UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Artificial hybrids of B.C. spruce species : growth, phenology and cold hardiness Kolotelo, David 1991

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-UBC_1991_A6_7 K64.pdf [ 6.38MB ]
Metadata
JSON: 831-1.0098510.json
JSON-LD: 831-1.0098510-ld.json
RDF/XML (Pretty): 831-1.0098510-rdf.xml
RDF/JSON: 831-1.0098510-rdf.json
Turtle: 831-1.0098510-turtle.txt
N-Triples: 831-1.0098510-rdf-ntriples.txt
Original Record: 831-1.0098510-source.json
Full Text
831-1.0098510-fulltext.txt
Citation
831-1.0098510.ris

Full Text

A R T I F I C I A L HYBRIDS O F B . C . S P R U C E SPECIES: G R O W T H , P H E N O L O G Y A N D C O L D HARDINESS By David Kolotelo B. Sc.F. University of New Brunswick A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF T H E REQUIREMENTS FOR T H E DEGREE OF M A S T E R OF SCIENCE in T H E FACULTY OF GRADUATE STUDIES FOREST SCIENCE We accept this thesis as conforming to the required standard T H E UNIVERSITY OF BRITISH COLUMBIA March 1991 © David Kolotelo, 1991 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, 1 agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Forest Science The University of British Columbia 2075 Wesbrook Place Vancouver, Canada V 6 T 1W5 Date: A B S T R A C T The usefulness of interspecific crosses between Sitka spruce (Picea sitchensis (Bong.) Carr.) and interior spruce (Picea glauca (Moench) Voss, Picea engelmannii Parry and their hybrids) was investigated in one coastal environment. For height growth and bud set most of the variation was at the regional and individual cross level, but very little variation was at the subregional level. The main genetic effects, male and female, accounted for a majority of the genetic variance and additive genetic effects are interpreted as the main factor in the determination of height growth and bud set. For bud set the maternal source of variation accounted for the majority of the genetic variance and a maternal influence on bud set is suggested. Some specific cross combinations were outstanding in height growth and non-additive genetic factors are considered important in these crosses. For bud break most of the variation was due to the residual error, although regions and crosses were statistically significant sources of variation. The Female*Male term was the most important genetic source of variation although bud break is not considered to have as much genetic variation as height and bud set. Large differences were found in the pattern of cold hardiness in the fall and it is considered that photoperiod plays a much larger role than previously thought, especially for interior spruce. Most of the variation was again at the regional and individual cross level. The intermediate performance of the hybrids suggests an inheritance of cold-hardiness based on additive genetic effects. Recommendations are given in the text for the use of these hybrids as well as the areas in which further research would be desireable. ii \ Table of Contents A B S T R A C T ii List of Tables vi List, of Figures viii Acknowledgements ix 1 I N T R O D U C T I O N 1 2 L I T E R A T U R E R E V I E W 7 3 M A T E R I A L S A N D M E T H O D S 18 3.1 H Y B R I D I N D E X 20 3.2 M E T H O D S 22 3.3 M E A S U R E M E N T S . 25 3.4 D A T A A N A L Y S I S 27 3.4.1 S E E D T R A I T S . . . 28 3.4.2 H E I G H T G R O W T H A N D P H E N O L O G Y 29 3.4.3 C O L D H A R D I N E S S . . . . . 36 4 R E S U L T S 39 4.1 S E E D T R A I T S 39 4.2 H E I G H T G R O W T H 42 4.2.1 H I E R A R C H I C A L A N A L Y S I S 43 i i i 4.2.2 F A C T O R I A L ANALYSIS 49 4.3 P H E N O L O G Y 50 4.3.1 H I E R A R C H I C A L ANALYSIS 55 4.3.2 F A C T O R I A L ANALYSIS 56 4.4 C O L D HARDINESS 59 5 DISCUSSION 66 6 C O N C L U S I O N S 82 7 R E C O M M E N D A T I O N S 84 L I T E R A T U R E C I T E D 86 A P P E N D I C E S 98 A The origins of the parent trees. 98 B The ingredients of the soil mixture 100 C Range of coefficients of the hierarchial analysis 101 D Range of coefficients of the factorial analysis 102 E Filled seed %, germination % and seed weight (g/100 seeds) of the crosses as defined in Figure 2. 103 F Sources of variation, degrees of freedom (DF) and mean squares (MS) for the hierarchial a) and factorial b) A N O V A for first year budset at three dates 104 iv G Sources of variation, degrees of freedom (DF) and mean squares (MS) for the hierarchial a) and factorial b) A N O V A for first year budbreak, at three dates 105. H Sources of variation, degrees of freedom (DF) and mean squares (MS) for the hierarchial a) and factorial b) A N O V A for second year budset, at three dates 106 v List o f Tables 3.1 Hybr id index values and taxonomic designation for parents from the East Kootenay ( E K ) and Prince George ( P G ) regions 21 3.2 Expected mean squares for the hierarchical A N O V A . . . 31 3.3 F-tests for the hierarchical A N O V A 32 3.4 Expected mean squares for the factorial A N O V A 35 3.5 F-tests for the factorial A N O V A 35 3.6 Expected mean squares for hierarchial cold-hardiness A N O V A . 1 38 3.7 F-tests for the hierarchial cold-hardiness A N O V A 38 4.8 Mean values of filled seed % ( F L L D S D % ) , germination % ( G E R M % ) and seed weight based on 100 seeds ( S D W T ) for the regional crosses 40 4.9 One-way analysis of variance for filled seed % ( F L L D S D % ) , germination % ( G E R M % ) and seed weight ( S D W T ) by female, male and region (variables as defin ed in Table 4.8) 41 4.10 The percentage of seed from each regional cross which was planted on a particular date 42 4.11 Mean height by region with range of cross means for 1988 and 1989 in cm. 2 44 4.12 Results of the hierarchical A N O V A for the 1988 and 1989 height measure-ments 44 4.13 General Combining Ab i l i t y ( G C A ) values for height of the Regions and Subregions in cm 45 vi 4.14 The General Combining Ability (GCA) values for height of each parent in cm 46 4.15 Specific Combining Ability (SCA) values for height of Regions and Subre-gions in cm 46 4.16 Specific Combining Ability (SCA) for height of Crosses in cm 48 4.17 Heterosis (HET) values for height of Regions and Subregions in cm. . . . '48 4.18 Heterosis (HET) values for height of the crosses in cm 49 4.19 Results of the factorial A N O V A For 1988 and 1989 height measurements 51 4.20 Results of the hierarchial A N O V A for first and second year phenology . . 56 4.21 Results of the factorial A N O V A for first and second year phenology . . . 58 4.22 Results of the hierarchial A N O V A for cold hardiness 65 vn List of Figures 3.1 The location of populations used in this study and the seed orchards where they are maintained.3 19 3.2 A hybrid index for interior spruce based on cone traits 20 3.3 The mating design for the hybrid spruce project.4 22 4.4 The pattern of first-year budset by regional cross 52 4.5 The pattern of first-year bud break by regional cross 53 4.6 The pattern of second-year budset by regional cross 54 4.7 L T 5 0 values for the regional crosses on six test dates 61 4.8 Percentage damage to shoots from regional crosses at three temperatures on September 23 62 4.9 Percentage damage to shoots from regional crosses at three temperatures on October 24 63 4.10 Percentage damage to shoots from regional crosses at three temperatures on November 21 64 v n i Acknowledgements I am deeply indebted to Don Lester, my research supervisor, for initiating this research project, providing funding, helping in all aspects of this thesis and always having the time for me as a student and friend. In the area of quantitative genetics I must recognize the valuable help of Judy Loo-Dinkins for continous help in this sometimes slippery field. John Worrall for being himself, Tony Kozak for help with statistics and the other members of my committee, Jack Woods and Ray Peterson, for helpful comments and patience. Fellow graduate students played a large role in data collection, cone extraction and are also appreciated for the insightful discussions of genetics and the possibility of any of us getting jobs will be remembered. The following people must be recognised for their contributions to this thesis: Barb Thomas, Marilyn Cherry, Greg O'Neill, Mathew Koshy, Salim Silim, Donna Robertson and John Russell with sincere gratitude and not just chocolates. Special thanks must also be given to Don Fowler who sparked my interest in this exciting field. Last but certainly not least I must thank Joy-Anne Zwierink for Love, patience and understanding during my struggle. ix Chapter 1 I N T R O D U C T I O N Hybridization can be denned as the act of crossing parents which are genetically unlike (Abercrombie et al, 1982). In forestry, hybridization is usually restricted to describe crosses between differing species, but varietal and provenance hybrids are also commonly referred to. Early tree breeders considered interspecific hybridization to be the answer to increasing the yield and pest resistance of forest trees. Although hybridization has great potential as a breeding tool, its major contributions to date have been in aiding our understanding of phylogenetic relationships (Wright, 1955; Fowler, 1983), crossing barriers (Mikkola, 1969) and natural hybridization (Manley, 1975) rather than producing improved seedlings for reforestation. While hybridization is still being pursued in some tree improvement programs, most agencies have abandoned it for more conventional selection practices. A good review of early work in forest tree hybridization is provided by Schmitt (1975). The primary reason used to justify hybridization in tree improvement is the possibility of non-additive genetic effects playing a major role in traits of economic importance.-If .the traits are inherited in an additive fashion, we would expect the hybrids to be intermediate. Deviations from intermediacy can be in the positive direction, properly called heterosis (Shull,1948), or the negative direction and clonal selection can produce good genetic improvement based on the heterotic genotypes. At the extreme, hybrids can surpass both parents and this is referred to as hybrid vigor. This phenomenon^ is quite rare in forest trees, but good examples are the Larix decidua X leptolepis and Populus 1 Chapter 1. INTRODUCTION 2 nigra X deltoides hybrids. Hybr id iza t ion is also very appealing for the introduction of specific traits from one species to another, such as disease resistance or frost hardiness. The Fi hybrids are usually backcrossed to the parental species into which one wishes to introduce the novel trait. This has proved successful in breeding for Chestnut blight resistance in the Amer-ican Chestnut (Castanea dentata (Marsh.) Borkh.) and with fusiform rust resistance in the Pinus taeda X echinata hybrid . The third rationale for hybridizat ion is to utilize the intermediacy of the hybrids. Al though separately the traits may be far superior in the original parents, the benefit of combining these traits in one type of tree may be economically appealing or biologically necessary. The most well known success is the tradeoff in form and hardiness with the Pinus rigida X taeda hybrid used in Korea (Hyun, 1972). The most obvious problem in species hybridization is that of incompatibility or in-viabi l i ty between species (Kr iebel , 1975). Pre- and post-zygotic barriers combine to determine the crossability of species. Many genera such as Acer, Picea, Pinus and Pop-ulus show distinct crossability patterns and sections based on these patterns have been suggested for these genera. In the spruces, there are contradictory views as to whether there should be sections. Classically, the genus has been divided into three sections (F lor in , 1960; Gaussen, 1966), but more recently authorities tend to agree that the genus should be divided into only two sections (Mikkola , 1969; Fowler, 1983). Others feel that the spruce gene pool has been relatively static and sections are unwarranted (Wright, 1955; Hoffmann & Kleinschmit , 1979). Whether or not we divide the genus into sections we cannot ignore the large body of evidence describing crossing barriers in Picea (Wright, 1955; M i k k o l a , 1969; Fowler, 1983). For forest trees, it has generally been assumed that i) isolation mechanisms leading to speciation have resulted from geographic separation followed by genetic differentiation Chapter 1. INTRODUCTION 3 and ii) the more closely related two species are the easier they can be crossed (Fowler, 1987). Therefore, crosses among species widely separated geographically are assumed to be more distantly related and to produce relatively few, if any, seedlings due to a lower amount of chromosome homology (Kleinschmit,1979). Relatively recent work has suggested that post-zygotic selection of hybrids is acting on sympatric spruce species in addition to gradual genetic differentiation resulting in pre-zygotic isolating mechanisms (Manley, 1975; Gordon, 1976; Fowler, 1983). If these post-zygotic crossing barriers are important then crossability may not be a good indicator of species relationships (Fowler, 1987). Besides species compatibility, hybridization poses problems in forest trees with respect to the amount of variation present in the offspring. The generation is considered to be very uniform when the crosses are between highly inbred, therefore largely homozygous, plants such as field crops. This is not the case with forest trees for it is known that most trees are highly heterozygous and that most traits are under multigenic control (Lester, 1975). Therefore, the Fj cross will not produce uniform offspring and a large amount of variability will be present within families. This is primarily due to the formation of gene combinations which would not occur naturally. The presence of additional variability, if present in the direction of selection, allows one to increase gain by selecting individuals within families, but causes problems with conventional seed production. As tree breeding goes beyond the second generation one can expect an increase in the breakup of linkage groups and increased variation to select from. Specific superior individuals are very difficult to reproduce in a conventional seed or-chard due to pollen contamination, heterozygosity and recombination within the genome. Vegetative propagation methods such as those proposed by Russell and Ferguson (1990) would probably be utilized to reproduce these superior individuals. Another complicating factor is the need to develop and improve two species to produce one crop species. For Chapter 1. INTROD UCT10N 4 spruce in B.C. , a large interior spruce (Picea glauca (Moench.) Voss, Picea engelmannii Perry and their natural hybrids) improvement program has identified superior genotypes, but very little has been done to date with Sitka spruce (Picea sitchensis Carr.). Hybrid programs are primarily aimed at producing first generation hybrids. There is concern about advanced generation breeding, as it is hypothesized that this will disrupt tightly linked gene complexes which have evolved over a very long time period. These co-adapted gene complexes have not been demonstrated in forest trees, although it is most probable that they do occur (Lester, 1975). Hybrid programs can be realized at various levels depending on the desired intensity of the program and the variance structure of the crop. If a large portion of the genetic variance is between provenances, then crosses between the best provenances of each species may be the most efficient strategy for a hybrid program. If there is a large degree of within-provenance variation, or of non-additive genetic variance, then one may wish to make crosses between the the very best individuals or take advantage of exceptional combinations to capture the greatest amount of genetic gain. Crosses between the best individuals from each species may not produce the greatest amount of increased vigour, but they are the logical trees to initiate a program with. Spruce species (Picea spp.) constitute a very important portion of the forest resource in British Columbia. They account for over one fifth of an annual harvest of about 56 000 000 m 3 of wood. In terms of planting, interior spruce is the most planted species group in the province with over 91 000 000 seedlings planted in 1988 (Richmond, 1989). The wood from interior spruce is used mainly for pulp and structural timber. The coastal spruce species, Sitka spruce, can be represented by very large trees, up to 100 m in height and 3 m in diameter. This would be a preferred species on many coastal sites, but use of Sitka spruce in reforestation has been limited due to the Sitka spruce weevil (Pissodes strobi (Peck)). This pest bores into and destroys the terminal C h a p t e r 1. INTRODUCTION 5 leader resulting in reduced growth and inferior stem form due to the frequent occurrence of mutiple leaders following attack. The planting effort in this species is about 6 625 000 seedlings annually, mostly in the Queen Charlotte Islands where the weevil is not present (Richmond, 1989). White and Engelmann spruce are very closely related species and hybridize naturally to a great extent at mid-elevations in the B.C. interior (Daubenmire, 1974). The species and hybrids are almost indistinguishable and this has led to them being combined into a single taxon called interior spruce, a category used by foresters throughout the province. Sitka spruce and white spruce also hybridize in the river valleys along the coast in the northern part of the province (Daubenmire, 1968). Taxonomically these are two quite distinct species and seedlots from areas of natural hybridization cause problems due to the unpredictable performance of the introgressed progeny. Hybridization between Sitka and Engelmann spruce has been suggested to occur in Southwestern B .C . (Krajina et al., 1982) but has not been well documented based on traditional taxonomy. Hybrids have been found using molecular markers, at B.C. Research, Vancouver, to occur between Sitka and Engelmann spruce, but the extent of hybridization has not been investigated (Ben Sutton, pers. comm.). In British Columbia, seed from the Nass-Skeena region of the province has proven to be a problem in nurseries. White spruce requires an extended photoperiod when grown in Southern B .C . nurseries in order to meet the height standards set out by the B .C. Ministry of Forests (BCMOF) . When Sitka spruce is grown with an extended photoperiod, the seedlings grow beyond target height and have weak stems in proportion to their height. With introgressed populations the problem is that neither the Sitka or interior spruce growing regime is ideal, although the Sitka spruce regime is considered to be preferential for the hybrids. This problem is currently being investigated by the B .C . Forest Service (Woods, 1988). Chapter 1. INTRODUCTION 6 With species which hybridize so easily, the potential exists to use hybridization as a breeding method to genetically improve spruce planting stock for reforestation. Hy-bridization is an attractive alternative when species have complementary traits which might be combined in one type of planting stock. In British Columbia, it may be pos-sible to increase growth in the interior by introducing genes from Sitka spruce for faster or prolonged growth. Sitka spruce cannot grow in the interior as it cannot tolerate the continental climate of interior B . C . Even if the hybrids are not suitable to the harsh interior climate, extension of the ecological amplitude of Sitka spruce to climates with less of a maritime influence would be desirable. With Douglas-fir Pseudotsuga menziesii Mirb.(Franco) a similar program with crosses between the coastal variety (var.menziesii) and the inland variety (var. glauca) have proven to be very successful. After 10 years of field testing the hybrids survived just as well as the inland parents, but were twice as tall (Rehfeldt, 1986). In addition, high within-family variability was found and this would result in even greater gains with efficient selection techniques. Douglas-fir varieties overlap much of the range of Sitka spruce, except the Queen Charlotte Islands, and a large portion of the range of interior spruce providing encouragement for the possible success of spruce hybridization. This thesis was initiated in order to assess the usefulness of hybridization as a prac-tical breeding method for spruce in British Columbia and is concerned with three major objectives: 1. To investigate the growth potential and adaptability of hybrids at various levels (Region, Subregion and Family). 2. To estimate genetic parameters for the hybrids. 3. To estimate the relative importance of additive and non-additive gene control in several traits. Chapter 2 L I T E R A T U R E R E V I E W Spruce species in western North America are thought to have originated from Asiatic progenitors, during the Cretaceous period, before continental separation. It is generally agreed that white spruce is the living taxon most representative of that migration. Two different phylogenies have been proposed for the evolution of the western spruces. The first proposes Engelmann spruce as the next offshoot from white spruce, followed much later by Sitka spruce and blue spruce (Picea pungens Engelm.) (Nienstaedt & Teich, 1971). The second theory proposes that Sitka spruce diverged early as a maritime eco-type. Engelmann spruce diverged later, from white spruce, and gave rise to blue spruce (Taylor & Patterson, 1980). Based on ecology, morphology, crossability, and the available fossil record it would appear that the second theory is more likely. In the western spruces, natural hybridization has been well documented between white and Engelmann spruce (Garman, 1957; Roche, 1969; Daubenmire, 1974). Roche (1969) even states that these two species represent extremes of a clinal pattern of variation which is caused by introgression at mid-elevation slopes where the species meet. An extreme view is taken by Taylor (1959) who recommends that the two species be considered as subspecies of a single species. The extent of this taxonomic problem is reflected in the use of 'interior spruce' to describe either of these species in areas of sympatry. References in the literature have also been made to Picea glauca var. albertiana (S. Brown) Sargent and Picea engelmannii var. glauca and these are most likely based on introgfessed specimens of white and Engelmann spruce (Taylor, 1959). 7 Chapter 2. LITERATURE REVIEW 8 Attempts have been made to identify unique characters in order to separate white and Engelmann spruce.. Daubenmire (1974) provides an excellent review and discussion in this area. The major characters separating the species are cone characters such as__ scale length, width, shape and type of margin, free scale length (length of scale beyond seed wing imprint) and bract shape. A recent index has been prepared by Dr. G.W. Douglas for differentiating B.C. spruce species based on cone characters and their ratios (/TI Coupe et al, 1982). Natural hybridization also occurs between white and Sitka spruce in Alaska (Copes and Beckwith, 1977) and in the Nass, Skeena and Bulkley river basins of British Columbia (Little, 1953; Garman, 1957; Daubenmire, 1968; Roche, 1969). The natural hybrid between these two species has been described and named Picea X lutzii (Little, 1953). Based on a large number of crosses, these species have been shown to be fairly compatible with a crossability of 55 % (Fowler, 1987) and are considered to be very closely related (Roche & Fowler, 1975). Patterns of isozyme distribution (Yeh & Arnott, 1986) and chloroplast D N A (Szmidt et a/., 1988) have been used successfully to separate white, Sitka and hybrid spruce seedlots. These methods may have promise in the future, but very few facilities currently have adequate equipment and expertise in these areas and the problem is currently not considered important enough to warrant the-required expenditure of funds. Chemotaxonomic studies have provided important additional information for study-ing introgressed populations as well as verifying natural and artificial hybrids in spruce. White and Engelmann spruce, although almost identical in leaf, twig and wood morphol-ogy, were found to differ substantially in leaf oil terpenes (Oglivie & Von Rudloff, 1968), cortical oleoresins (Habeck & Weaver,1969), and flavanoids and phenols (LaRoi & Dugle, 1969). Most of these differences are quantitative and exhibit clinal patterns of variation with elevation. Chapter 2. LITERATURE REVIEW 9 For hybrid verification the monoterpenes have been used to verify blue X white spruce hybrids (Bongarten & Hanover, 1982), while chromatography of leaf phenolics can be used to separate Sitka X white hybrids (Hanover & Wilkinson, 1970) and to verify the existence of the very rare white X black spruce hybrid (Riemenschneider & Mohn, 1975). In a study on variation patterns in white spruce using monoterpenes it was found, by cluster analysis, that the provenances grouped into a western and eastern grouping (Wilkinson et a/., 197]). This is in agreement with suggestions by Neinstaedt (1968) that the species is divided into two clines based on growth characteristics. While natural hybrids have been suggested to occur between Sitka and Engelmann spruce (Garman, 1957; Krajina et al.,1982) the evidence is not convincing. The white X Sitka spruce hybrids produce cones very similar to Engelmann spruce and this may explain these putative hybrids (Roche, 1969). Although these species are sympatric in some areas, a morphological sampling of seedlings and mature trees in southwestern B.C. was inconclusive with regard to the question of whether hybridization or introgression had taken place (Scagel, 1984). Blue and Engelmann spruce are believed to be very closely related species based on morphology (Wright, 1955; Daubenmire, 1972). The ranges of these species overlap greatly in the western United States, yet blue spruce occurs at lower elevations than Engelmann spruce. The opportunity for hybridization is present at intermediate eleva-tions where there is a slight altitudinal overlap. Studies have shown that introgression is minimal in sympatric populations (Taylor et al, 1959) and this is due to strong, yet incomplete, incompatibility between the two species (Kossuth & Fechner, 1973). A mor-phological index and a discriminant function have been used by Schaeffer and Hanover (1985) in the Scotch Creek Drainage in Colorado to differentiate blue and Engelmann spruce. The methods seem highly reliable as the groupings explained 95 % of the variance and should be useful in future studies of natural hybridization between these two species. Chapter 2. LITERATURE REVIEW 10 The white-Engelmann-Sitka spruce complex has received much taxonomic attention (Garman, 1957; Daubenmire, 1968, 1974; Roche, 1969), but relatively little work has been done in terms of improving these hybrids for reforestation in B .C . It is known that certain low elevation interior spruce provenances grow as well as Sitka spruce in coastal nurseries (Roche, 1970). The lack of work with Sitka-interior spruce hybrids is largely due to the lack of effort which has been given to the genetic improvement of Sitka spruce in British Columbia. This is mainly due to the white pine weevil (Pissodes strobi Peck.) which is why Sitka spruce has been planted on such a small scale in its natural range while it has done exceptionally well as an exotic in Britain. Studies in Washington state have shown that white X Sitka spruce hybrids are less susceptible to the weevil based on 16-year-old plantations (Mitchell et al., 1974). In British Columbia, based on local tests, the general consensus is that there do not appear to be any fully resistant-populations among Sitka spruce, white spruce and the hybrids (Lester, pers. comm.). There is evidence supporting the view that some Sitka spruce provenances from the dry maritime biogeoclimatic zone are above average in resistance (Woods, pers. comm.) and that large family differences with respect to.susceptibility exist in interior spruce (Kiss, pers. comm.). Spruce hybridization programs are underway in many areas and the major emphasis is on increasing growth rate while still maintaining an adequate level of cold hardiness. In Ontario, various western spruce hybrids were grown in a number of tests and it was found that the hardiness was highly dependant on the proportion of white spruce germ plasm in the hybrid. The results also support the view of polygenic inheritance of cold-hardiness. These tests ranged from 10 to 22 years of age (Ying & Morgenstern, 1982). A spruce hybridization program is currently underway in the Maritimes in order to improve upon the native species. Attempts have been made with a variety of species, but the most promising are the combinations of black spruce (Picea mariana (Mill.)B.S.P.) Chapter-2. LITERATURE REVIEW 11 with Sitka spruce (Fowler, 1983) or Serbian spruce (Picea omorika (Pancik) Purkyne) (Fowler, 1980) and the white X Sitka spruce hybrid (Fowler, 1987). In Ontario, work is also underway with hybrids to identify highly productive combinations for reforestation. Much of this work has focused on the Picea rubra Sarg. X omorika hybrid (Gordon, 1981). In Scotland, Sitka X white spruce hybrids are being investigated as an alternative [to lodgepole pine (Pinus contorta Dougl. ex Loud.) on exposed, frosty and/or dry sites that were considered marginal for Sitka spruce. The hybrids were all more frost hardy than the northernmost, Ketchikan, Alaska, Sitka spruce provenance in late summer and fall, but less hardy than the Masset provenance of Sitka spruce in early spring. The hybrids showed rapid growth, exceeding Sitka spruce in height by 10-20 % at age 10, and this was probably not solely explained by greater frost hardiness (Sheppard &i Cannell, 1985). In Germany, a large spruce hybridization program is underway in order to produce vigorous, heterotic. stock for reforestation. Hoffmann and Kleinschmit (1979) have found that many of their crosses show heterotic responses in terms of height. The development j of a large scale vegetative propagation program for spruce has made these hybrids a practical alternative in Germany, although work is still ongoing in order to determine the best species combinations (Hoffmann & Kleinschmit, 1979). Following glaciation it appears that selective pressures such as photoperiod, temper-ature and precipitation have been important in Picea based on studies in white spruce (Nienstaedt Sz Teich, 1971), black spruce (Morgenstern, 1969) and Sitka spruce (Kraus & Lines, 1976; Cannell & Smith, 1983). A simplified view of phenological adaptation in spruce is that budset is directly related to a decreasing photoperiod and bud break directly related to an accumulation of degree days above a certain threshold tempera-ture. This is considered a simplification as it has been shown that other factors can have an impact on phenology, especially bud break, such as winter chilling, photoperiod and Chapter 2. LITERATURE REVIEW 12 possibly soil temperature (Cannell Sz Smith, 1983). White spruce and Engelmann spruce are very similar in their annual growth of vege-tative buds (Owens et al., 1977; Harrison Sz Owens, 1983). Bud break is best explained using a 5°C threshold temperature from the time at which dormancy ends and quiescence begins during the winter (Owens et ai, 1977). It is usually assumed that differences in heat sums differentiate dates of bud break, but evidence has also been presented showing that high elevation provenances flush early due to a lower threshold temperature (Wor-rall, 1983). In a common environment, the high elevation provenances are the first to flush in the spring (Roche, 1969). For white spruce, the difference in bud burst between high and low elevation provenances can be as great as 6 weeks (Owens et al., 1977) and within a stand it can vary as much as 3 weeks (Nienstaedt & King, 1969). In Sitka spruce, budburst seems to also be mainly determined by heat sum accumu-lation, but genetic differences are not as obvious as with budset (Burley, 1966; Roche, 1969). Sitka spruce is said to have a high chilling requirement so that it does not break bud in the warm winter weather which often occurs in its maritime climate. Variation in budburst seems less in Sitka spruce as there is only a 5 day difference in flushing in most of the provenances grown in the United Kingdom (Cannell & Smith, 1983). It has been well recognized that within a tree the terminal bud breaks up to 3 weeks later than the laterals. The time lag seemed correlated with variability of last spring frost and it is suggested that this time lag has evolved in response to late spring frosts (Sweet, 1965). Photoperiod is considered to be the primary factor determining shoot growth cessa-tion, or bud set, in spruce (Morgenstern, 1969; Burley, 1966). The critical night length for bud set is considered to be under additive gene control in Picea abies L . , although specific crosses may show non-additive gene action (Dormling et al, 1974). Southern provenances moved North respond to the longer photoperiod by growing longer into the season and may be injured by fall frosts. Northern provenances moved South set buds Chapter 2. LITERATURE REVIEW 13 sooner than local stock and therefore do not make full use of the growing season. Other factors, such as temperature, appear to have an effect on bud set in Sitka spruce, black spruce and Engelmann spruce (Burley, 1966; Morgenstern, 1969). In other areas, such as the B .C. interior, growth cessation is caused by drought before photoperiod can play a role (Lavender, 1984). It must be noted that many researchers have referred to bud set as dormancy, although this is not necessarily the case. Dormancy is classically defined as any case in which a tissue is predisposed to grow but does not do so (Doorenbos, 1953 in Lavender, 1985). Dormancy has also been defined as the period in which there is no mitotic activity in the buds (Owens & Molder, 1973) and the period in which there is maximal resistance to stress (see Lavender, 1985 for a review of bud dormancy). The two points, bud set and bud break, commonly considered in experiments are just two points in a continuous annual growth cycle of trees. They are used mainly because of ease of measurement; they give us an idea of variability, and they are also correlated to stress resistance which is a major concern in seedling culture. A major stress on exotic, or hybrid trees in Canada is that of frost. Seedling transfer to more northern latitudes or higher elevations is often restricted by the amount of frost damage expected. Frost damage is usually associated with late spring frosts or early fall frosts. Glerum (1985) outlines 3 questions concerning frost, hardiness of hybrids or exotics: i) is the maximum winter hardiness adequate for the location? ii) do the trees become frb"st hardy early enough in the season and keep ahead of decreasing temperatures? and iii) do the trees retain frost hardiness in the spring long enough to avoid damage from mild winter or early spring dehardening? In order for the introduced seedlings to avoid frost damage these 3 conditions must, be met. Freezing injury is caused mainly due to intercellular freezing which causes dehydra-tion of the cells, cell shrinkage, an increased concentration of cell solutes, pH changes and Chapter 2. LITERATURE REVIEW 14 removal of water from macromolecules (Lyons et al, 1979). The primary effect of these changes which results in injury is membrane disruption. Hardy plants are able to resist cell membrane damage through chemical protection and membrane synthesis. This is accomplished through frost avoidance which is the delay of ice formation through super-cooling to -5° to -7°C and frost tolerance which is dependent on the hardening process in response to environmental conditions (Glerum, 1985). The current theory of frost hardiness is that it is a multiphase process with various environmental cues. Phase I is initiated in late August or September, in the North-temperate regions, in response to a decreased photoperiod. Growth cessation is concur-rent with this as well as many metabolic changes occurring in order to enable the plant to respond to low temperatures. During this phase only temperatures slightly below freezing can be tolerated and therefore the seedlings are not very hardy at this point. Phase II is initiated by mild frosts which trigger many metabolic changes within the plant and allow it to be exposed to very low temperatures without damage. An in-depth review of the metabolic changes occurring in the plant can be found in Weiser (1970). Moisture and nutrient status also have an effect on frost hardiness (see Glerum, 1985 for an excellent review of cold hardiness in coniferous seedlings) The frost hardiness of a species, or family, can be defined as the minimum temperature at which a certain percentage of individuals will survive. When it can be estimated, the lethal temperature for 50 % of the population (LT 5 0 ) is commonly used to define hardiness. Frost hardiness is directly related to stress resistance and is maximal from December through February when there is no mitotic activity. Many methods are available to determine frost hardiness (Ritchie, 1984). The classical ones have been to freeze seedlings or twig segments to desired temperatures and to then determine the proportion of damage. This can be done visually, called the browning method, or by more sophisticated techniques such as the impedance ratio or conductivity Chapter 2. LITERATURE REVIEW 15 of cell fluids. In Pinus sylvestris L. seedlings, autumn frost hardiness was under additive genetic control and no specific genetic interactions were evident following long-distance crossing (Norell et ai, 1986) In most physiological studies, and to some extent morphological studies, the samples used have usually been based on bulked provenances without any knowledge of the relat-edness of seedlings. In these cases general trends for provenances can be identified, but very little can be said of the genetic control of the traits of interest. In order to inves-tigate genetic control the experimental design must consider dividing the main effect of the provenance into hierarchial components, to determine where most of the variability lies, or into groups of related seedlings to investigate the type of genetic control for the trait in question. Quantitative genetics provides a large body of theory and literature on analysing various mating designs between different individuals. Genetic parameters such as addi-tive genetic variance and non-additive genetic variance can be estimated, following some assumptions, for individuals in various mating designs based on the covariances of rel-atives (see Falconer, 1981 or Becker, 1984 for a thorough explanation). It was found that with Sitka spruce, tree form characters (stem straightness, bud number and branch angle) were inherited in a predominantly additive fashion. For tree vigour (height, dry weight, branch number and length) additive, dominance and maternal genetic control mechanisms were suggested to occur (Samuel et ai, 1972). For Picea mariana (Mill.) B.S.P. it was found that for seedling characters, most of the genetic variation is found within families rather than populations or subpopulations. For phenological characters in black spruce, almost all of the genetic variance was explained by populations which Morgenstern (1969) concludes is due to the strong clinal pattern of variation for these characters. ' For crosses between populations it would be of interest if additive and non-additive Chapter 2. LITERATURE REVIEW 16 genetic variances for populations could be estimated. Unfortunately, the extension of quantitative genetics to the population level has not been theoretically accepted as the probability of gene identity by descent at the population level is difficult to conceptualize (Park, pers. comm.). The problem has been addressed in a more practical sense by estimating a general combining ability (GCA) value for each population and estimating specific combining ability (SCA) values for inter-population hybrids (Gerhold and Park, 1986). These pa-rameters are based on mean values and are used to compare various populations or inter-population crosses. They are not calculated on the basis of covarianc.es between relatives although they do give a general means of comparing the importance of additive and non-additive genetic variances at the population level. In their study on Scotch pine, Gerhold and Park (1986) found large differences among population hybrids and these were mainly explained by the GCA values of the parents although some large SCA values were found. A n additional parameter used to describe hybrid families is the heterosis value. It describes the hybrid family in reference to the two parental families as the deviation from the mid-parent value. The term heterosis was proposed for the increase in vigor following the union of dissimilar gametes (Shull, 1914). Therefore genetic diversity of the parents is a requirement for heterosis. Several hypothesis have been outlined to explain heterosis (Shull, 1948). One of the most widely accepted was the alternative dominant hypothesis in which detrimental recessive alleles in the hybrid are rendered ineffective by dominant alleles from the other parent. Based on quantitative genetic theory Crow (1948) has determined that if all homozygous recessives were replaced then the increase in vigour would only be a few percent. This, therefore, cannot fully explain the genetic mechanism of heterosis. Rather than masking the recessives, others argue that heterosis is due to the overdominance bf heterozygotes. Chapter. 2. LITERATURE REVIEW 17 In forest trees it is unclear whether heterosis is a masking of recessives therefore overcoming any inbreeding depression, or the positive interaction of alleles. Release from inbreeding depression would provide genetic, gain only in the first generation of breeding. Selection for more vigorous individuals, due to favourable gene interactions, could be continued throughout the entire breeding plan. Heterosis values were used by Park and Gerhold (1986) to evaluate the genetic po-tential of inter-population hybrids. While the mechanism for heterosis at the population level is not easily explained, the parameter gives us an indication of the usefulness of com-bining regions or subregions with presumably different gene frequencies. In their paper Park & Gerhold (1986) find some large heterosis values, but suggest that hybridization for heterosis solely may not be a wise decision. Chapter 3 M A T E R I A L S A N D M E T H O D S The parental trees (clones) used in this study were selected from four different areas of the province (Figure 3.1). Two populations, named Vancouver Island and Queen Charlotte Islands, represent spruce in the coastal region of B . C . and are Sitka spruce. Two populations, named East Kootenays and Prince George, represent interior spruce in B.C. Al l parents were selected from wild stands on the basis of phenotype, grafted, and are now maintained in the seed orchards or clone banks at locations illustrated in Figure 3.1. The interior parents are located at the Kalamalka Seed Orchard in Vernon, B . C . The Vancouver Island sources are in the Nootka Seed Orchard, Central Saanich, B . C . , and the Queen Charlotte Island sources are located in the Lost Lake Seed Orchard in Saanich, B . C . Parents for crossing were chosen primarily on the basis of availability of female and male buds in the spring of 1987. In all populations, reproductive buds were numerous on several clones. For interior spruce, data on phenology of vegetative budburst was used to identify genotypes across the range of variation in timing of budburst. Some parents were chosen specifically to represent early- and late flushing genotypes. Comparable variation in phenology was not observed in Sitka spruce. A listing of clones and their origins is presented in Appendix A. 18 Figure 3.1: The location of populations used in this study and the seed orchards were they are maintained.1 Regions are defined as: QCl=Queen Charlotte Islands; VI=Vancouver Island; EK=East Kootenays and PG=Prince George. The seed orchards are defined as K=Kalamalka Research Station, N = Nootka Seed Orchard and LL=Lost Lake Seed Orchard. 19 Chapter 3. MATERIALS AND METHODS 20 3.1 H Y B R I D I N D E X For the purposes of this study, it was desirable to classify the interior spruce parents into white, hybrid or Engelmann spruce. This should help to better explain the results and give us an idea of whether the crosses used in this thesis are Fj crosses. A key to the native Picea species and hybrids in B.C. is available (In Coupe et al, 1982), but it produced very unsatisfactory results. Many of the Prince George and East Koptenay parents keyed out to white X Sitka hybrids, which is a virtual impossibility as Sitka spruce does not occur naturally within 400 km of the parent trees. A simple hybrid index based on four cone traits was devised from a review of pertinent literature (Rehder, 1947; Wright, 1955; Garman, 1957; Den Ouden & Boom, 1982). Three qualitative and one quantitative character are used in this index (Figure 3.2). The three qualitative characters: % F R E E S C A L E 5.1-12 = 1 12.1-19 = 2 19.1-26 = 3 26.1-33 = 4 33.1-40 = 5 S C A L E S H A P E OBOVATE = 1 MEDIUM = 2 (1) RHOMBIC = 3 S C A L E M A R G I N ENTIRE = 1 MEDIUM = 2 SERRATE = 3 B R A C T S H A P E -SHORT, B L U N T = 1 MEDIUM = 2 (1) ft (3) LONG, POINTED = 3 WHITE SPRUCE 4.00 - 7.33 HYBRID SPRUCE 7.34 - 10.66 E N G E L M A N N SPRUCE 10.67 - 14.00 C7 - 0 (1) (3) 6 Figure 3.2: A hybrid index for interior spruce based on cone traits. Chapter 3. MATERIALS AND METHODS 21 ovuliferous scale shape, ovuliferous scale margin and bract shape have been mentioned in the literature to be distinct between the species. Samples were scored by comparison with published diagrams of 'pure' species ideotypes. For each trait, a value of 1 denotes white spruce, 2 denotes hybrid spruce and 3 denotes Engelmann spruce. The fourth character is the free scale percentage calculated as the percentage of the ovuliferous scale extending beyond the seed wing imprint. This character was considered highly reliable by Daubenmire (1974) and a similar character, the reciprocal was consid-ered a good character by Roche (1969). This fourth character in the index was given a higher weighting, 1 to 5 versus 1 to 3, due to high reliability. The interior spruce parents were characterized as 12 white spruce, 7 hybrid spruce and 1 Engelmann spruce based on this index (Table 3.1). The index seemed adequate in separating types and was in general agreement with expectations based on elevational trends (Appendix A) . Although all parents from a higher elevation were not closer to Engelmann spruce, it is reasonable to assume local geography plays a large role in the elevations at which the hybrid zone occurs. Table 3.1: Hybrid index values and taxonomic designation for parents from the East Kootenay (EK) and Prince George (PG) regions. P A R E N T S H Y B R I D I N D E X P A R E N T S H Y B R I D I N D E X EK 4 9.9 hybrid PG 21 6.3 white EK 48 4.0 white PG 134 5.8 white EK 23 5.2 white PG 106 5.3 white EK 81 6.7 white PG 32 8.5 hybrid EK 3 8.0 hybrid PG 122 6.7 white EK 20 5.1 white PG 25 7.5 hybrid EK 82 5.2 white PG 155 9.9 hybrid EK 6 6.0 white PG 4 6.3 white EK 89 10.1 hybrid PG 42 6.3 white EK 101 13.5 Engelmann PG 135 8.8 hybrid Chapter 3. MATERIALS AND METHODS 22 3.2 METHODS From each of the 4 seed orchards, there were 10 clones selected (5 used as males; 5 used as females). The crossing design is illustrated in Figure 3.3. The design has two regions (coast and interior) and they are crossed in all combinations to produce the four types of regional crosses which are referred to in the text simply as regions for brevity. For each region, two Subregions were selected and the combination of all of these subregions results in 16 hybrid groups which are referred to as subregions within the text. At the individual tree level there are 80 full-sib Crosses in the design. This design was chosen so that the analysis could be performed as a nested design to elucidate genetic hierarchy of spruce hybrids or as 5 sets of 4 X 4 factorials to look at genetic parameters. QCI VI EK PG 3 2 3 2- 1 1 1 1 1 1 9 1 1 7 7 3 9 2 5 5 2 8 8 0 2 5 4 3 3 1 2 4 6 0 1 0 3 0 2 6 9 1 5 5 4 1 1 5 220 1 11 41 51 Q 209 2 12 42 52 c 242 3 13 43 53 I 260 4 14 44 54 266 5 15 45 55 149 6 16 46 56 15 7 17 47 57 v 187 8 18 48 58 I 58 9 19 49 59 161 10 20 50 60 4 21 31 61 71 48 22 32 62 72 E 23 23 33 63 73 K 81 24 34 64 74 3 25 35 65 75 4 26 36 66 76 134 27 37 67 77 P 106 28 38 68 78 G 32 29 39 69 79 122 30 40 70 80 Figure 3.3: The mating design for the hybrid spruce project.2 2abbreviations refer to regions QCI=Queen Charlotte Islands; VI=Vancouver Island; EK—East Kootenays and PG=Prince George. Pareival numbers, females along top and males on side, are given Chapter 3. MATERIALS AND METHODS 23 For the coastal trees, pollen was obtained by picking individual pollen cones, prior to pollen shed, separately for each clone in the spring of 1987. The microstrobili were then extracted in the pollen extraction room at the Canadian Pacific Forest Products ( C P F P ) Research Centre in Saanich, B.C. For the interior trees, pollen.was collected and extracted in a similar manner at the Kalamalka Research Station in Vernon, B.C. (Figure 3.1) in the spring of 1985 or 1986 and stored in a dessicator at - 20°. Fresh pollen from the interior could not be used on the coast because pollen maturation in the interior occurs after megastrobili receptivity on the coast. Therefore, stored pollen was used for all crosses with interior pollen in order to maintain a common pollen pool. Female cones were enclosed in labelled paper pollination bags prior to strobili emer-gence in late March and early April . Any microstrobili on the branches were removed before closing the bags in order to avoid contamination. Several check bags were also placed on the clones in the orchard to test the procedures. When these bags were har-vested in the fall, no filled seed was found in any of the cones. Pollinations were conducted with pollinating syringes in April as each clone became receptive. Most bags were pol-linated twice to increase the probability of good seed set. After the strobili had closed, in May, the paper bags were replaced by fiberglass net bags to allow air flow around the strobili, keep insects away from the cones, avoid very high temperatures in the pollina-tion bags while ensuring that no seeds would be lost as cones became mature. Crosses were monitored for problems at least, once a month before harvest. The coastal trees experienced problems with aphid damage on the strobili. Aphids were controlled by a spray of Safer's insecticidal soap at a concentration of 25 cc per litre. In late August and early September, mature cones were collected from parent trees and placed in cloth bags. Seeds from crosses made at Kalamalka were extracted there and crosses made on the coast were extracted at the C P F P Research Centre. A l l strobili were with each subregional code and code numbers for each cross are given within the figure. Chapter 3. MATERIALS AND METHODS 24 air dried and seed was removed by shaking the bags. For the coastal crosses an additional wetting and oven-drying was required to open some strobili. Seed was separated into full and empty seed with a seed blower. Seed was then stored in a cooler at 0°C until the following spring. The seeds were stratified, in early May, by soaking them overnight in petri dishes, surface drying them and placing them between moist paper in a cooler at 0 °C for 21 days. The petri dishes were then placed on the germination bench with a 16-hour daylength. The germinants were dibbled into styroblock cavities using tweezers and a laboratory spatula from June 3-14, 1988. The styroblocks used in this study were CFS 415B's with 112 cavities per block and a cavity volume of 102 cm 3 . The soil mixture used in this experiment was a 3:2 peat:vermiculite mixture with pH modifiers and micronutrients added. The exact contents of the soil mixture are given in Appendix B . The experimental design was a randomized complete block with 8 blocks and 8-tree plots. Each replicate contained 7 styroblocks. In order to fill these blocks some crosses appeared more than once in a replicate. The experiment was not completely balanced as some crosses did not produce enough seed and/or seedlings to complete the experimental design. Additional seedlings for use in cold-hardiness testing were dibbled into an additional 44 styroblocks in a completely randomized design. These blocks contained the remaining germinants and were not completely filled. A l l crosses were randomized using a random number generator to generate styroblock and plot number. Seedlings were watered and fertilized with a solution of 20-20-20 three times weekly during the growing season. Survival was far from homogeneous as the 1st and 2nd replicate suffered severe mortality. Much of this appeared to be dessication and it was deduced that this was due to the very warm weather at the time of planting, but these blocks also showed signs of insect damage. After the first season of growth, the experiment was consolidated into 6 replicates and the extra blocks were also reduced in number. For Chapter 3. MATERIALS AND METHODS 25 each seedling transferred its initial block number was recorded so that block effects, if present, could be accommodated. The whole experiment was reduced from 100 sty-roblocks to 60. During the winter months, styroblocks were placed on the ground to avoid any frost damage to the roots. 3.3 M E A S U R E M E N T S The traits chosen to be analysed for this thesis are considered to be adaptive and of inter-est to tree breeders and physiologists. For measures of growth rhythm and adaptability, budset, budburst and autumn cold-hardiness were chosen. Although dehardening in the spring would be of interest it was not compatible with the scheduling of this Masters program. Phenology and cold-hardiness are characters that are important to the fitness of individuals. These characters have evolved to enable genotypes to become adapted to their local environments. Seed characters were used as an indication of very early growth and to assess the effect of maternal inheritance. Growth in height over two seasons was chosen to represent growth potential of the seedlings. Diameter measures were also to be included, but initial attempts resulted in seedling damage due to their narrow spacing in the styroblocks and the late growth cessation of Sitka spruce. Growth of these seedlings was monitored in a soil-less growth medium in styroblock containers in a manner similiar to the way most seedlings are grown in B .C. They were exposed solely to growing conditions found in the lower mainland which is appreciably milder than most, of this province. The conclusions and recommendations of this thesis are subject to these study limitations. The author feels that while the exact dates of budset or budburst and the values of height growth may differ the patterns found would not vary appreciably. Initial measurements on seed characters were made on the individual crosses. The Chapter 3. MATERIALS AND METHODS 26 mean seed weight of 100 seeds was measured to the nearest .1 mg, although presented to 0.01 of a gram, after seeds were dried to a moisture content of 5 %. Full-seed percentages were obtained by subjecting 100 seeds to X-rays and inspecting for full embryos on the photographic film. Germination percentages were obtained as the number of germinated seeds divided by the total number of seeds stratified. The dates when germinants were dibbled into the styroblocks were recorded and used as a measure of germination speed. This allows for only a crude comparison of germination speed as on some dates more seeds had germinated than were able to be placed in the styroblocks. In the first year, three traits were measured on the seedlings: budset, total height from soil-line and budburst. The experiment consisted of approximately 3400 seedlings although this number varied depending on mortality and errors in sampling. For height and phenological traits, family means were based on 3350 to 3430 seedlings with an average of 3390. Phenology measurements were based on the terminal bud exclusively as it was easier and quicker to measure. Budset was measured on three dates in the fall of 1988: September 11th, September 25th and October 9th. A seedling was considered to have set its bud if the terminal bud was fairly obvious and the bud scales were brown or in the process of turning brown. Data were recorded as the percentage of seedlings exhibiting budset in each plot. After shoot growth had ceased, seedling height was measured before and after consolidation of the experiment. In the spring, budburst was measured on April 12th and April 16th, 1989. The terminal bud was considered burst if the bud scales were opening and a substantial portion of green foliage appeared. Budset was estimated on July 12th, July 21st, August 18th and September 20th, 1989. Second-year height measurements were made after shoot growth had ceased in November. For estimates of cold hardiness, twig segments of 2-4 cm were obtained from a sub-sample of the experiment. Twig segments were moistened, placed in plastic bags and left Chapter 3. MATERIALS AND METHODS 2 7 overnight, in a darkened freezer ( 0 ° C ) to prevent the occurrence of premature softening (Sakai & Weiser, 1 9 7 3 ) . Five twig segments were placed in each bag and three bags for each cross were used to get damage estimates at three different temperatures. Bags were attached at their tips to cardboard allowing them to hang and have free airflow around them. These cardboard pieces were placed on racks in the controlled freezer. The tem-perature of the freezer was first equilibrated at O 0 C and then decreased at the rate of 5 °C per hour to the desired temperature. The temperatures were chosen based on results of the previous experiment and the weather. The freezing chamber was then maintained at this temperature for one hour after which the samples for that temperature were re-moved. This procedure was repeated starting at the current temperature and lowered to the next chosen test temperature. Samples removed from the chamber were again placed in a darkened freezer at 0 ° C for about 1 2 hours, after which they were then laid out on a table with indirect lighting for about 5 - 7 days. The samples were then visually assessed for foliar damage. A percent damage value was estimated, to the nearest 1 0 %, for each twig segment based on the proportion of browned or badly discoloured tissue. This method allows for a fast and fairly reliable means of scanning many crosses for frost hardiness at various temperatures. 3.4 D A T A ANALYSIS A l l data analyses for this thesis were performed on the U B C mainframe computer using version 5 of SAS ( 1 9 8 5 ) . Many assumptions are involved in analysis' of variance, which is the main statistical technique used in this thesis. These assumptions have been inves-tigated and will be presented in the results section without further coverage of methods used in this section. The A N O V A tables for the phenology data are simplified and pre-sented with 3 preselected levels of significance c * = 0 . 1 = * ; Q = 0 . 0 5 = **; and a = 0 . 0 1 = Chapter 3. MATERIALS AND METHODS 2 8 ***. This has been used to avoid boring repetition of tables. While a summary format is adequate for the text, many readers will be interested in more quantitative information for each analysis and this is why each analysis is presented in the appendices (Appendices D, E and F). The degrees of freedom and mean squares are presented and from this information readers can obtain their own probability values for their own interpretation. For items which are close to being statistically significant at Q = 0 . 1 it will be attempted to present these probability values in the text for discussion of their practical significance. Presentation of the results was influenced by Warren's paper ( 1 9 8 6 ) on the presentation of statistical analyses and most, but not all, of his comments have been incorporated into the text. For sources of variation for which there is no direct error term to test against, a Pseudo F-test has been constructed and the appropriate degrees of freedom calculated as shown by Satterthwaite ( 1 9 4 6 ) . The one exception to Satterthwaite's method is that in constructing an appropriate F-test, the mean squares are never subtracted, but added to avoid invalidating the F-test (Judy Loo-Dinkins, pers. comm.). Analysis of Variance is used throughout this thesis although for analysis of quantitative genetics more powerful techniques, Best Linear Predictor (BLP) and Best Linear Unbiased Predictor (BLUP) , are currently gaining in popularity. The familiarity of A N O V A with the author and tree breeders made it the logical technique for this thesis. These modern statistical techniques are very powerful in unbalanced situations or when one may wish to select parents based on very different tests and interested readers are referred to an excellent book by White and Hodge ( 1 9 8 9 ) on the application of these techniques to tree improvement. 3.4.1 S E E D TRAITS The seed data were based on unreplicated measurements of germination %, filled seed %, and seed weight. Multiple regression analysis was used to see if an equation using Chapter 3. MATERIALS AND METHODS 29 seed weight and filled seed % could be used to predict germination. Transformations with arcsine, squareroot and logarithmn were attempted to improve normality, but none of these worked and the raw data are used in the analysis. The data were subjected to one-way analysis of variance using maternal half-sib families, paternal half-sib families and regions as the dependant variable. There was only one value for the seed characters and half-sib families and regions were used in -the ANOVA to provide an error term for the analysis. The significance of sources could not be tested if an A N O V A of individual crosses was used. 3.4.2 H E I G H T G R O W T H A N D P H E N O L O G Y For height, the analyses were based on the 1st year height measurements performed after consolidation of the experiment (December, 1988) and 2nd year height measurements performed in November, 1989. Seedlings were transferred from blocks during consoli-dation, but it was decided that 1st year block means after consolidation need not be corrected for initial block means, as they only accounted for 1.5% of the variance of final height measurements. Two different models were used to analyse the height and phenology data. The first is intended to subdivide the total variation into a geographical hierarchy. The term "regions" (short for regional crosses) refers to variation associated with the mean performance of the four possible combinations of parents from the coast and interior. The term "subregions" (short for subregional crosses) refers to variation associated with the 16 combinations of crosses between the 4 populations sampled and "crosses" refers to the variation among individual full-sib families. In Figure 3.3 the hierarchy is apparent as we go from the 4 regional crosses, each of which has 4 subregional crosses nested within it to the 5 crosses nested within each subregional cross. The hierarchical model based on plot means is: Chapter 3. MATERIALS AND METHODS 30 Y1jM = + Bi + R,- + S(R) f e j + C(S(R)) i f e j + BR,j + BS{R)ikj + E,jkl where Y-iju is the plot mean of the Ith cross nested in the kih subregion nested in the j " 1 , region in the ith block; p is the experimental mean; B, is the effect of the iih block; Rj is the effect of the jth region; S(R)fcj is the effect of the kth subregion nested within the ] t h region; C(S(R));jtj is the effect of the Ith cross nested in the kth subregion nested in the ) t h region; BR i : ) is the effect of the interaction of the \ t h block and the ] t h region; BS(R),/tj is the effect of the interaction of the ] t h block with the kth subregion nested in the ] t h region and; "E-ijki is the residual error term which is equivalent to BC(S(R))^j the interaction of the iih block with the \ i h cross nested in the kth subregion nested in the 'fh region. In this model, the regions and subregions are considered fixed while all other effects are considered to be random. The expected mean squares for this model are given in Table 3.2. The coefficients in the expected mean squares are not exactly equal for all sources due to imbalance in the data with respect, to missing cells and unequal numbers The ranges of these coefficients for all of the data sets using this model are presented in Appendix C. The appropriate F-tests for the full hierarchical analysis are given in Table 3.3. For the analyses on height, the error term [B*C(S(R))], can be tested using an estimate of Chapter 3. MATERIALS AND METHODS 31 Table 3.2: Expected mean squares for the hierarchical A N O V A . S O U R C E E X P E C T E D M E A N S Q U A R E S B L O C K (B) REGION (R) SUBREGION(R) S(R) CROSS(S(R)) C(S(R)) B * R B*S(R) E R R O R 02E + C i c r | , S ( H ) + C20-2B*R + c3aC(S(R)) + 4>S(R) + <}>R 2 2 2 crE + CiCT B , 5 ( H ) 4- £ 3 ^ ( 5 ^ ^ -f (pS(R) C£ T C30"C(S(/?)) cr| + C\C~B*S(R) + C1°B*R 4 the within-plot variance obtained by a method outlined in the Quantitative Genetics Handbook (Judy Loo-Dinkins, pers. comm.). This method was used as the analysis, based on individual trees, was too large for SAS to accommodate in its memory. Chapter 3. MATERIALS AND METHODS 32 Table 3.3: F-tests for the hierarchical ANOVA. SOURCE F-TEST B L O C K (B) MSB + MSCIS[R)) + MSB,S(R) M S S ( R ) + M S B + MSB R E G I O N ' (R) M Sr+MSb.S(r) SUBREGION(R) S(R) CROSS(S(R) C(S(R)) MSC(S(R)) MSB B*R MSR,n M SB*S(R)) B*S(R) MSE Chapter 3. MATERIALS AND METHODS 33 The second model is based on the 5 sets of 4 X 4 factorials, allowing for an analysis of the various genetic components and a means of estimating genetic parameters. This model is obtained by recombining Figure 3.3 into 4X4 factorials. For example, in each subregional cross all of the first crosses in the diagonals will make up the first set or 4 X 4 factorial. In this model all sources are considered to be random and the analysis was performed on plot means. The factorial model is: Yijkim = p + B, + S, + F(S) f c j + M(S)tj + F M ( S ) W . + BS^ + BF(S)ikj + BM(S) t l j +E where Yijkim 1 S t h e plot mean in the cross between the \ i h male and the kth female in the j : " 1 set in the ith block; p is the experimental mean; B, is the effect of the ith block; Sj is the effect of the jth set; F(S)^ isthe effect of the kth female within the ] t k set; M(S)/j is the effect of the ] t h male within the jth set; F M ( S ) « j is the effect of the interaction between the kth female and the Ith male in the ith set; BSjj is the effect of the interaction between the iih block and the jih set; BF(S)ifcj is the effect of the interaction between the \ t h block and the kth female within the jth set; Chapter 3. MATERIALS AND METHODS 34 BM(S),7j is the effect of the interaction between the \ i h block and the Ith male within the ] t h set and; Eijfc/m is the residua] error which is equivalent to B F M ( S ) , j H m which is the interaction of the \ t h block and the kth female and Ith male in the jth set. The expected mean squares for the factorial A N O V A are given in Table 3.4 and the appropriate F-tests given in Table 3.5. The ranges of coefficients in the factorial model, for all data sets, are presented in Appendix D. Statistical significance of the plot error term, B*F*M(S), is testable with the height analysis based on an estimate of within-plot variance. For these two analyses, the generalized least squares method was used to compute the sums of squares due to the imbalance of the data. The type 4 sums of squares was used as it "is considered to be appropriate when there are missing cells, as well as imbalance in numbers between cells (Freund, 1986). For the analysis of the height data, size restraints in memory allocation made it impossible for the full analyses to be run on the U B C mainframe computer. The model was therefore split into an analysis based on plot means and an analysis based on within-plot variability. Both of these analyses are combined in the thesis and appear as one analysis for the height data. The phenology data are plot means and therefore a similar problem was not encountered, but the highest order interaction term becomes the error term. For the factorial analysis, the components for each of the sources of variation were computed with P R O C V A R C O M P in SAS for the plot means model. These components enable one to compare components with respect to the portion of the total variation which they account for. Variance components were not calculated for the hierarchical model as two of the genetic sources of variation, regions and subregions, are fixed effects which have no variance components. Chapter 3. MATERIALS AND METHODS 35 Table 3.4: Expected mean squares for the factorial A N O V A . SOURCE EXPECTED MEAN SQUARES BLOCK (B) SET (S) FEMALE) S) F(S) MALE(S) M(S) F'Ml'Sj ; B'S B"F(S) B"M( E j ERKOR °B 4 c2"f?.M(S) 4 C1°B.F|S) 4 c 3 a B . S + c<4 4 + C!AB.M(S) 4 C'"B..F(S) 4 C3"B.S + C*4.M(S) + c6aM(S) 4 C'4(S) 4 C8<4 °£ 4 [i4.M(S) 4 C1°B.F(S) 4 C7"^(S) 4 4 C&<4.M(S) 4 £ 2°B.M(S| 4 '6°M(S) "JE 4 'S^.MISI °B 4 C?AB.MlS) 4 'l'B.F(S) 4 '4 f fB.S aE 4 C 1°B.F(5) 4 4 C2°B.M(S) Table 3.5: F-tests for the factorial ANOVA. S O U R C E F - T E S T B L O C K (B) MSB.S SET (S) MSs + MSB,M{s)+MSB,F(S)+MSF.M{S) MSFi S) + M S M(S) + MSB.s+MSE F E M A L E ( S ) F(S) MSF( 5|+M5j M A L E ( S ) M(S) F*M(S) MSF.M MSE B*S MSB,F(S)+MSB,M{S) B*F(S) M SB'F(S) MSB B*M(S) MSB Chapter 3. MATERIALS AND METHODS 36 Other genetic parameters were calculated for 2nd-year height growth. Mean values were obtained for all parents and crosses at all levels. The general combining ability ( G C A ) and the specific combining ability (SCA) were then calculated for each parent and cross respectively. The G C A value was calculated as the mean value for that region, subregion or individual parent minus the overall mean for the experiment. A positive value denotes a superiority in cm above the mean height, while a negative value denotes inferiority with respect to the overall mean. The SCA values were calculated as the mean value of the group of interest, minus the overall mean, minus the G C A values of the maternal and paternal group. These values can also be positive and negative. Whereas G C A values estimate additive genetic effects, SCA values estimate deviations from additivity. Another genetic parameter calculated was the heterosis value. This is easily explained and calculated as the deviation from the mid-parent value. The mean of the two parental trees, subregions or regions is subtracted from the group mean. The value can be either positive or negative. 3.4.3 C O L D H A R D I N E S S Cold-hardiness was measured on six dates. On the first four dates, all samples were tested at the same 3 temperatures. For the last two dates, different temperatures were used for different regions because the regions exhibited large differences in L T 5 0 values. To obtain intermediate levels of damage on these two dates, temperatures had to be nested within regions. On all dates samples of some of the crosses were not placed in the freezer and used as controls, for comparison and evaluation of damage in the frozen samples. These controls were shoots from some of the crosses to be tested, were given the same pre- and post-freezing treatments and were extremely useful in the scoring of damage. The hierarchical model for the first four dates is similar to the one used for height Chapter 3. MATERIALS AND METHODS 37 and phenology, but an extra term, temperature, must be accounted for in the model and a completely randomized design without blocking was used due to size constraints in the freezer and the assumption, that conditions are uniform within the freezer. As in the hierarchical analysis of height and phenology, regions and subregions are considered to be fixed effects while all other sources of variation are considered random.The hierarchical moclel for cold hardiness is: Yijklm = p+ R, + S(R) 0 . + C(S(R) ) : ; , + T , + TR,-, + TS(R)ijt + T C ( S ( R ) ) .JFC, + E i j k l m where Yijkim is the value of the mth seedling tested at the Ith temperature from the kih cross nested in the jth subregion nested in the ith region; p is the experimental mean; R T is the effect of the \ t h region; S(R) , J is the effect of the j * ' 1 subregion nested within the ith region; . C(S(R)),jj. is the effect of the k f / l cross nested in the jth subregion nested in the ith region; T ; is the effect of the Ith temperature; T R , | is the effect of the interaction of the \ i h temperature and the \ t h region; TS(R)JJ7 is the effect of the interaction of the ] i h temperature with the j t K subregion nested in the \ t h region; TC(S(R)) , J T c ; is the effect of the interaction of the 1th temperature with the kth cross nested in the ) t h subregion nested in the ith region and; Eijkim is t h e residua] error term; Chapter 3. MATERIALS AND METHODS 38 The expected mean squares for these analyses are given in Table 3.6 and the appropriate F-tests are presented in Table 3.7. Table 3.6: Expected mean squares for hierarchial cold-hardiness A N O V A . 3 SOURCE EXPECTED MEAN SQUARES REGION R + C^TC(S(F)) + CZ°TR + C ' r C + SUREGION(R) S(R) -> " E + C l ( , T C ( S ( f i J ] + . C J 4 s ( R ) + C*°C + CROSS(S(R)) C(S(R)) ° E + C l 4 c ( S ( R ) ) + C t [ , C TEMP T + C l 4 c ( S l S ) ) + C i 4 T*R -> + C l C r T C ( S i R ) ) + Ci°TR T*S(R) + C l " r c ( s | R | | + C 2 ' r s ( R ) T*C(S(R)) -> °E I 2 + C 1 C T T C ( S ( R ) ) ERROR 4 Coefficients: cj = 5; c.2 ranges from 10.93-13.57; C3 ranges from 32.74-41.79; C4 ranges from 11.43-13.46 and c5 =140. Table 3.7: F-tests for the hierarchial cold-hardiness A N O V A . SOURCE F-TEST REGION R M SK + M S-rrr M £7. fi -f- M SUBREGION(R) S(R) M S T S + M S C CROSS(S(R)) C(S(R)) M S C M S R C TEMP T MST hA S<j> ^ T*R M ST R T*S(R) M STS(R) M ST Q T*C(S(R)) M STc(S{R)) M S T P Chapter 4 RESULTS 4.1 SEED T R A I T S The average filled seed percentage, based on x-ray analysis, was fairly high (87.2 %). The crosses which were performed in the interior had a higher percentage of filled seed compared to the crosses on the coast (Table 4.8). Some crosses produced no filled seed and others with substantial numbers of seed had very low filled-seed percentages (e.g. cross 53 X 3 = 11 %). The results of the seed characteristics for each cross are presented in Appendix Q. The germination percentage was lower than the filled-seed percentage on average, (73.4 %), and in 70 of the 80 crosses. The crosses which were performed in the interior had a higher germination percentage compared to those on the coast (83.3 versus 63.2). The germination percentage of the Coast X Coast crosses was surprisingly low (67.6 %) for intraspecific crosses, but the Coast X Interior crosses had an even lower germination percentage (59.3) (Table 4.8). Throughout, this thesis crosses will be presented with the standard format of maternal parent X the paternal parent. The average weight, based on 100 seeds, was 0.24 grains, with a range from 0.10 to 0.30 grams (Appendix 3). Seed weight appeared normally distributed, germination per^ cent was slightly negatively skewed but filled seed % had large deviations from normality based on stem leaf and normal probability plots produced with the P R O C UNIVARI-A T E procedure in SAS. Transformations with arcsine, squareroot and logarithm did not 39 Chapter 4. RESULTS 40 Table 4.8: Mean values of filled seed % (FLLDSD%), germination % (GERM%) and seed weight based on 100 seeds (SDWT) for the regional crosses R E G I O N F L L D S D % G E R M % S D W T (g) C O A S T X COAST 89.2 67.6 0.23 C O A S T X INTERIOR 73.8 59.3 0.22 INTERIOR X COAST 92.2 80.0 0.25 INTERIOR X INTERIOR 93.7 86.6. 0.25 improve the normality and were therefore not used. Bartlett's test was performed on the class variables, male, female and region, for all three traits. Heteroscedasticity was statistically significant in all cases examined. For the regression equation, the failure to meet, the assumptions has no effect on the b values or the use of this equation for prediction, but it will affect the inference we can make (e.g. confidence limits will be affected). For the A N O V A it is recognized that the error variance will be affected, but it is assumed that this is minimal and that the F-test is robust to these deviations. The multiple regression equation to predict germination ( G E R M ) based on seed weight (SDWT) and filled seed % (FLLDSD%) is presented in equation 4.1. GERM = —5.211 + 0.535 * FLLDSD% + 136.637 * SDWT (4.1) The equation was statistically significant at the 0.01 alpha level, but for practical purposes it was not satisfactory as it only explained 41.6 % of the total variation (R2 = 0.416). The seed variables were found to be much more affected by the maternal than the paternal parent (Table 4.9). A l l three variables: filled seed %, germination % and seed weight were found to differ at the 0.01 alpha level (p=0 .0001) for females. The males did not differ at the 0.1 level for alpha(GERM [p=.28]; F L L S D W T [p=.90] and SDWT[p=.99]). Chapter 4. RESULTS 41 The one-way analysis of variance based on regional crosses showed that there were no statistically significant differences among regions for seed weight. There were differences for proportion of filled seed and with the Student Newman Keuls (SNK) mean separation test it was found that the coast X interior crosses had less filled seed then the other regions. Crosses with a coastal maternal parent were lower in germination percentage than those with an interior maternal parent (Table 4.9). Table 4.9: One-way analysis of variance for filled seed % (FLLDSD%), germination % (GE RM %) and seed weight (SDWT) by female, male and region (variables as defined in Table 4.8). VARIABLE F L L D S D % G E R M % SDWT' (g) DF MS SIGN. MS SIGN. DF MS SIGN. F E M A L E (%) 19 983.90 *** 1170.52 *** 3 0.7063 *** M A L E (%) 19 263.07 n.s. 519.54 n.s. 3 0.0472 n.s. REGION 3 1374.45 *** 2609.97 *** 3 0.3859 n.s. MEAN 86.7 73.2 .2354 The regional crosses showed a consistent pattern with respect to date of planting which is an estimate of germination rate. The interior crosses germinated much more quickly than the coastal crosses (Table 4.10). By the third planting day 58 % of the interior X interior seeds had germinated while only 5 % of the coast X coast seedlings had. Interior X coast crosses also germinated more rapidly over the first few days than the coast X interior crosses. The mean squares for seed weight are multiplied by 10 2. Chapter 4. RESULTS . 4 2 Table 4.10: The percentage of seed from each regional cross which was planted on a particular date. R E G I O N J U N E 3 4 5 6 7 8 9 10 13 14 CO'AST X COAST 0 1 4 6 16 16 24 13 18 2 COAST X INTERIOR 4 10 8 13 15 13 21 5 11 0 INTERIOR X COAST 7 16 13 16 10 11 11 5 9 2 INTERIOR X INTERIOR 14 22 22 4 7 7 11 4 6 2 4.2 H E I G H T G R O W T H After one year, the seedlings averaged 5.3 cm in height and individual seedlings ranged from 0.8 to 11.7 cm in height. The overall mean height did not. differ significantly between regions after the first year, ranging from 4.4 to 5.9 cm (Table 4.11). After two years, the overall average height was 16.2 cm. The largest seedling was 32.5 cm and the shortest only 1.5 cm. Based on family means, the tallest family (cross 4) had a mean of 22.4 cm while the shortest family (cross 77) was 10.8 cm. The distribution of the data was investigated at all levels (set, region, subregion and cross) for departures from normality. A l l data sets of height appeared normally distributed, based on normal probability plots and stem leaf plots, with the exception of some of the full-sib families which deviated from normality. At this level the number of observations was sometimes quite small and it becomes difficult to test for departures from normality. It is assumed that for the 1988 and 1989 height measurements there are no significant departures from normality. Bartlett's test and the maximum F-ratio test were performed on the height data to 2means with the same letter are not significantly different. Chapter 4. RESULTS 43 test for homogeneity of variances. With Bartlett's test, heteroscedasticity was statisti-cally significant at all levels (regions, subregions and crosses). These differences are not practically significant in all cases as when the maximum F-test is performed the F-value, as the ratio of largest to smallest variance, can be as low as 1 . 5 . When the maximum F-test is performed on regions the degrees of freedom are very large, between 3 5 0 and 6 0 0 , and at this point any difference at all between regions is statistically significant as one is essentially comparing populations. 4.2.1 H I E R A R C H I C A L ANALYSIS For the hierarchical A N O V A , the results are presented for 1 9 8 8 and 1 9 8 9 height measure-ments in Table 4.12. The results for the two years are identical in terms of the statistical significance of sources of variation. The interaction terms were all non-significant for both years. Subregions within regions were also a non-significant source of variation in both years. Blocks, Regions and Crosses within subregions within regions each differed ( Q — 0 . 0 1 ) for both years. By the use of the Student Newman Keuls (SNK) multiple com-parison technique it was shown that for first year height only one block (number 6 ) was significantly different from the rest. In the second year the blocks were arranged in three groups. The Interior X Interior crosses were smaller than the rest in the first year although actual differences between regions were not great. The Interior X Interior crosses averaged 4 . 4 cm, while the Coast X Coast crosses, the tallest, measured 5 . 9 cm in height. After the second-year differences were more pronounced and the 4 regions separated out into 3 groups. The 2 interspecific regional crosses were not significantly different from each Chapter 4. RESULTS 44 Table 4.11: Mean height by region with range of cross means for 1988 and 1989 in cm. 2 R E G I O N 1988 H E I G H T 1989 H E I G H T MEAN N U M . R A N G E N U M . M E A N RANGE COAST X COAST 680 5.9 a 5.1 - 7.4 680 20.2 a 18.2 - 22.4 COAST X INTERIOR 575 5.3 a 4.1 - 6.2 568 17.0 b 15.3 - 19.2 INTERIOR X COAST 947 5.8 a 3.6 - 7.4 940 16.7 b 11.7 - 19.5 INTERIOR X INTERIOR. 1186 4.4 a 3.0 - 6.1 1171 12.2 c 10.8 -14.8 OVERALL 3388 5.3 3359 16.2 Table 4.12: Results of the hierarchical A N O V A for the 1988 and 1989 height measure-ments. SOURCE DF MS F-VALUE SIGN. DF MS F-VALUE SIGN. 1988 HEIGHT 1989 HEIGHT BLOCK (B) 5 5.851 5.47 *#* 5 39.942 5.98 *** REGION (R) 3 41.976 12.45 *** 3 961.451 51.13 »•* SUBREGION(R) S(R) 12 3.350 1.30 n.s. 12 8.809 0.70 n.s CROSS(S(R) C(S(R)) 51 2.377 2.53 • * * 51 13.031 3.21 **» B*R 15 1.070 1.17 n.s. 15 4.067 1.35 n.s. B*S(R) 60 1.655 0.98 n.s. 60 4.342 1.07 n.s. B*C(S(R)) 230 0.938 0.19 n.s. 231 4.067 0.31 n.s. ERROR 3008 2.874 2979 9.006 TOTAL 3387 3358 Chapter 4. RESULTS 45 other, but different from the Interior X Interior and Coast X Coast crosses. It seems as though divergence is occurring among regional crosses and maternal effects are no longer important as evidenced by the lack of any difference between the interspecific regional crosses. Genera] combining ability (GCA) values were similar within the two species groups used in this experiment. The coastal Sitka spruce mainly showed positive GCA values while all of the interior spruce GCA values were negative. The female G C A values were higher in the coastal crosses and in 3 of the 4 subregions, although the values were quite similar. The coastal subregions all had positive G C A values which ranged from 2.09 to 2.54 cm, while the interior sources were all negative ranging from -1.96 to -1.25 cm (Table 4.13). When looking at the G C A values on an individual cross basis, one can see a very Table 4.13: General Combining Ability (GCA) values for height of the Regions and Subregions.in cm. G R O U P SEX G C A G R O U P SEX G C A REGIONS COAST F 2.34 INTERIOR | F -1.76 M 2.21 M -1.62 SUBREGIONS Q U E E N F 2.34 EAST F -1.96 C H A R L O T T E S | M- 2.41 K O O T E N A Y M -1.84 V A N C O U V E R | F 2.54 PRIN CE F -1.50 ISLAN D M 2.09 G E O R G E M -1.25 large amount of variability within the subregions (Table 4.14). Even on a cross basis all of the interior parents had negative GCA values. The values for the interior parents ranged from -4.97 to -0.40 cm in height with both the best and worst parents found in the East Kootenay region. The Queen Charlotte sources contained 4 of the 5 best parents, although at the low Chapter 4. RESULTS 46 Table 4.14: The General Combining Ability (GCA) values for height of each parent in cm. QUEEN CHARLOTTES VANCOUVER ISLAND EAST KOOTENAY PRINCE GEORGE PARENT GCA PARENT GCA PARENT GCA PARENT GCA 260 4.14 121 3.55 89 -0.40 25 -0.51 374 3.99 50 2.95 4 -0.57 106 -0.86 293 3.64 187 2.75 101 -0.62 42 -1.44 220 3.07 130 2.68 3 -1.48 134 -1.66 212 2.41 58 2,57 20 -1.68 32 -1.82 209 1.92 192 2 12 6 -2.16 4 -2.06 311 1.56 149 2.10 48 -2.86 155 -2.26 .266 0.87 15 1.69 23 -3.08 122 -2.39 276 0.28 53 0.23 81 -3.15 135 -3.87 242 -0.91 161 -0.63 82 -4.97 21 -3.60 end, -0.91 was estimated for Q C l parent number 242. The best parent (QCl 260) was over 4 cm better than the average parent. The Vancouver Island sources also had only one negative GCA value and the average G C A value was almost identical to that of the Queen Charlottes (2.00 vs 2.10). Specific. Combining Ability (SCA) values- were negative at the regional level for intraspecific crosses and slightly negative, but basically zero(-0,02), for the interspecific crosses (Table 4.15). Most of the subregional crosses (12/16) had negative values and none of the positive values were above 1.0. The largest were the Prince George X Vancouver Island crosses with a value of 0.76 (Table 4.15 ). In Table 4.16 Table 4.15: Specific Combining Ability (SCA) values for height of Regions and Subregions in cm. i FEMALE MALE SCA FEMALE MALE SCA REGIONS COAST COAST -0.61 INTERIOR INTERIOR -0.61 COAST INTERIOR 0.02 INTERIOR COAST 0.02 SUBREGIONS QCl QCl -0.91 EK QCl -0.26 QCl VI -1.16 EK VI -0.65 QCl EK -0.01 EK EK -0 44 QCl PG 0.67 EK PG -0.06 VI QCl -0.43 PG QCl 0.19 VI VI -0.68 PG VI 0.76 VI EK 0.06 PG EK -1.03 VI PG -0.68 PG PG -1.34 Chapter 4. RESULTS 47 the SCA values are shown for individual crosses. The blank spaces in this table represent crosses in which the number of seedlings was less than 10. The table is in a form similar to the mating design in Figure 3 .1 , except that crosses are ordered sequentially by SCA value for each subregion. One can see a great deal of variability within each subregion for SCA values. For example, the Prince George X Queen Charlottes crosses had values ranging from -1.62 to 1.79 cm. The largest positive SCA value in the experiment was 4.54 cm between Prince George 135 and Vancouver Island 161, while the largest negative value was -2.84 cm between Queen Charlottes 374 and Vancouver Island 58. For individual crosses within regions, the coast X interior crosses averaged 0.66 cm with 9 crosses positive and 6 crosses negative (9-f , 6-). The interior X coast crosses averaged 0.40 cm (8 + , 10-); interior X interior crosses averaged about zero (0 .002)(8 - f , 10-) and the coast X coast crosses averaged -1.26 cm ( + 1, 14-). Heterosis values are given in Table 4.17 for regions and subregions. The regional hybrids showed positive heterotic responses of 0.76 and 0.49. Most subregions exhibited positive heterosis values ( 13 /16 ) . The largest was between the Queen Charlottes and Prince George subregions (1.88). Relatively large heterosis values were also found when Prince George was used as the male parent and crossed with the Queen Charlottes (1.23) and Vancouver Island (1.47). Two of the three negative heterosis values were between intraspecific hybrids. On the basis of individual crosses, the heterosis values ranged from -1 .59 to 2.61 (Table 4.18). For heterosis values there is a great deal of variability between full-sib families within subregions. For example, the east Kootenay X Vancouver Island crosses ranged in heterosis values from 2.61 to -1.46. 1 apter 4. RESULTS Table 4.16: Specific Combining Ability (SCA) for height of Crosses in cm. FEM MAL MAL FEM FEM MAL FEM MAL QCl QCl SCA VI QCl SCA EK QCl SCA PG QCl L SCA 212 242 0.45 130 220 -0.90 101 266 • 1.47 135 266 1.79 311 209 -0.67 192 209 -1.09 82 209 0.46 4 242 1,52 374 260 -2.00 50 260 -1.36" 6 106 0.41 25 220 0.40 293 220 121 242 89 260 -0.19 42 260 -1.62 276 266 — 53 266 20 220 -1.00 155 209 QCl VI SCA VI VI SCA EK VI SCA PG VI SCA 212 187 -0.94 50 58 -0.03 6 187 0.47 135 161 4.54 311 15 -1.13 192 15 -0.34 20 149 0.46 42 58 1.01 293 149 -2.70 130 149 -1.48 101 161 -0.08 4 .187 0.32 374 58 -2:84 121 187 -1.66 89 58 -1.09 25 149 -0.67 276 161 53 161 82 15 -1.11 155 15 QCl EK SCA VI EK SCA EK EK SCA PG EK SCA 276 3 0.78 53 3 3.74 82 48 2.46 155 48 2.06 311 48 0.02 192 48 1.40 6 23 1.00 25 4 -0.03 374 81 -0.14 130 4 -0.04 20 4 -1.13 4 23 -0.81 212 , 23 -0.22 121 23 -0.26 101 3 -1.22 42 81 -1.50 293' 4 50 32 -2.41 89 81 135 3 -1.54 QCl PG SCA VI PG SCA EK PG SCA PG PG SCA 276 122 3.06 192 134 0.91 6 187 0.47 155 134 2.46 212 106 2.11 130 21 0.85 20 149 0.46 4 106 1.00 311 134 53 0.33 101 161 -0.08 25 21 -1.13 374 32 50 32 -0.28 89 58 -1.09 135 122 -1.22 293 21 121 106 82 15 -1.11 42 32 Table 4.17: Heterosis (HET) values for height of Regions and Subregions in cm. F E M A L E M A L E H E T F E M A L E M A L E H E T REGIONS COAST INTERIOR 0.76 INTERIOR COAST 0.49 SUBREGIONS Q C l VI -0.79 VI Q C l 0.46 Q C l E K 0.69 VI E K 0.74 Q C l PG 1.88 VI PG 0.52 E K Q C l 0.39 P G Q C l 1.23 E K VI -0.54 PG VI 1.47 E K PG 0.89 P G EK -0.20 Chapter 4. RESULTS 49 Table 4.18: Heterosis (HET) values for height of the crosses in cm. F E M M A L F E M M A L F E M M A L VI QCI H E T E K QCI H E T P G QCI H E T 130 220 0.96 20 220 1.52 25 220 1.23 50 260 -0.08 82 209 0.67 4 242 -0.41 192 209 -0.18 6 106 -1.83 155 209 121 242 89 260 -1.83 42 260 53 266 101 266 135 266 QCI VI H E T E K VI H E T P G VI H E T 212 187 0.93 20 149 2.61 4 187 0.71 293 149 -0.85 6 187 0.56 25 149 -0.20 311 15 -1.02 89 58 -0.27 155 15 374 58 -2.08 82 15 -1.46 42 58 276 161 101 161 — 135 161 QCI E K H E T VI E K H E T P G EK H E T 212 23 2.06 192 48 2.08 155 48 0.08 374 81 1.41 121 23 1.82 4 23 > -0.01 311 48 0.46 130 4 -0.41 135 3 -0.55 293 4 50 32 -1.59 25 4 -1.42 276 3 53 3 42 81 QCI P G H E T VI P G H E T E K P G H E T 212 106 2.58 130 21 3.00 20 149 2.41 311 134 192 134 1.11 82 15 1.22 293 21 121 106 101 161 0.39 374 32 50 32 — 6 187 0.36 276 122 53 122 89 58 4.2.2 F A C T O R I A L A N A L Y S I S For the factorial analyses, the results are presented in Table 4.19 for the 1988 and 1989 height measurements. In terms of significance of sources of variation, the results are iden-tical for the two analyses. The interactions of the Blocks with the Sets, Females, Males and Female*Male interaction (equivalent to cross) were all non-significant for height, in Chapter 4. RESULTS 50 both year's analyses, therefore the crosses performed similarily over all blocks. Differ-ences between the five factorial sets were shown to be non-significant sources of variation. The Blocks, Females, Males and the Female*Male interaction were all highly significant -at a—0.01. Although the Female*Male (F*M) interaction term is significant, a more careful look at the actual variance components allows one to discuss the main effects with confidence. In the second year the F * M interaction is statistically significant, but. it only accounts for 5.6 % of the total variation. The female component accounts for 27.0 % and the male component accounts for 37.3 % of the total variation. The Set and Block*Set terms had negative variance components and these are assumed to be zero. Blocks accounted for only 2.6 % of the variation, Block*Female interaction for 1.5 % and Block*Male interaction for 1.2 % of the variation. The remaining 24.8 % of the variation was accounted for by the residual error term which is equivalent to the Block*Female*Male interaction term. 4.3 P H E N O L O G Y Very large differences in budset were observed between the regional crosses in 1988 (Figure 4.4) and,, based on general observations, it was seen that many of the Interior X Interior seedlings had set buds by early August. On September 11th, 70 % of the Interior X Interior seedlings had set bud while only 1 % of the Coast X Coast seedlings had set bud. The Interior X Interior crosses had increased to 95 % of the seedlings having terminal buds, while the Coast X Coast crosses increased only to 4 % by September 25th. The frequency of bud set in the hybrids was initially quite low on September 11th, but increased dramatically over the next 4 weeks. Over the time frame examined, the Interior Chapter 4. RESULTS 51 Table 4.19: Results of the factorial A N O V A For 1988. and 1989 height measurements S O U R C E D F M S F - V A L U E S I G N . D F M S F - V A L U E S I G N . 1988 H E I G H T 1989 H E I G H T B L O C K (B) 5 4.013 5.34 *»* 5 21.675 5.46 **• S E T (S) 4 4.411 0.66 n.s. 4 19.255 0.24 n.s. F E M A L E ( S ) F(S) 15 3.810 1.29 15 64.976 5.22 *** MALE(S) M(S) 16 6.880 2.14 *•* 16 78.935 6.36 *** F*M(S) 32 2.524 3.02 *** 32 8.582 2.18 *** B'S 20 0.751 0.74 n.s. 20 3.970 0.87 n.s. B*F(S) 75 1.078 1.29 n.s. 75 4.611 1.17 n.s. B'M(S) 77 1.075 1.29 n.s. 76 4.454 1.13 n.s B*F*M(S) 135 0.836 0.15 n.s. 136 3.929 2.67 n.s. E R R O R 3008 2.874 2979 9.006 T O T A L 3387 3358 X Coast crosses had a higher percentage of budset than the Coast X Interior crosses. The budbreak pattern in the spring was very different from budset. The process of budbreak occurred very rapidly compared to budset and there were no large differences between regions (Figure 4.5). The Coast X Coast and Interior X Interior crosses had approximately the same percentage of flushed buds on the two dates examined. The hybrids appeared to burst their buds earlier and the Interior X Coast crosses had a consistently higher percentage of budbreak. For second-year budset, the pattern appeared very similar to the results of 1988 (Figure 4.6). In the second year, budset occurred earlier compared to the first year. These second year dates for budset are probably more indicative of future performance of the group. The late budset of the first year seedlings is due to the late planting of these seedlings in the spring. The Interior X Interior seedlings had set over 80 % of their Chapter 4. RESULTS 52 Figure 4.4: The pattern of first-year budset by regional cross. Chapter 4. RESULTS BUD B U R S T % 100 APRIL 12 APRIL 16 D A T E C X c C X I I X c I X I Figure 4.5: The pattern of first-year bud break by regional cross. Chapter -1. RESULTS 54 BUD SET % 100 80 60 40 20 -0 I JUL 12 JUL 21 AUG 18 DATE SEP 20 c x c C X I I X c I X I Figure 4.6: The pattern of second-year budset by regional cross. Chapter 4. RESULTS 55 buds before the end of July while the Coast X Coast seedlings had set only about 2 % of their buds by this date. The trend for Interior X Coast seedlings to set their buds earlier was still consistent, but differences were reduced compared to the first year. For phenology, transformations were attempted with arcsine, squareroot and logarithm, but no improvement was made on the data so the raw data were used in all of the analyses on budset and budbreak. 4.3.1 H I E R A R C H I C A L ANALYSIS For phenology, the results of the hierarchical ANO VA's are presented in Table 4.20. For first-year budset the Block*Subregion interaction was statistically non-significant, but the Block*Region interaction was significant at a—0.01 on two of the three observation dates. i Crosses and subregions differed (a=0.01) on two of the three dates examined. Regions differed on all dates and blocks were not significant in all cases. Again it appears as if regions are the primary consistent source of variation, while crosses and subregions appear equally as important for year 1. The results of the hierarchical ANOVA for second-year budset show that the Block* Subregion interaction was significant (a=0.05) on two of the four dates. Block*Region was significant (a=0.1) for the first date and non-significant thereafter. Crosses differed on all dates as well as Regions. Subregions were non-significant on the first three dates, yet highly significant on the fourth. For second-year budset the pattern agrees well with height growth in that the important sources of variation appear to be regions and crosses. Blocks differed (Q=0.1) on the first two dates and were non-significant thereafter. In the hierarchical analysis of budbreak, the interaction terms of Blocks with Subre-gions were not statistically significant and for regions the second date was significant at a=0.1. These interactions are assumed to not affect interpretation of the main effects. The blocks were not significant as a source of variation. For the genetic sources of Chapter 4. RESULTS 56 variation the crosses seemed the most important factor as crosses were highly significant on both dates examined. The subregions were significant at a=0.1 on one date and not significant on the other, while regions were significant at a=0.05 and a=0.1 at dates one and two respectively. The lack of highly significant differences for regions is surprising in view of the major differences in climate between the coast and interior. Figure 4.5, however shows that very little variation existed at the regional level for bud break. Table 4.20: Results of the hierarchial A N O V A for first and second year phenology B U D S E T 88 B-URST 89 B U D S E T 89 S O U R C E 1 2 3 1 2 1 2 3 4 B L O C K B n.s. n.s. n.s. n.s. n.s. * * n.s. n.s. R E G I O N R *** *** ** * ** ** * *** *** *** S U B R E G I O N ( R ) S(R) *** ** * n.s. n.s. * n.s. n.s. n.s. *** CROSS(S(R)) C(S(R)) *** n.s. ** * *** *** ** * *** *** *** B * R *** *** n.s. n.s. * n.s. n.s. n.s. B*S(R) n.s. n.s. n.s. n .s. n.s. n.s. * * ** n.s. 4.3.2 F A C T O R I A L A N A L Y S I S The factorial A N O V A results are presented in Table 4.21 for first- and second-year phe-nology measurements. The Block*Set and Block*Female components were considered non-significant for all dates. The Block*Male component was highly significant (a==0.01) on the first budset date and on both flushing dates examined. The interactions of the blocks with the genetic main effects were minimal as only 2 of 21 interactions were con-sidered significant for budset. The Female*Male interaction was highly significant in all cases. This illustrates that for budset, specific combinations of parents account for a large Chapter 4. RESULTS 57 portion of the variance. The Female and Male components were also highly significant but their interpretation is better illustrated by looking at the proportion of variation accounted for by each term. An average of the variance components for all seven dates was derived to indicate which sources accounted for the largest portion of the budset variation. The female component was clearly the most important source of variation accounting for 46.2 % of the total variability. Males (22.1 %) and the Female*Male interaction (11.6 %) were also important sources of variation. The residual error term (B*F*M) accounted for 17.9 % of the variation and none of the other components accounted for more than 1 % of the variation. For budbreak the Block*Set and Block*Female interactions were considered non-significant. The Block*Male interaction was significant at a— 0.05 for the first date and ct=0.01 for the second. This result is surprising as there is no explanation for a consistent Block*Male interaction term. Blocks, Sets and Males were considered to be non-significant while Females were significant at 0.05 for the second date. In terms of variance components averaged over both dates the residual error accounted for the majority (50.8 %) of the variation present in bud break. Genetic, effects were mainly due to the Female*Male interaction (19.1 %), but the male (8.5 %) and female (6.7 %) components also accounted for some variability. The Block*Male term was greater than the genetic main effects (9.7 %), while the Block*Female component accounted for only 2.5 % of the variability. The remaining variability (2.5 %) was accounted for by the effects of individual sets as the component for Blocks was considered to be zero. a.pter4. RESULTS Table 4.21: Results of the factorial A N O V A for first and second year phenology B U D S E T 88 B U R S T 89 B U D S E T 89 S O U R C E 1 2 3 1 2 1 2 3 4 B L O C K B n.s. n.s. n .s. n.s. n.s. n.s. n.s. n.s. n.s. S E T S n.s. n.s. n .s. n.s. n.s. n .s. n.s. n.s. n.s. F E M A L E ( S ) F(S) *** * ** *** n.s. * * *** ** * *** *** M A L E ( S ) M(S) ** * * ** *** n.s. n.s. *** ** * ** * *** F*M(S) ** * * ** *** *** • * *** *** ' *** *** B*S n.s. n.s. n .s. n.s. n.s. n.s. n.s. n.s. n.s. B*F(S) n.s. n.s. n.s. n.s. n.s. * n.s. n.s. n.s. B*M(S) ** * n.s. n .s. ** *** n.s. n.s. n.s. n.s. Chapter 4. RESULTS 59 4.4 C O L D HARDINESS The most common method of presenting frost hardiness results is by the use of the LT50, which is the temperature at which 50 % of the samples are considered dead due to freezing damage. The results of this experiment over 6 freezing dates will be presented in this manner although there are several problems. The first is that if the test temperatures are not low enough to kill 50 % of the samples a L T 5 0 value can only be estimated by extrapolation which can have a large error associated with it. The other extreme is when all test temperatures result in damage which exceeds 50 % making it impossible to estimate a L T 5 0 . Both results were obtained with a variety of samples on several dates of testing. The problem arose as a consequence of huge differences in terms of frost hardiness based on evolutionary responses to the very different climates in which Sitka and interior spruce are found. In Figure 4.7 the obtained L T 5 0 values are plotted over the experimental period for the regional crosses. The most obvious result is that an LT50 value for Interior X Interior crosses was only obtained on the first date and thereafter only minimal damage was experienced by these crosses. No trend can be extrapolated from these results, but the lowest temperatures used are presented, in the dotted line, to illustrate that the L T 5 0 must be below this line on all dates. By mid November these crosses had suffered only 28 % damage when tested at -50 0 C. The quick drop in L T 5 0 was also confirmed by Salim Silim who found a quick increase in frost hardiness during early September in interior spruce (pers. comm.). The Coast X Coast crosses were highly susceptible to fall frosts and did not increase greatly in cold hardiness until after November 8th. This increase,in hardiness occurred at the same time as the first freezing temperatures below -2° C. It was found that very little difference in temperature accounts for the difference between 50 % kill and total Chapter 4. RESULTS 60 kill in Sitka spruce. For the Coast X Interior crosses a LT50 of -34 ° C was found by the end of this experiment on November 215'. The lethal temperature began decreasing after October 7th even though no temperatures below zero were experienced by the seedlings. For the reciprocal hybrid, Interior X Coast cross results were even more promising. The LT50 reached a value of -39 0 C by the end of the experiment. As in Coast X Interior crosses, very little increase in cold hardiness was found before October 1th. Before this date, the hybrids appeared quite similar, but by the end of the experiment the divergence of the two types was obvious. Due to the absence of L T 5 0 values a comparison of cross types was developed using actual percentages of injury plotted over test temperatures for three separate dates, each approximately one month apart. On the September 23 date (Figure 4.8) there is good separation among the regional crosses. In this early stage of hardening, the temperature difference is not great between the regional crosses, but the Interior X Interior crosses are quite hardy. For practical purposes the hybrids appear to have a high degree of hardiness even for early frosts in the interior. By October 24, the pattern of the regional crosses is similar, but divergence seems to be occurring in terms of cold hardiness. Very little damage has occurred in the Interior X Interior crosses even at temperatures as low as -20 0 C. At this stage the hybrids appear closer to the Interior X Interior crosses rather than the Coast X Coast crosses. On the final experimental date (November 21) the separation of regional crosses is very large. The Interior X Interior crosses show less than 30 % damage at -50 ° C. Coastal crosses are gaining hardiness rapidly and show tolerance to temperatures which would normally not be experienced in the coastal environment. The hybrids are quite cold hardy, but the Interior X Coast crosses are much hardier at this point than at any other dates in the experiment. Chapter 4. RESULTS 01 FROST HARDINESS (LT50) 9/8 9/23 10/9 10/24 DATE 11/7 11/21 C X C c x i 1 X C I X Figure 4.7: L T 5 0 values for the regional crosses on six test dates. C'hapter 4. RESULTS % DAMAGE 1U0 -12 -15 T E M P E R A T U R E ( C ) i x I i x c C X I C X c Figure 4.8: Percentage damage to shoots from regional crosses at three temperatures September 23. Chapter 4. RESULTS % DAMAGE - 8 -14 - 2 0 T E M P E R A T U R E ( C ) I X ! - + - I X C - * ~ C X I ~ * - C X C Figure 4.9: Percentage damage to shoots from regional crosses at three temperatures October 24. Chapter 4. RESULTS % D A M A G E 80 i 0 1 1 ' - J 1 -J i i I -10 -15 -20 - 2 5 - 3 0 - 3 5 - 4 0 - 4 5 -50 H A R D I N E S S — i x i - + - r x c - * - c x i - + - C X C Figure 4.10: Percentage damage to shoots from regional crosses at three temperatures on November 21. Chapter 4. RESULTS 65 The data for cold-hardiness show severe departures from normality and transforma-tions with arcsine, squareroot and logarithm did not improve normality. The analysis presented here used the raw data, but it must be recognized that the assumption of normality has not been met and interpretation assumes that the robustness of ANOVA can accommodate these departures. The results of the first four freezing dates, in which all crosses were tested at the same temperatures, are presented in Table 4.24. Most of the sources of variation are not as easily explained in terms of consistency of statistical significance as are the height and phenology data. Significant interactions were found between temperature and regions on all dates, between temperature and crosses on 3 of the 4 dates and the interaction of temperature with subregions was not significant on any of the dates investigated. Temperature was highly significant on all dates, but it's effects should be investigated more thoroughly on a regional or cross basis due to the significance of the Tempera-ture*Region interaction. The subregions were not involved in a significant interaction and are not significant as a main effect component of variance. Table 4.22: Results of the hierarchial A N O V A for cold hardiness SOURCE D A T E 1 D A T E 2 D A T E 3 D A T E 4 REGION R n.s. ** * * ** *** SUBREG ION(R) S(R) | n.s. n.s. n.s. n.s. CROSS(S(R)) C(S(R)) | * ** * * ** ** T E M P T ** * ** * * ** *** T*R ** * ** * * ** *** T*S(R) n.s. n.s. n.s. n .s. T*C(S(R)) ** * n.s. * ** *** Chapter 5 DISCUSSION The X-ray method, used to determine filled seed percentage, overestimated the actual germination percent and this was more pronounced in the coastal crosses. Dr. George Edwards at Forestry Canada, Victoria, B.C. noted that the X-ray method is not a very good method for predicting germination especially for someone without extensive expe-rience with the technique. It may have been a good measure of filled seed %, but, even so, some of these seeds with fully formed embryos did not germinate. An overall experimental germination percentage of 73.4 % was low compared to what one would expect in a commercial nursery. Part of this is explained by the fact that at least half of the crosses are between different species. Interspecific crosses between Sitka spruce and white spruce had a crossability of 54 % based on a large sample (Fowler, 1987). This crossability average is based on ethanol immersion of the seed, a method'of determining filled seed % which should give results similar to the X-ray analysis of filled seeds. These methods are quick measures of germination % and based on the results of the multivariate equation, the X-ray method is not a particularly good predictor of germination in this experiment. The three seed traits (filled seed %; germination % and seed weight) differed for maternal half-sib families, but not for paternal half-sib families. This suggests a strong maternal effect for proportion of filled seeds, germination percent and seed weight. Ger-mination has been shown to have a strong maternal influence in Virginia Pine (Bramlett 66 Chapter 5. DISCUSSION 67 et al., 1983) and the results of this study are therefore not surprising. The early impor-tance of the maternal parent is due to the fact that only the embryo, which constitutes 10- 20 % of the seed weight, contains genes from the pollen parent. The female parent determines the seed weight, seed coat thickness, stratification requirements and polyem-bryony which are correlated with early progeny vigor (Perry, 1976). These effects are common in forest trees but their influence is thought to be practically insignificant after the first season of growth in spruce (Logan & Pollard, 1978). Among the four regional crosses, the surprising result is the germination percentage (67.6 %) of the Coast x Coast crosses. The Provincial Seed Center in Surrey, B.C. found that Sitka spruce had a mean germination of 89 % while white spruce averaged 76 %. These results were based on large numbers of seeds tested over many years and therefore provide a better germination estimate than this experiment. The discrepancy of intraspecific germination results must be explained by treatment differences between the interior and coast. On the coast, poor cone opening made seed extraction by cone destruction necessary. Due to this technique more seeds of lower quality were probably recovered from the cones compared to the interior crosses. Seeds from the interior crosses were heavier than the coastal crosses and this agrees with the average weights presented by Fowells (1965). Seed weights among the species, however, have large overlapping ranges so it is not safe to generalize. In this experiment, the heavier seeds were associated with the higher germination percentages and higher germination rates. The analysis on males and females included crosses with both interior and coastal pollen sources and therefore would only be biased, by unequal gamete con-tribution, if any of the four crosses were not included. Larger seeds have been found to germinate more quickly and produce larger germinants after 28 days (Dunlap and Bar-nett, 1983). The extraction technique in the interior was probably selective for heavier seeds with a greater germinative capacity. Chapter 5. DISCUSSION 68 The other crosses performed on the coast, Coast X Interior, had even lower germina-tion (59.3 %), but these were statistically not different from the Coast X Coast crosses. The one-year-old interior pollen did not contribute to this problem as the interior crosses which were pollinated with this pollen yielded high germination percentages. The other major factor was an infestation of aphids on the coastal strobili while seed formation was occurring. It therefore appears reasonable to attribute the seed character differences to both the technique of seed extraction and the aphid infestation and not to major genetic differences between the species. The interior crosses showed a clear superiority in germination speed. The speed of germination also appears to be under maternal influence as the regional hybrids exhibited behaviour closer to that of the maternal parent. Even though regions were not statis-tically different in terms of seed weight, a definite association between seed weight arid speed of germination is apparent. It must be stressed again that germination speed was estimated by date of planting and the error of planting a seedling on a date different from that on which it germinated is present. Even with this considered, the trend seems too strong to ignore. The initial advantage in height for Interior X Interior crosses was very short-lived. The Interior X Interior seedlings were quite small after first-year growth. Because only final height measurements were obtained, growth rates cannot be compared between regions. Growth rate is important to tree breeding as it is desirable for growth to occur during the period of minimal risk of frost damage. Rapid growth during the summer is therefore advantageous to avoid damage from frost and to ensure that conditions are favourable for the formation of next year's buds. Sitka spruce grows long into the fall and initiates growth early in the spring so rapid growth during the summer is a good selection criterion for Sitka spruce or hybrids moved inland. In some areas of the interior, drought and not frost will be the limiting factor to growth and should be investigated. Chapter 5. DISCUSSION 69 This will probably be important in hybrids with a portion of Sitka spruce genes, as Sitka spruce has evolved in a moist environment. The discussion of the results of height and bud set in this thesis will tend to con-centrate more on the second-year results as they are less influenced by maternal effects, better reflect genetic differences for predetermined growth and are therefore believed to be a better indicator of adult performance. The main performance trait, height, and bud set were highly correlated. The interior X interior crosses set bud very early in the available growing season re-sulting in quite small seedlings. In the interior, trees have adapted to local environmental factors to set bud early in the season. Early fall frosts and summer drought periods are considered to be two major factors contributing to early bud set in interior spruce. The timing mechanism responsible for invoking growth cessation and budset of interior spruce is thought to be the decreasing photoperiod (increasing night length) experienced during late summer and fall. The other factor involved in this experiment, is the elevational change the seedlings encountered by being grown in a coastal nursery. Movement of seedlings from high to low elevations is considered by many to have the same effect as moving seedlings from North to South: they set buds earlier in the season, compared to local seedlings, and cannot take full advantage of the local growing season. The interior sources were moved to a southern location at a much lower elevation where the criti-cal night length will be reached sooner than in their natural environment and therefore budset should be earlier than what one would expect if these trees were planted in the interior. While this will alter the values of the investigated characters it is believed that the pattern of variation will not be greatly altered by growing the seedlings in an interior nursery. Drought, as an environmental cue, was not a consideration as all the seedlings were watered regularly. Early bud set as an adaptation to drought, however, could influence Chapter 5. DISCUSSION 70 the differentia] bud set noted in this experiment. Photoperiodic response seems an ad-equate explanation for early bud set in the interior spruce as early growth cessation is caused by a shortened photoperiod (Roche, 196 9) even when conditions are ideal for... growth (available water and nutrients). Most of the work done on the photoperiodic response of budset has been done on species from continental climates so this agrees with the findings of this experiment. On the other extreme, the Coast X Coast crosses had a long period of shoot elonga-tion even though the latitudes were virtually identical to the interior sources. It appears that factors other than photoperiod must play a large role in bud set for Sitka spruce. Prolonged growth of Sitka spruce in this experiment is in agreement with comparitive shoot growth curves of B.C. spruces produced by Roche (1969). With Sitka spruce, shoot growth cessation is triggered by photoperiod, but budset may be under temper-ature control. Southern provenances tend to set their buds quickly after growth ceases while northern sources prolong the bud formation process (Kraus & Lines, 1976). The data collected on growth cessation were actually budset and this would explain why pho-toperiod did not seem to be critical for Sitka spruce. If growth rates had been obtained for Sitka spruce, one would expect growth cessation to occur in early fall, although bud formation would not occur until later in the fall. The nursery situation is probably closer to the natural habitat of Sitka spruce than white spruce with no water deficit and fairly warm temperatures. The interspecific hybrids were intermediate between the two parental species in terms of budset, but consistently closer to the maternal parent type. The Coast X Interior crosses set buds later than the Interior X Coastal crosses. It therefore appears as if a maternal influence is present in budset. These differences were more pronounced in first season and were still present, but much reduced, after the second growing season. In' the hierarchical analyses for height and budset there did not appear to be significant Chapter 5. DISCUSSION 71 interactions of blocks with regions, subregions and crosses. This shows that these main effects performed similarly over blocks even though the blocks were significantly different for the height analysis. In budset the block effect was only significant in 2 of 7 dates over the two-year experimental period. The reasons for a significant block effect are unclear as the experiment was in a total area of approximately 5 m X 8 m and there was no apparent gradient or reason to explain differences in blocks. The results for height and budset show that most of the variation is occurring at the wide macrogeographic (regional) level and at the individual cross level. For height the reciprocal hybrid crosses at the regional level were not significantly different from each other, but were significantly different than the intraregional crosses. If second-year height were a good predictor of future performance then a tree breeder would tend to concentrate efforts on regional crosses and individual tree crosses, but put very little or no emphasis on crosses between subregions within a region. For second-year growth, the top ten full-sib families were all from coast X coast crosses. The 11th cross was from a interior X coast cross and within the top 25 crosses there were 5 interior X coast and 5 coast X interior crosses. While Sitka X Sitka crosses were generally superior, there were interspecific crosses which approached and exceeded many of the coastal intraspecific. crosses in height. Regions showed predictable GCA differences as the coastal parents were superior to the interior parents. The interior parental groups all had negative GCA values. GCA values were calculated by subtracting the overall experimental mean from the parental mean. This was done in order to compare crosses at all levels for selection potential. The G C A estimates include inter- and intraspecific matings. If only intraspecific crosses were used then only two crosses would be used to calculate the parental GCA values and this was considered insufficient. While the cross type is confounded within the GCA Chapter 5. DISCUSSION 72 estimated this is not considered to have an adverse effect on the interpretation. The par-ent trees GCA values/are actually average GCA values of inter- and intraspecific crosses. Selection based on this G C A value will allow parents to be selected for both a species and hybrid breeding program simultaneously. The hybrid program would probably be of less importance and this can be incorporated by giving a lower weight to interspecific crosses. These GCA values are all relative to the overall experimental mean, rather than group mean (regional cross for example), and this allows for direct comparison of hybrid types with non-hybrid types which was the main emphasis of this thesis. Large within-subregion variability is apparent for G C A values when the parents are investigated individually. At the cross level the differing sets could bias the results, but a correction was not used as sets were statistically non-significant with an F-value of 0.24. From the Prince George region parents 4, 89 and 101 were higher than the lowest coastal i. sources and have high selection potential. The performance of Prince George parents was not correlated to how they grouped according to the hybrid index presented in Figure 3.2. Two Prince George sources ( 25 and 21) from lower elevations were ranked 1 s t and I01h in G C A value, and therefore an explanation of performance seems more complex than solely an elevational transfer. From the East Kootenays, parent 25 had a GCA value greater than the lowest coastal sources and parent 106 was higher than the lowest Queen Charlotte Islands parent. From the East Kootenay region, 4 of the top 5 parents were considered hybrids and the other, ranked third, was considered to be Engelmann spruce. The remaining five parents were considered to be white spruce. The 5 tallest parents were from the area encircling the town of Spillamachbeen, B .C . and further selection in this area should be considered for interior spruce. The trees in this area are considered to be hybrids, based on cone morphology, and perhaps these hybrids would show superiority in other traits due to unique gene combinations beneficial to the genus. Chapter 5. DISCUSSION 73 The coastal parents exhibited some excellent general combiners with a GCA range over 5 cm and 12 of 20 parents had values above 2 cm after 2 seasons of growth. The Queen Charlotte Island parents produced the best general combiners (4 of the 5 best) yet choosing untested trees from this region does not neccesarily mean success as one of the parents had a negative G C A value and two were below 1 cm. Parents 260 and 374 were extremely good, both with GCA values at about 4 cm. The Vancouver Island sources performed well and had a selection differential of 4.2 cm. Based on geography, no logical subdivision of coastal parents into zones of superior performance was possible. The parents on the coast were chosen from a limited range with the Queen Charlottes occurring over an area representing less than half a degree of latitude and longitude, while the Vancouver Island parents were selected over one degree of latitude and longitude. Elevation also varied, but it did not provide a trend for explaining the superior performance of some of the parents. If G C A values for height are to be the unit of selection, then the choice of individual parents should be the primary consideration in tree improvement. It was at this level that the selection differential and the potential gain were the highest. The strategy of the B.C. Forest Service has been to select parents in each species based on phenotype and to cull inferior parents from seed orchards as breeders gain confidence in progeny test results. The parent trees selected for this experiment were a small subset of the available 'plus' trees and within the entire breeding population one can expect a much greater selection differential available to the breeder for increasing forest yield than found in this experiment. Two growing seasons is a short test period especially with something as radical as the use of species hybrids in B . C . The emphasis on such a short term study must be on the potential value of hybridization in B.C. based on the amount of variability, or opportunity for increasing yield by selection. These results should provide basic information on the Chapter 5. DISCVSSION 74 feasibility of hybridization and aid in the development of longer term tests. Seedlings from this experiment, have been planted in 2 interior sites, Skimikin and Red Rock, and one coastal site, Mahatta River, for further investigation of longer-term potential and adaptation. These outplantings will address some of the questions related to the adaptability of the hybrids in an operational environment, especially in the interior. On a regional basis interspecific crosses showed no appreciable Specific Combining Ability (SCA) (0.02 cm) after two years growth in a coastal environment. Therefore crosses between randomly chosen 'plus' trees from the coast and interior will not produce any increase in yield in comparison to the overall experimental mean. Crosses between specific subregions showed some large gains although 12/16 of the combinations produced negative SCA values. None of these positive values exceeded 1 cm after two seasons and are therefore considered to be of minimal practical value. Several full-sib families were enough above the mean in height to warrant attention. Interspecific crosses between QCI 276 and PG 122 (3.06 cm); QCI 212 and PG 106 (2.11 cm); VI 53 and E K 3 (3.74) and PG 135 and VI 161 (4.54) were far above the mean and show potential for selection. These are mean values for the crosses and greater gain can be achieved by selection within these full-sib families for the remainder of the genetic variance. Some of the interior X interior sources also had high SCA values; EK 82 and E K 48 (2.46); P G 135 and PG 266 (1.79); PG 4 and PG 134 (1.79); P G 155 and PG 134 (2.46) At the regional level, the hybrids showed heterosis values of 0.76 cm for the coast X interior crosses and 0.49 cm for the interior X coast crosses. This number describes the amount by which the hybrids were superior to the mid-parent value. For subregions, the values range from -0.79 to 1.88 cm with 13 of the 16 subregions producing positive values. Based on heterosis, it appears as if subregional crosses can produce small gains in yield in comparison to the mid-parent value one would expect with these hybrids. The Chapter 5. DISCUSSION 75 hybrid subregional crosses with Prince George as one of the parents produced 3 of the 4 highest heterosis values at this level. For cross comparisons, values ranged from -2.08 cm to 3.00 cm. Large variability within subregions was evident and indicated that while subregions as a whole may pro-duce heterotic stock, within these subregions some crosses are greater than the mid-parent value while others are less. The heterosis values are not always consistent with the SCA values in terms of ranking of families. The heterosis values illustrate whether the hybrid performance is superior, above the mid-parent value, without reference to the population as a whole. They there-fore illustrate whether the hybrid progeny is better than one would expect with solely additive gene effects. The SCA values are the cross performance minus the G C A values for the parent trees. In the calculation of the GCA values the population mean has been subtracted and SCA values identify specific crosses superior to the average performance of their parents. SCA values are of interest if the parents already meet some predeter-mined performance level as the combination of two poor general combiners may result in a high SCA value, but from a practical standpoint the cross is still useless. In the factorial analysis, the interactions of blocks with sets, females and males were not significant sources of variation for height or budset. As with the hierarchical anal-ysis, the main effects performed similarly over the blocks. Block effects again are unex-plainably significant for height, but not significant on all dates for budset. Sets were a non-significant source of variation. The Female*Male interaction term is highly significant in the factorial analysis and this term is equivalent to the cross term in the hierarchical analysis.' Since each factorial is composed of equal numbers of interior and coastal parents one half of the crosses are between very different genotypes and it is not surprising that crosses were a significant source of variation. Although the Female*Male interaction term is significant, when Chapter 5. DISCUSSION 76 the variance components are examined the major part of the genetic variance is in the main effects, Female and Male. Together the Female (27.0) and Male (37.3) component account for almost 65 % pf the total phenotypic variation in height. For budset, the females account for 46.2 % of the variability and males for 22.1 %. Therefore, additive genetic variance appears to be the primary mode of gene action for height growth and budset and for selection purposes GCA values would provide better gains than selection based on SCA. The residual error term (Block*Female*Male) accounted for a substantial portion of the variance 24.8 % for height and 17.9 % for budset. This illustrates that crosses performed differently over blocks. Non-additive genetic variance due to dominance and epistatic. gene action account for only 5.6 % of the total variation in height and 11.6 % in budset. This interaction is important, but its importance appears to be restricted to certain vigorous families. In hybrid production, if concerns are with frost damage in the fall, the use of an interior female would be of value as it accounts for a much larger portion of the variation in bud set. This additional variability may be due to a maternal effect and one would therefore prefer to use the more cold hardy interior species as the female parent. This maternal effect was shown to decrease from year 1 to 2, so this information should be used with caution until studies of a longer time period are available. The proportion of variation accounted for by the specific crosses (SCA) should be monitored as Boyle (1987) has shown that SCA can increase over time in black spruce. Gerhold and Park (1986) found that in Scotch pine most of the variability from population hybridization was due to GCA, although several families exhibited large SCA values. The results with Scotch pine agree very well with results in this experiment and both indicate that production of hybrids based on parental GCA values is the first logical step for a hybrid spruce program in B.C. Certain specific cross combinations were exceptional, but at this level their superiority Chapter 5. DISCUSSION does not seem adequate to alter tree improvement practices for specific crosses. The utilization of specific crosses involves additional work in the form of seed production, more complicated testing and analysis and with this comes a much greater expense. The solution to many of these problems, if one desires to capture the additional gain offered by specific individuals, is the utilization of vegetative propagation. Methods are given by Russell and Ferguson (1990) for the commercial production of stecklings (plantable rooted cuttings) for interior spruce and the methods should be equally appli-cable to the hybrids. Through vegetative propagation one can reproduce the best hybrid individuals without losing the gain achieved by specific gene combinations, as may occur with recombination in a traditional seed orchard. Clones will also allow for more precise testing of site specificity and increase the efficiency of within-family selection. It seems apparent from these figures that if hybrids are to be used in severe cli-mates, with a high risk of frost damage, then one must select for early budset, directly or indirectly, and it appears that selection based on GCA values for budset would be appropriate. The problem of late spring frosts may also cause problems and selection for genotypes which break buds late would be desirable. Observed patterns of variation in budbreak are not as easily explained as budset and this has been found by many investigators (Kraus & Lines, 1976; Roche, 1969). This is due to the fact that genetic differences in budbreak in spruce have not been as clearly demonstrated as in budset (Roche and Fowler, 1975). The process of flushing occurred very quickly as the proportion of flushed buds practically doubled from 46 % to 91 % over a four day period in this experiment. Differences were not large between regions, although some variability was present, at the subregion and cross level. It is surprising that the interspecific hybrids, at the regional level, had a consistently higher proportion of flushed buds on the two dates examined. The variability in budbreak was most probably condensed into a very sort time frame as from Apri l 9th to April 16th 80 degree days, Chapter 5. DISCUSSION 78 based on a 5° C threshold temperature, had accumulated. With budbreak, the genetic components did not account for as great a portion of the total variance as in height growth or budset. Al l three genetic terms combined accounted for about 34 % of the variability and most of this was the interaction term (19.1 %). The major part of the variability was accounted for by the residual error term showing that large differences were found for crosses in different blocks. The Block*Male term accounted for 50.8 % of the variation and was more important than the female or male component. With height growth and budset, the traits appeared to be under additive genetic control as most crosses were intermediate to their parents. This does not appear to be the case with budbreak. Very little variability was found for budbreak in comparing Sitka, white, Engelmann spruce and their hybrids. One cannot attribute the pattern to dominance as the hybrids are not closer to either of the parents, but exceed both. The significance of the full-sib families and the interaction between female and male parents suggests that some form of non-additive genetic control or other mechanism is present and controlling budbreak. Closely tied to phenology is cold-hardiness. After growth ceases and bud formation occurs, the seedlings enter dormancy, which is a state more resistant to stress. In our climate, the stress is low temperatures in the fall and winter. It was very clear that interior spruce was hardy early in the fall and could tolerate temperatures as low as -15 ° C two months before freezing temperatures were encountered at the nursery. Hardiness to temperatures this low is reported to be the result of mild freezing temperatures triggering a rapid increase in hardiness. This does not seem to be the case for interior spruce and additional experiments by Salim Silim (pers. comm.) with white spruce, indicate that cold-hardiness acclimation in the fall is mainly in response to decreasing photoperiod in interior spruce and not in reaction to mild frosts. Chapter 5. DISCUSSION 79 The trend in Sitka spruce shows that seedlings do not acquire appreciable cold-hardiness until late in the autumn. This is easily explained as bud formation did not occur for many of the Sitka spruce families until November. Decreases in L T 5 0 occurred following the first fall frosts and agreed with the entry of the seedlings into phase II of cold hardiness (Glerum, 1985). The seedlings were hardy to -21 0 C by mid-November, although ambient temperatures only went as low as - 4. This L T 5 0 provides evidence that Sitka spruce can survive winters much more severe than those frequently experienced on the coast. The interior X coast hybrids revealed a greater degree of hardiness in comparison to the coast X interior and this difference increased over the experiment so that by November 21 a practically important difference was found between these regional hybrids (-39 versus -34). Cold hardiness may have a component of maternal influence, but this difference may be sampling error based on the specific crosses used in the freezing study. These values show extremely good potential for the hybrids with respect to avoidance of damage from fall frosts. This agrees with the findings of Sheppard and Cannell (1985) who found Sitka X white spruce hybrids to be hardy to below - 2 0 ° C in September and October, lower than all of the Sitka spruce provenances tested. In the paper by Sheppard and Cannell (1985), no type of genetic control was attributed to the hybrids, although the hardiness appeared to be intermediate to the parental types and therefore appeared to be under additive genetic control. This agrees with findings for the polygenic inheritance of the fall frost hardiness (Norell et al., 1986; Ying & Morgenstern, 1982). From the hierarchical analysis, it appears as if interactions of temperature with regions and crosses are important. Differential damage to the two regions which account for significant portions of the variability at differing temperatures is not surprising. The regions have been shown to differ considerably in LT 5 o values and this would explain the significant Temperature*Region interaction. The significant Temperature*Cross(S(R)) Chapter 5. DISCUSSION SO interaction shows that opportunities exist on a cross basis for selection of differential cold hardiness. Freezing-test temperature was always an important source of variation and it seems intuitive that temperature will play a significant role as it is the resistance to this factor which is being measured. Again, among the genetic sources of variation, cross and region appear to be the most important, while subregions were an insignificant source of variation on all dates examined. Although not considered in this thesis, the role of spring frost hardiness or the rate at which dehardening occurs is very important to a spruce hybridization project in B.C. This has been a problem in Britain, with Sitka spruce, as many plantations have suffered from late spring frosts at the time of vegetative budburst (Cannell, 1984). In the hybrids studied by Sheppard and Cannell (1985) , a major concern was the fact that the hybrids were less hardy than the Masset (Q.C.I.) provenance in early spring. The hybrids de-hardened a week earlier in the spring and this correlates well with the higher proportion of flushed buds on the hybrids in this experiment. Based on the results of this thesis, spruce hybrids have potential for reforestation in the B.C. interior. Sitka spruce, in almost all cases, will outperform the hybrids on the coast and it is unlikely the hybrids will be useful in this area. In the interior the climate is too severe for Sitka spruce, but increases in growth can be made by incorporating Sitka spruce genes, for rapid growth, into the genotypes to be deployed in this area, while still maintaining a high level of cold hardiness. The interspecific hybrids have potential in the boreal, sub-boreal, interior cedar-hemlock and montane spruce biogeoclimatic zones. The use of hybrids, which are relatively expensive to produce, would therefore probably be restricted to the best sites in these zones, with high fertility and adequate moisture, for reforestation. The climate of the Engelmann spruce-subalpine fir zone may prove to be too severe for the seedlings, and tree improvement in this area will probably not be Chapter 5. DISCUSSION 81 a profitable venture. The transition area between Sitka and white spruce, in the coastal western hemlock zone, is another area of the province in which hybrids may be a practical alternative. Hybrids occur naturally in this area, the climate is relatively mild and the introduction of interior spruce genes, for cold-hardiness, allows the forester to extend the range of Sitka spruce. As this area is milder than the interior, tree breeders may wish to backcross the hybrids to Sitka spruce and select the most cold hardy genotypes for use in this area. Chapter 6 C O N C L U S I O N S This thesis has addressed many questions into whether spruce hybridization is a rea-sonable tree breeding strategy for yield improvement in the B.C. interior. This project was meant as an exploration into the genetical and physiological characteristics and con-straints of using spruce hybrids. The results are based on seedlings grown in one coastal site with the main emphasis on showing the feasibility and desirability of using hybrids. The main conclusions of this work are summarized below. 1. Strong maternal effects on filled seed %, germination capacity, seed weight and rate of germination were found for the families in this experiment. 2. For height growth, most of the variability occurred at the regional, macrogeo-graphic, level and at the individual cross level while subregions were not a large source of variation. -3. Several parents were superior in G C A values, which accounted for the greatest amount of variation, from all regions and it is concluded that height growth is mainly under additive genetic control. 4. Several specific crosses, hybrids and interior X interior, exhibited outstanding per-formance due to non-additive genetic effects and, although not the main source of variation, potential for large gains in yield are available through the use of SCA in clonal programs. 82 Chapter 6. CONCLUSIONS 83 5. For bud set most of the variation was at the regional and cross level while subregions were not a large source of variation. 6. Bud set. is considered to be due to additive genetic effects, although the maternal parent appears to account for a larger portion of the variability. This maternal effect did decrease from year one to year two in this experiment although it is still very obvious in year two. 7. For bud break the regions and crosses were significant sources of variation, but did not contribute as much variation as in height and bud set. Most of the variability was due to residual error suggesting that genetic control is weaker for bud burst. 8. Large differences between regions were found for cold-hardiness and the hybrids showed intermediate behaviour illustrating the predominance of additive gene con-trol for frost hardiness. Chapter 7 R E C O M M E N D A T I O N S This experiment has investigated some of the questions involved in the production of spruce hybrids for use in B.C. No work can cover all aspects of an area and in this section recommendations from the thesis will be given, as well as listing some of the areas of investigation which have not been addressed by this thesis, but are considered to be important to hybrid spruce production. 1. Plantations established from the seedlings of this experiment should be maintained and monitored for information on the field performance and environmental speci-ficity of hybrids in direct comparison with the parental species. 2. Selection of parents based on GCA should be at the individual tree level for the production of hybrids. It is recommended that due to the suggested maternal influence on budset, the interior sources be used as the maternal parent to try and reduce the risk of frost damage in the fall. 3. The area of Spillamachbeen, B .C . should be investigated as an area of outstanding genetic value for interior spruce. 4. The specific superior crosses found in this experiment should be reproduced and placed out with standard progeny tests in the province. 84 Chapter 7. RECOMMENDATIONS 85 In future experiments on spruce hybridization the following are various areas in need of research: • Growth rate and the timing of this growth should be investigated more intensively throughout the summer. • The effects of drought on the hybrids must be considered and tests placed on dry areas may yield a great deal of information on differences in drought tolerance. • The effect of Genotype X Environment interaction needs to be addressed as many unexplained interactions and significant block effects were found in this experiment. L I T E R A T U R E CITED Abercrombie, M . , C.J . Hickman, and M . L . Johnson. 1982. The Penguin Dictionary of Biology. Seventh Ed. Penguin Books. 323 pp. Becker, W.A. 1984. Manual of quantitative genetics. 4th Ed. Academic Enterprises, Pullman, WA. 188 pp. Bongarten, B.C and J.W. Hanover. 1982. Hybridization among white, red, blue and white X blue spruces. For. Sci. 28:129-134. Boyle, T.J .B. 1987. A diallel cross in black spruce. Genome 29:180-186. Bramlett, D.L. , T.R. Dell and W.D. Pepper. 1975. Genetic and maternal influences on Virginia pine seed germination. Silvae Genetica 32: 1-4. Burley, J. 1966. Genetic variation in seedling development of Sitka spruce Picea sitchensis (Bong.) Carr. Forestry 39:68-94. Cannell, M.G.R. 1984. Spring frost damage on young Picea sitchensis 1. Occurrence of damaging frosts in Scotland compared with western North America. Forestry 57:159-175. Cannell, M.G.R.and R.I. Smith. 1983. Thermal time, chill days and prediction of bud-burst in Picea sitchensis. J. Appl. Ecol. 20:951-963. 86 LITERATURE CITED Copes, D.L. and R.C. Beckwith. 1977. Isoenzyme identification of Picea glauca, P. sitchensis and P. lutzu populations. Bot. Gaz. 134:512-521. Coupe, C.A., R.A. Comeau, M.V. Ketcheson and R . M . Annas. 1982. A guide to some common plants oj the Skeena. area, British Columbia. B . C . Min. Forests, Land Management Handbook 4. 215 pp. Crow, J.F. 1948. Alternative hypothesis of hybrid vigor. Genetics 33:477-486. Daubenmire, R. 1968. Some geographic variations in Picea sitchensis and their ecological interpretation. Can. J. Bot. 46:787-798. r 19 72. On the relation between Picea pungens and Picea. engelmannn in the Rocky Mountains. Can. J. Bot. 50:733-742. 19 74. Taxonomic and ecological relationships between Picea glauca and Picea engelmannii. Can. J . Bot. 52:1545-1560. Den Ouden, P. and B . K . Boom. 1982. Manual oj cultivated conifers hardy in the cold and warm-temperate zone.3rd Ed . Martinus Ninjhoff/Dr. Junk Publishers. The Hague. 520 pp. Doorenbos, J . 1953. Review of the literature on dormancy in buds of woody plants. Medelingen Landbouwhogeschool te Wageningen/Nederland 53:1-24. Dormling, I., I. Ekberg, G. Eriksson and D. von Wettstein. 1974. The inheritance of the critical night length for budset in Picea abies (L.) Karst. Proc. Joint IUFRO Meetings, S.02.04 .1, Stockholm, Sweden. Session VI . pp 439-448. LITERATURE CITED 88 Dunlap, J.R. and J.P. Barnett. 1983. Influence of seed size on germination and early development in Loblolly pine (Pinus taeda L.) germinants. Can. J . For. Res. 13:40-44. Falconer, D.S. 1981. Introduction to quantitative genetics. 2nd Ed. Longman Inc., New York, N . Y . 340 pp. Fowells, H.A. 1965. Si]vies of forest trees of the United States. U.S. Dep. Agric. Forest Service Agriculture Handbook No. 271, 762 pp. Florin, R. I960., The distribution of conifer and taxad genera in time and space. Acta Horti Bergiania 20:122-144. Fowler, D.P. 1980. Hybridization of black and Serbian spruce. Can. For. Serv., Mar. For. Res. Cen. Inf. Rep. M-X-112. 30 pp. —: 1983. The hybrid black X Sitka spruce: implications to phylogeny of the genus Picea. Can. J. For. Res. 13:108-115. 1987. The hybrid white X Sitka spruce: species crossability. Can. J . For. Res. 17:413-417. Freund, R.J . 1986. Use and misuse of G L M . SUGI 11 Tutorial, Texas A & M University, pp 30-35. Garman, E . H . 1957. The occurrence of spruce in the interior of British Columbia. B.C. For. Serv. Tech. Publ. T49. 31 pp. LITERATURE CITED 89 Gaussen, H . 1966. Les gymnospermes actuelles et fossiles. Fasc. 8 genre: Pseudolanx, Keteleria. Larix, Pseudotsuga, Pitiytes, Picea, Cathya, Tsuga. Trav. Lab. For., Toulouse, T.II, Sec. 1, Vol. I, Partie II. 481-672. Gerhold, H.D. and Y.S. Park. 1986. Population hybridization in Scotch pine (Pinus sylvestns L.): II. combining ability comparisons. Silvae Genetica 35:195-201. Gordon, A . G . 1976. The taxonomy and genetics of Picea rubens and its relationship to Picea marxana. Can. J . Bot. 54:781-813. 1981. Genetics of genecology of spruce. Proc. 18th Can. Tree Imp. Assoc. Part 1:112-116. Glerum, C. 1985. Frost hardiness of coniferous seedlings: principles and applications. In Evaluating seedling quality: principles, procedures, and predictive ability of major tests, Duryea, M . L . , Ed., Forest Research Laboratory, Oregon State University, Corvallis, Oregon, pp 7-15. Habeck, J.R. and T.W. Weaver. 1969. A chemosystematic analysis of some hybrid spruce (Picea) populations in Montana. Can. J. Bot. 47:1565-1570. Hanover, J .W. and R.C. Wilkinson. 1970. Chemical evidence for introgressive hybridiza-tion in Picea. Silvae Genetica 19:17-22. Harrison, D.L.S. and J.N. Owens. 1983. Bud development in Picea engelmanniil. Vegeta-tive bud development, differentiation, and early development of reproductive buds. Can. J. Bot. 61:2291-2301. LITERATURE CITED 90 Hoffmann, D. and J. Kleinschmit. 1979. An utilization program for spruce provenance and species hybrids. In Proc. I t lFRO Joint Meet. Work. Parties, Norway spruce. Sect. 2.03.11-2.02.11. Bucharest, Romania, pp 216-234. Hyun, S.K. 1972. The possibility of F2 utilization in pine hybridization. Proc. IUFRO Genet. - S A B R A O Jt. Symp. Gov. For. Exp. Stn. Jap. C-4(I). 10 pp. Kleinschmit, J. 1979. Present knowledge in spruce provenance and species hybridization potential. Proc. IUFRO Joint Meet. Work. Parties, Norway spruce Publ. Lower Saxony For. Res. Inst., Fed. Repub. Germany. 187-201. Kossuth, S.V. and G.H. Fechner. 19 73. Incompatibility between Picea pungens Engelm. and Picea engelmannii Perry. For. Sci. 19:50-60. Krajina, V . J . , K . Klinka and J. Worrall. 1982. Distribution and ecological characteristics of trees and shrubs of British Columbia. Faculty of Forestry, University of British Columbia, Vancouver. 131 pp. Kraus, J.F. and R. Lines. 1976. Patterns of shoot growth, growth cessation and bud set in a nursery test of Sitka spruce provenances. Scottish Forestry 30:16-24. Kriebel, H.B. 1975. Interspecific incompatibility and inviability problems in forest trees. In Proc. 14th Meet. Can. Tree Imp. Assoc., Fredericton, N.B. , Aug 28-31, 1973, pp 67-75. LaRoi, G. and J.R. Dugle. 1969. A systematic and genecological study of Picea glauca and P. engelmannii, using paper chromatograms of needle extracts. Can. J . Bot. 46:649-687. LITERATURE CITED 91 Lavender, D.P. 1984. Plant physiology and nursery environment: interactions affecting seedling growth. In Forest Nursery Manual: Production of Bareroot Seedlings. Duryea, M.L. and T.D. Landis (Eds.) Forest Research Laboratory, Oregon State University, Corvallis, Oregon, pp 133-141. :— 1985. Bud dormancy. In Evaluating Seedling quality: principles, pro-cedures, and predictive abilities of major tests, Duryea, M.L . , Ed., Forest Research laboratory, Oregon State University, Corvallis, Oregon, pp 7-15. Lester, D.T. 1975. The role of interspecific hybridization in forest tree breeding. In Proc. 14th Meet. Can. Tree Imp. Assoc., Fredericton, N.B., Aug. 28-31, 1973. pp 85-98. Little, E.L., Jr. 1953. A natural hybrid spruce in Alaska. Jour. For. 51:745-747. Logan, K . T . and D.F.W. Pollard. 1978. Components of growth and their relationship to early testing. C O J F R C Symposium Proceedings O-P-7: pp 181-188. Lyons, J . M . , J .K. Raison and P.L. Steponkus. 1979. The plant membrane in response to low temperature: An overview. In Low temperature stress in crop plants. J .M. Lyons, D. Graham and J .K. Raison, Eds. Academic Press. New York, pp 1-24. Manley, S.A.M. 1975. Genecology of hybridization in red spruce (Picea rubens Sarg.) and black spruce (Picea mariana (Mill.) B.S.P.). PhD dissertation, Yale University. 154 pp. LITERATURE CITED 92 Mikkola, L. 1969. Observations on interspecific sterility in Picea. Annales Botanici Gennici 6 :285-339. Mitchell, R .G . , N.E . Johnson and K . H . Wright. 1974. Susceptibility of 10 spruce species and hybrids to the white pine weevil ( = Sitka spruce weevil) in the Pacific Northwest. USDA Forest Service. Res. Note PNW -225 . 8 pp. Morgenstern, E .K . 1969. Genetic variation in seedlings of Picea mariana (Mill.) B.S.P. I. Correlation with ecological factors. II. Variation patterns. Silvae Genetica 18:151-167. Neinstaedt, H. 1968. White spruce seed source variation and adaptation to 14 planting sites in Northeastern United States and Canada. Proc l\th Comm. For. Tree Breed, in Can.:183-184. Quebec, Canada. Neinstaedt, H . and J.P.King. 1969. Breeding for delayed budbreak in Picea glauca (Moench) Voss - potential frost avoidance and growth gains. Proc. Second World Consult. For. Tree Breeding, Washington, D.C. 1:61-80. Neinstaedt, H. and A. Teich. 1971. The genetics of white spruce. USDA For. Serv. Res. Pap. W O - 1 5 , 24 pp. Norell, L. , G. Eriksson, I. Ekberg and I. Dormling. 1986. Inheritance of autumn frost hardiness in Pinus sylvestris L. seedlings. Theor. Appl. Genet. 72 :440 -448 . Oglivie, R.T. and E. Von Rudloff. 1968. Chemosystematic studies in the genus Picea. (Pinaceae). IV. The introgression of white and Engelmann spruce as found along the Bow River. Can J. Bot. 46 :901 -908 . LITERATURE CITED 93 O w e n s , J.N. and M . Molder. 1973. A study of DNA and mitotic activity in the vegetative apex of Douglas-fir during the annual growth cycle. Can. J. Bot. 51:1395-1409. Owens, J.N., M . Molder and H. Langer. 1977. Bud development in Picea glauca. I. Annual growth cycle of vegetative buds and shoot elongation as they relate to da te and temperature sums. Can. J. Bot. 55 :2728-2745. Park, Y.S. and H.D. Gerhold. 1986. Population hybridization in Scotch pine (Pinus sylvestris L.):I genetic variance components' and heterosis. Silvae Genetica 35 :159-165 . Perry, T.O. 1976. Maternal effects on the early performance of tree progenies. In Tree Physiology and Yield Improvement. Cannell, M.G.R. and F.T. Last. Eds. Academic Press, L o n d o n , U K . pp 473 -481 . Rehder, A. 1947. Manual of cultivated trees and shrubs hardy in North America. 2nd Ed. The MacMillan Company, New York. 996 pp. Rehfeldt, G.E. 1986. Performance of Douglas-fir intervarietal hybrids after 10 years of field testing. USDA For. Serv. Res. Note INT -355. 2pp . Richmond, C. 1989. Annual report of the Ministry of Forests and Lands for the fiscal year ended March 31, 1989. 91 pp. Riemenschneider, D. and C.A. Mohn. 1975. Chromatographic analysis of an open-pollinated Rosendahl spruce progeny. Can. J. For. Res. 5 :414-418 . LITERATURE CITED 94 Ritchie, G.A. 1984. Assessing seedling quality. In Forest Nursery Manual: Production of Bareroot seedlings, Duryea, M.L. and T.D. Landis (Eds.) Forest Research Laboratory, Oregon State University, Corvallis, Oregon, pp 243-259. Roche, L. 1969. A genecological study of the genus Picea in British Columbia. New Phytol. 68:505-554. 1970. The silvicultural significance of geographic variation in the white Engelmann spruce complex in British Columbia. For. Chron. 46:1-10. Roche, L. and D.P. Fowler. 1975. Genetics of Sitka spruce. USDA For. Serv. Res. Pap. WO-26. 155 pp. Russell, J . and C. Ferguson. 1990. Production of genetically improved stecklings of interior spruce. F R D A Research Report 110. 15 pp. Sakai, A. and C.J. Weiser. 1973. Freezing resistance of trees in North America with reference to tree regions. Ecology 54:118. Samuel, C.J .A. , R .C.B. Johnstone and A . M . Fletcher. 1972. A diallel cross in Sitka spruce: Assessment of first year characters in an early greenhouse test. Theor. Appl . Gen. 42: 53-61. SAS Institute Inc. 1985. SAS User's Guide: Statistics, Version 5, Gary, NC. 956 pp. Satterthwaite, F .E. 1946. An approximate distribution of estimates of variance compo-nents. Biometrics Bull . 2:110-114. LITERATURE CITED 95 Scagel, R.L. 1984. Morphological and anatomical variation of Picea in northwestern British Columbia. MSc. Thesis, Dept. of Botany, University of British Columbia. Schaffer, P.R. and J.W. Hanover. 1985. A morphological comparison of blue and En-gelmann spruce in the Scotch Creek drainage, Colorado. Silvae Genetica 34:105-111. Schmitt, D . M . 1975. Interspecific hybridization in forest trees: potential not realized. In Proc. 14th Meet. Can. Tree Imp. Assoc., Fredericton, N.B. Aug 28-31, 1973, pp 57-66. Sheppard, L.J.and M.G.R. Cannell. 1985. Performance and frost hardiness of Picea sitchensis X Picea glauca hybrids in Scotland. Forestry 58:70-74. Shull, G.H. 1914. Duplicate genes for capsule form in Bursa bursa-pastoms. Z.I.A.V. 12:97-149. Shull, G.H. 1948. What is 'heterosis'? Genetics 33:439-446. Sweet, G.B. 1965. Provenance differences in Pacific Coast Douglas-fir. Silvae Genetica 14:46-56. Szmidt, A .E . , El-Kassaby, Y . A . , Sigurgeirsson, A. , Alden, T., Lindgren, D. and J.E. Hallgren. 1988. Classifying seedlots of Picea sitchensis and P. glauca in zones of introgression, using restriction analysis of chlorplast DNA. Theor. Appl. Genet. 76:841-845. LITERATURE CITED 96 Taylor, T .M.C. 1959. The taxonomic relationship between Picea glauca (Moench) Voss and Picea. engelmannii Parry. Madrona 15:111-115. Taylor, T . M . C . and T.F. Patterson. 1980. Biosystematics of Mexican spruce species and populations. Taxon 29:421-469. Warren, W.G. 1986. On the presentation of statistical analysis: reason or ritual. Can. J. For. Res. 16:1185-1191. Weiser, C J . 1970. Cold resistance and injury in woody plants. Science 169:1269-1277. White, T .L . and G.R. Hodge. 1989. Predicting Breeding Values with Applications in Forest Tree Improvement. Kluwer Academic. Publishers. Norwell, M A . 367 PP-Wilkinson, R . C , J.W. Hanover, J .W. Wright and R.H. Flake. 1971. Genetic variation in the monoterpene composition of white spruce. For. Sci. 17:83-90. Woods, J. 1988. Nursery trials of Sitka-lnterior spruce hybrids. F R D A Research Memo •No. 59. 2 pp. Worrall, J . 1983. Temperature-bud-burst relationships in amabilis and subalpine fir provenance tests replicated at different elevations. Silvae Genetica 32:203-209. Wright, J .W. 1955. Species crossability in spruce in relation to distribution and taxon-omy. For. Sci. 1:319-349. LITERATURE CITED 97 Yeh, F.C. and J .T . Arnott. 1986. Electrophoretic and morphological differentiation of Picea sitchensis, Picea glauca, and their hybrids. Can. J . For. Res. 16:791 — 798. Ying, C C . and E . K . Morgenstern. 1982. Hardiness and growth of western spruce species and hybrids in Ontario. Can. J . For. Res. 12:1017-1020. Appendix A 98 Appendix A. The origins of the parent trees. 99 The origins of the parent trees; T R E E L O C A T I O N E L E V ( m ) L A T . LONG. SEX | QUEEN C H A R L O T T E ISLANDS 293 Pacofi Bay 335 52°51' ]31°54' F 311 Gogit Pass 260 52° 39' 131°30' F 212 Talunkwan 20 52°50' 131°50' F 374 Alliford 10 53° 10' 132° 04' F 276 Selwyn Point 15 52°51' 131°51' F 220 Louise Point 20 52°58' 131°54' M 209 Talunkwan 20 52°49' 131°42' M 242 Tanu Island 20 52°46' 131°41' M 260 Richardson Island 20 52°43' 131°45' M 266 Carmichael Passage 5 52°49' 131°54' M V A N C O U V E R I S L A N D 130 Waukwass Cr. 30 50° 35' 127°23' F 192 Kwokwesta Cr. 88 50° 31' 127°33' F 121 Yellow Bluff 213 49°52' 127°08' F 50 Caycuse Cr. < 30 50° 23' 127°29' F 53 Easy Cr. 8 50° 08' 127°20' F 149 Porntt Cr. 30 50°55' 127°10' M 15 Kaouk River 30 50°05' 126°59' M 187 Cleogh Cr. 119 50° 27' 127°45' M 58 Easy Cr. 30 50°09' 127°21' M 161 Holberg Inlet 152 50°41' 128°11' M EAST K O O T E N A Y 20 1515 50°06' 114°50' F '82 1060 51°15' 116° 59' F 6 1730 50°28' 116°31' F 89 1375 51°02' 116° 40' F 101 1640 51°27' 116°17' M 4 1525 50°46' 116°42' M 48 1800 50°02' •116°11' M 23 955 . 49°23' 115°00' M 81 1350 50°43' 115°35' M 3 1350 50°43' 115°35' M P R I N C E G E O R G E 25 Hwy 16 East 763 53°53' 122°12' F 155 Swift R. Rd. 1143 52°58' 121°50' F 4 Aleza Lake 610 54°05' 122°05' F 42 Spring Mt. 1006 53°47' 122°06' F 135 Ahbau-Nave 1006 53°25' 122°14' F 21 Hwy 16 East 732 53°54' 122°02' M 134 Ahbau-Nave 1006 53°25' 122°13' M 106 Ahbau-Nave 1006 53°24' 122° 04' M 32 Willor R. 823 53°46' 122°15' M 122 Ahbau-Nave 1143 53°18' 122° 08' M Appendix B The ingredients of the soil mixture 2 bales of peat moss 1.33 bales vermiculite 1.8 kg Nutracote 16 -10 -10 (180 days) 0.5 kg dolomite lime 0.4 kg gypsum Micronutrients 21.8 g (12.0 %) C a 19.8 g (10.9 %) S 9.1 g (5.0 %) Mg 6.4 g (3.5 %) Mn 12.7 g (7.0 %) Fe 3.6 g (2.0 %) Zn 0.9 g (0.5 %) Cu — (0.03 %) B — (0.03 %) Mo 100 A p p e n d i x C Range of coefficients of the hierarchial analysis D A T A S E T R A N G E C l c 2 c 3 HT88 low 3.39 12.31 4.37 high 3.81 14.22 5.51 HT89 low 3.41 12.35 4.38 high 3.83 14.29 5.53 SET88.1 low 3.26 12.32 4.38 high 3.62 13.55 6.90 SET88.2 low 3.20 11.44 5.41 high 3.56 13.14 6.73 SETS 8.3 low 3.06 10.17 4.78 high 3.45 12.61 6.47 SET89.1 low 3.42 12.32 4.38 high 3.78 14.21 5.45 SETS 9.2 low 3.28 11.73 4.23 high 3.67 13.67 5.38 SETS 9.3 low 3.37 12.25 4.35 high 3.79 14.12 5.47 SETS 9.4 low 3.09 11.26 3.99 high 3.68 13.21 5.33 Appendix B. The ingredients of the soil mixture 102 Appendix D Range of coefficients of the factorial analysis DATA R A N G E C2 c 3 c 4 c 5 c 6 c7 HT88 low 1.70 1.88 5.57 2.45 8.47 8.84 33.41 high 2.75 2.80 8.16 5.22 13.29 13.02 33.41 HT89 low 1.69 1.88 5.59 2.46 8.50 8.87 33.53 high 2.79 2.81 9.01 5.31 13.33 13.10 33.53 SET88.1 low 1.38 1.55 4.18 2.49 8.70 8.87 33.47 high 2.60 2.60 7.99 6.25 15.66 14.45 33.47 SET88.2 low 1.41 1.57 4.26 2.53 8.87 9.04 34.12 high 2.53 2.52 7.94 5.97 15.69 14.19 34.12 SETS 8.3 low 1.29 1.25 3.52 2.01 7.03 7.17 27.03 high 2.44 2.49 7.10 5.69 14.02 13.25 27.03 SETS 9.1 low 2.76 2.76 8.80 4.30 15.60 15.20 52.77 high 2.12 2.12 8.50 5.12 13.26 13.29 52.77 SETS 9.2 low 1.47 1.63 4.81 2.17 7.48 7.64 28.86 high 2.08 2.08 8.33 5.00 13.19 11.63 28.86 SETS 9.3 low 1.79 1.97 5.83 2.59 9.09 9.26 34.99 high 2.81 2.79 9.18 5.20 14.07 12.83 34.99 SETS 9.4 low 1.56 1.71 7.91 2.25 7.87 8.01 30.27 high 2.68 2.68 5.04 4.94 13.32 12.32 30.27 Appendix E Filled seed %, germination % and seed weight (g/100 seeds) of the crosses as defined in Figure 2. CROSS F L L D S D G E R M S D W T CROSS F L L D S D G E R M SDWT l 83 79 .21 41 78 79.2 .23 2 68 42 .25 42 71 93 .25 3 96 94 .28 43 99 92 .30 4 92 70 .28 44 93 91 .26 5 89 27 .22 45 77 75 .29 6 97 43 .23 46 91 84 .25 7 90 78 .27 47 83 85 .21 8 99 77 .27 48 100 96 .31 9 91 83 .28 49 99 91 .28 10 36 8 .23 50 96 31 .26 11 97 66 .22 51 93 78 .27 12 100 88 .21 52 95 95 .25 13 89 29 .17 53 97 76 .22 14 94 71 .21 54 99 83 .22 15 43 23 .15 55 94 74 .24 16 94 65 .25 56 — — — 17 100 88 .22 57 — — — 18 97 84 .16 58 99 81 .20 19 93 86 .23 59 100 70 .16 20 — — — 60 99 82 .26 21 56 50 .20 61 99 78 .27 22 63 57 .24 62 99 92 .27 . 23 90 64 .28 63 100 97 .26 24 89 75 .30 64 100 85 .26 25 94 55 .22 65 100 94 .30 26 90 75 .22 66 100 93 .25 27 40 95 .29 67 74 77 .21 28 67 84 .32 68 100 97 .25 29 92 57 .21 69 97 94 .28 30 — — 70 96 93 .30 31 99 54 .24 71 75 .28 32 55 63 .16 72 99 95 .27 33 100 81 .18 73 40 72 .22 34 77 49 .20 74 99 89 .16 35 11 14 .10 75 98 91 .23 36 97 74 .26 76 98 73 .27 37 91 75 .23 77 95 87 .25 38 100 67 .16 78 93 84 .22 39 88 81 .23 79 96 74 .15 40 12 14 .11 80 100 80 .22 103 Appendix F Sources of variation, degrees of freedom (DF) and mean squares (MS) for the hierarchial a) and factorial b) A N O V A for first year budset at three dates D A T E 1 D A T E 2 D A T E 3 S O U R C E D F MS D F MS D F MS B L O C K B 7 .0418 7 .1324 7 .0516 R E G I O N R 3 9.5714 3 15.1986 3 10.1438 SUBREGION(R) S(R) 12 0.2102 12 0.2666 12 0.1145 CROSS(S(R)) C(S(R)) •51 0.0837 51 0.0550 51 0.0855 B*R 21 0.0533 21 0.0789 21 0.0856 B*S(R) 84 0.0214 83 0.0304 83 0.0484 E R R O R 301 0.0351 292 0.0576 279 0.0421 ) - D A T E 1 D A T E 2 D A T E 3 S O U R C E D F MS D F MS D F MS B L O C K B 7 .0440 7 .0738 7 .0310 S E T S 4 0.1049 4 0.3195 4 0.2496 F E M A L E ( S ) F(S) 15 0.5119 15 1.2122 154 0.8370 M A L E ( S ) M(S) 15 0.75 95 15 1.1530 15 0.4662 F*M(S) 32 0.1677 32 0.0953 32 0.1636 B*S 28 0.0322 28 0.0553 28 0.0507 B*F(S) 105 0.0328 105 0.0448 101 0.0402 B*M(S) 105 0.04 49 104 0.0567 104 0.0417 E R R O R 168 0.02 96 159 0.0152 150 0.0432 104 Appendix G Sources of variation, degrees of freedom (DF) and mean squares (MS) for the h ierarchial a) and factorial b) A N O V A for first year budbreak, at three dates a) — D A T E 1 D A T E 2 S O U R C E DF MS F - V A L . D F MS F - V A L . B L O C K - B 5 0.00 90 1.19 5 0.0025 1.01 R E G I O N R 3 0.34 60 3.12 3 0.0939 2.53 S U B R E G I O N (R) S(R) | 12 0.0945 0.69 12 0.0224 0.77 CROSS(S (R)) C(S(R)) | 51 0.1581 4.23 51 0.0348 2.75 B * R 15 0.02 73 0.81 15 0.0189 1.79 B*S(R) 60 0.03 36 0.90 60 0.0106 0.84 E R R O R 230 0.03 73 - 219 0.0126 -b) — D A T E 1 D A T E 2 S O U R C E D F MS F - V A L . D F MS F - V A L . B L O C K B 5 0.006 7 0.28 5 0.0014 0.08 S E T S 4 0.1378 0.86 4 0.04 63 1.00 F E M A L E ( S ) F(S) 15 0.1284 1.12 15 0.0437 2.02 M A L E ( S ) M(S) 15 0.198 7 1.50 15 0.0181 0.90 F * M 32 0.1046 3.29 32 0.0148 1.52 B*S 20 0.023 9 0.65 20 0.0176 0.98 B*F(S) 75 0.038 2 1.20 75 0.0116 1.20 B*M(S) 75 0.0471 1.48 75 0.0162 1.67 E R R O R 135 0.0318 124 0.0097 105 Appendix H Sources of variation, degrees of freedom (DF) and mean squares (MS) for the hierarchial a) and factorial b) A N O V A for second year budset, at three dates a) D A T E 1 D A T E 2 D A T E 3 D A T E 4 S O U R C E D F MS D F MS D F MS D F MS B L O C K B 5 0.1012 5 0.0113 5 0.0314 5 0.0308 R E G I O N R 3 8.6124 3 10.2537 3 11.4187 3 6.4040 SUB REGION(R ) S(R) 12 0.0841 12 0.1751 12 0.1608 12 0.1320 CROSS(S(R)) C(.S(R)) | 51 0.0886 50 0.3501 51 0.1565 51 0.04 60 B * R 15 0.0351 15 0.0417 15 0.0494 15 0.0183 B*S(R) 60 0.0202 60 0.0268 60 ' 0.0325 60 0.0140 E R R O R 227 0.0203 219 0.0190 228 0.0211 221 0.0156 b) D A T E 1 D A T E 2 D A T E 3 D A T E 4 S O U R C E D F MS D F MS D F MS D F MS B L O C K •B 5 0.0300 5 0.0040 5 0.0257 5 0.0295 S E T S 4 0.1003 4 0.1731 4 0.3923 4 0.1778 F E M A L E ( S ) F(S) 15 0.5555 15 1.4800 15 1.1027 15 0.6799 M A L E ( S ) M(S) 15 0.744 8 15 0.5389 15 0.7454 15 0.3813 F * M 32 0.1159 31 0.0878 32 0.1133 32 0.1888 B*S 20 0.0260 20 0.0262 20 0.0223 20 0.0152 B*F(S) 75 .0269 75 0.0249 75 0.0251 75 0.0177 B*M(S) 75 0.0202 75 0.0199 75 0.02 79 75 0.0115 E R R O R 132 0.0196 124 0.0201 134 0.0227 126 0.0156 106 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0098510/manifest

Comment

Related Items