES OF THE RESULTS OF THE STATISTICAL ANALYSIS USING MSU CDC CYBER 750 COMPUTER 87 88 Table Bl.--Samp1e of an analysis of variance table for height growth. . Sum of Mean Signif. Source of variation squares df square F of F Main effects 300027.451 18 16668.192 3.023 .003 TRT 232941.176 16 14558.824 2.640 .009 REP 67086.275 2 33543.137 6.083 .006 Explained 300027.451 l8 16668.192 3.023 .003 Residual 176447.059 32 5513.971 Total 476474.510 50 9529.490 Table BZ.--Sample of an analysis of variance table for height growth of the 25% tallest trees. . Sum of Mean Signif. Source of variation squares df square F of F Main effects 224168.627 18 12453.813 2.516 .011 TRT 192800.000 16 12050.000 2.435 .016 REP 31368.627 2 15684.314 3.169 .055 Explained 224168.627 18 12453.813 2.516 .011 Residual 158364.706 32 4948.897 Total 382533.333 50 7650.667 89 Table 83.-—Samp1e of an analysis of variance table for diameter at breast height. . . Sum of Mean Signif. Source of variation squares df square F of F Main effects 8277.569 18 459.865 2.273 .021 TRT 7148.039 16 446.752 2.208 .028 REP 1129.529 2 564.765 2.791 .076 Explained 8277.569 18 459.865 2.273 .021 Residual 6475.137 32 202.348 Total 14752.706 50 295.054 Table B4.--Samp1e of an analysis of variance table for forked trees. . . Sum of Mean Signif. Source of variation squares df square F of F Main effects 8.471 18 .471 1.855 .062 TRT 6.588 16 .412 1.623 .119 REP 1.882 2 .941 3.710 .036 Explained 8.471 18 .471 1.855 .062 Residual 8.118 32 .254 Total 16.588 50 .332 90 Table B5.--Sample of an analysis of variance table for ant damage. . . Sum of Mean Signif. Source of variation squares df square F of F Main effects 252.784 18 14.044 1.784 .075 TRT 172.745 16 10.797 1.371 .217 REP 80.039 2 40.020 5.083 .012 Explained 252.784 18 14.044 1.784 .075 Residual 251.961 32 7.874 Total 504.745 50 10.095 Table B6.--Samp1e of an analysis of variance table for foxtail. . . Sum of Mean Signif. Source of variation squares df square F of F Main effects 34.784 18 1.932 2.581 .009 TRT 32.745 16 2.047 2.733 .008 REP 2.039 2 1.020 1.362 .271 Explained 34.784 18 1.932 2.581 .009 Residual 23.961 32 .749 Total 58.745 50 1.175 mmmsmu ucm u pz< meme» umxeoc u gxoa mmmgp pmm_Pmp fimm Lo pgmwmg u mmez Seaman Lo mace; u szmc uemucmgm u P=>ahm “no.8; came u F=zohm amm.ua mae.ua .oo.ua _oo.ua _oo.ua ego.ua mmoF. ¢_oo.- mmmm. Kmmm. come. coco._ emo~.- p=zohm h=z~< '\' I \ ,\ 'r\ X '1’ X —‘-r X Jr >< Ar Inner guard are 103 i\ Equivalent n9 trees per hectare .1- (124) I ‘7‘. L (247) .i 7%. (494) .1. (865) ..i- + (1234) ~ I -1. (1606); 7" (1977) , X ‘~ I + (2347) ’ 7‘ 7L )( (2718 Y {43089 ~ ,‘ ‘ ~1- ~,L 7’. a/ 7‘\ Figure C3.--Layout of PRODEPEF's plan B. 104 Analysis Nelder (1962) discussed the main principles underlying the analysis of growth traits from systematic designs. He recommended: 1. to express the trait-density-rectangularity relation by a suitable function and estimate the unknown parameters in it for each "grid" (circle, or sector or replicate). 2. to analyze the parameter values from the individual grids as if they constituted a random sample. Treating the parameter values from the grids as if they were a random sample from some p0pulation is an important assumption to be justified. Nelder affirmed that the assumption was clearly analogous to that of taking a centric systematic area-sample (Milne, 1959) and treating it as random. The main danger in this procedure lies in the possibility that certain kinds of unsuspected periodic variation may be present. Milne concluded from an examination of 50 populations that this danger was negligible. provided that the experimenter takes certain precautions in laying out his experiment to avoid any risk of unwanted periodicity from known sources. Nelder agreed in assessing the risk from unknown sources as negligible. Yield-density functions for fixed rectangularity have been dis-I cussed by Bleasdale and Nelder (1960), Namkoong (1966), and others. They presented the following relationship: 6'“ = a + bD where G is growth rate, 0 is the plant density, and x, a, and b are constants. 105 The mathematical treatment of the data, although of value, is not essential to the interpretation of the results for practical purposes. Meaningful curves can often be drawn through the series of points on a graph, plotting the growth trait against density. The data from each grid should be examined separately before combining the data from several grids (Bleasdale, 1967). A suggested analysis of variance table for PRODEPEF's plan A design considering the whole grid is: Source of variation D$32§§§m°f Distance (density) 9 Spoke (provenance) 46 Error 414 Total 469 It is desirable to consider separately each sector (species) of the circle. The analysis of variance for an individual sector is similar to the analysis for the whole grid. For loblolly pine (16 provenances), for example, it is: 106 Source of variation Diggggngf Distance (density) 9 Spoke (provenance) 15 Error 135 Total 159 ANOVA tables for other species in the grid and for plan B can be constructed in a similar manner, taking into account the number of provenances under test. Provenances were not replicated in Series 1, so if the F value for spoke (provenance) is significant, this means that the spokes are different from each other because of either posi- tion or provenance effect. Likewise, if the F value for distance (density) is significant, this means that the distances are different from each other because of either position or density effect. A very meaningful part of the analysis would be calculating the mean for each density, plotting those means against distance, and inspecting the graph. Another method would be to choose an equation that seems to fit the curve the best and calculate the variances due to regression and error. Analyzing the same design (Plan A) and accounting the four plantations (locations) as replicates, then the ANOVA table is: 107 Degrees of Source of variation freedom F-test Spoke 187 Provenance 46 use A Plantation 3 use A (A) Plant. X Provenance 138 Distance (Density) 9 use B Distance X Spoke . 1683 Distance X Provenance 414 use 8 Distance X Plantation 27 use 8 (B) Dist. X Plant. X Prov. _____ 1242 Total 1897 In Series 2 plantations, provenances were replicated. When two or more species were tested in the same circle, different species were confined to sectors of it. Within each sector (species) at least three replicates were planted as subsectors. Position of each provenance within replicates was randomly assigned. The ANOVA table for a species represented by 5 provenances, 3 replicates, and 10 den- sities within spokes is: 108 Degrees of Source of variation freedom F-test Spoke l4 Provenance 4 use A Replicate 2 use A (A) Provenance X Replicate 8 Distance (Density) 9 use 8 Distance X Spoke 126 Distance X Provenance 36 use B Distance X Replicate 18 use B (B) Dist. X Replicate X Prov. ____ 72 Total 149 Discussion and Conclusions The desirability of including density in experimental designs as an important variable of the cultural environment is obvious. Nelder-Bleasdale circular grids make it possible. They avoid the dif- ficulties of the rectangular plots and still study density response over a wide range. Circular grids, however, have the disadvantage of not being fit to easy mechanical planting, cultivation, and maintenance. Nelder-Bleasdale circular designs, used in PRODEPEF trials, achieved their original purpose of demonstration of another type of experimental design which can be employed in forest trials. They certainly are appropriate for silvicultural experiments, and especially advantageous for spacing experiments. 109 Two of the basic Nelder-Bleasdale systematic designs can be used for genetic experiments. They are designs "1a" and "1b“ (and their adaptations). I do not believe, however, that they could in a satis- factory manner replace the traditional randomized complete block designs for most genetic experiments. Those systematic designs have the disadvantages that the estimates of means might be biased and that there may be no satisfactory estimate of error variance. These difficulties may to some extent be met by orienting the replicates differently and by having enough replications that analyses of parameters estimated from each replication may provide suitable esti- mates of error. Replication, randomization, and orientation of repli- cates should consequently be factors of careful consideration when using systematic designs. BIBLIOGRAPHY 110 BIBLIOGRAPHY Andersen, P. (1967). Provenance trials: Pinus elliottii and Pinus taeda. Silvic. Res. Rec. Silvic. Res. Sta. 14, 1-15. Barrett, W. H. (1974). Variaci6n geografica en Pinus elliottii Engelm. y P. taeda L. 11. Cinco afios de crecimiento en el nordeste argentino. IDIA Suplemento Forestal 8, 18-39. Barrichelo, L. E. G., Kageyama, P. Y., Speltz, R. M., Bonish, H. J., Brito, J. 0., and Ferreira, M. (1978). Estudos de procedéncias de Pinus taeda visando seu aproveitamento industrial. IPEF Boletim Informativo 18, 1-14. Beattie, W. D. , and Ferreira, J. M. (1978). Analise financeira e socio- economica do reflorestamento no Brasil. Série: Estudos Perspectivos para o Periodo 1979- 85, 1- 158. Bleasdale, J. K. A. (1967). Systematic designs for spacing experi- ments. Expl. Agric. 3, 73-85. Bleasdale, J. K. A., and Nelder, J. A. (1960). Plant population and crop yield. Nature Lond. 188, 342. Burgess, I. P. (1973). Provenance trial of P. elliottii and P. taeda at age 4 years in New South Wales. In T‘Tropical Provenance and Progeny Research and International Cooperation" (J. Burley and D. G. Nikles, eds. ),p pp 150-152. Commonwealth Forestry Insti- tute, Oxford, England. Burley, J. (1966). Review of variation in slash pine (Pinus elliottii Engelm. ) and loblolly pine (P. taeda L.) in relation to prove- nance research. Res. Commonw. For. Rev. 45(4): 126, 322-338. Canavera, D. (1971). "Results from the Pinus Provenance Tests Installed at Rio Negro, Parana" (letter dated October 5,1971, to Dr. Carl E. Ostrom). Curitiba, Parana. Conkle, M. T. (1963). The determination of experimental plot size and shape in loblolly and slash pines. North Carolina State Coll. Sch. Forest., Tech. Rep. No. 17, pp.71151. Conselho de Desenvolvimento Econ6mico. (l974). "Programa Nacional de Papel e Celulose. " CDE,Bra51lia. 111 112 Crow, A. B. (1964). Ten-year results from a local geographic seed source study of loblolly pine in southeastern Louisiana. La;_ State Univ. For. Notes 57, 1-4. Dillewijn van, F. J. et a1. (1966). "Inventario do Pinheiro do Parana. Distribuicao de Florestas de Araucaria angustifolia no Sudoeste do Parana." Codepar, FAD, Escola de Florestas, Curitiba. Dorman, K. W. (1976). "The Genetics and Breeding of Southern Pines." Agr. Handb. 471. USDA, Forest Serv., Washington, D.C. Falkenhagen, E. R. (1978). Thirty five-years results from seven Pinus elliottii (Engelman.) and Pinus taeda (L.) provenance trials in South Africa. South African Forest. J. 107, 22-36. Fonseca, S. M. da, Kageyama, P. Y., Ferreira, M., and Jacob, W. S. (1978). Sintese do programa de melhoramento genético de Pinus §pp_que vem sendo conduzido, sob a coordenacao do IPEF, na regiao sul do Brasil. I£££_Bpletim_1nfprmatiyo_18, 45-60. Gansel, C. R., Brendelmuehl, R. H., Jones, E. P. Jr., and McMinn, J. W. (1971). Seed source effects in 5- and lO-year-old test plantings of slash pine in Georgia and Florida. Forest Sci. 17, 23-30. Goddard, R. E. (1964). The frequency and abundance of flowering in a young slash pine orchard. Silvae Genet. 13, 184-186. Goddard, R. E., and Cole, 0. E. (1966). Variation in wood production of six-year-old progenies of selected slash pines. Tappi 49, 359-362. Goddard, R. E., Peters, W. J., and Strickland, R. K. (1962). C00pera- tive forest genetics research program, fourth progress report. Univ. Fla. Sch. For. Res. Rep. 7, 1-16. Goddard, R. E., and Smith, W. H. (1969). Progeny testing for inten- sive management. South.Forest Tree Improve. Conf. Proc. 10th., 1969 pp. 76-83. Goggans, J. F., Lynch, K. 0., and Garin, G. I. (1972). Early results of a loblolly pine seed source study in Alabama. Auburn Univ. Agric. Exp. Sta. Circ. 194, 1-19. Golfari, L. (1971). Confferas aptas para reflorestamento nos estados do Parana, Santa Catarina e Rio Grande do Sul. Brasil Florestal Ed. Especial Boletim Técnico No. 1, 1-71. Golfari, L., Caser, R. L., and Moura, V. P. G. (1978). Zoneamento ecoldgico esquematico para reflorestamento no Brasil (2? aproxi- macao). PNUD/FAO/IBDF/BRA-45 Série Técnica No. 11, 1-66. 113 Grigsby, H. C. (1973). South Carolina best of 36 loblolly pine seed sources for southern Arkansas. USDA Forest Serv. Res. Pap. SO—89. 1-10. Iwakawa, M., Watanabe, M., and Mikami, S. (1964). A provenance trial of loblolly pine. Five-year results at Dneyama test plantation. For. Exp, Sta. Bull. 170, 85-98. Krall, J. (1969). Growth of loblolly pine in Uruguay. J. Forest. 67, 481. Krall, J. (1973). Introduction of provenances of Pinus taeda in Cerro Largo, Uruguay. Ig_"Tropica1 Provenance and Progeny Research and International Cooperation" (J. Burley and D. G. Nikles, eds.), pp. 146-149. Commonwealth Forestry Institute, Oxford, England. Kraus, J. F. (1967). A study of racial variation in loblolly pine in Georgia--tenth-year results. South. Forest Tree Improve. Conf. Proc. 9th, 1967 pp. 78-85. Lantz, C. W., and Hofmann, J. G. (1969). Geographic variation in growth and wood quality of loblolly pine in North Carolina. South. Forest Tree Improve. Conf. Proc. 10th, 1969 pp. 175-188. Miller, W. F. (1970). "Report to the Brazilian Government." Mississippi State Contract, USAID/RID/ARDD. Milne, A. (1959). The centric systematic area-sample treated as a random sample. Biometrics 15, 270-297. Namkoong, G. (1966). Application of Nelder's designs in tree improve- ment research. South. Forest Tree Improve. Conf. Proc. 8th, 1965 pp. 24-37. Nanson, A. (1972). The provenance seedling seed orchard. Silvae Genet. 21, 243-249. Nelder, J. A. (1962). New kinds of systematic designs for spacing experiments. Biometrics 18, 283-307. Nikles, D. G. (1962). "Tree Breeding in Queensland 1957 to 1962." Brisbane, Australia. Perez, C. A. (1967). "Disease and Insect Susceptibility and Species Adaptability of Some North American Forest Tree Species Planted in Colombia.“ Medellin, Colombia. 114 PrevOst, M. J., Barnes, R. D., and Mullin, L. J. (1973 a). Pinus elliottii Engelm. provenance trials in Rhodesia. Ip_"Tropical Provenance and Progeny Research and International Cooperation" (J. Burley and D. G. Nikles, eds.), pp. 153-162. Commonwealth Forestry Institute, Oxford, England. PrevOst, M. J., Barnes, R. D., and Mullin, L. J. (1973 b). Pinus taeda L. provenance trials in Rhodesia. Ip_"Tropical Provenance and Progeny Research and International C00peration" (J. Burley and D. G. Nikles, eds.), pp. 134-145. Commonwealth Forestry Institute, Oxford, England. PRODEPEF. (l973). "IBDF/FAD Provenance Trials." Centro de Pesquisas Florestais da Regiao Sul, Curitiba. (Mimeographed handout). PRODEPEF. (1976). Centro de Pesquisas Florestais da Regiao Sul- Programacao Técnica. PNUD/FAO/IBDF/BRA-45 Série Divulgacao No. 12, 1-144. Shimizu, J. Y. (1978). Crescimento juvenil do Pinus elliottii Engelm de alto e baixo rendimento em resina no sul de 530 Paulo. Brasil Florestal 33, 24-27. Siqueira, J. D. P. (1977). Tabelas de volume para povoamentos nativos de Araucaria angustifolia (Bert.) O. Ktze, no sul do Brasil. Floresta 8 (1), 7-12. Slee, M. U., and Reilly, J. J. (1966). Recent highlights of tree breeding in Queensland. 6th World For. Congr. Proc. Vol. II, 1773-1779. Snyder, E. 8., Wakeley, P. C., and Wells, 0. O. (1967). Slash pine provenance tests. J. Forest. 65, 414-420. Squillace, A. E. (1966 a). Combining superior growth and timber quality with high gum yield in slash pine. South. Forest Tree Improve. Conf. Proc. 8th, 1965 pp. 73-76. Squillace, A. E. (1966 b). Geographic variation in slash pine. Forest Sci. Monogr. 10, 1-56. Streets, R. J. (1962). "Exotic trees in the British Commonwealth." Clarendon Press, Oxford. van Buijtenen, J. P. (1969). The impact of state-industry cooperative programs of tree planting. For. Farmer 29(2), 14, 20, 22. Wakeley, P. C. (1944). Geographic source of loblolly pine seed. J. Forest. 42, 23-32. 115 Wakeley, P. C. (1961). Results of the southwide pine seed source study through 1960-1961. South. Forest Tree Improve. Conf. Proc. 6th,_1961 pp. 10-24. Wakeley, P. C., and Bercaw, T. E. (1965). Loblolly pine provenance test at age 35. J. Forest. 63, 168-174. Wells, 0. O. (1969). Results of the southwide pine seed source study through 1968-1969. South. Forest Tree Improve. Conf. Proc. 10th, 1969 pp. 117-129. Wells, 0. O. (1975). Geographic variation in southern tree species. LSU For. Sym. Proc. 24th, 1975 pp. 19-31. Wells, 0. 0., and Switzer, G. L. (1971). Variation in rust resistance in Mississippi loblolly pine. South. Forest Tree Improve. Conf. Proc. 11th, 1971 pp. 25-30. Wells, 0. O., and Wakeley, P. C. (1966). Geographic variation in survival, growth and fusiform-rust infection of planted loblolly pine. Forest Sci. Monogr. 11, 1-40. Wright, J. W. (1963). The design of field tests. Invited paper at World Consul. Forest Genet. Tree Improve. 1963 Stockholm. FAO, Rome. Wright, J. W. (1978). A simplified design for combined provenance and progeny testing. Silvae Genet. 27, 68-70. Wright, J. W. (1980). The role of provenance testing in tree improvement. To be published In "Kedharnath Commemoration Volume on Forest Genetics," India. Wright, J. W., and Freeland, F. 0. Jr. (1959). Plot size in forest genetic research. Pap. Mich. Acad. Sci., Arts, Letters Vol. XLIV, 1959 (1958 meeting), 177-182.