UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Quantitative genetics and QTL mapping of growth and wood quality traits in coastal Douglas-fir Ukrainetz, Nicholas K. 2006

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

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
831-ubc_2006-0330.pdf [ 9.26MB ]
Metadata
JSON: 831-1.0074997.json
JSON-LD: 831-1.0074997-ld.json
RDF/XML (Pretty): 831-1.0074997-rdf.xml
RDF/JSON: 831-1.0074997-rdf.json
Turtle: 831-1.0074997-turtle.txt
N-Triples: 831-1.0074997-rdf-ntriples.txt
Original Record: 831-1.0074997-source.json
Full Text
831-1.0074997-fulltext.txt
Citation
831-1.0074997.ris

Full Text

Q U A N T I T A T I V E G E N E T I C S A N D Q T L M A P P I N G O F G R O W T H A N D W O O D Q U A L I T Y T R A I T S I N C O A S T A L D O U G L A S - F I R by N I C H O L A S K . U K R A I N E T Z B S c , The University of British Columbia, 2002 A THESIS S U B M I T T E D I N P A R T I A L F U L F I L L M E N T O F T H E R E Q U I R E M E N T S F O R T H E D E G R E E O F M A S T E R OF S C I E N C E in T H E F A C U L T Y OF G R A D U A T E S T U D I E S (Forestry) T H E U N I V E R S I T Y OF B R I T I S H C O L U M B I A March 2006 © Nicholas K . Ukrainetz, 2006 A b s t r a c t Coastal Douglas-fir (Pseudotsuga menziesii (Mirb.) Franco var menziesii) is one of the most commercially important tree species on the west coast of British Columbia and the United States due largely to its superior wood quality. Several growth and wood quality traits, including height, diameter, volume, earlywood, latewood and average density, latewood proportion, fibre length and coarseness, and complete cell wall chemistry, were measured for 600, 26 year old Douglas-fir trees from 15 full-sib families located on four sites in southwestern British Columbia. B u d samples were collected and D N A isolated for molecular marker analysis. Family data were used to calculate broad-sense heritabilities, and genetic, phenotypic and family mean correlations for all traits. Amplified fragment length polymorphism ( A F L P ) data from eight families was used to generate a linkage map employing a joint likelihood function, which contained 19 linkage groups with 120 markers spanning 938.6cM of the Douglas-fir genome. A Q T L map for commercially important traits was created using a combination of interval mapping by sib-pair analysis and single locus analysis by analysis of variance. Fibre properties have the lowest heritability (0.10 - 0.18), while glucose content has the highest heritability (0.98). Growth traits are under moderate genetic control (0.23 - 0.30) as is microfibril angle (0.20). For growth ring traits, earlywood density and average density have moderately high heritabilities (0.54 and 0.47, respectively) followed by latewood proportion (0.30) and latewood density (0.21). Average density is influenced primarily by the proportion of latewood (0.85) and earlywood density (0.74). Growth traits, which are important components of tree improvement programs, are generally positively correlated with fibre traits, microfibril angle and lignin content, but negatively correlated with density and cell wall carbohydrate content. Twenty-two Q T L s were detected for compound traits that explained i i between 3.6% and 17.7% of the phenotypic variation. A n additional 78 individual ring density Q T L s were identified which form 11 Q T L clusters and 11 independent QTLs . The resulting map should serve as a valuable scaffold for future metabolite Q T L mapping and comparative genomic projects. i i i Tab le o f C o n t e n t s Abstract » Table of Contents iv List of Tables vii List of Figures ix Acknowledgements x Chapter 1 1 LITERATURE REVIEW 1 DOUGLAS-FIR 1 Douglas-fir Breeding Program 2 POPULATION STATISTICS 5 Heritability 5 Correlations 8 GROWTH AND WOOD QUALITY TRAITS 11 Growth Traits 11 Fibre Properties 14 Microfibril Angle 15 Wood Density 17 Wood Chemistry 19 MOLECULAR MARKERS 21 AFLP Concepts 21 AFLP Versus Other Molecular Marker Techniques 25 LINKAGE MAPPING 27 Linkage Mapping Theory 27 Linkage Maps for Trees 33 Douglas-fir Linkage Maps 35 QTL ANALYSES 36 QTL Analysis Theory : 36 QTL Analyses in Trees 39 QTL Analyses in Douglas-fir 43 APPLICATIONS OF LINKAGE MAPPING AND QTL ANALYSES 45 Marker Aided Selection 45 QTL Effects and the QTL Concept 47 Comparative Genome Mapping 48 Candidate Gene Mapping : 49 THESIS STATEMENT 50 REFERENCES 51 Chapter 2 62 BROAD-SENSE HERITABILITY, PHENOTYPIC AND GENETIC CORRELATIONS OF WOOD QUALITY TRAITS IN COASTAL DOUGLAS-FIR 62 INTRODUCTION 62 MATERIALS AND METHODS 64 Sample population 64 Phenotyping 66 Growth Traits and Core Sampling 66 Fibre Properties 66 Wood Density 67 iv Microfibril Angle 67 Wood Chemistry 69 Statistical Analysis 71 RESULTS 74 Growth and Yield 74 Fibre Properties 76 Wood Density 79 Microfibril Angle 80 Lignin 82 Carbohydrates 82 Selected Families 83 DISCUSSION 85 Heritability and Genetic Control 85 Relationships Among Traits 90 Implications for Tree Improvement 95 REFERENCES 100 Chapter 3 105 A N A F L P L I N K A G E M A P FOR COASTAL DOUGLAS-FIR USING EIGHT FULL-S IB FAMILIES 105 INTRODUCTION 105 MATERIALS AND METHODS 108 Plant Material 108 DNA Isolation 108 AFLP Template Preparation and Reactions 109 Detection and Scoring of AFLP Fragments 112 Linkage Analysis 114 Marker Distribution and Family Contributions 115 RESULTS '. 117 AFLP Polymorphisms 117 Map Results 118 Marker Distribution and Family Contributions 119 DISCUSSION .' 124 Primer Screening and Segregation 124 Map Comparisons 126 Unmapped Loci 128 Applications 129 REFERENCES 131 Chapter 4 136 IDENTIFICATION OF QUANTITATIVE TRAIT Loci FOR W O O D QUALITY AND G R O W T H TRAITS FOR EIGHT F U L L -SIB COASTAL DOUGLAS-FIR FAMILIES 136 INTRODUCTION 136 MATERIALS AND METHODS 140 Sample Population 140 Phenotypic Data 141 Growth Traits and Core Sampling 141 Fibre Length and Coarseness 141 Wood Density 142 Microfibril Angle 142 Wood Chemistry 144 Genotypic Data and Map Construction 145 QTL Analysis 146 Expected Number of Undetected QTLs 148 RESULTS 149 Number and Effect of Detected QTLs for Compound Traits 149 Number and Effect of Detected QTLs for Ring Density Traits 152 Number and Effect of Undetected QTLs for Compound Traits 153 DISCUSSION 157 v QTLs and QTL Effects 157 Expected Number of QTLs 160 Temporal Stability of QTLs 162 REFERENCES ...164 Chapter 5 171 Conclusions 171 RECOMMENDATIONS FOR IMPROVEMENT 174 RESEARCH SIGNIFICANCE 176 FUTURE RESEARCH PROSPECTS 176 Appendix 1 178 A N O V A TABLES 178 Appendix 2 181 FAMILY M E A N S AND STANDARD DEVIATIONS 181 vi L i s t o f Tab les Page 1.1 Narrow-sense heritability (h 2) estimates for several tree species and traits reported in the literature. 12 2.1 Components of variance. 72 2.2 Broad-sense heritability (H 2 ) (with standard error estimates in parentheses), and F-values (p-values in parentheses) for family, site and family-by-site variation for growth and wood quality traits in coastal Douglas-fir. 75 2.3 Phenotypic correlations for coastal Douglas-fir (top number is the correlation coefficient and the bottom number is the p-value significant at a = 0.05; bold values are significant). 77 2.4 Genetic correlations (above diagonal; top number is the correlation coefficient and the bottom number is the standard error) and family mean correlations (below diagonal; top number is the correlation coefficient and the bottom number is the p-value significant at a = 0.05; bold values are significant) for Coastal Douglas-fir wood quality traits. 78 2.5 Phenotypic, genetic and family mean correlations between microfibril angle and ring 17 traits (respective p-values for phenotypic and family mean correlations, and standard errors (s.e.) for genetic correlations, are in parentheses). 81 3.1 Number of polymorphic loci generated by each primer combination for primer screening. 113 3.2 Backcross, intercross and the total number of markers generated by each primer combination in the final A F L P analysis. 118 3.3 Backcross, intercross and total number of markers generated for each full-sib family. 119 3.4 Marker density per linkage group. The expected number of markers Q^) is equal to the observed number of markers (m;) i f the Poisson two-tailed p-value is > 0.025. 121 3.5 Frequency data and results of the coefficient of dispersion analysis for marker distribution. 122 3.6 Observed and expected (in brackets) number of markers contributing to linkage from each family. 123 v i i 4.1 Average height (HT), diameter (DBH) , volume ( V O L ) , earlywood density (EWD), latewood density ( L W D ) , latewood proportion (LWP) and average density ( A D ) of the full-sib Douglas-fir families employed for Q T L analysis. 141 4.2 Components of variance and degrees of freedom used to calculated the proportion of variation explained by each Q T L for the corresponding marker. 148 4.3 Q T L s detected using interval mapping. The F is the F-value for the regression analysis and Fo.os and Fo.oi are critical F-values for significance at the corresponding a levels. Marker is the closest A F L P marker to the Q T L position and is used to assess the proportion of phenotypic variation explained by the Q T L . P-values are reported for the significance of genotype nested within family for the single marker Q T L analysis. 150 4.4 Ring density interval mapping QTLs . F is the F-value for the regression analysis and Fo.os and Fo.oi are the critical F-values at the corresponding a value. Marker is the closest marker to the Q T L position and " P " is the p-value of the single locus analysis testing genotype nested within family. Year is the year for which each ring corresponds. 155 4.5 Number of undetected Q T L , the average effect of undetected Q T L and the expected number of Q T L for composite traits. 157 v i i i L i s t o f F i g u r e s Page 1.1 Schematic representation of the major breeding activities of a typical tree improvement program. Each cycle begins with the selection of a selected population (adapted from White, 1987). 4 1.2 Schematic of the amplified fragment length polymorphism ( A F L P ) procedure using the adapters and primers from the protocol employed in the Genome Development Centre at U B C (procedure adapted from Remington et al. 1999). Blue text indicates adapter sequences, red text indicates primer sequences and green text indicates selective nucleotides. 22 1.3 A n example of an amplified fragment length polymorphism ( A F L P ) polyacrylamide gel image using E + A C G / M + C C G A . Lanes labeled " M " are ladders and " P " are parents; all other lanes are progeny from a single family. L o c i identified with " A " are backcross markers segregating in a 1:1 ratio with only one parent carrying the allele. L o c i identified with a " B " are intercross markers segregating in a 3:1 ratio with both parents carrying the allele. 23 2.1 Image o f a tracheid cell showing: A ) microfibril angle ( M F A ) and B ) standard curve. 69 2.2 Least square means for families designated as good (G), intermediate (I) and poor (P). Groups with the same letter are statistically the same at a = 0.05. 84 3.1 A F L P linkage map for integrated segregation data from eight full-sib families of coastal Douglas-fir. Marker names are the E+3/M+4 primer combination and size. Map distances are in c M estimated by the Kosambi mapping function. 120 4.1 Q T L map for composite and ring density traits for coastal Douglas-fir. Q T L s marked " D E N " contain Q T L s detected in multiple years from all four wood density traits. The numbers in brackets are the number of Q T L s detected within the Q T L location. The scale is in c M (Kosambi map function). 154 ix A c k n o w l e d g e m e n t s This work would have never been completed without the help and support of many friends and colleagues. M y gratitude is extended to the members o f the Mansfield lab, particularly Kyu-Young Kang who aided in the field work and completed the fibre quality, density and wood chemistry phenotyping, and Ian Cullis, who was an instrumental part of the field portion of the project. Special thanks is extended to Tony Kozak for his much appreciated advice concerning the statistical analysis o f the project, and Carol Ritland for training and support involving D N A isolation and molecular markers. Sally Aitken was a guiding force in the calculation of heritabilities and correlations, and Kermit Ritland was patient in lending advice concerning linkage mapping and Q T L analysis. Finally, extra thanks must be extended to my supervisor, Shawn Mansfield, for his dedication, support and advice throughout the course of the project. His hard work is an inspiration to his students. C h a p t e r 1 L i t e r a t u r e R e v i e w Douglas-fir The northern Pacific coast of North America is well known for producing lumber of superior quality, and much of this reputation is a result of the vast stands of Douglas-fir (Pseudotsuga menziesii (Mirb.) Franco). In 1791, the Scottish naturalist Menzies was the first European to identify and document this unique species in British Columbia. The common name is taken from David Douglas, who introduced Douglas-fir to Britain in 1827 (van Gelderen 1989). Douglas-fir is a member of the Pine family (Pinaceae) and the genus Pseudotsuga. Members of the genus can be found in North America, China and Japan. The native range of Douglas-fir extends from 55°N in British Columbia, Canada, to southern New Mexico, U . S . A . , and from the west slopes of the Rocky Mountains to the Pacific coast. It is most abundant in Washington, Oregon and British Columbia on the western foothills o f the Cascade Mountains (Department of Scientific and Industrial Research 1957). There are two varieties of Douglas-fir: a coastal variety (var menziesii) and an interior variety (var glauca). The coastal variety (var menziesii) extends from southwest British Columbia to southern California, while the interior variety (var glauca) occurs in the interior of British Columbia and extends south to Pico de Orizaba on the border of Puebla and Veracruz states, Mexico (Rushforth 1987). In the 1970's and 80's, it 1 was estimated that coastal Douglas-fir covered 3.3 mill ion ha in Oregon, 1.6 million ha in Washington, 698,000ha in California and 900,000ha in British Columbia (Oswald et al. 1986). Historically, Douglas-fir has been identified as a species with inherent high quality wood properties, and ideal for heavy construction and building (Department o f Scientific and Industrial Research 1957; Forest Products Research Laboratory 1964). Mature Douglas-fir retains branches only on the uppermost part of its stem resulting in a large proportion of clear wood, has high density (second only to larch in the region), and is not prone to drying defects, making it ideal for high value structural lumber. However, the wide rings produced by fast grown Douglas-fir, and the abrupt transition in density from earlywood to latewood, decreases veneer quality, as veneer quality is largely influenced by wood homogeneity. A s our mature forests diminish, the proportion o f juvenile wood employed in the wood products industry w i l l increase. Douglas-fir juvenile wood has lower wood density, lower latewood percentage, shorter fibres and a larger microfibril angle than mature wood, all o f which reduce the quality o f the timber produced (Megraw 1986). The transition to higher proportions of juvenile wood wi l l have a significant influence on the value and demand for superior Douglas-fir wood products. Douglas-fir Breeding Program The goal of operational tree improvement programs is to deliver maximum genetic gain to plantations. This is done through selection and breeding. Selection is based on the principle that groups of selected individuals w i l l have higher mean genetic 2 value than for the entire population. Most traits of commercial importance are quantitative and controlled by many genes. The goal of selection is to increase the frequency of favourable alleles in the parents of the next generation. Trees with superior phenotypes (termed "plus trees") are chosen from a base population (Figure 1.1), which is the population to which improvement is being applied and consists of all o f the individuals available to selection. The genetic quality of the base population improves after each round in the breeding cycle. Plus trees form the selected population and are generally retained by grafting cuttings into clone banks (White 1987). Some or all o f the selected individuals can be employed in the generation of genetically improved seed stocks for plantations. The selected population can also be used as a breeding population to produce offspring for genetic tests. Breeding individuals in the selected population results in a new mix o f alleles and facilitates recombination (White 1987). In British Columbia, the coastal Douglas-fir breeding program is directed at the Maritime seed zone between 0 and 700m elevation, and is currently focused on recurrent selection for traits o f commercial value and performance stability. The first generation o f testing and the selection of the second generation is nearly complete (Woods et al. 1996). The second generation is divided into sublines containing 1 0 - 2 0 parents, which can be selected with emphasis on different trait combinations. This type of design allows for the control of inbreeding, while maintaining sub-populations with higher frequencies of superior alleles. Furthermore, sub-populations can be used as infusion populations to introduce new alleles to other sublines. The second generation o f tests w i l l employ a polymix to estimate parental breeding values and full-sib families to provide material for selection of parents for the next generation (Woods et al. 1996). 3 ' N J 1 > Production Packaging Operational Population Plantations Figure 1.1: Schematic representation of the major breeding activities of a typical tree improvement program. Each cycle begins with the selection of a selected population (adaptedfrom White 1987). 4 Population Statistics Heritability Heritability is an estimate of the amount of variation in a population that is attributable to differences among relatives. It is an estimate of the degree to which parents pass characteristics to their offspring and can be used to predict gains in selection programs (Zobel and Talbert 1984). There are two types of heritability estimates: individual-tree heritability and the heritability of family means. Heritability can be estimated by a variety of methods including, sib-pair analysis and parent-offspring regression (Zobel and Talbert 1984). Only individual-tree heritability derived from sib-pair analyses w i l l be discussed here. Understanding the contributions of additive and non-additive genetic variation to a population for a given trait is essential to estimating heritability. Genetic variation ( G 2 G ) is caused by three factors: additive genetic variance ( O 2 A ) , epistatic genetic variance (O 2 E) and dominance genetic variance (O 2 D) , and can be expressed as: o-2G = a 2 A + a 2 E + c 2 D . [1.1] Additive genetic variance is a results of multiple alleles that contribute to a quantitative trait. It is described as "additive" since the sum of the contribution of each allele should equal the final phenotypic value for any given individual. Epistatic variation results from an interaction between alleles at different loci, while dominance variation is caused by an interaction of alleles at a common locus. Both epistatic and dominance variation are 5 referred to as non-additive effects. Additive variation is the variation that selection primarily acts upon and is most interesting to tree breeders (Falconer and Mackay 1996; Zobel and Talbert 1984). There are two types of heritability estimates: narrow-sense heritability (h2) and broad-sense heritability (H2). Narrow-sense heritability is the ratio of additive genetic variance to phenotypic variance (a2p): h2= q 2 A [1.2] Since narrow-sense heritability is an estimate of the amount of additive genetic variation that contributes to the overall phenotypic variation of a trait, it is useful to predict gains from breeding efforts. In contrast, broad-sense heritability (H2) is the ratio of the total genetic variation to phenotypic variation and is calculated as: q 2A + q 2 E+g 2 D [1.3] Broad-sense heritability gives an indication of the amount of variation in a population that is attributed to genetics (Falconer and Mackay 1996; Zobel and Talbert 1984). Additive and non-additive genetic variation must be estimated from the relationships among sibs. With random mating and equal variances for progeny resulting from both males and females, the variance due to half-sib families (O 2 F) is equal to a quarter of the additive genetic variation (O 2 F = V4O2A). Variation due to full-sib families is equal to one half of the additive genetic variance and one quarter of the non-additive H 2 = a 2 G -6 genetic variance (O 2F = V20 2A + V402NA). A S a result, analyses with full-sib families cannot isolate additive genetic variance and only broad-sense heritability can be estimated. Phenotypic variation is calculated by adding the components of variance for a particular experimental design. For instance, if a design incorporates families and replications, phenotypic variation is equal to the sum of family variation (O2F), family by replication interaction (CJ2FR) and within-plot variation (O2E). For this type of analysis, narrow-sense heritability is calculated as follows (Zobel and Talbert 1984): h 2= 4o2F [1.4] 2 2 2 O F + C J F R + C T E And similarly, for full-sib families, broad-sense heritability would be estimated as (Zobel and Talbert 1984): H 2 = 2Q-2F [1.5] 2 2 2 O F + C F R + 0"E When individuals from families are replicated in different environments, an interaction term is added to the heritability calculation. If the variation due to location is ignored, some of the environmental variation is incorporated in the family component, and consequently leads to an overestimate. It must be noted that heritability estimates are restricted to the specific conditions and population in which they are derived. As a result, they can only be related to a particular population in a specific environment at a given time (Falconer and Mackay 1996; Zobel and Talbert 1984). 7 The assumption of random mating is made for heritability calculations. However, for populations derived from tree improvement programs in which parents are pre-selected based on phenotype, this assumption is not met. Heritability estimates are affected by assortative mating and selection of parents. Falconer and Mackay (1996) suggest that the correlation between breeding values for individuals mated by phenotype increase additive genetic variance and heritability. However, using families derived from parents that were chosen as extremes for a particular trait may improve precision for heritability estimated by regression. Correlations Correlations are a useful tool to determine relationships among traits. There are three main categories of correlations: phenotypic, genetic and environmental. Phenotypic correlations indicate the current relationship of two traits in the sample population, while genetic correlations are a result of the actions of common genes on two related traits and indicate the degree of pleiotropy or linkage disequilibrium within the population. Environmental correlations measure the effects of common environmental conditions on two traits. Both phenotypic and genetic correlations will be described in more detail below. Correlations are generally calculated as the ratio of the covariance of two traits (covxy) over their respective standard deviations, (varx)/2 and (vary)'/2 (Falconer and Mackay 1996; Zobel and Talbert 1984): rxy= COVxy [1.6] (varx vary)2 8 Phenotypic correlations indicate the direction (positive or negative) and magnitude of the relationship between traits. For example, a correlation of 0.42 between earlywood density and average density indicates that within the sample population, trees with higher average density generally have higher earlywood density, and the relationship is moderate. Genetic correlations are important for three reasons. First, they give an indication of the relationship among traits due to the pleiotropic effects of genes or gene clusters. Second, they provide valuable information about the changes brought on by selection, and third, they can provide information about the relationship between a metric character and fitness in natural populations for predictions about natural selection and adaptation (Falconer and Mackay 1996). Genetic correlations are more difficult to derive than phenotypic correlations. The covariance component for genetic correlations is calculated by taking the product of the trait values for each individual and partitioning the sum of products according to the source of variation (i.e. families). This differs from the derivation of the covariance component in the phenotypic correlation calculation which partitions the sums of squares according to the source of variation. The covariance component for half-sibs is equal to one quarter of the additive covariance (covariance attributed to additive genetic variation for the two traits). For full-sibs, the covariance component is equal to one half of the additive genetic variance, but contains a non-additive (dominance) component. Since there are four possible alleles at any locus in the genome, the probability that two individuals share the same genotype, and the same 9 dominance interactions, is V*. As a result, the covariance component for full-sibs is equal to Vicovx + Vicovo (Falconer and Mackay 1996). Genetic correlations for half-sib families are an estimate of the additive genetic correlation between traits, while estimates from full-sib families reflect a broad genetic correlation due to the incorporation of non-additive effects. Similar to heritability estimates, the genetic correlations for full-sibs are likely to be overestimates of the true additive genetic correlation and should be used with caution when making inferences about changes due to selection of related traits. Genetic correlations are known to be plagued by large sampling errors and are difficult to estimate precisely. They are also influenced by allele frequencies making estimates unique to specific populations. As such, genetic correlations are mere approximations of the actual effects of genetics acting on related traits. The poor precision of genetic correlations can be supplemented using family mean correlations. The mean of full-sib families is equal to the mean of the breeding values of each of the parents which is equivalent to V£C 2 A + V40 2 NA (Falconer and Mackay 1996). The correlations between full-sib family means are an indirect way of measuring the contributions of additive and non-additive genetic effects to the relationships among traits. For half-sibs, family mean correlations describe the additive effects of genes on the two traits. 10 Growth and Wood Quality Traits Growth Traits Growth traits have been a high priority for tree improvement programs since initiation in the mid 1900's (Yanchuk and Kiss 1993). Volume has been recognized as a major contributor to the whole-tree dollar value of trees, and specifically Douglas-fir. Aubry et al. (1998) compared the effect of wood density, stem volume and branch diameter on the whole-tree dollar value based on both visual and machine stress rated (MSR) grading rules. They found that in both cases, stem volume was the single most important factor contributing to the merchantable value of trees due to the greater number of pieces extracted from larger logs. Although wood quality characteristics have been recognized as important factors for improvement programs (Aubry et al. 1998), tree height, diameter and volume are still the most important factors for tree value. Typically, growth traits in trees are under moderate to high genetic control. Published heritability estimates for Douglas-fir (Table 1.1) range between 0.13 and 0.17 for height, and 0.27 to 0.32 for diameter and volume, respectively (St.Clair 1994; Yeh and Heaman 1987). These values are consistent with heritability estimates reported in the literature for other species. Estimates for height growth range between 0.18 and 0.74, but are typically near 0.30 (Ericsson and Fries 2004; Hannrup et al. 2000; Hannrup et al. 1998; Ivkovich 1996; Ivkovich et al. 2002a; Pot et al. 2002; Yanchuk and Kiss 1993). 11 Table 1.1: Narrow-sense heritability (h ) estimates for several tree species and traits reported in the literature. Trait Species h 2 Age Reference SG Picea glauca x engelmanii 0.47 15 Yanchuk and Kiss 1993 SG Pinus taeda 0.4 7 Talbert et al. 1983 SG Pinus taeda 0.76 10 Talbert et al. 1983 SG Pinus taeda 0.44 20 Talbert et al. 1983 SG Pinus taeda 0.3 25 Loo et al. 1985 AD Picea glauca x engelmanii 0.52 20 Ivkovich et al. 2002a AD Pinus elliottii 0.16 15-25 Hodge and Purnell 1993 AD Pinus radiata 0.47 4 Nicholls et al. 1980 AD Pinus sylvestris 0.5 33 Hannrup et al. 2000 AD Pinus sylvestris 0.5 33 Hannrup et al. 1998 AD Populus balsamifera 0.46 12-13 Ivkovich 1996 AD Pseudotsuga menziesii 0.55 15 Vargas-Hernandez and Adams 1991 AD Pseudotsuga menziesii 0.52 18 St. Clair 1994 AD Pseudotsuga menziesii 0.71 12 Loo-Dinkins and Gonzalez 1991 EWD Pinus elliottii 0.13 15-25 Hodge and Purnell 1993 EWD Pseudotsuga menziesii 0.51 15 Vargas-Hernandez and Adams 1991 LWD Pseudotsuga menziesii 0.46 15 Vargas-Hernandez and Adams 1991 LWP Picea glauca x engelmanii 0.51 20 Ivkovich et al. 2002a LWP Pseudotsuga menziesii 0.39 15 Vargas-Hernandez and Adams 1991 HT Picea glauca * engelmanii 0.3 15 Yanchuk and Kiss 1993 HT Picea glauca x engelmanii 0.74 20 Ivkovich et al. 2002a HT Pinus pinaster 0.46 14 Pot et al. 2002 HT Pinus sylvestris 0.29 13 Hannrup et al. 2000 HT Pinus sylvestris 0.28 25 Ericsson and Fries 2004 HT Pinus sylvestris 0.32 33 Hannrup et al. 1998 HT Pinus sylvestris 0.32 33 Hannrup et al. 1998 HT Populus balsamifera 0.18 12-13 Ivkovich 1996 HT Pseudotsuga menziesii 0.13 7 YehandHeaman 1987 HT Pseudotsuga menziesii 0.17 18 St. Clair 1994 DBH Picea glauca x engelmanii 0.11 15 Yanchuk and Kiss 1993 DBH Picea glauca x engelmanii 0.34 20 Ivkovich et al. 2002a DBH Pinus sylvestris 0.18 13 Hannrup et al. 2000 DBH Pinus sylvestris 0.17 25 Ericsson and Fries 2004 DBH Pinus sylvestris 0.27 33 Hannrup et al. 2000 DBH Populus balsamifera 0.19 12-13 Ivkovich 1996 DBH Pseudotsuga menziesii 0.27 18 St. Clair 1994 VOL Pseudotsuga menziesii 0.32 18 St. Clair 1994 12 Table 1.1 continued: Trait Species h 2 Age Reference FL Pinus pinaster 0.172 14 Pot et al. 2002 FL Pinus sylvestris 0.31 11 Hannrup et al. 2000 FL Pinus sylvestris 0.31 25 Ericsson and Fries 2004 FL Pinus sylvestris 0.48 31 Hannrup et al. 2000 FL Pinus taeda 0.11 25 Loo et al. 1985 FL Populus balsamifera 0.13 12-13 Ivkovich 1996 CS Pinus pinaster 0.374 14 Pot et al. 2002 MFA Picea abies 0.12-0.20* 5 Hannrup et al. 2004 MFA Picea abies 0.11-0.29* 12 Hannrup et al. 2004 MFA Picea glauca x 0.29-0.57 20 Ivkovich et al. 2002b engelmanii 0.71* MFA Pinus radiata 1 Donaldson and Burdon 1995 MFA Pinus radiata 0.47* 5 Donaldson and Burdon 1995 MFA Pinus radiata 0.09* 10 Donaldson and Burdon 1995 MFA Pinus radiata 0.72* 15 Donaldson and Burdon 1995 MFA Pinus taeda 0.17-0.41 4 Myszewski et al. 2004 MFA Pinus taeda 0.33 5 Myszewski et al. 2004 MFA Pinus taeda 0.39 19 Myszewski et al. 2004 MFA Pinus taeda 0.51 20 Myszewski et al. 2004 CEL Pinus pinaster 0.343 14 Pot et al. 2002 TL Pinus pinaster 0.471 14 Pot et al. 2002 SG = specific gravity; AD = average density; EWD = earlywood density; LWD = latewood density; LWP = latewood percent; HT = height; DBH = diameter; VOL = volume; FL = fibre length; CS = coarseness; Cel = cellulose; TL = total lignin; MFA = microfibril angle. * Broad-sense heritability estimates based on clonal data. Diameter estimates are usually lower, ranging between 0.11 and 0.34 (Ericsson and Fries 2004; Hannrup et al. 2000; Hannrup et al. 1998; Ivkovich 1996; Ivkovich et al. 2002a; Yanchuk and Kiss 1993). 13 Fibre Properties Fibre length and coarseness are important wood fibre traits which significantly influence paper quality, and thus have implications for the pulp and paper industry. Fibre morphology and cell wall structure directly influence fibre flexibility, plasticity and resistance to processing (Makinen et al. 2002). Coarseness is important to paper strength, which decreases with increasing coarseness (Seth and Kingsland 1990). It is also expected that improvements to growth and wood density through breeding activities will directly result in changes to fibre dimensions. The correlations and heritability estimates for these traits must be determined to make accurate predictions for fibre traits. Fibre length and coarseness seem to be under moderate genetic control. Heritability estimates for fibre length range from 0.11 to 0.48 (Ericsson and Fries 2004; Hannrup et al. 2000; Hannrup et al. 1998; Ivkovich 1996; Loo et al. 1985; Pot et al. 2002), but have been estimated to be as high as 0.85 and 0.97 for summerwood and 0.54 and 0.77 for springwood fibres as measured for two loblolly pine progeny test populations in Louisiana and Georgia, respectively (Goggans 1964). Heritability has been estimated for fibre coarseness in maritime pine at 0.37 (Pot et al. 2002). Via et al. (2004) report that heritability estimates for tracheid length and coarseness vary greatly in the literature, and since measuring and quantifying fibre properties is tedious and costly, most studies have been conducted on relatively few families with small sample sizes, often from a single site and a single ring. In an extreme case, the effect of increasing growth rate due to fertilizer application on fibre properties was measured in Norway Spruce (Picea abies (L.) Karst.) 14 (Makinen et al. 2002). Measurements were taken on trees involved in a long-term nutrient-optimisation experiment in northern Sweden. Trees from control plots were compared to trees from irrigation/fertilization plots. The authors found that in the presence of unlimited water and nutrients, growth rate increased resulting in a concurrent increase in lumen diameter and cell wall thickness, and decrease in cell wall proportion and fibre length. The fibre properties of the fertilizer trials were compared to older trees of similar diameter and were found to possess similar traits at equal distances from the pith. This suggests that fibre morphology may be controlled by maturation and the number of cell divisions (Makinen et al. 2002). In Scots pine (Pinus sylvestris L.), Ericsson and Fries (2004) report a negative phenotypic correlation between fibre length and diameter. However, strong positive phenotypic and genetic correlations were observed between fibre length, coarseness and height in maritime pine, and fibre length, height and diameter growth in Scots pine (Hannrup et al. 2000; Pot et al. 2002). Furthermore, a strong relationship between wood density and coarseness and a weak relationship between wood density and fibre length have been documented by Via et al. (2004). Microfibril Angle Microfibril angle is the angle of cellulose microfibrils, primarily in the S2 layer of the cell wall, measured against the long axis of fibre cells. It is an important determinant of wood strength and elasticity (Cave and Walker 1994). Microfibril angle has been shown to explain a large portion of the variation of longitudinal modulus of elasticity 15 (EL) in loblolly pine (Cramer et al. 2005) and Eucalyptus delegatensis R.T. Baker (Evans and Ilic 2001). Cramer et al. (2005) reports that up to 75% of the variation in E L can be explained by microfibril angle and specific gravity when earlywood and latewood data are combined, while Evans and Ilic (2001) report an R 2 of 0.956 relating E L to the ratio of microfibril angle and density. There has been an increase in the recognition of the importance of incorporating wood quality traits in tree improvement programs, however, until recently, microfibril angle has been largely ignored, probably due to measurement difficulty (Cave and Walker 1994). There is little doubt that microfibril angle is an important contributor to wood strength, yet the genetic control and relationship with other traits are largely unexplored. The more recent advances in X-ray diffraction technology have made the processing of large numbers of samples possible (Batchelor et al. 2000; Cave 1997a, b; Long et al. 2000; Sahlberg et al. 1997). Past analyses were limited to light microscopy and were difficult and time consuming, resulting in small sample sizes which make heritability and correlation estimates less precise. Within individual tracheids, microfibril angle changes little from tip to tip (Anagnost et al. 2002), however, the trend is for decreasing angles from the first earlywood cell to the final latewood cell within a growth ring (Anagnost et al. 2002; Wimmer et al. 2002). Microfibril angle decreases from pith to bark and with increasing height in the tree, and there is a strong relationship with the number of rings from the pith (Donaldson 1992; Erickson and Arima 1974). Heritability estimates are variable, which is probably a result of small sample sizes and lack of site replication. Earlywood and latewood microfibril angle heritability estimates for loblolly pine range from 0.17 to 0.51 and 0.31 to 0.40, respectively (Myszewski et al. 2004), while similar 16 estimates were reported for interior spruce {Picea glauca (Moench) Voss, P. engelmanii Perry x Engel., and their hybrids) of 0.29 - 0.57 (Ivkovich et al. 2002b). Clonal estimates of broad-sense heritability in Norway spruce were 0.11 - 0.29 and 0.16 - 0.36 for earlywood and latewood, respectively (Hannrup et al. 2004), and 0.70 for radiata pine (Donaldson and Burdon 1995). It is widely accepted that microfibril angle is related to fibre length (Barnett and Bonham 2004) with an r2 of 88.4% for Douglas-fir (Erickson and Arima 1974) and significant negative correlations reported by Ivkovich et al. (2002b) and Hannrup et al (2004). There is evidence for negative relationships between microfibril angle and wood density, but no significant correlations were observed for growth traits (Hannrup et al. 2004; Myszewski et al. 2004). Wimmer et al. (2002) subjected a plantation of Eucalyptus nitens to different irrigation regimes and used a fine-scale method for analyzing the relationship between growth and microfibril angle. The authors found evidence that microfibril angle follows growth patterns with positive correlations to growth rate, which provides strong evidence that microfibril angle is linked to growth. Wood Density Wood density is the single most important physical property influencing the quality of wood. It is an excellent predictor of strength, stiffness and hardness, and can be related to pulp quantity and quality (Aubry et al. 1998; Loo et al. 1985; Pot et al. 2002). Aubry et al. (1998) identified wood density as a major contributor to the whole tree dollar value along with branch angle and volume. Wood density results from three interacting components: earlywood density, latewood density and the proportion of latewood 17 (Louzada and Fonseca 2002; Vargas-Hernandez et al. 1994). Earlywood is produced early in the growing season until shoot elongation terminates, then latewood is deposited until cambial activity ceases in late summer (Vargas-Hernandez et al. 1994). Douglas-fir is a determinant species and forms all of the vegetative material for the next growing season in the previous summer. Dendrochronological data suggests that there is a connection between earlywood production and vegetative growth influenced by climate in the previous year. Douglas-fir earlywood ring width is correlated more often to the previous years June, July and August precipitation, whereas latewood width is correlated more often to June precipitation in the active growing season (Watson and Luckman 2002). Douglas-fir wood density traits have been thoroughly analyzed and described (St.Clair 1994; Vargas-Hernandez and Adams 1991). Vargas-Hernandez and Adams (1991) estimate average density heritability to be 0.55, while St. Clair (1994) estimated heritability to be 0.52 in coastal Douglas-fir. Furthermore, Vargas-Hernandez and Adams (1991) separated average density into its components of earlywood density, latewood density and latewood proportion. Heritability estimates for these traits were 0.51, 0.46 and 0.39, respectively. However, the heritability estimates for these studies may be upwardly biased due to a lack of site replication and thus no consideration of site by family interactions. Estimates for density components reveal that earlywood density and average density are more strongly influenced by genetics, compared to latewood density and latewood proportion. The heritability estimates for average density in Douglas-fir are consistent with estimates for other species. Most estimates range between 0.46 and 0.52 for average core density (Hannrup et al. 2000; Hannrup et al. 1998; Ivkovich 1996; 18 Ivkovich et al. 2002a; Nicholls et al. 1980) and 0.30 - 0.47 for core specific gravity (Loo et al. 1985; Talbert et al. 1983; Yanchuk and Kiss 1993). However, rather high heritability estimates (0.76) have been described for loblolly pine (Talbert et al. 1983) and coastal Douglas-fir (0.71) (Loo-Dinkins and Gonzalez 1991) and low estimates (0.16) reported for slash pine {Pinus elliottii) (Hodge and Purnell 1993). Despite these discrepancies, it seems that the genetic control of wood density is fairly conserved across species with h2 estimates near 0.49 on average. Patterns of high average and earlywood density and low latewood density and latewood proportion heritability estimates were also reported for 12 individual rings in maritime pine (Louzada and Fonseca 2002). Both Louzada and Fonseca (2002) and Zamudio et al. (2002) report similar time trends for heritability estimates in density components, which appear to be quite variable from year to year. One of the major obstructions to the inclusion of wood density traits in tree improvement programs has been their negative correlation with growth traits. Vargas-Hernandez and Adams (1991) and St. Clair (1994) report negative genetic and phenotypic correlations between growth and density traits for coastal Douglas-fir. Similar trends were also noted for interior spruce and balsam poplar (Ivkovich 1996; Ivkovich et al. 2002a; Yanchuk and Kiss 1993). Wood Chemistry Wood cell walls are composed of cellulose, hemicellulose and lignin (Baucher et al. 2003). Cellulose is a homopolymer of P-l-4-linked glucan molecules aligned to form 19 microfibrils that dominate the cell wall (Baucher et al. 2003; Saxena and Brown 2005). In contrast, lignin is a heterogeneous hydrophobic phenolic polymer that encrusts the cell wall and contributes to its overall strength (Baucher et al. 2003). Hemicellulose is a heterogeneous material that forms the matrix of the cell wall (Baucher et al. 2003) and is composed of arabinose, galactose, glucose, mannose and xylose. Cellulose is the most abundant cell wall component followed by lignin and hemicellulose (Sjostrom and Raimo 1999). Since the majority of the cell wall is made of cellulose (glucose polymers), the majority of glucose in the cell wall can be attributed to cellulose and can approximate cellulose content. Loehle and Namkoong (1987) warn of the dangers of changing the energy allocation patterns of trees through breeding efforts. They suggest that breeding for growth can lead to an unstable combination of traits that impact tree health and survival. Lignin reduction in agricultural crops has had both positive and negative affects on plant fitness, which suggest that optimum lignin content is essential for plant health, survival and function (Pedersen et al. 2005). A strong inverse genetic relationship (~ -1) between cellulose and lignin content was observed in maritime pine (Pot et al. 2002). Wood chemical components ultimately contribute to the strength of wood and affect pulping (Brown et al. 2003; Pot et al. 2002). During the production of high quality paper, lignin is removed from the polysaccharide component of wood, which must be balanced with cellulose degradation (Baucher et al. 2003). Chantre et al. (2002) conducted a comprehensive study on the feasibility and relevance of genetic selection for pulping potential in Douglas-fir. They report strong positive relationships between density traits and cellulose content, but negative relationships with lignin. They also 20 report a strong positive relationship between modulus of elasticity (MOE), an indicator of lumber strength, and cellulose (0.606), but a negative relationship with lignin (-0.548). In maritime pine, lignin has a positive (0.395) and cellulose has a negative (-0.366) genetic correlation with height growth. As well, there is a strong, positive genetic correlation between cellulose and density (0.624) and a strong negative genetic correlation (-0.544) between lignin and density (Pot et al. 2002). The literature suggests that both lignin and cellulose may be under moderate to strong genetic control with heritability estimates of 0.471 and 0.343, respectively, in maritime pine (Pot et al. 2002). Molecular Markers AFLP Concepts In 1995, Vos et al. (1995) developed a molecular marker technique utilizing selective polymerase chain reaction (PCR) amplification of restriction fragments of unknown sequence for molecular markers. The DNA fingerprinting technique, known as AFLP (amplified fragment length polymorphism), involves three steps: i) restriction digest and ligation of oligonucleotide adapters, ii) selective amplification of sets of restriction fragments (Figure 1.2), and iii) visualization of fragments through gel analysis (Figure 1.3). The restriction digest utilizes two different restriction enzymes. The first is a frequent cutter (typically Msel), which controls the size of restriction fragments and ensures an optimal size for resolution during gel analysis. The second is a rare cutter 21 Digestion EcoRI Msel 3' TTAA; • i J ;AATT Ligation 3' CTGCTGACGCATGGTTAA 5' AACGACGACTGCGTACCAATT P re-Amplification 3' CTGCTGACGCATGG TTAAGTG 5' GACGACTGCGTACC AATTCAC . 1 AAT T ;TAA AATGAGTCCTGAGTAGCAG 5' TTACTCAGGACTCAT 3' AATGAGTCCGAGTAGCAG 5' GGTTACTCAGGCTCATCGTC 3' CCAATGAGTCCTGAGTAG 5' Final Amplification 3' CTGCTGACGCATGGTTAAGTGC CTCCAATGAGTCCTGAGTAG 5' 5' CACGACGTTGTAAAACGACGACTGCGTACCAATTCACG p. GAGGTTACTCAGGACTCATC 3' 3' G T G C T G C A A C A T T T T G C T G C T G A C G C A T G G T T A A G T G C ^ CTCCAATGAGTCCTGAGTAG 5' 5' CACGACGTTGTAAAACGA h. M13 Label Various Sizes Final Product 5' CACGACGTTGTAAAACGACGACTGCGTACCAATTCACG GAGGTTACTCAGGACTCAT 3' M13 Label Figure 1.2: Schematic of the amplified fragment length polymorphism (AFLP) procedure using the adapters and primers from the protocol employed in the Genome Development Centre at UBC (procedure adaptedfrom Remington et al. 1999). Blue text indicates adapter sequences, red text indicates primer sequences and green text indicates selective nucleotides. 22 M P P - P P , P P M W* *wW » * . ~Ww * * * * * ' * W A ^ * a * WW» WW § ** *** W*»* Figure 1.3: An example of an amplified fragment length polymorphism (AFLP) polyacrylamide gel image using E+ACG/M+CCGA. Lanes labeled "M" are ladders and "P" are parents; all other lanes are progeny from a single family. Loci identified with "A" are backcross markers segregating in a 1:1 ratio with only one parent carrying the allele. Loci identified with a "B" are intercross markers segregating in a 3:1 ratio with both parents carrying the allele. (typically EcoRI or PstT), that directs the number of fragments being resolved. Paglia and Morgante (1998) suggest that using Pstl as the rare cutter for A F L P analyses reduces the amount of signal to noise of resolved fragments, and is more appropriate for complex genomes compared to EcoRI , which was reported in the initial A F L P protocol by Vos et al. (1995). The double restriction digest results in three types of D N A fragments with different restriction cut sites at their ends: fragments with two frequent cutter (here we 23 will describe EcoRI as the rare cutter) cut sites, fragments with two rare cutter (here we will describe Msel as the frequent cutter) cut sites, and fragments with one frequent and one rare cutter cut site. Vos et al. (1995) report that >90% of the fragments are expected to be Msel -Msel (twice as many fragments as there are EcoRI restriction sites should have Msel - EcoRI ends), and only a small number of fragments are expected to have EcoRI - EcoRI fragment ends. Adapters are ligated to the single-stranded overhangs created by the restriction digest (Figure 1.2), which are designed with regions that are complementary to the restriction sites and primers used in subsequent amplification stages. Amplification of restriction fragments occurs via PCR reaction with selective primers designed with three regions: i) a 5' region corresponding to the adapter sequence, ii) a restriction site sequence, and iii) one or more selective nucleotides added to the 3' end. The number of selective nucleotides added to the 3' end of the primer controls the number of fragments amplified (i.e. more selective nucleotides results in fewer resolved fragments). In organisms with complex genomes, it is often necessary to add three or more selective nucleotides to reduce the number of resolved fragments to a manageable number. Adding more than three selective nucleotides results in greater tolerance of mismatches, the amplification of alternative fragments, and a loss in selectivity and reproducibility (Cervera et al. 2000; Vos et al. 1995). To compensate for this, a pre-amplification step can be used to add < 3 selective nucleotides. This step reduces the number of fragments in the final amplification and improves resolution. The final primers include selective pre-amplification nucleotides and extra 3' selective nucleotides for 24 further fragment reduction. Pre-amplification and final amplification result in amplification of fragments that have Msel - EcoRI cut sites (Figure 1.2). AFLP Versus Other Molecular Marker Techniques APLPs are a proven technology and are highly reproducible. Vos et al. (1995) conducted experiments using simple genomes with known sequences to confirm results. They used phage X, and AcNPV DNA of known sequence in which all restriction sites were identified. When amplified using the AFLP technique, all predicted restriction fragments were detected. Adding selective nucleotides to the AFLP primers reduced the number of bands four-fold and, more importantly, always resulted in a subset of the original fingerprint. Vos et al. (1995) conclude that the AFLP technique is an efficient way to amplify large numbers of fragments simultaneously, that the amplified fragments are indeed restriction fragments, that the number of fragments increase with genome size, and that selective nucleotides at the ends of primers reduce the number of amplified fragments as expected. Reproducibility between labs is important for analysis comparison (Cervera et al. 2000; Jones et al. 1997; Vos et al. 1995). Jones et al. (1997) compared RAPDs (randomly amplified polymorphic DNA), AFLPs and SSRs (simple sequence repeats) for reproducibility between labs in Europe. RAPDs use primers and PCR reactions to amplify random fragments from genomic DNA. SSRs use specially designed primers to flank areas of highly repetitive DNA, which are highly mutable and often result in 25 multiple alleles. Genetic screening packages and standardized procedures were created for each marker technique and sent to various labs throughout Europe. They concluded that the AFLP and SSR techniques were equally reproducible, while RAPD profiles were difficult to reproduce between labs with many discrepancies. Cabrita et al. (2001) compared the AFLP technique to RAPDs and isozymes in the ability to distinguish between clones of dried fig (Ficus carica L.) using two cultivars (11 clones from Sarilop and one from Sarizeybek). The authors report that the AFLP technique produced more polymorphic loci per reaction when compared to the RAPDs and isozymes. Although isozymes were successful in distinguishing cultivars, and RAPDs for distinguishing cultivars and two Sarilop groups, the AFLP technique successfully separated all 11 Sarilop clones. Cervera et al. (2000) used various inter and intra-specific crosses to assess the nature and inheritance of polymorphic loci generated using AFLPs and found that between 17% - 46% and 28% - 38% of the fragments analyzed in intra-specific and inter-specific crosses were polymorphic, respectively. They also acknowledge that every AFLP fragment generated in inter and intra-specific full-sib progenies was inherited from at least one of the parents. Although AFLPs are superior when considering reproducibility and generating large numbers of polymorphic DNA per reaction (Cabrita et al. 2001; Cervera et al. 2000; Jones et al. 1997; Vos et al. 1995), the technique is reduced to producing dominant markers which are less informative than codominant markers. AFLPs are scored as the presence or absence of a band, which result in only two possible alleles per locus. The heterozygosity of an individual at any locus can only be predicted by determining the 26 segregation. Codominant markers, such as SSRs, reveal the exact size and number of alleles at each locus. Dominant markers are generally less informative, but easier and cheaper to produce and thus make them ideal for saturating linkage maps. Codominant markers are more expensive to produce, but are more reliable for aligning maps and can be used for comparative genome projects within and between species. Linkage Mapping Linkage Mapping Theory Genetic linkage is the association of genes located on the same chromosome. The statistical definition of linkage is the association or non-independence among alleles at more than one locus (Liu 1998). The remainder of this section, unless otherwise referenced, is based on a comprehensive understanding of the descriptions presented by Liu (1998). Linkage analysis has five main steps: single-locus analysis, two-locus analysis, linkage grouping, gene ordering and multi-point analysis. The following discussion will focus on the analysis of backcross data and dominant markers. Using dominant markers, backcross loci are defined as loci that segregate in a 1:1 ratio in the offspring and are derived from parents that are +-/-- (Figure 1.3), where + is the presence of a band and - is the absence of a band. Backcross loci for dominant markers are equally informative as codominant markers because banded offspring are heterozygous for the "+" allele, while individuals that are missing a band are homozygous for the "-" allele. 27 The first step in linkage analysis is to distinguish between backcross and intercross markers, and to test individual loci for segregation distortion. This is known as single-locus analysis as it explores each locus individually. The segregation ratio (backcross or intercross locus) can be determined using parental genotypes. For backcross markers, one parent is + while the other is - and the markers in the offspring segregate in a 1:1 ratio. For intercross markers, both parents have + genotypes and the markers in the progeny segregate in a 3:1 ratio. Once backcross loci have been distinguished from intercross loci, segregation distortion can be tested. Segregation distortion is the deviation from the expected genotypic ratio and is tested using a chi squared test with one degree of freedom or a log likelihood test statistic. Data from the two-locus analysis is the basis for the remaining steps of the linkage analysis. For this step, linkage statistics (LOD scores and recombination rates) are calculated for all combinations of loci. An estimate of recombination rate is necessary before LOD scores can be determined. Recombination rate (0) is the probability that a gamete caries a non-parental or recombinant gene arrangement. For backcross loci, recombination rate can be determined by counting the number of recombinant genotypes for a pair of loci and dividing by the total number of samples. LOD scores are then calculated based on the estimated recombination rate as: LOD = (non-recombinants)Logio[(l-0)/O.5] + (recombinants)Logio[0/O.5] [1.7] Joinmap (Van Ooijen and Voorrips 2001), a software package used for linkage mapping, calculates LOD scores as: 28 LOD = TL o log (o/e) [1.8] where "o" and "e" are the observed and expected number of individuals in a cell for a two-way contingency table, respectively. Under the null hypothesis, this statistic has degrees of freedom equal to the number of rows minus one, multiplied by the number of columns minus one. Van Ooijen and Voorrips (2001) suggest that this statistic is not affected by segregation distortion and is less prone to spurious linkage. The linkage grouping step uses the LOD scores and recombination rates from the two-locus analysis to place markers into linkage groups. The statistical definition of a linkage group is a group of loci inherited together according to certain statistical criteria. The criteria are threshold LOD (minimum) and recombination rates (maximum) that are created to define linkage between loci. A minimum LOD threshold of three has been commonly used in human genome mapping to define linkage between markers. An LOD of three means that linkage at the estimated 9 value is 1000 times more likely than 0 = 0.5. Some iteration is, however, required when trying to define acceptable grouping criteria, and using the criteria to determine grouping pattern. The following factors can be used as a general guideline when assessing linkage groupings: 1. There is high confidence if the grouping pattern does not change over a wide range of criteria; 2. If a grouping pattern meets biological expectations (i.e. number of haploid chromosomes), then there is high confidence; 29 3. Unexpectedly large linkage groups suggests false linkage and the criteria need to be more restrictive (i.e. higher LOD); 4. A large number of unlinked markers is a sign of poor quality data, small population size or a small number of genetic markers. Joinmap (Van Ooijen and Voorrips 2001) groups markers based on LOD scores. A threshold LOD value is defined, above which linkage is considered significant (Stam 1993). Once the markers have been grouped, they must be ordered. Liu (1998) presents a variety of procedures for ordering markers after grouping. Joinmap (Van Ooijen and Voorrips 2001) uses a procedure outlined by Stam (1993) for ordering markers on linkage groups. The procedure first chooses the most informative pair of loci and calculates the map distance. The next marker for ordering is chosen from the remaining unmapped markers based on LOD scores with the mapped loci. The new marker is placed on the map in the most suitable position without changing the current order of markers. Finally, the order of markers is "reshuffled" and the best fitting order is determined. This process is repeated until all of the markers are placed on the map. The optimal position is measured by calculating goodness-of-fit statistics. When a marker is added to the map, the goodness-of-fit is re-calculated. If the change is higher than the defined "jump" (allowable change) or the map distance is negative at the best position, the marker is rejected and tried in a second round of additions. The "reshuffle" procedure is referred to as a "ripple" in Joinmap. Joinmap takes groups of three adjacent markers and calculates the goodness-of-fit and map statistics for all permutations to derive the best order. This is continued along the linkage group until all groups of three markers have been assessed. 30 The final step in the generation of a linkage map is a multi-point analysis. In this step, map distances are calculated between loci. In Joinmap (Van Ooijen and Voorrips 2001), this step occurs simultaneously with marker ordering. Map distances between ' markers on a linkage map are determined from recombination rates. However, the two-locus recombination rates are not additive, even for loci linearly arranged on a linkage map. As a result, information regarding multiple loci on a linkage group must be considered and integrated to determine map distances. Map distances are generally related to crossover events and the rate of recombination, and less to actual physical distances between genes measured in nucleotide bases. For humans, one cM (centi-Morgan) is equal to approximately 1 lOOkb. The relationship between physical and map distances are species dependent. The calculation of map distance (my) is a function of recombination rate (ru): There are several mapping functions that can be used to estimate map distances which are listed by Liu (1998). Joinmap (Van Ooijen and Voorrips 2001) provides the option of using Haldane's or Kosambi's map function. Haldane's map function assumes that crossovers occur randomly along the length of the chromosome and is defined by: my =F(ry) [1.9] m = F(r)= -Vi log (l-2r) for 0<r<0.5 [1.10] 00 for r> 0.5 31 where m is map distance and r is the recombination rate. However, we know that crossovers do not occur randomly along the length of a chromosome. In contrast, Kosambi's map function incorporates crossover interference and is defined by: m = F(r)= ] /4 log [(l+2r)/(l-2r)] for0<r<0.5 [1.11] = co for r> 0.5 where m is map distance and r is the recombination rate. Joinmap (Van Ooijen and Voorrips 2001) uses a weighted least squares approach with the square of LOD scores as weights to calculate the final map distances. This approach gives greater weight to stronger data, as outlined in Stam (1993). Using this approach, Joinmap is able to combine linkage maps from multiple populations with common markers. Hu et al. (2004) have proposed a unique way of integrating linkage maps from different experiments using a joint likelihood function. The method outlined by Stam (1993) requires linkage maps to be created for each experiment before integration. The joint likelihood function proposed by Hu et al. (2004) integrates marker data at the two-locus stage. The procedure calculates "average" LOD and recombination rates for common markers across experiments or populations. Hu et al. (2004) compared the joint likelihood function to the procedure outlined by Stam (1993) and found that regardless of marker type (codominant, dominant or a combination), the joint likelihood function gave superior results (smaller standard deviations); the best results were obtained for dominant markers. Hu et al. (2004) report that sample sizes are greatly 32 reduced using the joint likelihood function for which 50 individuals from 10 crosses is sufficient, compared to > 100 individuals for the procedure outlined by Stam (1993). Employing APLPs to genotype 100 individuals from each population (except for one in which 257 individuals were genotyped), they report an integrated linkage map for seven populations of Linanthus jepsonii and Linanthus bicolor from three families (Hu et al. 2004). This procedure resulted in 162 scored markers with linkage among 79 (49.8% of the total number of markers scored). They found nine linkage groups with greater than four markers, three doubles and one triplet with a total map distance of 864.7cM and an average distance between markers of 13.3cM. This procedure is a new and unique way to integrate dominant, molecular marker information across populations. Linkage Maps for Trees Linkage mapping has been widely used in forest trees. Several linkage maps have been generated for both conifers and hardwoods, and a variety of strategies have been employed. Linkage maps have been developed for individuals using haploid megagametophyte tissue and AFLP markers for Pinyon pine {Pinus edulis) (Travis et al. 1998) and loblolly pine {Pinus taeda L.) (Remington et al. 1999). Remington et al. (1999) developed a saturated linkage map (508 markers) with 12 linkage groups (equal to the haploid number of chromosomes in loblolly pine). The linkage map presented by Travis et al. (1998) contained 338 markers mapped to 25 linkage groups using 40 megagametophytes. The difference in linkage group number is probably related to the number of polymorphic markers and sample size. 33 A variety of crosses have been used to generate linkage maps for species of Populus. Cervera et al. (2001) employed two separate crosses (a, Populus deltiodes female with Populus nigra and Populus tirchocarpa pollen donors) to generate two sets of F2 hybrids for linkage analysis. The analysis was conducted using AFLPs to saturate and microsatellite markers to align the individual Populus deltoides maps. The result was a linkage map for each species. Similarly, a female hybrid clone (Populus tomentosa x Populus bolleana) crossed with a male Populus tomentosa was employed to create two linkage maps with AFLP markers (Zhang et al. 2004). One map reflected the genome of Populus tomentosa, while the other modeled the genome of the female hybrid. In this way, linkage maps for combinations of species can be created to study the effects of hybridization. Many of the pedigrees available to mapping studies in trees are full-sib families. Traditional linkage mapping requires the construction of maps for both the maternal and paternal parents. Parental maps have been developed (Yin et al. 2003) for Scots pine and European beech (Fagus sylvatica L.) (Scalfi et al. 2004). Parental maps can be combined using intercross markers common to each parent. Consensus linkage maps have been developed for Populus deltoides (Wu et al. 2000) and maritime pine (Pinuspinaster Ait.) (Chagne et al. 2002). In a similar manner, 108 orthologous markers common to at least two mapping populations were used to align linkage maps from two outbred, three-generation pedigrees of loblolly pine (Sewell et al. 1999). It has been reported that the average genome length was higher for males (1983.7cM) than females (1339.5cM), and that the meiotic rate of recombination was higher for males compared to females. The final map consisted of 357 markers (278 RFLPs, 67 RAPDs and 12 isozymes) and 12 34 linkage groups accounting for 1175cM. Another two linkage groups were merged in one of the pedigrees, while six small single-population linkage groups remained unmerged. Douglas-fir Linkage Maps Linkage mapping in Douglas-fir has been conducted using RAPDs (Krutovskii et al. 1998) and a combination of RAPD and RFLP markers (Jermstad et al. 1998) from pedigrees developed for QTL analyses. Krutovskii et al. (1998) created linkage maps for two female hybrid individuals (var menziesii x var glauca), where each hybrid tree was crossed with the other plus unrelated coastal and interior males. The population structure was designed to study the influence of the paternal genetic background on QTLs donated from the maternal parent. Complementary DNA (cDNA) was isolated from megagametophyte tissue and was used to create one map (clone 3B) that had 18 linkage groups with 244 markers that spanned 2279cM with an average distance between markers of 10. lcM, and a second map (clone 10D) that had 20 linkage groups with 210 markers over 2468cM with an average of 12.9cM between markers. The authors explain clustering within their maps as locations of low recombination, such as centromeric regions. Jermstad et al. (1998) developed a linkage map from a three generation outbred pedigree that was designed to segregate for the date of bud flush. DNA was collected from four grandparents, two parents and 240 progeny (192 outplanted and 48 greenhouse plants). The resulting map consisted of 17 linkage groups with 141 markers spanning 1062cM with an average distance between markers of 7.5cM. A remaining 32 markers 35 formed two-marker linkage groups that covered 150cM. In an attempt to bridge the gap between unlinked markers or small, two-marker linkage groups, the authors added 50 supplementary RFLP markers, and report that the additions were largely unsuccessful in combining groups. QTL Analyses QTL Analysis Theory The goal of quantitative trait locus analyses is to identify the location of genes that contribute to the phenotypic distribution of a quantitative trait in a population. Quantitative traits are thought to be influenced by a number of genes that contribute to the phenotypic value within individuals. At the population level, the frequency distribution of trait values resembles a normal distribution, and genes that influence quantitative traits are called quantitative trait loci (QTLs). QTL analyses attempt to locate the position of trait loci relative to polymorphic molecular markers. As a result, the analysis gives no information about the actual function of a gene or co-located genes, but can reveal the location and strength of their effects. There are two categories of QTL analyses: single-locus analysis and interval mapping. Single-locus analyses test each locus independently for segregation with a QTL, while interval mapping tests the distance spanning two markers for the location of a QTL. Interval mapping is superior to single-locus mapping which is limited by low statistical power and confounding estimates of QTL location and effects (Beavis 1998; 36 Liu 1998). The advantage of single-locus analysis is its computational simplicity. Liu (1998) describes four types of single-locus analyses: simple t-test, analysis of variance (ANOVA), linear regression and a likelihood approach. These approaches were used in early studies of associations between trait values and marker segregation patterns, and follow linear models such as: yj = u + flmarkerj) + 8j [112] where yj is the trait value of the ]th individual, u is the population mean, f(markerj) is a function of the marker genotype and 8j is the residual associated with the ]th individual (Liu 1998). For dominant markers and backcross progeny, the mean trait values of when the allele is present (+) or absent (-) can be tested under the null hypothesis that the difference is equal to zero. A significant difference between genotypes suggests linkage of a QTL to the locus in question. Using these methods, QTL regions can be identified and complex experimental designs can easily be incorporated to identify other effects related to site and family differences (Liu 1998). Interval mapping is more powerful than single locus analysis and can identify more exact QTL locations between segregating markers. However, these approaches are often statistically and computationally intensive. Many interval mapping approaches have been developed in recent years; most fall into three categories: likelihood, regression or a combination of both approaches. These analyses use the recombination rates for two markers and between each marker and the proposed QTL, to estimate its position (Liu 1998). 37 For human populations in which single, large families are difficult to obtain, sib-pair regressions have been used for QTL studies. For this analysis, a regression is performed on the probability of identity by descent and the squared difference of trait values between two sibs at a given locus. Theoretically, if two sibs are genetically similar at a locus, they will have common trait values if a QTL is located nearby (Haseman and Elston 1972). Visscher and Hopper (2001) investigated the power of regression and maximum likelihood methods to map QTLs in sib pairs and found that using the regression approach reported by Haseman and Elston (1972) was as powerful as maximum likelihood analyses. They also reported that using a regression integrating the squared difference (D2; = [Yu - Y,2]2) and the mean corrected squared sum (S2i = {[Yn -Mil] + [Yi2 - Ui2]}2) produced the best results. The computer program QTL Express (Seaton et al. 2002), for example, employs this approach to analyze sib-pair data. The primary constraints for QTL studies are experimental design and genetic complexity (Beavis 1998). The traditional inferences about significance of test statistics cannot be applied due to the large number of non-independent test statistics calculated in each QTL experiment. These statistical tests are not independent because markers are genetically linked. Improvements have been made by using permutations to estimate appropriate significance thresholds (Beavis 1998). Furthermore, interval mapping is based on the null hypothesis of no QTL, however, unless multiple QTL are added to the model and effects estimated simultaneously, estimates of genetic effects and inferences about significance will be biased (Beavis 1998). The result is little power, precision and accuracy for most studies: given small sample sizes, there is a tendency to underestimate QTL number and overestimate effect size (Beavis 1998; Otto and Jones 2000). This has 38 been described as the "Beavis Effect" (Otto and Jones 2000), which adds uncertainty about QTL results. Otto and Jones (2000) acknowledge the limitations of current QTL studies to detect the actual number of QTLs, and have developed a QTL-based estimator to calculate the number of QTL of small effect that go undetected assuming an exponential distribution of effect sizes. QTL Analyses in Trees The identification of QTL is a powerful tool for investigating quantitative traits and has become popular among forest geneticists. Commercially important traits, as well as several adaptive traits, have been used to identify approximate genome locations for genes affecting quantitative traits. For example, in most studies, only a small number of growth trait QTLs are detected suggesting that these traits are controlled by few genes with large effect, or many genes of small effect that are below the threshold of detection. Yoshimaru et al. (1998) studied growth, flower bearing and rooting ability in Japanese -cedar (Cryptomeria japonica D. Don) using a three generation pedigree with progeny from a selfed Fi individual, created from the cross of two local cultivars. The authors used both interval mapping and single-locus QTL analysis, and identified two QTL regions for height growth and one for diameter. Height and diameter QTLs at age four and five were found to co-locate on a linkage group, while a second QTL region for height growth at age 14 mapped to the end of a second linkage group. The QTLs for height explained 18.6% - 48.8% of the phenotypic variation while the diameter QTLs explained 23.6% - 45.4%. Yoshimaru et al. (1998) suggest that the cluster of height and 39 diameter QTLs with a QTL for female fertility is evidence for pleiotropic effects. In a similar study in Eucalyptus grandis, two unique QTLs were detected using both interval and single locus analyses for circumference at breast height which explained 3.9% -6.6% of the variation (Grattapaglia et al. 1996). A comprehensive study of maritime pine (Pinuspinaster Aiton.) conducted by Markussen et al. (2003) identified seven QTLs for diameter and three for height. The QTLs explained 4.93% - 14.51% of the variation for each trait. A study of hybrid larch (Larix deciduas * Larix kaempgeri) was used to examine the underlying genetic effects of several wood density related traits (Arcade et al. 2002). Significant QTLs were identified for ring width and ring area explaining less than 8% of the total variation. Two studies were conducted on a two generation cross of Populus trichocarpa x Populus deltoides for growth, form and phenology traits for the first three years of growth (Bradshaw and Stettler 1995; Wu 1998): both studies identified QTLs for total height growth on the same linkage group. A separate QTL was detected for year three total height growth associated with the same marker as a QTL for height increment. Two linkage groups had significant QTLs for total basal area: year 2 and year 3 basal area were located on the same linkage group, while year 1 basal area produced a unique QTL. No QTLs were detected for volume. The year 2 height QTL explained 25.9% of the phenotypic variation while the basal area QTLs explained 24.4% - 27.2%. Bradshaw and Stettler (1995) identified two QTL clusters for several growth, branch and leaf area traits, which they suggest are a result of pleiotropic effects. Additionally, several growth form characteristics have been studied in Populus (Bradshaw and Stettler 1995; Wu 1998) and European beech (Scalfi et al. 2004) with many unique QTLs detected. 40 Wood property traits have received a great deal of attention in recent years and have been identified as important traits for tree breeding. Wood quality traits seem to be under the control of several genes with minor effect. Several QTL analyses have provided information regarding the underlying genetic control of these traits. Grattapaglia et al. (1996) identified five unique QTLs for specific gravity explaining 3.4% - 10.2% of the phenotypic variation. Markussen et al. (2003) used 80 full-sibs from a cross of a clone chosen for growth and one for straightness in maritime pine and identified eight QTLs for mean wood density, three for minimum wood density and six for maximum wood density. The QTLs for wood density traits explained between 5.09% and 14.51% of the phenotypic variation. A single locus analysis in hybrid larch detected multiple QTL for earlywood density, latewood density and average density from multiple years co-locating on several linkage groups explaining 3.4% - 6.2% of the variation (Arcade et al. 2002). Sewell et al. (2000) located 93 QTLs for specific gravity, volume percentage of latewood and microfibril angle in loblolly pine, and several of these QTLs co-located to the same position and were thought to be independent verification of the same QTL. The QTLs explained less than 20% of the phenotypic variation, and reveal temporal patterns of QTL activity and independent QTL for earlywood and latewood traits. Four to eight QTLs were detected for earlywood specific gravity, three to six for latewood specific gravity, two for a combination of QTLs affecting each trait and two to five affecting earlywood or latewood microfibril angle (Sewell et al. 2000). Two significant QTL regions for microfibril angle were detected in two full-sib pedigrees of Eucalyptus globulus (Thamarus et al. 2004). 41 Wood chemical composition is an important parameter for the pulp and paper sector. Wood chemistry traits seem to be controlled by a moderate number of genes that affect more than one trait. Two studies have identified QTLs for wood chemical composition in conifers. Markussen et al. (2002) detected QTLs for alpha-cellulose content (5), lignin content (6), pulp yield (7) and extractive content (2). Sewell et al. (2002) identified 29 QTLs associated with cellulose, lignin and hemicellulose content in loblolly pine. Several of these chemical QTLs co-located. As a result, most of the independent QTLs were combined into unique QTLs affecting wood chemistry reducing the number of independent, unique QTLs to eight. The pursuit of knowledge about the underlying genetic control of adaptive traits is even more important given the effects of climate change. Studies have revealed that adaptive traits, such as bud flush, bud set and frost tolerance, are under strong genetic control (Bradshaw and Stettler 1995; Jermstad et al. 2003; Wheeler et al. 2005). Frewen et al. (2000) studied QTL for bud flush and bud set in Populus and identified three unique QTLs affecting bud set that explained 6% - 12.2% of the phenotypic variation. Two of the three QTLs were found to reduce the days to bud set. The authors also detected six unique QTLs for bud flush explaining 5.9% - 16.8% of the phenotypic variation, of which, four delayed the time to bud flush. All bud set QTLs were mapped to linkage groups with bud flush QTLs. Bradshaw and Stettler (1995) detected five QTLs affecting bud flush in Populus. All five QTLs occurred on different linkage groups and accounted for 28.7% - 51.5% of the phenotypic variation. Frost tolerance was studied for Eucalyptus nitens using a single locus analysis (Byrne et al. 1997). Only two independent QTLs were detected for frost tolerance, which are predicted to map 40cM apart on a 42 single linkage group. The two QTLs explained 7.7% and 10.2% of the phenotypic variation. Finally, an allele responsible for seedling death was identified in radiata pine using segregation distortion of RAPD and microsatellite markers (Kuang et al. 1998). Kuang et al. (1998) selfed a plus tree and employed megagametophyte tissue to study the segregation distortion in surviving seedlings. The authors suggested that severe segregation distortion was evidence that a lethal gene was linked to the affected markers. They identified one marker locus that segregated with a lethal allele, which they referred to as SDPr (seedling death in Pinus radiata). The authors concluded that the viability of homozygous recessive SDPr individuals is near zero, the lethal allele is completely recessive, SDPr accounts for all death of selfed progeny in the first month after germination, and that the allele is essential for early seedling development, but not seed development or germination (Kuang et al. 1998). QTL Analyses in Douglas-fir QTL analyses of adaptive traits have been conducted for Douglas-fir (Jermstad et al. 2003; Jermstad et al. 2001a; Jermstad et al. 2001b; Wheeler et al. 2005). Thirty-three unique QTLs were found to affect the timing of spring bud flush. The authors found that maternal by paternal effects differed little from zero, suggesting that the genes are mostly additive in effect and that most genes controlling the onset of spring bud flush are expressed annually. They also found that the expression of QTLs was affected by environment, and report a different combination of QTLs detected on each site. Eleven and 15 unique QTLs affecting fall and spring cold-hardiness were identified, 43 respectively. QTLs for spring cold-hardiness explained a higher proportion of variation compared to fall cold-hardiness (5.8% compared to 4.1%, respectively). Like the bud flush QTLs, many of the cold-hardiness QTLs appear to be additive in effect. The interaction effects of growth initiation and growth cessation with various environmental signals was studied in the same Douglas-fir pedigree (Jermstad et al. 2003). The authors detected < 90 QTLs controlling growth initiation and growth cessation, however, like the previous studies they report that, a modest number of QTLs (5-10) with small effect (5% -10%) control adaptive traits in Douglas-fir. They also report QTL by treatment interactions for winter chill and flushing temperature. The QTL interacting with winter chill co-locates with a QTL interacting with site in field tests. Furthermore, several interacting QTLs with day length and moisture stress were identified. Jermstad et al. (2003) used replicated field trials to test QTL by site interactions and found that, unlike the previous studies, most QTLs were detected with confidence in both test environments. Cold-hardiness QTLs were verified (Wheeler et al. 2005) using two cohorts from the same three generation outbred pedigree. In total, six QTLs were identified for spring cold hardiness in cohort 1 and eight in cohort 2. Four of the eight QTLs identified in cohort 2 co-localized with QTLs from cohort 1 (Wheeler et al. 2005). This provides verification for the co-localized QTLs and implies that as many as 10 QTLs may act to control spring cold hardiness. Although extensive research has been conducted on adaptive traits in coastal Douglas-fir, at present there exists a significant gap in QTLs for wood quality and growth traits. 44 Applications of Linkage Mapping and Q T L Analyses Marker Aided Selection Given the advances in molecular marker technology, and the expansion of linkage mapping and QTL analyses to forest trees, the utilization of marker aided selection (MAS) in tree improvement is becoming a powerful option for tree breeders (Chagne et al. 2003). Although MAS has been attempted in the agriculture community, it has been met with cautious scepticism by tree breeders (Johnson et al. 2000; Strauss et al. 1992). In theory, individuals with markers that are strongly associated with trait improving QTLs can be selected as parents for the next generation (Lande and Thompson 1990). The feasibility of MAS to supplement traditional breeding efforts for Douglas-fir has been studied through computer simulation by Johnson et al. (2000), and the authors concluded that justification for MAS in Douglas-fir is difficult because: 1) the montane habitat of Douglas-fir results in small breeding zones limiting the area over which gains can be spread. 2) rotation ages of 40 - 60 years are common so that financial gains for MAS are discounted for long periods compared to selection costs which occur early and are discounted little. 3) The large amount of genetic diversity suggests that QTL alleles may differ substantially between families. 4) There is no evidence that traits of economic importance are controlled by QTL of large effect suggesting the need for large sample sizes and several markers. 45 5) The amount of non-additive genetic variance will be limited due to the lack of use of clonal propagation in commercial programs. QTLs are affected by epistasis, dominance and environment (Johnson et al. 2000). Environment by QTL interactions can be addressed by replicating QTL experiments in a variety of environments. However, epistatic and dominance interactions among QTLs and other loci will diminish in subsequent generations requiring new maps to be generated in each generation. There is also evidence that QTL are dependent on genetic background and differ between families (Neale et al. 2002). As a result, family specific maps are necessary. Furthermore, in order to capture QTL of small effect, large sample sizes are necessary, adding to the cost associated with these approaches (Lande and Thompson 1990). It has been suggested that MAS will be most effective for traits with low h2 where a large fraction of the additive genetic variation is associated with a marker (Johnson et al. 2000; Lande and Thompson 1990; Strauss et al. 1992; Wu 2002). MAS may be most effective in identifying QTL associated with severe threats to forest health, in making use of extremely high-valued families in situations where a limited number of clones or genotypes will be identified for use, and in improving gains within and among families in conjunction with phenotypic selection (Strauss et al. 1992). In order to be effective, markers associated with a particular QTL must explain the majority of additive variance within families and traits with low h2. 46 QTL Effects and the QTL Concept Strauss et al. (1992) suggest that a potential application of QTL identification and mapping is to explore the current QTL model. The study of seasonal expression, temporal stability and gene action of QTLs will help re-work the QTL model and aid in verifying previous hypotheses (Strauss et al. 1992; van Buijtenen 2001). Different environmental conditions may trigger a variety of gene combinations to produce altered phenotypes. For example, Jermstad et al. (2003, 2001a) identified QTL by environment interactions for adaptive traits in Douglas-fir. Knowledge of the relationship of QTLs among families, populations and species is of great importance to tree breeders. Neale et al. (2002) attempted to verify QTLs and explore family effects on QTL expression for wood property traits in loblolly pine. The authors found that 43% of QTLs detected in related families and only 16% of QTLs detected in unrelated families were common to multiple families. These results suggest that the interaction of QTL with genetic background are significant and cannot be ignored. The stability of QTL through time has been detected for wood specific gravity traits in loblolly pine (Brown et al. 2003; Sewell et al. 2000). Brown et al. (2003) report that 62% of QTLs for individual rings were detected for more than one growing season. They suggest that QTLs detected in multiple years reflect the determinants of a particular trait, while those identified for single rings may represent QTL activated in response to biotic or abiotic variation. Thus, QTL temporal variation can be explored quite easily by mapping traits in consecutive years. 47 Comparative Genome Mapping The application of linkage mapping and QTL analyses to comparative genome mapping provides a mechanism to explore the nature of genome organization and composition between species. To compare linkage maps of two species, common markers must be identified on each species map and subsequently aligned. By aligning linkage groups, identified QTLs for common traits can be compared. The comparative conifer genome project (CCGP) have developed a series of ESTPs (expressed sequence tag polymorphisms) for use as anchor loci for comparing linkage maps (Chagne et al. 2003). Chagne et al. (2003) used ESTPs common to linkage maps for loblolly and maritime pine to compare synteny, colinearity and QTL positions. They found that both synteny (the conservation of gene content) and colinearity (the conservation of gene order) were conserved. Some slight differences in gene order were attributed to mapping errors rather than chromosome rearrangements. Furthermore, the authors report the synteny and colinearity of two QTLs controlling wood density and cell wall components. Krutovsky et al. (2004) used orthologous RFLP and ESTP markers to compare linkage maps for Douglas-fir and loblolly pine, and reported ten homologous linkage groups. For most groups, the orthologous markers were co-linear. In two cases, two Douglas-fir linkage groups aligned with a single loblolly pine linkage group. As a result, they suggest that the th 13 chromosome in Douglas-fir may have originated from the break-up of a single pine chromosome in a common ancestor. The future of comparative genome mapping may 48 help unlock the secret behind the 13th Douglas-fir chromosome, and may aid in the identification of significant QTL for multiple species. Candidate Gene Mapping QTLs are statistical entities that represent a region in the genome of a possible gene(s) that affects a particular trait. QTL analyses say nothing of the actual type or identity of the gene(s) in question, while association genetics attempts to identify genes that affect quantitative traits. Candidate gene mapping places genes that have the potential to affect a particular trait onto a QTL map. Genes that co-locate with QTL for an associated trait are identified for further analysis by association genetics. Candidate genes can be identified by known function, genetic expression studies or by the co-location of unknown genes or ESTs with known QTL (van Buijtenen 2001). Association genetics then attempts to identify molecular differences within candidate genes that contribute to phenotypic differences (Wheeler et al. 2005). Positional cloning of candidate genes has become an attractive alternative to map-based cloning in species with complex genomes (Brown et al. 2003). Candidate gene mapping has begun in loblolly pine for genes relating to wood property traits (Brown et al. 2003) and in Douglas-fir for adaptive traits (Wheeler et al. 2005). Brown et al. (2003) mapped 18 candidate genes associated with the production of cell wall components. Several of these genes mapped near QTL identified for earlywood wood specific gravity, latewood specific gravity and latewood percent. Wheeler et al. (2005) mapped 29 candidate genes for cold tolerance in Douglas-fir. Seventeen of these 49 candidate genes mapped within the 95% confidence intervals of six QTL regions. Candidate gene mapping is an ideal way to prioritize genes for association studies, but by no means implies a definitive association (Wheeler et al. 2005). Thesis Statement This thesis contains a comprehensive analysis of Douglas-fir quantitative genetics for growth and wood quality traits. Traits associated with growth, wood density, fibre properties, and wood chemistry are described in terms of traditional quantitative genetics (heritabilities and correlations) and the identification and description of QTLs. The research presented here aims to: 1. Identify the degree of genetic control and correlations among traits using heritabilities and correlations; 2. Develop a comprehensive linkage map using AFLPs from eight full-sib families to be used as a base map for QTL analysis; 3. Determine the number of QTLs of major effect, and the amount of explained phenotypic variation, for growth and wood quality traits in coastal Douglas-fir. 50 References Anagnost, S.E., Mark, R.E., and Hanna, R.B. 2002. Variation of microfibril angle within individual tracheids. Wood and Fiber Science. 34 (2): 337-349. Arcade, A., Faivre-Rampant, P., Paques, L.E., and Prat, D. 2002. Localisation of genomic regions controlling microdensitometric parameters of wood characteristics in hybrid larches. Annals of Forest Science. 59 (5-6): 607-615. Aubry, C.A, Adams, W.T., and Fahey, T.D. 1998. Determination of relative economic weights for multitrait selection in coastal Douglas-fir. Canadian Journal of Forest Research. 28(8): 1164-1170. Barnett, J.R., and Bonham, V. A. 2004. Cellulose microfibril angle in the cell wall of wood fibres. Biological Reviews. 79: 461-472. Batchelor, W.J., Conn, A.B., and Parker, I.H. 2000. Comparison of techniques to measure the fibril angle. Appita Journal. 53 (6): 432-437. Baucher, M., Halpin, C , Petit-Conil, M., and Boerjan, W. 2003. Lignin: Genetic engineering and impact on pulping. Critical Reviews in Biochemistry and Molecular Biology. 38 (4): 305-350. Beavis, W.D. 1998. QTL analyses: power, precision and accuracy. In Molecular dissection of complex traits. CRC Press, Raleigh, North Carolina, pp. 145-162. Bradshaw, H.D., and Stettler, R.F. 1995. Molecular genetics of growth and development in Populus. IV. Mapping QTLs with large effects on growth, form, and phenology traits in a forest tree. Genetics. 139: 963-973. Brown, G.R., Bassoni, D.L., Gill, G.P., Fontana, J.R., Wheeler, N.C., Megraw, R.A., Davis, M.F., Sewell, M.M., Tuskan, G.A., andNeale, D.B. 2003. Identification of quantitative trait loci influencing wood property traits in loblolly pine (Pinus taeda L.). III. QTL verification and candidate gene mapping. Genetics. 164 (4): 1537-1546. 51 Byrne, M., Murrell, J.C., Owen, J.V., Williams, E.R., and Moran, G.F. 1997. Mapping of quantitative trait loci influencing frost tolerance in Eucalyptus nitens. Theoretical and Applied Genetics. 95 (5-6): 975-979. Cabrita, L.F., Aksoy, U., Hepaksoy, S., and Leitao, J.M. 2001. Suitability of isozyme, RAPD and AFLP markers to assess genetic differences and relatedness among fig (Ficus caricaL.) clones. Scientia Horticulture. 87 (4): 261-273. Cave, I.D. 1997a. Theory of X-ray measurement of microfibril angle in wood. Parti. The condition for reflection. Wood Science and Technology. 31: 143-152. Cave, I.D. 1997b. Theory of X-ray measurement of microfibril angle in wood. Part 2. The diffraction diagram. Wood Science and Technology. 31: 225-234. Cave, I.D., and Walker, J.C.F. 1994. Stiffness of wood in fast-grown plantation softwoods: the influence of microfibril angle. Forest Products Journal. 44 (5): 43-48. Cervera, M.T., Remington, D., Frigerio, J.M., Storme, V., Ivens, B., Boerjan, W., and Plomion, C. 2000. Improved AFLP analysis of tree species. Canadian Journal of Forest Research. 30(10): 1608-1616. Cervera, M.T., Storme, V., Ivens, B., Gusmao, J., Liu, B.H., Hostyn, V., Van Slycken, J., Van Montagu, M., and Boerjan, W. 2001. Dense genetic linkage maps of three Populus species (Populus deltoides, P. nigra and P. trichocarpa) based on AFLP and microsatellite markers. Genetics. 158 (2): 787-809. Chagne, D., Brown, G., Lalanne, C , Madur, D., Pot, D., Neale, D., and Plomion, C. 2003. Comparative genome and QTL mapping between maritime and loblolly pines. Molecular Breeding. 12 (3): 185-195. Chagne, D., Lalanne, C , Madur, D., Kumar, S., Frigerio, J.M., Krier, C , Decroocq, S., Savoure, A., Bou-Dagher-Kharrat, M., Bertocchi, E., Brach, J., and Plomion, C. 2002. A high density genetic map of maritime pine based on AFLPs. Annals of Forest Science. 59 (5-6): 627-636. Chantre, G., Rozenberg, P., Baonza, V., Macchioni, N., Le Turcq, A., Rueff, M., Petit-Conil, M., and Heois, B. 2002. Genetic selection within Douglas fir (Pseudotsuga menziesii) in Europe for papermaking uses. Annals of Forest Science. 59 (5-6): 583-593. 52 Cramer, S., Kretschmann, D., Lakes, R., and Schmidt, T. 2005. Earlywood and latewood elastic properties in loblolly pine. Holzforschung. 59: 531-538. Department of Scientific and Industrial Research. 1957. A handbook of softwoods. Her Majesty's Stationary Office, London, England. 73 pgs. Donaldson, L.A. 1992. Within- and between- tree variation in microfibril angle in Pinus radiata. New Zealand Journal of Forestry Science. 22 (1): 77-86. Donaldson, L . A , and Burdon, R.D. 1995. Clonal variation and repeatability of microfibril angle in Pinus radiata. New Zealand Journal of Forestry Science. 25 (2): 164-174. Erickson, H.D., and Arima, T. 1974. Douglas-fir wood quality studies. Part II: Effects of age and stimulated growth on fibril angle and chemical constituents. Wood Science and Technology. 8: 255-265. Ericsson, T., and Fries, A. 2004. Genetic analysis of fibre size in a full-sib Pinus sylvestris L. progeny test. Scandinavian Journal of Forest Research. 19 (1): 7-13. Evans, R., and Ilic, J. 2001. Rapid prediction of wood stiffness from microfibril angle and density. Forest Products Journal. 51 (3): 53-57. Falconer, D.S., and Mackay, F.C. 1996. Introduction to quantitative genetics. 4th ed. Longman Group Ltd, Burnt Mill, Harlow, Essex, England. 464 pgs. Forest Products Research Laboratory. 1964. Douglas-fir. Her Majesty's Stationery Office, London, England. 12 pgs. Frewen, B.E., Chen, T.H.H., Howe, G.T., Davis, J., Rohde, A., Boerjan, W., and Bradshaw Jr., H.D. 2000. Quantitative trait loci and candidate gene mapping of bud set and bud flush in Populus. Genetics. 154: 837-845. Goggans, J.F. 1964. Correlation and heritability of certain wood properties in loblolly pine (Pinus taeda L.). Tappi. 47 (6): 318-322. Grattapaglia, D., Bertolucci, F.L.G., Penchel, R., and Sederoff, R.R. 1996. Genetic mapping of quantitative trait loci controlling growth and wood quality traits in 53 Eucalyptus grandis using a maternal half-sib family and RAPD markers. Genetics. 144: 1205-1214. Hannrup, B., Cahalan, C , Chantre, G., Grabner, M., Karlsson, B., Le Bayon, I, Jones, G.L., Muller, U., Pereira, H., Rodrigues, J.C., Rosner, S., Rozenberg, P., Wilhelmsson, L., and Wimmer, R. 2004. Genetic parameters of growth and wood quality traits in Picea abies. Scandinavian Journal of Forest Research. 19: 14-29. Hannrup, B., Ekberg, I., and Persson, A. 2000. Genetic correlations among wood, growth capacity and stem traits in Pinus sylvestris. Scandinavian Journal of Forest Research. 15: 161-170. Hannrup, B., Wilhelmsson, L., and Danell, O. 1998. Time trends for genetic parameters of wood density and growth traits in Pinus sylvestris L. Silvae Genetica. 47 (4): 214-219. Haseman, J.K., and Elston, R.C. 1972. Investigation of linkage between a quantitative trait and a marker locus. Behavior Genetics. 2 (1): 3-9. Hodge, G.R., and Purnell, R.C. 1993. Genetic parameter estimates for wood density, transition age, and radial growth in Slash Pine. Canadian Journal of Forest Research. 23 (9): 1881-1891. Hu, X.-S., Goodwillie, C , and Ritland, K.M. 2004. Joining genetic linkage maps using a joint likelihood function. Theoretical and Applied Genetics. 109: 996-1004. Ivkovich, M. 1996. Genetic variation of wood properties in balsam poplar (Populus balsamiferaL). Silvae Genetica. 45 (2-3): 119-124. Ivkovich, M., Namkoong, G., and Koshy, M. 2002a. Genetic variation in wood properties of interior spruce. I. Growth, latewood percentage, and wood density. Canadian Journal of Forest Research. 32 (12): 2116-2127. Ivkovich, M., Namkoong, G., and Koshy, M. 2002b. Genetic variation in wood properties of interior spruce. II. Tracheid characteristics. Canadian Journal of Forest Research. 32: 2128-2139. Jermstad, K.D., Bassoni, D.L., Jech, K.S., Ritchie, G.A, Wheeler, N.C., and Neale, D.B. 2003. Mapping of quantitative trait loci controlling adaptive traits in coastal 54 Douglas fir. III. Quantitative trait loci-by-environment interactions. Genetics. 165 (3): 1489-1506. Jermstad, K.D., Bassoni, D.L., Jech, K.S., Wheeler, N.C., and Neale, D.B. 2001a. Mapping of quantitative trait loci controlling adaptive traits in coastal Douglas-fir. I. Timing of vegetative bud flush. Theoretical and Applied Genetics. 102 (8): 1142-1151. Jermstad, K.D., Bassoni, D.L., Wheeler, N.C., Anekonda, T.S., Aitken, S.N., Adams, W.T., and Neale, D.B. 2001b. Mapping of quantitative trait loci controlling adaptive traits in coastal Douglas-fir. II. Spring and fall cold-hardiness. Theoretical and Applied Genetics. 102: 1152-1158. Jermstad, K.D., Bassoni, D.L., Wheeler, N.C., and Neale, D.B. 1998. A sex-averaged genetic linkage map in coastal Douglas-fir (Pseudotsuga menziesii [Mirb.] Franco var 'menziesii') based on RPLP and RAPD markers. Theoretical and Applied Genetics. 97: 762-770. Johnson, G.R., Wheeler, N.C., and Strauss, S.S. 2000. Financial feasibility of marker-aided selection in Douglas-fir. Canadian Journal of Forest Research. 30 (12): 1942-1952. Jones, C.J., Edwards, K.J., Castaglione, S., Winfield, M.O., Sala, F., van de Wiel, C , Bredemeijer, G., Vosman, B., Matthes, M., Daly, A., Brettschneider, R., Bettini, P., Buiatti, M., Maestri, E., Malcevschi, A., Marmiroli, N., Aert, R., Volckaert, G., Rueda, J., Linacero, R., Vazquez, A , and Karp, A. 1997. Reproducibility testing of RAPD, AFLP and SSR markers in plants by a network of European laboratories. Molecular Breeding. 3 (5): 381-390. Krutovskii, K.V., Vollmer, S.S., Sorensen, F.C., Adams, W.T., Knapp, S.J., and Strauss, S.H. 1998. RAPD genome maps of Douglas-fir. Journal of Heredity. 89 (3): 197-205. Krutovsky, K.V., Troggio, M., Brown, G.R., Jermstad, K.D., and Neale, D.B. 2004. Comparative mapping in the Pinaceae. Genetics. 168 (1): 447-461. Kuang, H., Richardson, T.E., Carson, S.D., and Bongarten, B.C. 1998. An allele responsible for seedling death in Pinus radiata D. Don. Theoretical and Applied Genetics. 96:' 640-644. 55 Lande, R., and Thompson, R. 1990. Efficiency of marker-assisted selection in the improvement of quantitative traits. Genetics. 124: 743-756. Liu, B.-H. 1998. Statistical genomics: linkage, mapping and QTL analysis. CRC Press, New York, New York. 611 pgs. Loehle, C , and Namkoong, G. 1987. Constraints on tree breeding - growth tradeoffs, growth strategies, and defensive investments. Forest Science. 33 (4): 1089-1097. Long, J.M., Conn, A.B., Batchelor, W.J., and Evans, R. 2000. Comparison of methods to measure fibril angle in wood fibres. Appita Journal. 53 (3): 206-209. Loo, J.A., Tauer, C.G., and McNew, R.W. 1985. Genetic variation in the time of transition from juvenile to mature wood in loblolly pine (Pinus taeda L). Silvae Genetica. 34 (1): 14-19. Loo-Dinkins, J.A., and Gonzalez, J.S. 1991. Genetic control of wood density profile in young Douglas-Fir. Canadian Journal of Forest Research. 21 (6): 935-939. Louzada, J., and Fonseca, F.M.A. 2002. The heritability of wood density components in Pinus pinaster Ait. and the implications for tree breeding. Annals of Forest Science. 59 (8): 867-873. Makinen, H., Saranpaa, P., and Linder, S. 2002. Effect of growth rate on fibre characteristics in Norway spruce (Picea abies (L.) Karst). Holzforschung. 56 (5): 449-460. Markussen, T., Fladung, M., Achere, V., Favre, J.M., Faivre-Rampant, P., Aragones, A., Perez, D.D., Harvengt, L., Espinel, S., and Ritter, E. 2003. Identification of QTLs controlling growth, chemical and physical wood property traits in Pinus pinaster (Ait.). Silvae Genetica. 52 (1): 8-15. Megraw, R.A. 1986. Douglas-fir wood properties. In Douglas-fir: stand management for the future. College of Forest Resources, Seattle, Washington, pp. 81-97. Myszewski, J.H., Bridgwater, F.E., Lowe, W.J., Byram, T.D., and Megraw, R.A. 2004. Genetic variation in the microfibril angle of loblolly pine from two test sites. Southern Journal of Applied Forestry. 28 (4): 196-204. 56 Neale, D.B., Sewell, M.M., and Brown, G.R. 2002. Molecular dissection of the quantitative inheritance of wood property traits in loblolly pine. Annals of Forest Science. 59.(5-6): 595-605. Nicholls, J.W.P., Morris, J.D., and Pederick, L.A. 1980. Heritability estimates of density characteristics in juvenile Pinus radiata wood. Silvae Genetica. 29 (2): 54-61. Oswald, D.D., Hegyi, F., and Becker, A. 1986. The current status of coast Douglas-fir timber resources in Oregon, Washington, and British Columbia. In Douglas-fir: stand management for the future. College of Forest Resources, Seattle, Washington, pp. 26-32. Otto, S.P., and Jones, CD. 2000. Detecting the undetected: Estimating the total number of loci underlying a quantitative trait. Genetics. 156: 2093-2197. Paglia, G., and Morgante, M. 1998. PCR-based multiplex DNA fingerprinting techniques for the analysis of conifer genomes. Molecular Breeding. 4: 173-177. Pedersen, J.F., Vogel, K.P., and Funnell, D.L. 2005. Impact of reduced lignin on plant fitness. Crop Science. 45 (3): 812-819. Pot, D., Chantre, G., Rozenberg, P., Rodrigues, J.C, Jones, G.L., Pereira, H , Hannrup, B., Cahalan, C , and Plomion, C. 2002. Genetic control of pulp and timber properties in maritime pine (Pinuspinaster Ait.). Annals of Forest Science. 59: 563-575. Remington, D.L., Whetten, R.W., Liu, B.-H, and O'Malley, D M . 1999. Construction of an AFLP genetic map with nearly complete genome coverage in Pinus taeda. Theoretical and Applied Genetics. 98: 1279-1292. Rushforth, K.D. 1987. Conifers. Bath Press Ltd., London, England. 232 pgs. Sahlberg, U . , Salmen, L., and Oscarsson, A. 1997. The fibrillar orientation in the S2-layer of wood fibres as determined by x-ray diffraction analysis. Wood Science and Technology. 31 (2): 77-86. Saxena, I.M., and Brown, R.M. 2005. Cellulose biosynthesis: Current views and evolving concepts. Annals of Botany. 96 (1): 9-21. 57 Scalfi, M., Troggio, M., Piovani, P., Leonardi, S., Magnaschi, G., Vendramin, G.G., and Menozzi, P. 2004. A RAPD, AFLP and SSR linkage map, and QTL analysis in European beech (Fagus sylvatica L.). Theoretical and Applied Genetics. 108 (3): 433-441. Seaton, G., Haley, C.S., Knott, S.A., Kearsey, M., and Visscher, P.M. 2002. QTL express: mapping quantitative trait loci in simple and complex pedigrees. Bioinformatics. 18 (2): 339-340. Seth, R.S., and Kingsland, M.A. 1990. The reinforcing properties of softwood kraft pulps. Pulp and Paper Canada. 91 (7): 68-72, 74-75. Sewell, M.M., Bassoni, D.L., Megraw, R.A., Wheeler, N.C., and Neale, D.B. 2000. Identification of QTLs influencing wood property traits in loblolly pine (Pinus taeda L.). I. Physical wood properties. Theoretical and Applied Genetics. 101 (8): 1273-1281. Sewell, M.M., Davis, M.F., Tuskan, G.A, Wheeler, N.C., Elam, C.C., Bassoni, D.L., and Neale, D.B. 2002. Identification of QTLs influencing wood property traits in loblolly pine (Pinus taeda L.). n . Chemical wood properties. Theoretical and Applied Genetics. 104 (2-3): 214-222. Sewell, M.M., Sherman, B.K., and Neale, D.B. 1999. A consensus map for loblolly pine (Pinus taeda L.). I. Construction and integration of individual linkage maps from two outbred three-generation pedigrees. Genetics. 151 (1): 321-330. Sjostrom, E., and Raimo, A. 1999. Analytical methods in wood chemistry, pulping and papermaking. Springer, Berlin, New York. 316 pgs. St.Clair, J.B. 1994. Genetic variation in tree structure and its relation to size in Douglas-fir. I. Biomass partitioning, foliage efficiency, stem form, and wood density. Canadian Journal of Forest Research. 24 (6): 1226-1235. Stam, P. 1993. Construction of integrated genetic linkage maps by means of a new computer package: JoinMap. Plant Journal. 3 (5): 739-744. Strauss, S.H., Lande, R., and Namkoong, G. 1992. Limitations of molecular-marker-aided selection in forest tree breeding. Canadian Journal of Forest Research. 22 (7): 1050-1061. 58 Talbert, J.T., Jett, J.B., and Bryant, R.L. 1983. Inheritance of wood specific gravity in an unimproved loblolly pine population: 20 years of results. Silvae Genetica. 32 (1-2): 33-37. Thamarus, K., Groom, K., Bradley, A., Raymond, C.A., Schimleck, L.R., Williams, E.R., and Moran, G.F. 2004. Identification of quantitative trait loci for wood and fibre properties in two full-sib pedigrees of Eucalyptus globulus. Theoretical and Applied Genetics. 109 (4): 856-864. Travis, S.E., Ritland, K., Whitham, T.G., and Keim, P. 1998. A genetic linkage map of Pinyon pine (Pinus edulis) based on amplified fragment length polymorphisms. Theoretical and Applied Genetics. 97 (5-6): 871-880. van Buijtenen, J.P. 2001. Genomics and quantitative genetics. Canadian Journal of Forest Research. 31: 617-622. van Gelderen, D.M. 1989. Conifers. 2nd ed. Timber Press, Portland, Oregon. 369 pgs. Van Ooijen, J.W., and Voorrips, R.E. 2001. JoinMap 3.0, software for the calculation of genetic linkage maps. Plant Research International, Wageningen, the Netherlands. Vargas-Hernandez, J., and Adams, W.T. 1991. Genetic variation of wood density components in young coastal Douglas-fir - Implications for tree breeding. Canadian Journal of Forest Research. 21 (12): 1801-1807. Vargas-Hernandez, J., Adams, W.T., and Krahmer, R.L. 1994. Family variation in age trends of wood density traits in young coastal Douglas-fir. Wood and Fiber Science. 26 (2): 229-236. Via, B.K., Stine, M., Shupe, T.F., So, C.-L., and Groom, L. 2004. Genetic improvement of fiber length and coarseness based on paper product performance and material variability - a review. IAWA Journal. 25 (4): 401-414. Visscher, P.M., and Hopper, J.L. 2001. Power of regression and maximum likelihood methods to map QTL from sib-pair and DZ twin data. Annals of Human Genetics. 65: 583-601. 59 Vos, P., Hogers, R., Bleeker, M., Reijans, M., Vandelee, T., Homes, M., Frijters, A., Pot, J., Peleman, J., Kuiper, M., and Zabeau, M. 1995. AFLP - a new technique for DNA-fmgerprinting. Nucleic Acids Research. 23 (21): 4407-4414. Watson, E., and Luckman, B.H. 2002. The dendroclimatic signal in Douglas-fir and ponderosa pine tree-ring chronologies from the southern Canadian Cordillera. Canadian Journal of Forest Research. 32 (10): 1858-1874. Wheeler, N.C., Jermstad, K.D., Krutovsky, K., Aitken, S.N., Howe, G.T., Krakowski, J., and Neale, D.B. 2005. Mapping of quantitative trait loci controlling adaptive traits in coastal Douglas-fir. IV. Cold-hardiness QTL verification and candidate gene mapping. Molecular Breeding. 15 (2): 145-156. White, T.L. 1987. A conceptual framework for tree improvement programs. New Forests. 4: 325-342. Wimmer, R., Downes, G.M., and Evans, R. 2002. Temporal variation of microfibril angle in Eucalyptus nitens grown in different irrigation regimes. Tree Physiology. 22: 449-457. Woods, J., Pomeroy, N., and Bird, K. 1996. Coastal Douglas-fir. In Tree improvement council of British Columbia progress report to 1996. Ministry of Forests, Tree Improvement Program, Saanichton, British Columbia, pp. 10. Wu, H.X. 2002. Study of early selection in tree breeding. 4. Efficiency of marker-aided early selection (MAES). Silvae Genetica. 51 (5-6): 261-269. Wu, R.L. 1998. Genetic mapping of QTLs affecting tree growth and architecture in Populus: implications for ideotype breeding. Theoretical and Applied Genetics. 96: 447-457. Wu, R.L., Han, Y.F., Hu, J.J., Fang, J.J., Li, L., Li, M.L., and Zeng, Z.B. 2000. An integrated genetic map of Populus deltoides based on amplified fragment length polymorphisms. Theoretical and Applied Genetics. 100 (8): 1249-1256. Yanchuk, A.D., and Kiss, G.K. 1993. Genetic variation in growth and wood specific gravity and its utility in the improvement of interior spruce in British Columbia. Silvae Genetica. 42 (2-3): 141-148. 60 Yeh, F.C., and Heaman, J.C. 1987. Estimating genetic parameters of height growth in 7-year old coastal Douglas-fir from disconnected diallels. Forest Science. 33 (4): 946-957. Yin, T.M., Wang, X.R., Andersson, B., and Lerceteau-Kohler, E. 2003. Nearly complete genetic maps of Pinus sylvestris L. (Scots pine) constructed by AFLP marker analysis in a full-sib family. Theoretical and Applied Genetics. 106 (6): 1075-1083. Yoshimaru, H., Ohba, K., Tsurumi, K., Tomaru, N., Murai, M., Mukai, Y., Suyama, Y., Tsumura, Y., Kawahara, T., and Sakamaki, Y. 1998. Detection of quantitative trait loci for juvenile growth, flower bearing and rooting ability based on a linkage map of sugi (Cryptomeria japonica D. Don). Theoretical and Applied Genetics. 97: 45-50. Zamudio, F., Baettyg, R., Vergara, A., Guerra, F., and Rozenberg, P. 2002. Genetic trends in wood density and radial growth with cambial age in a radiata pine progeny test. Annals of Forest Science. 59: 541-549. Zhang, D., Zhang, Z., Yang, K., and Li, B. 2004. Genetic mapping in (Populus tomentosa x Populus bolleana) and P. tomentosa Carr. using AFLP markers. Theoretical and Applied Genetics. 108 (4): 657-662. Zobel, B., and Talbert, J. 1984. Applied forest tree improvement. Waveland Press, Inc, Prospect Heights, Illinois. 252-264 pgs. 61 Chapter 2 Broad-sense heritability, phenotypic and genetic correlations of wood quality traits in Coastal Douglas-fir Introduction Coastal Douglas-fir (Pseudotsuga menziesii (Mirb.) Franco var menziesii) is an important economic tree species along the west coast of British Columbia and the United States, which has also been cultivated in New Zealand and Europe (Aubry et al. 1998; Gartner et al. 2002). The native range of Douglas-fir extends from the Pacific Coast to the Rocky Mountains and from Mexico to central British Columbia (USDA 2000). Its high value wood, with superior strength and density properties, is ideal for construction materials and was integral in establishing and maintaining the reputation of high quality lumber for the west coast of North America (Aubry et al. 1998). However, supplies of old growth material are limited and second growth resources are being grown at increasingly shorter rotations to meet demands and compete with plantation forestry in the southern hemisphere. The result is a higher proportion of juvenile wood entering the manufacturing sector and a concurrent reduction in value and quality (Gartner et al. 2002; Loehle and Namkoong 1987). A solution to maintaining wood quality is to incorporate "desired" wood traits into tree improvement programs. Improvement programs, therefore, require reliable information regarding the genetic control and relationships among growth and yield attributes and wood quality traits. Armed with this fundamental knowledge, advanced tree improvement programs can make important gains to increase growth and 62 yield, while maintaining superior wood traits (Goggans 1964). However, Loehle and Namkoong (1987) warn of the dangers of forcing trees to change their energy allocation patterns and suggest that selecting for growth may lead to an unstable combination of traits. Information about the physical and genetic interaction of wood quality and growth traits is imperative to develop more efficient breeding programs (Goggans 1964). The focus of early tree improvement programs on growth and yield traits is driven by their ease of quantification, and the fact that volume has been shown to be a significant determinant of the whole tree dollar value (Aubry et al. 1998). In general, most studies that have investigated growth and wood quality attributes often focus on the inheritance and relationships among a small number of traits. In coastal Douglas-fir, wood density is under strong genetic control (narrow-sense heritability of 0.52 - 0.71), while growth and yield traits show slightly weaker (0.13 - 0.32) levels of inheritance (Loo-Dinkins and Gonzalez 1991; St.Clair 1994; Vargas-Hernandez and Adams 1991; Yeh and Heaman 1987). Similar values have been detected in other tree species suggesting a rather conserved mode of inheritance for these traits (Ericsson and Fries 2004; Hannrup et al. 2000; Hannrup et al. 1998; Hodge and Purnell 1993; Ivkovich 1996; Ivkovich et al. 2002a; Loo et al. 1985; Nicholls et al. 1980; Pot et al. 2002; Talbert et al. 1983; Yanchuk and Kiss 1993). Although increasingly tree breeders are becoming concerned with wood quality traits such as wood density and microfibril angle, the often negative genetic relationship between growth and wood density prevents significant gains in both sets of traits. As Douglas-fir is harvested primarily for lumber production, it is imperative that the inheritance and relationships among important wood quality traits, such as wood density, microfibril angle and volume, as well as fibre and wood chemistry 63 characteristics, are understood in order to predict changes to maintain tree health and wood quality for future markets. The purpose of this study is to provide comprehensive information relating coastal Douglas-fir wood quality traits to growth characteristics, including: height, diameter, volume, fibre length, fibre coarseness, microfibril angle, earlywood density, latewood density, latewood proportion, average density, lignin, arabinose, galactose, glucose, xylose and mannose content. The genetic control of growth and wood quality traits was estimated via broad-sense heritability, while the relationships among these traits are reported in the form of phenotypic, genetic and family mean correlations. Materials and methods Sample population The test population was sampled from the British Columbia Ministry of Forests second-generation progeny test program for coastal Douglas-fir. Fifteen full-sib families were chosen to maximize phenotypic variation. Based on previous growth data, five "good" (families 2, 26, 38, 62 and 75), five "poor" (families 7, 46, 56, 92 and 130) and five "intermediate" (families 150, 151, 154, 155 and 156) families were chosen for this study. The intermediate families were chosen to span the gap of phenotypic values between the good and poor families. The sample population was selected from four progeny test sites located in south-western British Columbia: two sites were chosen on Vancouver Island and two on the mainland. The sites were selected to provide phenotypic variation; two high productivity 64 sites (Adam River and Lost Creek) and two low productivity sites (Gold River and Squamish River) selected based on previous growth data. Lost Creek and Squamish River are located on the B.C. mainland in the CWHvm and CWHds subzones of the Biogeoclimatic Ecosystem Classification System, respectively. Lost Creek (latitude: 49° 22' 15; longitude: 122° 14' 10) is 424m above sea level and has a lush understory (e.g. Rubus spectabilis, Vaccinium parvifloris, Pteridium munitum and Dryopteris expansd), while Squamish River (latitude: 50° 12' 05 ; longitude: 123° 22' 30) is 470m above sea level and has very little understory vegetation, occurring in patches (R. spectabilis, D. expansa, and Tiarella spp.). Adam River and Gold River are located on Vancouver Island in the CWHvm and CWHxm subzones, respectively. Adam River (latitude: 50° 24' 00; longitude: 126° 10' 00) is 576m above sea level and has very little understory vegetation, while Gold River (latitude: 49° 51' 30; longitude: 126° 04' 45) is 561m above sea level and has a major component of Vaccinium spp. The trees were planted in October and November of 1977 from 1+0 styro 8 plugs at 3.05m spacing. Each site contains 165 crosses (families). Each family is represented by 16 individuals replicated in four blocks of four-tree row plots. Sampling was conducted in summer and fall (August to November) of 2003. Trees were selected for sampling randomly within sites using Excel's random number generator. When a selected tree was found to be dead, a replacement tree was selected randomly from the site map. Ten individuals were selected randomly from each family within each site without regard for blocking (total of 600 individuals; 10 individuals from 15 families located on four sites). Individuals were later assigned to their appropriate blocks after sampling was complete. 65 Phenotyping Growth Traits and Core Sampling Height (HT) was measured using a Vertex instrument (Vertex III; Haglof, Sweden). Diameter (DBH) was assessed at breast height (1.3m) from the upper side of the tree using a diameter tape. Volume was calculated using Schumacher's equation for Douglas-fir: VOL = 0.000047966*(DBH1'81382)*(HT1'04242) [2.1] where VOL is volume, DBH and HT are diameter at breast height and tree height, respectively, at 26 years of age. A 10mm increment core sample was taken from bark to bark at breast height in a north-south direction. Half of the core was used for wood density measurements and the other half for chemistry and fibre quality. Fibre Properties Fibre length (FL) and coarseness (CS) were analyzed using the southern half of the increment core for three rings corresponding to ages 15-17. The samples were macerated in Franklin solution (1:1, 30% hydrogen peroxide:glacial acetic acid) for 48 hours at 70°C. The solution was decanted and the remaining fibrous material was washed under vacuum with de-ionized water until a neutral pH was achieved. The samples were dried overnight at 50 °C, and the moisture content measured to determine fibre mass for coarseness measurements. Accurately weighed sub-samples were then re-suspended in 10 mL of de-ionized water and fibre properties determined on a Fibre Quality Analyzer 66 (LDA02, OpTest Equipment Inc., Canada). All samples were run in triplicate. Fibre length is recorded in millimeters and fibre coarseness as mass per unit length (mg/m). Wood Density Wood density traits (average density, earlywood density, latewood density and latewood proportion) were measured using the core from the north side of the tree. The cores were sawn to produce 1.67mm thick sections using a precision pneumatic saw with the radial face exposed for X-ray densitometry. The sections were soxlet-extracted with acetone and allowed to acclimate to 7% moisture. Each section represented rings from the pith to the bark. The samples were analyzed using an X-ray densitometer (QTRS-01X, Quintek Measurement Systems Inc., USA). Average density (AD), earlywood density (EWD), latewood density (LWD) and latewood proportion (LWP) were measured as the average for the entire core. Latewood proportion was determined as the ratio of the latewood width to ring width for each ring and averaged for each core. Microfibril Angle Microfibril angle was measured on the earlywood portion of the growth ring corresponding to age 17 by X-ray diffraction (Megraw et al. 1998). Six core sections, as described for wood density measurements, were screened for 002 diffraction arc T-values using a Bruker D8 Discover X-ray diffraction unit equipped with an area array detector (GADDS) on the radial face of the earlywood portion of individual growth rings. Wide-angle diffraction was used in the transmission mode, and the measurements were performed with CuKai radiation (X = 1.54A), the X-ray source fit with a 0.5mm 67 collimator and the scattered photon collected by the GADDS detector. Both the X-ray source and detector were set to theta = 0°. The average T-value of the two 002 diffraction arc peaks was used as the T-values to generate the standard curve for microfibril angle calculations. Thirty-two individual growth rings from six sample trees were selected based on X-ray diffraction T-value distribution and profile symmetry. The growth rings were cut from the core sections processed using a pneumatic saw, and were treated for analysis by compound light microscopy (Wang et al. 2001). The same rings were sectioned using a microtome (20um - 30pm) and placed in lmL of 5% (wt/vol) cobalt chloride (C0CI2). The solution and wood sections were heated to 80°C for 2 hours then floated in a sonicator (47KHz) for a subsequent 2 hours. The sections were rinsed with de-ionized water, mounted on microscope slides and allowed to dry overnight. The sections were then viewed at 400* magnification using differential interference contrast (DIC) microscopy (Figure 2.1 A). The sections from three rings were poor and discarded from further analysis. Fifteen to 46 fibres from the remaining twenty-nine rings were collected using Qcapture and saved as TIFF files. Individual fibre microfibril angles were measured from the image files using the computer program ImageJ. The resulting mean microfibril angles measured by microscopy had 95% confidence intervals ranging from 0.97° to 3.16°. The standard curve (Figure 2. IB) for microfibril angle measurements was generated using linear regression (R2 = 0.92). Microfibril angle of all wood samples was then estimated by measuring the intensity of the 002 peak for the earlywood portion of each sample at ring 17 using the procedure for X-ray diffraction described above, and comparing T-values to the best-fit 68 linear relationship generated for known (measured microscopically) microfibril angles and their respective T-values. Wood Chemistry Wood chemical analyses were modified from the T A P P I Standard Methods (Huntley et al. 2003). The core from pith to bark of the north side of the tree was ground in a Wiley mi l l to pass a 0.4mm screen (40 mesh). The ground wood was soaked in lOOmL of acetone for 8 days (with repeated changes) to remove extractable components and to minimize the formation of "pseudolignin" during Klason analysis. The extracted lignocellulosic material was air-dried to remove the solvent and analyzed for sugar and lignin composition. A 0.2g sample of extracted wood was transferred to a 15mL reaction vial cooled on an ice bath. A 3mL aliquot of 72% (w/w) 69 H2SO4 was added to the sample and thoroughly mixed for one minute. The reaction vial was immediately transferred to a water bath maintained at 20°C, and mixed for one minute every 10 minutes. After 2 hours of hydrolysis, the contents of each test tube were transferred to a 125mL serum bottle using 112mL of de-ionized water to rinse all residue and acid from the reaction vial. The serum bottles (containing 115mL 4% (w/w) H2SO4 plus wood) were sealed and autoclaved at 121°C for 60 minutes. Samples were allowed to cool, and the hydrolysates vacuum-fdtered through pre-weighed medium coarseness sintered-glass crucibles. Each sample was washed with 200mL of warm (~50°C) de-ionized water to remove residual acid and sugars, and dried overnight at 105°C. The dry crucibles were re-weighed to determine Klason lignin (acid-insoluble lignin) gravimetrically. The fdtrate was then analyzed for acid-soluble lignin by absorbance at 205 nm using UV/VIS Spectrometer (Lambda 45, PerkinElmer Instruments Inc., USA) according to TAPPI Useful Method UM250 (Tappi Useful Method 1991). The concentration of sugars (arabinose, galactose, glucose, mannose and xylose) in the hydrolysate was determined using High Performance Anion Exchange Liquid Chromatography (HPLC). The HPLC system (Dionex DX-600, Dionex, USA) was equipped with an ion-exchange PA1 (Dionex) column, a pulsed amperometric detector with a gold electrode, and a Spectra AS50 autoinjector (Spectra-Physics, USA). Prior to injection, samples were fdtered through 0.45um HV fdters (Millipore, USA). A 20uL volume of sample was loaded containing fucose as an internal standard. The column was equilibrated with 250mM NaOH and eluted with de-ionized water at a flow rate of 1.0 mL/min. Total lignin (TL) was calculated as the sum of acid insoluble and acid soluble lignin. Total lignin (TL), arabinose (ARA), galactose (GAL), glucose (GLU), mannose 70 (MAN) and xylose (XYL) were measured as the proportion of the initial mass of the wood sample used in the analysis. Statistical Analysis To compensate for missing values and the unbalanced, incomplete block design, SAS's GLM (general linear model) procedure was used (SAS Version 9.1). The following linear model was used for the analysis of all phenotypic traits Yjjip = p + Fi + Si + Bj(i) + FBij(i) + FSii + Ep(ij,) [2.2] where, Yyip is the individual phenotypic observation, u, is the overall mean, F; is the fixed family effect, Si is the random site effect, Bj(i) is the random block effect, FBy(i) is the random family-by-block interaction nested within site, FSa is the random family-by-site interaction and Ep(jji) is the random residual effect. Each trait was tested for the assumptions of analysis of variance: normal distribution of the sample population and homogeneous variance. SAS's univariate procedure was used to test the normal distribution of residuals, while Bartlett's test was used to test for equal variances of the residuals among families. If the assumptions were not met, several power transformations were attempted and the data re-tested. If the transformed data did not pass the assumptions, untransformed data were used for the analysis and the resulting p-values interpreted with caution. The correct F-tests were derived from the components of variance in Table 2.1. 71 Table 2.1: Components of variance. • df Components of Variance Family (f-1) c E + no FB + bna FS + bcna F Site (s-1) a 2 E + fna 2 B + fbno 2 s Family*Site (f-l)(s-l) o 2 E + no 2 F B + bno 2 F S Block(site) s(b-l) o 2 E + fno 2 B Family*Block(Site) s(b-l)(f-l) a 2 E + n a 2 F B Sampling Error sfb(n-l) a 2 E F = family; B = block; S = site; f = # of families; b = # of blocks; s = # of sites; n = # of trees Differences between family classes (good, poor and intermediate) were tested using the linear model (equation 2.2), where families were replaced with class. Families that were designated as good, intermediate or poor were grouped and tested using the GLM procedure. Class was designated a random effect and consequently tested using the mean square of class by site as the proper error term. The lsmeans option was used to test for differences between individual classes. Classes were considered statistically different if p*m was < 0.05 where p is the p-value output by SAS and m is the number of comparisons ([k-l]k/2; k = the number of classes). Using full-sib families, additive genetic variance cannot be isolated from non-additive genetic variance and, therefore, narrow-sense heritability cannot be estimated. As a result, broad-sense heritability (H2) was estimated. The components of variance for broad-sense heritability estimates were calculated using SAS's Varcomp procedure and the REML (restricted maximum likelihood) method to compensate for missing values and the unbalanced nature of the design. Heritabilities were calculated using the following formula: 72 H 2 = 2O_F [2.3] _2 , _2 T"~2 T"Z2 O F + C J F S + O F B + C J E where C 2 F is family variance, O 2 F S is the variance of family-by-site interaction, O 2 F B is the variance of family-by-block nested within site and O 2 E is the residual variance. Standard errors for heritability estimates were calculated according to Falconer and Mackay (1996): SH 2= {2[ 1 +(f-1 )t]2( 1 -t)/f(f-1 )(F-1) >1/2 [2.4] where SH2 is standard error of heritability, f is the number of individuals per family, F is the number of families and t is the intrafamily variation (H2). Phenotypic correlations were calculated using SAS's Corr procedure. Genetic correlations were estimated according to Falconer and Mackay (1996): rA = a2xy [2.5] / 2 * 'l sV* ( O x O y ) where TA is the genetic correlation between traits x and y, a2xy is the covariance of traits x and y, o 2 x is the variance of trait x and c 2 y is the variance of trait y. Standard error for VA was calculated following Falconer and Mackay (1996) as: o(rA) = [(l-rA 2)/(2)1 / 2]*[a (H2x)0 (H2y/H2 x*H2 y] [2.6] 73 where C(rA) is the standard error of the genetic correlation estimate between traits x and y, <3(H2x) and cj(H2y) are standard errors of the heritability estimates for traits x and y, respectively, and H 2 X and H 2 y are broad-sense heritability estimates for traits x and y, respectively. Genetic correlations are difficult to estimate and have large standard errors. In order to supplement genetic correlations, family mean correlations were calculated. The arithmetic family mean for each trait was calculated and used to generate correlations using the SAS Corr procedure. Due to the unbalanced nature of the design, least square (LS) means were generated using the SAS GLM procedure, and compared to arithmetic means. There was an average difference of 0.55% between arithmetic family means and LS family means. Due to the small difference between family means generated arithmetically and by LS analysis, arithmetic means were used to estimate family mean correlations. Results Growth and Yield Both height and diameter showed significant family and site variation, but no family-by-site interaction (Table 2.2). In contrast, volume showed a significant family-by-site interaction, as well as significant family and site variation (see appendix 1 for ANOVA tables). As such, a site-by-site analysis was conducted: all sites, except for Gold River, showed significant family variation (data not shown). Broad-sense heritability estimates were moderate, ranging from 0.23 to 0.30 (Table 2.2). 74 Table 2.2: Broad-sense heritability (H2) (with standard error estimates in parentheses), and F-values (p-values in parentheses) for family, site and family-by-site variation for growth and wood quality traits in coastal Douglas-fir. Trait Site n1 Family Site Family x Site HT 0.23 (0.077) 4.09 (0.00) 16.68 (0.00) 1.32(0.11) DBH 0.24 (0.079) 5.24 (0.00) 46.68 (0.00) 1.22 (0.19) VOL 0.30 (0.089) 5.03 (0.00) 33.82(0.00) 1.67 (0.01) FL 0.10(0.045) 2.63 (0.01) 19.41 (0.00) 1.28 (0.14) CS 0.18 (0.066) 5.35 (0.00) 26.53 (0.00) 1.02 (0.45) EWD 0.53 (0.101) 13.09 (0.00) 11.44 (0.00) 1.04 (0.41) LWD 0.21 (0.073) 4.42 (0.00) 31.97 (0.00) 1.02 (0.45) AD 0.47 (0.102) 9.63 (0.00) 17.89 (0.00) 1.29 (0.13) LWP 0.30(0.085) 6.79 (0.00) 28.72 (0.00) 1.10(0.34) MFA 0.20 (0.070) 4.18(0.00) 78.05 (0.00) 1.11 (0.32) TL 0.04 (0.023)* 1.25 (0.28) 10.98 (0.00) 3.00 (0.00) AR 0.22 (0.105)* 1.71 (0.09) GR 0.87 (0.127) 6.29 (0.00) LC 0.36 (0.135) 2.84 (0.00) SQ 0.83 (0.139) 5.15 (0.00) ARA 0.25 (0.079) 2.28 (0.02) 16.90 (0.00) 4.42 (0.00) GAL 0.16(0.061) 4.16(0.00) 22.94 (0.00) 0.89 (0.67) GLU 0.98 (0.009) 14.58 (0.00) 17.68 (0.00) 2.41 (0.00) XYL 0.60 (0.097) 6.67 (0.00) 16.08 (0.00) 2.48 (0.00) MAN 0.24 (0.078) 3.10(0.00) 14.61 (0.00) 2.07 (0.00) HT = height; DBH = diameter at breast height; VOL = volume; FL = fibre length; CS = fibre coarseness; EWD = earlywood density; LWD = latewood density; AD = average density; LWP = latewood proportion; ARA = arabinose; GAL = galactose; GLU = glucose; XYL = xylose; MAN = mannose; TL = lignin; AR = Adam River; GR = Gold River; LC = Lost Creek; SQ = Squamish River ' No significant family variation, therefore, broad-sense heritability estimate should be interpreted with caution. Phenotypic correlations reveal positive relationships between height, fibre length and coarseness, and between volume and fibre coarseness (Table 2.3). Growth and yield traits were positively correlated with total lignin and microfibril angle. However, in contrast, they are generally negatively correlated to wood density traits and carbohydrates. 75 Genetic correlations (Table 2.4) show similar trends to phenotypic correlations. Fibre length is positively correlated to height and diameter, while volume is negatively correlated to all density traits. Lignin is positively correlated with diameter and volume, whereas glucose content is negatively correlated with all three growth and yield traits. Height is positively associated with galactose and xylose content, while mannose is positively correlated to diameter but negatively to volume. Only glucose showed significant negative family mean correlations to all growth and yield traits which are consistent with genetic correlations, while arabinose was positively associated with height and volume. Fibre Properties Fibre length and coarseness have significant family and site variation, but no family-by-site interaction. Heritability estimates were low for fibre length and coarseness (0.10 and 0.18, respectively). There was a strong positive phenotypic correlation between fibre length and coarseness. Fibre length and fibre coarseness are negatively correlated to density traits with the exception of a small, but significant positive correlation between latewood density and fibre coarseness. Both fibre length and coarseness are negatively correlated with microfibril angle, lignin content, and galactose and xylose content. Fibre length alone is negatively correlated to arabinose. Both traits are positively correlated to glucose and mannose. 76 Table 2.3: Phenotypic correlations for coastal Douglas-fir (top number is the correlation coefficient and the bottom number is the p-value significant at a = 0.05; bold values are significant). D B H H T V O L F L C S L W P E W D L W D A D T L A R A G A L G L U X Y L M A N M F A D B H 0.72 0.96 0.00 0.06 -0.27 -0.06 -0.07 -0.24 0.20 0.03 -0.07 -0.12 -0.29 -0.05 0.22 (0.00) (0.00) (0.94) (0.18) (0.00) (0.13) (0.08) (0.00) (0.00) (0.44) (0.10) (0.00) (0.00) (0.20) (0.00) H T 0.80 0.19 0.19 -0.07 -0.12 -0.04 -0.10 0.12 -0.07 -0.11 -0.05 -0.34 0.03 0.09 (0.00) (0.00) (0.00) (0.09) (0.00) (0.28) (0.01) (0.00) (0.10) (0.01) (0.22) (0.00) (0.41) (0.03) V O L 0.02 0.10 -0.22 -0.07 -0.09 -0.22 0.17 0.02 -0.09 -0.14 -0.31 -0.03 0.21 (0.67) (0.02) (0.00) (0.07) (0.03) (0.00) (0.00) (0.62) (0.02) (0.00) (0.00) (0.54) (0.00) F L 0.57 -0.11 -0.23 0.02 -0.13 -0.16 -0.14 -0.22 0.15 -0.23 0.12 -0.37 (0.00) (0.01) (0.00) (0.62) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) C S -0.06 -0.23 0.08 -0.09 -0.20 -0.04 -0.31 0.11 -0.17 0.22 -0.23 (0.14) (0.00) (0.05) (0.04) (0.00) (0.32) (0.00) (0.01) (0.00) (0.00) (0.00) L W P 0.46 0.16 0.85 0.03 -0.09 0.11 0.02 -0.02 0.02 0.12 (0.00) (0.00) (0.00) (0.42) (0.03) (0.01) (0.57) (0.70) (0.63) (0.00) E W D 0.24 0.74 -0.01 0.09 0.18 -0.11 -0.02 -0.02 0.09 (0.00) (0.00) (0.72) (0.03) (0.00) (0.01) (0.55) (0.69) (0.03) L W D 0.47 (0.00) -0.19 (0.00) 0.08 (0.06) 0.06 (0.17) -0.02 (0.57) 0.05 (0.25) 0.13 (0.00) -0.10 (0.01) A D -0.05 (0.22) -0.02 (0.69) 0.15 (0.00) -0.02 (0.65) -0.01 (0.89) 0.06 (0.15) 0.06 (0.15) T L -0.02 (0.62) 0.53 (0.00) -0.51 (0.00) 0.11 (0.01) -0.65 (0.00) 0.15 (0.00) A R A 0.02 (0.69) -0.28 (0.00) 0.32 (0.00) 0.06 (0.13) 0.07 (0.09) G A L -0.40 (0.00) 0.22 (0.00) -0.59 (0.00) 0.05 (0.21) G L U -0.07 (0.08) 0.54 (0.00) -0.05 (0.23) X Y L -0.21 (0.00) 0.02 (0.56) M A N -0.05 (0.22) DBH = diameter at breast height; HT = height; VOL = volume; FL = fibre length; CS = fibre coarseness; LWP = latewood proportion; EWD = earlywood density; LWD = latewood density; AD = average density, TL = total lignin; ARA = arabinose; GAL = galactose; GLU = glucose; X Y L = xylose; MAN = mannose; MFA = microfibril angle Table 2.4: Genetic correlations (above diagonal; top number is the correlation coefficient and the bottom number is the standard error) andfamily mean correlations (below diagonal; top number is the correlation coefficient and the bottom number is the p-value significant at a= 0.05; bold values are significant) for Coastal Douglas-fir wood quality traits. ._ A D T L A R A G A L G L U X Y L M A N M F A -0.19 2.39 7.87 1.32 -0.54 1.33 0.42 -0.11 (0.18) (1.50) (13.94) (0.18) (0.03) (0.12) (0.19) (0.24) -0.18 0.97 3.73 0.53 -0.58 0.43 -0.19 -0.02 (0.18) (0.02) (2.98) (0.18) (0.03) (0.13) (0.22) (0.24) -0.63 0.88 1.41 0.14 -0.74 0.03 -0.46 -0.26 (0.11) (0.07) (0.21) (0.23) (0.02) (0.16) (0.17) (0.21) -2.41 0.57 0.15 0.07 -0.20 -0.33 -0.69 0.58 (1.04) (0.25) (0.26) (0.28) (0.04) (0.17) (0.14) (0.18) -11.71 0.98 0.63 -0.14 -0.62 -0.37 -0.27 3.86 (26.97) (0.02) (0.14) (0.26) (0.03) (0.15) (0.23) (3.50) 0.93 -1.42 -0.22 -0.06 0.33 -0.35 0.63 -0.60 (0.02) (0.30) (0.21) (0.24) (0.03) (0.14) (0.13) (0.15) 0.91 -1.09 -0.06 -0.22 0.00 -0.43 0.25 -0.12 (0.03) (0.04) (0.17) (0.18) (0.03) (0.10) (0.16) (0.18) 0.71 -1.25 0.16 -0.49 -0.05 -0.51 0.63 0.49 (0.10) (0.19) (0.23) (0.19) (0.04) (0.12) (0.14) (0.18) -1.33 -0.11 -0.23 0.14 -0.46 0.51 0.01 (0.20) (0.18) (0.19) (0.03) (0.10) (0.14) (0.19) -0.54 0.15 0.34 -1.00 0.96 -1.10 -7.90 (0.04) (0.31) (0.30) (0.00) (0.02) (0.07) (20.2) -0.04 0.19 0.15 -1.00 0.20 -0.86 -29.7 (0.90) (0.49) (0.24) (0.00) (0.16) (0.06) (209.1) -0.17 0.19 0.07 0.12 0.36 -0.37 -5.31 (0.53) (0.49) (0.82) (0.04) (0.15) (0.21) (6.99) 0.13 -0.48 -0.75 0.10 -0.08 0.52 -0.16 (0.64) (0.07) (0.00) (0.73) (0.03) (0.03) (0.04) -0.37 0.40 0.20 0.31 -0.04 -0.65 -5.13 (0.17) (0.14) (0.48) (0.25) (0.88) (0.09) (4.28) 0.40 -0.71 -0.39 -0.36 0.45 -0.44 -3.21 (0.14) (0.00) (0.16) (0.19) (0.09) (0.10) (2.22) 0.08 -0.59 -0.23 -0.16 0.47 0.10 0.60 I.U.03J (v./y) y u . o j ; yu.,?, (0.78) (0.02) (0.42) (0.56) (0.08) (0.72) (0.02) DBH = diameter at breast height; HT = height; VOL = volume; FL = fibre length; CS = fibre coarseness; LWP = latewood proportion; EWD = earlywood density; LWD = latewood density; AD = average density, TL = total lignin; ARA = arabinose; GAL = galactose; GLU = glucose; XYL = xylose; MAN = mannose; MFA = microfibril angle. D B H H T V O L F L CS L W P E W D L W D D B H 0.92 0.99 0.58 3.04 -0.23 0.00 0.12 (0.03) (0.01) (0.18) (2.00) (0.21) (0.17) (0.23) H T 0.88 0.99 0.54 2.39 -0.36 0.00 0.15 (0.00) (0.00) (0.19) (1.14) (0.19) (0.18) (0.23) V O L 0.98 0.94 0.11 0.28 -0.35 -0.23 -1.13 (0.00) (0.00) (0.25) (0.21) (0.18) (0.16) (0.06) F L 0.13 0.25 0.20 0.62 -0.03 -0.37 -5.27 (0.65) (0.37) (0.48) (0.17) (0.25) (0.17) (7.29) C S 0.15 0.22 0.24 0.61 0.05 -2.44 -23.79 (0.60) (0.43) (0.38) (0.02) (0.23) (0.91) (140.44) L W P -0.41 -0.36 -0.38 -0.08 0.02 0.74 0.51 (0.13) (0.19) (0.16) (0.79) (0.95) (0.07) (0.17) E W D 0.00 0.00 0.01 0.04 -0.06 0.69 0.61 (1.00) (1.00) (0.98) (0.89) (0.84) (0.00) (0.11) L W D 0.06 0.22 0.11 -0.30 0.00 0.47 0.51 (0.84) (0.43) (0.70) (0.28) (0.99) (0.08) (0.05) A D -0.23 -0.15 -0.19 -0.07 0.02 0.92 0.88 0.66 (0.42) (0.60) (0.49) (0.80) (0.94) (0.00) (0.00) (0.01) T L 0.31 0.18 0.27 0.08 0.29 -0.51 -0.48 -0.52 (0.27) (0.51) (0.34) (0.77) (0.30) (0.05) (0.07) (0.05) A R A 0.49 0.58 0.59 0.05 0.40 -0.11 -0.03 0.16 (0.06) (0.02) (0.02) (0.86) (0.14) (0.69) (0.92) (0.56) G A L 0.07 0.04 0.05 -0.03 -0.09 -0.06 -0.14 -0.37 (0.80) (0.88) (0.86) (0.91) (0.75) (0.83) (0.61) (0.18) G L U -0.62 -0.55 -0.64 -0.15 -0.49 0.28 0.01 -0.03 (0.01) (0.03) (0.01) (0.58) (0.06) (0.31) (0.98) (0.90) X Y L -0.02 -0.11 -0.08 -0.30 -0.31 -0.25 -0.39 -0.41 (0.95) (0.69) (0.77) (0.29) (0.26) (0.36) (0.16) (0.12) M A N -0.39 -0.41 -0.39 -0.41 -0.23 0.45 0.20 0.47 (0.15) (0.13) (0.15) (0.13) (0.40) (0.09) (0.48) (0.07) M F A -0.13 -0.07 -0.14 -0.35 -0.35 0.07 -0.04 0.40 (0.65) (0.79) (0.63) (0.20) (0.20) (0.79) (0.88) (0.14) Fibre length has a negative genetic correlation with earlywood density. Both traits show strong positive correlations with lignin content and negative correlations with glucose content. There is a moderate positive correlation between fibre length and microfibril angle, although family mean correlations are opposite, but not significant. Arabinose is positively correlated to fibre coarseness, while mannose is negatively correlated with fibre length and xylose is negatively correlated with both fibre traits. The only significant family mean correlation was between fibre length and coarseness. Wood Density All wood density attributes, including earlywood density, latewood density, latewood proportion and average density, had significant family and site variation, but showed no significant family-by-site interaction. Heritability estimates were moderate to high with values of 0.21 and 0.22 for latewood density and latewood proportion, respectively, and 0.54 and 0.47 for earlywood and average density, respectively. Wood density attributes show positive phenotypic correlations to one another. Average density was most strongly correlated with latewood proportion followed by earlywood density and latewood density. Both latewood proportion and earlywood density are positively associated with microfibril angle, however, the opposite is true for latewood density. Earlywood density is negatively correlated with glucose content, while latewood density is negatively correlated with lignin content. Galactose is positively correlated to latewood proportion, earlywood density and average density. 79 Genetic correlations follow similar trends to the phenotypic correlations, where average density was mostly correlated with latewood proportion followed by earlywood and latewood density. Microfibril angle is negatively correlated with latewood proportion and positively with latewood density. There are strong negative correlations among all density traits and lignin content. Galactose is also negatively correlated to latewood density, and glucose positively correlated to average density. Xylose is negatively associated with all density traits, while mannose is positively associated with all density traits. Family mean correlations confirm the relationships between average density and the other density components. Latewood proportion was significantly correlated with earlywood density. There are also significant negative correlations between total lignin and latewood and average density. Microfibril Angle Microfibril angle has significant family and site variation but no significant family-by-site interaction. Earlywood microfibril angle at age 17 had a heritability of 0.20. Microfibril angle is positively associated phenotypically with lignin content, but shows no significant relationships with wood carbohydrates. Genetic correlations for microfibril angle seem inconsistent in many cases with phenotypic correlations. There is a significant negative family mean correlation with lignin content and a small, negative genetic correlation with glucose content. There is also a significant positive family mean correlation with mannose content. 80 Table 2.5: Phenotypic, genetic andfamily mean correlations between microfibril angle and ring 17 traits (respective p-values for phenotypic and family mean correlations, and standard errors (s.e.) for genetic correlations, are in parentheses). Trait Phenotypic Genetic Family Mean Correlations Correlations Correlations (p-values) (s.e.) (p-values) FL -0.37 (0.00) 0.57(0.18) -0.35 (0.20) CS -0.23 (0.00) 3.86 (3.50) -0.35 (0.20) LWFT7 -0.06 (0.13) 0.12(0.26) 0.03 (0.91) EWD17 0.07 (0.08) -0.49(0.19) -0.04 (0.88) LWD17 -0.21 (0.00) 0.41 (0.23) 0.36(0.18) AD17 -0.09 (0.02) -0.76 (0.11) 0.09 (0.73) LWW17 0.28 (0.00) -0.30 (0.26) -0.29 (0.29) RW17 0.38 (0.00) -0.51 (0.21) -0.39(0.15) FL = fibre length; CS = coarseness; LWP 17, EWD 17, LWD 17, AD17, LWW17 and RW17 = ring 17 latewood proportion, earlywood density, latewood density, average density, latewood width and ring width, respectively. In order to get a more accurate depiction of individual ring relationships with microfibril angle, phenotypic, genetic and family mean correlations for age 17 wood quality traits are given in Table 2.5. For this growth ring, microfibril angle is negatively correlated with both fibre properties. However, the genetic correlation for fibre length is opposite in sign and inconsistent with family mean correlations. There is a negative relationship with latewood and average density and a strong negative genetic correlation with the later. Earlywood microfibril angle is positively associated phenotypically and negatively associated genetically with latewood width and ring width. 81 Lignin Lignin content has significant family-by-site interaction and site variation but no significant family variation. All sites, except for Adam River (high productivity site), showed significant family variation for lignin content. The lack of family variation and significant family-by-site interaction makes it difficult to estimate heritability for all sites combined. Therefore, site-by-site estimates were determined and shown to range from 0.22 to 0.87. Lignin content has positive phenotypic associations with xylose and galactose and negative correlations with glucose and mannose. Lignin content is also shown to be genetically associated with xylose, but demonstrates strong negative relationships with glucose and mannose. The negative relationship between lignin and mannose is confirmed by family mean correlations. Carbohydrates The major wood carbohydrates, arabinose, galactose, glucose, xylose and mannose, were quantified for all individuals, and except for galactose, showed significant family-by-site interaction. All traits show significant family and site variation. Heritability estimates varied significantly, from 0.16, 0.24 and 0.25 for galactose, mannose and arabinose, respectively, to 0.60 and 0.98 for xylose and glucose. 82 Glucose is negatively associated phenotypically with arabinose and galactose, but positively with mannose. Xylose is positively associated with arabinose and galactose, but negatively associated with mannose. Glucose has a strong negative genetic correlation with arabinose which was confirmed by family mean correlations. Mannose is negatively associated with arabinose, while both mannose and galactose are positively correlated with glucose. Xylose is positively associated with galactose and negatively correlated to mannose. Selected Families The families designated as "good" (see appendix 2 for arithmetic family means and standard deviations for all traits), based on previous growth data, had significantly higher height, diameter and volume than both intermediate and poor classes (Figure 2.2). However, these families had significantly lower proportion of latewood and average density when compared to the poor families. Glucose content was also significantly lower in the good families compared to intermediate and poor families, while latewood density was statistically the same in both good and poor families and showed a significant difference between intermediate and poor family groups. 83 Class Class S I i Bill i l Class I 600 J 5 Figure 2.2: Least square means for families designated as good(G), intermediate (I) and poor (P). Groups with the same letter are statistically the same at a = 0.05. 84 Discussion Heritability and Genetic Control Many of the broad-sense heritability estimates reported in the current study are consistent with narrow-sense estimates for similar traits in Douglas-fir and other conifer species. Broad-sense estimates are expected to be larger than narrow-sense estimates due to the inclusion of non-additive effects, however, many of the broad-sense estimates generated in this study are either similar to, or lower, than narrow-sense estimates from the literature. This suggests that the non-additive variation involved in phenotypic variation is minimal and has little effect on heritability estimation. Stonecypher et al. (1996) showed that the ratio of non-additive to additive variation for Douglas-fir juvenile height growth is 0.46, suggesting that additive effects are more important contributors to phenotypic variation, which may account for the similarity of broad-sense and narrow-sense estimates. Growth parameters are relatively easy to measure and account for a significant proportion of whole tree dollar value (Aubry et al. 1998). As such, much of the early breeding activities have focused on height, diameter and volume as traits of interest. These traits are under moderate genetic control, and are influenced significantly by environment. Our current broad-sense heritability estimates were between 0.23 and 0.30 and are consistent with previously reported narrow-sense estimates for Douglas-fir (St.Clair 1994) of 0.17 for height, 0.27 for diameter and 0.32 for volume, but are higher than estimates for height (0.13) by Yeh and Heaman (1987). The influence of genetics on 85 growth traits is similar across species. For example, narrow-sense heritability was estimated to be 0.30 for height and 0.11 for diameter for interior spruce (Picea glauca (Moench) Voss, P. engelmanii Perry x Engel., and their hybrids) (Yanchuk and Kiss 1993), 0.28 - 0.32 for height and 0.17 - 0.27 for diameter in Scots pine (Pinus sylvestris L.) (Ericsson and Fries 2004; Hannrup et al. 2000; Hannrup et al. 1998), 0.18 for height and 0.19 for diameter of balsam poplar (Populus balsamifera L.) (Ivkovich 1996), 0.56 for height and 0.30 - 0.37 for diameter for interior spruce (Ivkovich et al. 2002a) and 0.456 for height in maritime pine (Pinuspinaster Ait!) (Pot et al. 2002). Douglas-fir growth is significantly affected by site, however, families react to variable site conditions in a consistent manner for height and diameter. There is a small but significant interaction between site and family for volume. As indicated in this study, these results suggest that while families react to site quality in a similar manner for height and diameter, there may be optimal genotypes for specific environmental conditions when considering tree volume. The moderate heritability estimates reveal that over 50% of the variation in the data for height, diameter and volume can be explained by factors other than genetics. These results suggest that phenotypic plasticity allows individuals to adapt to local conditions and that this occurs in a similar manner regardless of parents. Fibre properties are under moderate genetic control and have the lowest heritability estimates of the wood quality traits measured in this study. Our broad-sense estimates for fibre length (0.10) are in agreement with previous examples of narrow-sense estimates for balsam poplar of 0.13 (Ivkovich 1996), loblolly pine of 0.11 (Loo et al. 1985), and maritime pine of 0.172 (Pot et al. 2002), but are much lower than estimated in other studies (Ericsson and Fries 2004; Goggans 1964; Hannrup et al. 2000). 86 Sometimes an overestimation of heritability can result from a lack of site representation in the experimental design, which does not seem to be the case in the current study. The low heritability estimates, and link to growth rate, suggest that fibre properties in coastal Douglas-fir are heavily influenced by environment and will adapt to local conditions. The lack of family-by-site interaction for fibre properties suggests a conserved change in morphology given common site conditions. The estimates of broad-sense heritability presented in this study for average wood density (0.46) agree with previous estimates of 0.52 (St.Clair 1994) and 0.59 (Vargas-Hernandez and Adams 1991) for coastal Douglas-fir, although these estimates may be upwardly biased due to lack of consideration of family-by-site interactions. Examples of similar estimates have been reported for radiata pine (0.47) (Pinus radiata D. Don) (Nicholls et al. 1980), loblolly pine (0.30 - 0.76) (Loo et al. 1985; Talbert et al. 1983), Scots pine (0.35 - 0.50) (Hannrup et al. 2000; Hannrup et al. 1998), balsam poplar (0.46) (Ivkovich 1996), and interior spruce (0.47 - 0.52) (Ivkovich et al. 2002a; Yanchuk and Kiss 1993). However, our estimate is considerably lower than that of Loo-Dinkins and Gonzalez (1991) of 0.71 for 12 year old Douglas-fir. Goggans (1964) and van Buijtenen (1962) report rather high estimates for loblolly pine (0.76 - 0.87 and 0.84, respectively), while average density heritability was estimated to be 0.16 in slash pine (Pinus elliottii Englem.) (Hodge and Purnell 1993). Despite these few discrepancies, heritability of average wood density seems fairly conserved across species and is in agreement with the estimate presented in the current study, suggesting that the non-additive component of the broad-sense heritability estimate contributes little to the value. 87 Microfibril angle is a difficult trait to measure and, therefore, little information is available regarding the heritability of this trait and relationships between it and other traits. The broad-sense heritability estimate presented in the current study is for the earlywood component of a single growth ring and is somewhat low compared to examples from other studies. Clonal estimates of broad-sense heritability for microfibril angle in radiata pine (Donaldson and Burdon 1995) and Norway spruce (Hannrup et al. 2004) range from 0.09 - 0.72 and 0.11 - 0.29, respectively. Narrow-sense heritability for loblolly pine range between 0.17 and 0.51 depending on the growth ring analysed (Myszewski et al. 2004). Mirofibril angle decreases substantially from pith to bark (Donaldson 1992; Erickson and Arima 1974). Early in growth, the bole of the tree needs to be flexible to avoid breaking, which is aided by high microfibril angles and low Young's modulus. As the tree increases in size, increasing support is required resulting in a decrease in microfibril angle and increased Young's modulus in the outer wood (Barnett and Bonham 2004). The change in microfibril angle from pith to bark to compensate for increased biomass is good evidence for genetic control of microfibril angle (Barnett and Bonham 2004), however, the low heritability detected in the current study suggests that environmental factors, such as wind speed, air temperature (Wimmer et al. 2002) and growing environment (i.e. soil type, moisture), have a large impact on the angle of the cellulose microfibres in the cell wall. Lignin was the only wood trait that had no significant family variation for combined sites, but showed significant family-by-site interaction. A site-by-site analysis revealed significant family variation for three of the four sites. Only Adam River, one of the most productive sites in the study, had no significant family variation for lignin 88 content. Heritability estimates by site suggest a wide range of genetic control for lignin content. The two most productive sites in this study, Adam River and Lost Creek, had the lowest heritabilities (0.22 and 0.36 respectively), while the two low productivity sites, Squamish River and Gold River, had the highest heritability estimates (0.87 and 0.83 respectively). It is apparent that a higher proportion of the variation in lignin content can be explained by genetics on less productive sites. The heritability estimates are probably biased upwards due to a lack of consideration of family-by-site interaction, but reveal vastly altered genetic control that may relate to site conditions. Previously, Pot et al. (2002) report a narrow-sense heritability estimate for lignin of 0.47 for maritime pine, which is within the range of heritabilities reported in the current study and is probably a more accurate estimate of the degree of genetic control of lignin content. Cell wall carbohydrate content consistently showed significant family-by-site interaction (all carbohydrates except galactose). The four sites used in this study were chosen to represent more and less productive sites based on previous growth data! The chemical analysis suggests that carbohydrate biosynthesis is not triggered by the same environmental cues as growth traits, and may be related to other, less obvious stimuli. Glucose content had the highest heritability estimate of all traits in this study (0.98). Glucose is an important constituent of cell wall formation and is used in the construction of the cellulose microfibrils. The proper production of glucose is vital to plant survival. It is intuitive that such an important process is controlled more strongly by genetics. The hemicellulose sugars have considerably lower heritability estimates, and may be influenced by extraneous factors and alternate pathways. 89 Relationships Among Traits Fibre length and coarseness are positively correlated with height growth, while coarseness correlates with volume (phenotypic correlations). These results are in contrast to the results of Ericsson and Fries (2004) who report negative phenotypic and genetic correlations between height, diameter and fibre length for Scots Pine, but are supported by studies in balsam poplar (Ivkovich 1996) and maritime pine (Pot et al. 2002). The effect of growth rate on several fibre properties in fertilizer and irrigation trials of Norway Spruce (Picea abies (L.) Karst.) demonstrated that trees with faster growth rates tend to have shorter, wider fibres with thinner cell walls and more lumen space (Markinen et al. 2002). Consequently, trees with more resources and faster growth rates tend to change cell structure to accommodate water and nutrient uptake. As such, it is reasonable to suggest that given equal nutrient and moisture regimes as experienced by trees in the current study, trees that are healthier, grow faster, and have longer, more developed fibre cells. Density traits are negatively correlated to growth traits. This is consistent with previous reports for Douglas-fir (St.Clair 1994; Vargas-Hernandez and Adams 1991) and for balsam poplar (Ivkovich 1996), Scots pine (Hannrup et al. 2000), and interior spruce (Ivkovich et al. 2002a). Yanchuk and Kiss (1995) report negative phenotypic correlations between height, diameter and specific gravity for hybrid spruce, but show negligible genetic correlations. Diameter is negatively correlated to latewood proportion and average density. The relationship between diameter and average density is probably a result of the relationship between diameter and latewood proportion since latewood proportion is significantly related to average density. A negative correlation between 90 diameter and latewood proportion suggests a positive correlation between diameter and the proportion of earlywood. Since earlywood is less dense than latewood, having more earlywood would result in lower average density. This suggests that diameter may be more strongly influenced by earlywood production. Height growth is shown to be correlated to earlywood density: earlywood is produced during active growth in the spring and production ends when height growth stops. Dendrochronological data for Douglas-fir (Watson and Luckman 2002) suggests that earlywood is more closely related to conditions late in the growing season of the previous year (July-August). This type of relationship is expected for a determinant species such as Douglas-fir in which buds form in the previous growing season and contain all of the necessary material for growth in the subsequent year. Conditions in the current year can affect bud formation and growth in the next year. The correlations presented in this study combined with the dendrochoronological data suggest that earlywood production is linked to height growth and may be a major influence on diameter growth for Douglas-fir. Average density seems to be more strongly related to latewood proportion and earlywood density than to the actual latewood density given the strong phenotypic and genetic correlations between these traits. The significant positive phenotypic correlations between growth at year 26 and earlywood microfibril angle at year 17 suggests a relationship between these traits, however, the lack of genetic correlations suggests that these traits may not be associated genetically. Wimmer et al. (2002) reports similar phenotypic trends between growth and microfibril angle, while no significant genetic correlations were detected in clonal and full-sib tests in Norway spruce (Hannrup et al. 2004). This relationship is further 91 supported by the individual ring correlations for microfibril angle and ring width (0.38). However, at age 17, there seems to be a negative genetic correlation with ring width suggesting a possible decrease in microfibril angle given selection for increased ring width. It is well accepted that fibre length is negatively associated with microfibril angle both phenotypically and genetically (Barnett and Bonham 2004; Hannrup et al. 2004; Ivkovich et al. 2002b). The current study reports negative phenotypic correlations for both fibre properties, but a positive genetic relationship with fibre length. Hannrup et al (2004) reports positive genotypic correlations at one location for Norway spruce between the two traits, which is consistent with the genetic correlations reported in the current study. The genetic relationship between fibre length and microfibril angle may be complex and site dependent. Microfibril angle is also significantly associated with wood density traits. A combination of wood density or specific gravity and microfibril angle can be used to explain the majority of variation in wood stiffness data (Cave and Walker 1994; Cramer et al. 2005; Evans and Ilic 2001). Earlywood microfibril angle at age 17 is positively correlated phenotypically to average core latewood proportion (0.12) and earlywood density (0.09), and negatively to latewood density (-0.10). The low correlation coefficients may be due to correlating traits from different growth rings. A more accurate estimate of trait relationships can be seen when comparing traits from the same growth ring. Both latewood density and average density for the year 17 growth ring are negatively correlated with microfibril angle. There is a strong negative genetic correlation with average density (-0.76). It has been shown that microfibril angle decreases within individual growth rings from earlywood to latewood (Anagnost et al. 2002; Wimmer et al. 2002), which follows the trend of decreasing microfibril angle in regions of higher 92 wood density. The angle of cellulose microfibrils is positively associated with lignin content suggesting trees with higher total lignin content, have higher earlywood microfibril angles. The lack of significant relationships between microfibril angle and wood carbohydrate content suggests that the angle of the microfibrils is associated more with growth and development than biochemical processes. It is clear that the genetic and phenotypic relationships of microfibril angle with growth and wood quality traits is very complex. Lignin content has positive genetic and phenotypic correlations with height, diameter and volume suggesting that larger trees have more lignin. These results are consistent with those obtained for maritime pine (Pot et al. 2002). Trees with faster growth rates, therefore, allocate carbon to lignin production in order to increase bole strength and withstand growth stresses. In contrast, carbohydrates, namely galactose, glucose and xylose, are generally negatively correlated with growth traits. The negative relationship between growth traits and glucose suggest a negative relationship between growth and cellulose content. Our results suggest that larger trees tend to have more lignin and less cellulose. Lignin may be produced at the expense of carbohydrates in larger trees to add strength and support. There is some evidence that the regulation and control of cellulose and hemicellulose material may be linked to lignin biosynthesis and may result from an increase in available carbon when lignin production is lower (Hu et al. 1999). This relationship is further supported by a strong negative phenotypic correlation between lignin and glucose (-0.51) and a strong positive correlation between lignin, galactose and xylose (0.53 and 0.11 respectively). We also report strong negative genetic 93 correlations between lignin and glucose (-1.00) and positive genetic correlations between lignin and many of the hemicellulose constituents. Fibre length and coarseness are strongly correlated with each other. This strong correlation has implications for paper quality. Paper strength is related to the surface area available between fibres for bonding. Fibres with higher length and lower coarseness produce stronger paper (Mansfield et al. 2004). The phenotypic relationship between fibre length and density traits (earlywood density, latewood proportion and average density) are consistent with previously documented relationships between fibre length, density and growth rate (Via et al. 2004). However, the negative relationships between fibre coarseness, earlywood density and average density are counterintuitive and inconsistent with reports by Via et al. (2004). It is possible that in Douglas-fir, trees with faster height growth and longer fibres also have higher fibre coarseness, but a corresponding increase in lumen volume which would account for a decrease in wood density and the need for greater volumes of water used in elevated growth rates. Similar findings were observed by Markinen et al. (2002). Fibre properties show a strong relationship to wood chemistry data: fibre length and coarseness were negatively related phenotypically to total lignin, arabinose, galactose, and xylose, but positively associated with glucose and mannose. Trees with longer fibres may allocate more carbon to glucose production for cellulose arid hemicellulose biosynthesis at the expense of lignin and other hemicellulose carbohydrates. Careful inspection of the phenotypic correlations reveals significant positive correlations between galactose and several density traits (latewood proportion, earlywood density and average density). Galactose is found more abundantly in reaction wood, 94 which is also associated with higher density and shorter fibres. Lignin appears to have negative genetic and family mean correlations with density traits, whereas glucose has positive correlations. This type of relationship suggests that an increase in density may result in a favourable shift to more glucose and less lignin in subsequent generations. The reciprocal correlation between lignin, glucose and density traits adds further support to the trade-off in carbon allocation between lignin and glucose. Implications for Tree Improvement Present-day breeders are faced with the challenge of increasing volume production while maintaining wood quality and tree health. In order to overcome these challenges, breeders must understand the complex physical and genotypic relationships between traits and the influence of genetics and environment. Traditionally, tree improvement programs have focused on increasing volume through an increase in height and diameter growth. Aubry et al. (1998) investigated the importance of log volume, density, and knot size and frequency on the whole-tree dollar value. They found that volume was the most important factor in the absolute dollar value of the tree due to the increased lumber recovery. Wood density was less important due to the difficulty in estimation during the grading process. However, when combined with microfibril angle, wood density is an important contributor to wood stiffness (Cramer et al. 2005; Evans and Ilic 2001). It has only been in recent years that wood quality has been incorporated into many tree improvement programs (Aubry et al. 1998; Vargas-Hernandez and Adams 1991). The most informative wood quality trait is wood density. In order to ensure health 95 and vigour in tree improvement programs, tree breeders must understand the relationships between these and other wood quality traits. Breeding for one set of traits may have indirect impacts on less obvious traits that can lead to decreases in tree health and wood quality for alternative uses such as pulp and paper production. This study has provided reliable information to aid in the selection of plus trees (phenotypic correlations), make prediction about possible gains (heritability estimates) and forecast changes in other wood quality traits given changes in a trait of interest (genetic and family mean correlations). The most effective way to select plus trees for use in tree improvement programs is by visual inspection of easily observable traits. Given reliable estimates of phenotypic correlations, height and diameter can be used to predict the values of less obvious traits such as wood density, fibre quality, lignin and carbohydrate content. From this study, it is apparent that in general, taller Douglas-fir trees have longer fibres with higher mass to unit length ratio, lower earlywood and average density, higher microfibril angle, more lignin and less carbohydrates than shorter trees of the same age. Trees of greater diameter have a lower proportion of latewood, lower average density, higher microfibril angle, more lignin and less carbohydrates than smaller trees of the same age. Although there is significant site variation for all traits, only wood chemistry data consistently showed significant interactions between family and site. As a result, lignin and carbohydrate content may vary from site to site in a way that is inconsistent with changes in other traits, while the general correlations estimated for wood density and fibre quality should remain consistent. Armed with this information, more reliable selections can be made for less obvious traits. However, these estimates of phenotypic correlations are simply 96 generalizations. There is variation within the population that does not conform to these estimates. For example, it is possible to find "correlation breakers" - individuals with superior height and diameter growth as well as good wood density. With this in mind, selecting plus trees solely based on estimates of phenotypic correlations may not be suitable for developing the desired mix of traits required in breeding programs. It may be necessary to select a range of growth traits to capture the necessary variation to develop desired phenotypes. For example, breeding smaller trees with higher average density with larger trees and lower average density may result in a few individuals with the desired mix of density and growth phenotypes (Burley 2001; Namkoong 1976). The expected gain for a given trait is proportional to narrow-sense heritability estimates; traits with higher heritability will respond better to breeding activities. The estimates presented in this study are for broad-sense heritability. Broad-sense heritability estimates incorporate additive and non-additive genetic variation, whereas narrow-sense heritability is estimated using additive effects alone. Additive genetic variation is the variation acted upon during breeding activities. Therefore, narrow-sense heritability is ideal for estimating gains from selection. As a result of the use of full-sib families in this study, additive genetic variance could not be isolated from non-additive variance and narrow-sense heritability could not be estimated. Narrow-sense heritability estimates can be used to indicate the probability that selected individuals will be genetically superior based on phenotypic traits (Zobel and Talbert 1984). Although it is not suitable to estimate gains from broad-sense heritability estimates, we can make predictions about the response to selection relative to other traits assuming similar ratios of additive to non-additive genetic variation and similar phenotypic variation. For this population of 97 Douglas-fir, the highest heritability estimates were for glucose content (0.98) and lignin content at Gold River (0.87) and Squamish River (0.82). Several of the carbohydrate traits also had relatively high estimates of heritability. These traits are of less concern for wood quality, but do have implications in pulp production as with increasing amounts of juvenile wood entering the industry (less old growth material available), more juvenile wood residue will enter pulp manufacturing (Baucher et al. 2003). Of the remaining traits, earlywood density (0.54) and average density (0.47) had the highest heritability estimates. Consequently, earlywood density and average density can be expected to respond better to breeding activities compared to latewood proportion (0.30) and latewood density (0.21). Growth traits had slightly lower heritability estimates than that of wood density traits. Of the growth traits, volume had the highest heritability (0.30). The lowest heritability estimates were for fibre properties. Fibre length and fibre coarseness and microfibril angle are not expected to respond well to breeding activities relative to the other traits. Given the importance of volume and wood density in the value of trees for lumber production, these traits should be of primary focus in breeding efforts. Given their moderate heritability estimates, significant gains can be expected for these traits. It is also necessary to acknowledge correlated responses to selection given a change in a trait of interest (Falconer and Mackay 1996). Family mean correlations estimate the correlations between traits given the additive and non-additive genetic variation inherited by the progeny in each family, and is another way to estimate genetic relationships between traits. In most breeding programs, selection is based on growth traits. If this were the case for Douglas-fir, we would expect an increase in microfibril 98 angle, average fibre length and fibre coarseness, and a decrease in overall average density, an increase in lignin and hemicellulose content and a decrease in glucose (cellulose) content in subsequent generations. Clearly, this would have negative implications for wood quality. The five good families selected for this study were chosen based on growth and yield data and showed significant increases in growth traits over poor and intermediate families. However, these families had lower average density, glucose content and latewood proportion. These results are consistent with the results expected based on genetic correlations and heritability estimates. Assuming that genetic correlations and heritability in the parental generation are consistent with those presented here, this provides good evidence supporting changes in alternative traits given selection. Selecting for improved wood density would result in lower growth and yield. The optimum is a compromise between selecting for improved wood density and superior growth and yield properties or to search for individuals with optimal growth and yield and acceptable wood density. An alternative approach may be to develop separate breeding populations for each set of traits and use each as infusion populations to introduce new variation into the respective breeding programs. 99 References Anagnost, S.E., Mark, R.E., and Hanna, R.B. 2002. Variation of microfibril angle within individual tracheids. Wood and Fiber Science. 34 (2): 337-349. Aubry, C.A., Adams, W.T., and Fahey, T.D. 1998. Determination of relative economic weights for multitrait selection in coastal Douglas-fir. Canadian Journal of Forest Research. 28(8): 1164-1170. Barnett, J.R., and Bonham, V. A. 2004. Cellulose microfibril angle in the cell wall of wood fibres. Biological Reviews. 79: 461-472. Baucher, M., Halpin, C , Petit-Conil, M., and Boerjan, W. 2003. Lignin: Genetic engineering and impact on pulping. Critical Reviews in Biochemistry and Molecular Biology. 38 (4): 305-350. Burley, J. 2001. Genetics in sustainable forestry: the challenges for forest genetics and tree breeding in the new millennium. Canadian Journal of Forest Research-Revue Canadienne De Recherche Forestiere. 31 (4): 561-565. Cave, I.D., and Walker, J.C.F. 1994. Stiffness of wood in fast-grown plantation softwoods: the influence of microfibril angle. Forest Products Journal. 44 (5): 43-48. Cramer, S., Kretschmann, D., Lakes, R., and Schmidt, T. 2005. Earlywood and latewood elastic properties in loblolly pine. Holzforschung. 59: 531-538. Donaldson, L.A. 1992. Within- and between- tree variation in microfibril angle in Pinus radiata. New Zealand Journal of Forestry Science. 22 (1): 77-86. Donaldson, L.A., and Burdon, R.D. 1995. Clonal variation and repeatability of microfibril angle in Pinus radiata. New Zealand Journal of Forestry Science. 25 (2): 164-174. Erickson, H.D., and Arima, T. 1974. Douglas-fir wood quality studies. Part II: Effects of age and stimulated growth on fibril angle and chemical constituents. Wood Science and Technology. 8: 255-265. 100 Ericsson, T., and Fries, A. 2004. Genetic analysis of fibre size in a full-sib Pinus sylvestrisL. progeny test. Scandinavian Journal of Forest Research. 19 (1): 7-13. Evans, R., and Ilic, J. 2001. Rapid prediction of wood stiffness from microfibril angle and density. Forest Products Journal. 51 (3): 53-57. Falconer, D.S., and Mackay, F.C. 1996. Introduction to quantitative genetics. 4th ed. Longman Group Ltd, Burnt Mill, Harlow, Essex, England. 464 pgs. Gartner, B.L., North, E.M., Johnson, G.R., and Singleton, R. 2002. Effects of live crown on vertical patterns of wood density and growth in Douglas-fir. Canadian Journal of Forest Research. 32 (3): 439-447. Goggans, J.F. 1964. Correlation and heritability of certain wood properties in loblolly pine (Pinus taeda L.). Tappi. 47 (6): 318-322. Hannrup, B., Cahalan, C , Chantre, G., Grabner, M., Karlsson, B., Le Bayon, I., Jones, G.L., Muller, U., Pereira, H., Rodrigues, J.C., Rosner, S., Rozenberg, P., Wilhelmsson, L., and Wimmer, R. 2004. Genetic parameters of growth and wood quality traits in Picea abies. Scandinavian Journal of Forest Research. 19: 14-29. Hannrup, B., Ekberg, I., and Persson, A. 2000. Genetic correlations among wood, growth capacity and stem traits in Pinus sylvestris. Scandinavian Journal of Forest Research. 15: 161-170. Hannrup, B., Wilhelmsson, L., and Danell, O. 1998. Time trends for genetic parameters of wood density and growth traits in Pinus sylvestrisL. Silvae Genetica. 47 (4): 214-219. Hodge, G.R., and Purnell, R.C. 1993. Genetic parameter estimates for wood density, transition age, and radial growth in Slash Pine. Canadian Journal of Forest Research. 23(9): 1881-1891. Hu, W.J., Harding, S.A., Lung, J., Popko, J.L., Ralph, J., Stokke, D.D., Tsai, C.J., and Chiang, V.L. 1999. Repression of lignin biosynthesis promotes cellulose accumulation and growth in transgenic trees. Nature Biotechnology. 17 (8): 808-812. 101 Huntley, S.K., Ellis, D., Gilbert, M., C., C , and Mansfield, S.D. 2003. Significant increases in pulping efficiency in C4H-F5H-transformed poplars: improved chemical savings and reduced environmental toxins. Journal of Agricultural and Food Chemistry. 51: 6178-6183. Ivkovich, M. 1996. Genetic variation of wood properties in balsam poplar (Populus balsamiferaL). Silvae Genetica. 45 (2-3): 119-124. Ivkovich, M., Namkoong, G., and Koshy, M. 2002a. Genetic variation in wood properties of interior spruce. I. Growth, latewood percentage, and wood density. Canadian Journal of Forest Research. 32 (12): 2116-2127. Ivkovich, M., Namkoong, G., and Koshy, M. 2002b. Genetic variation in wood properties of interior spruce. II. Tracheid characteristics. Canadian Journal of Forest Research. 32: 2128-2139. Loehle, C , and Namkoong, G. 1987. Constraints on tree breeding - growth tradeoffs, growth strategies, and defensive investments. Forest Science. 33 (4): 1089-1097. Loo, J.A., Tauer, C.G., and McNew, R.W. 1985. Genetic variation in the time of transition from juvenile to mature wood in loblolly pine (Pinus taeda L). Silvae Genetica. 34 (1): 14-19. Loo-Dinkins, J.A, and Gonzalez, J.S. 1991. Genetic control of wood density profile in young Douglas-Fir. Canadian Journal of Forest Research. 21 (6): 935-939. Mansfield, S.D., Kibblewhite, R.P., and Riddell, M.J.C. 2004. Characterization of the reinforcement potential of different softwood kraft fibers in softwood/hardwood pulp mixtures. Fiber Science. 36 (3): 344-358. Markinen, H , Saranpaa, P., and Linder, S. 2002. Effect of growth rate on fibre characteristics in Norway spruce (Picea abies (L.) Karst.). Holzforschung. 56 (5): 449-460. Megraw, R.A., Leaf, G., and Bremer, D. 1998. Longitudinal shrinkage and microfibril angle in loblolly pine. In Microfibril angle in wood. University of Canterbury Press, Christchurch, New Zealand, pp. 27-61. 102 Myszewski, J.H., Bridgwater, F.E., Lowe, W.J., Byram, T.D., and Megraw, R.A. 2004. Genetic variation in the microfibril angle of loblolly pine from two test sites. Southern Journal of Applied Forestry. 28 (4): 196-204. Namkoong, G. 1976. Multiple-index selection strategy. Silvae Genetica. 25 (5-6): 199-201. Nicholls, J. W.P., Morris, J.D., and Pederick, L.A. 1980. Heritability estimates of density characteristics in juvenile Pinus radiata wood. Silvae Genetica. 29 (2): 54-61. Pot, D., Chantre, G., Rozenberg, P., Rodrigues, J.C., Jones, G.L., Pereira, H., Hannrup, B., Cahalan, C., and Plomion, C. 2002. Genetic control of pulp and timber properties in maritime pine (Pinus pinaster Ait.). Annals of Forest Science. 59: 563-575. SAS. Version 9.1. SAS System for Windows. Copyright (c) 2002 - 2003 SAS Institute Inc. SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc., Cary, NC, USA. St.Clair, J.B. 1994. Genetic variation in tree structure and its relation to size in Douglas-fir. I. Biomass partitioning, foliage efficiency, stem form, and wood density. Canadian Journal of Forest Research. 24 (6): 1226-1235. Stonecypher, R.W., Piesch, R.F., Helland, G.G., Chapman, J.G., and Reno, H.J. 1996. Results from genetic tests of selected parents of Douglas-fir (Pseudotsuga menziesii [Mirb.] Franco) in an applied tree improvement program. Forest Science Monograph 32. 42(2): 1-35. Talbert, J.T., Jett, J.B., and Bryant, R.L. 1983. Inheritance of wood specific gravity in an unimproved loblolly pine population: 20 years of results. Silvae Genetica. 32 (1-2): 33-37. Tappi Useful Method. 1991. UM-250: Acid-soluble lignin in wood and pulp. USD A. 2000. Wood handbook: wood as an engineering material. University Press of the Pacific, Madison, Wisconsin. 463 pgs. van Buijtenen, J.P. 1962. Heritability estimates of wood density in loblolly pines. Tappi. 45 (7): 602-605. 103 Vargas-Hernandez, J., and Adams, W.T. 1991. Genetic variation of wood density components in young coastal Douglas-fir - Implications for tree breeding. Canadian Journal of Forest Research. 21 (12): 1801-1807. Via, B.K., Stine, M., Shupe, T.F., So, C.-L., and Groom, L. 2004. Genetic improvement of fiber length and coarseness based on paper product performance and material variability - a review. IAWA Journal. 25 (4): 401-414. Wang, H.H., Drummond, J.G., Reath, S.M., Hunt, K., and Watson, P.A. 2001. An improved fibril angle measurement method for wood fibres. Wood Science and Technology. 34: 493-503. Watson, E., and Luckman, B.H. 2002. The dendroclimatic signal in Douglas-fir and ponderosa pine tree-ring chronologies from the southern Canadian Cordillera. Canadian Journal of Forest Research. 32 (10): 1858-1874. Wimmer, R., Downes, G.M., and Evans, R. 2002. Temporal variation of microfibril angle in Eucalyptus nitens grown in different irrigation regimes. Tree Physiology. 22: 449-457. Yanchuk, A.D., and Kiss, G.K. 1993. Genetic variation in growth and wood specific gravity and its utility in the improvement of interior spruce in British Columbia. Silvae Genetica. 42 (2-3): 141-148. Yeh, F.C., and Heaman, J.C. 1987. Estimating genetic parameters of height growth in 7-year old coastal Douglas-fir from disconnected diallels. Forest Science. 33 (4): 946-957. Zobel, B., and Talbert, J. 1984. Applied forest tree improvement. Waveland Press, Inc, Prospect Heights, Illinois. 252-264 pgs. 104 Chapter 3 An AFLP Linkage Map for Coastal Douglas-fir Using Eight Full-Sib Families Introduction Douglas-fir (Pseudotsuga menziesii (Mirb.) Franco) is arguably one of the most important commercial tree species along the west coast of Canada and the United States (Aubry et al. 1998; Gartner et al. 2002). Its native range extends from the Rocky Mountains to the Pacific Ocean and from central British Columbia to Mexico (USDA 2000). Two varieties of Douglas-fir occur throughout its range: coastal Douglas-fir (var menziesii) and interior Douglas-fir (var glauca). The coastal variety occurs mainly west of the Cascade and Coastal Mountains, and its wood is highly valued for lumber production due to its inherent high density and exceptional strength properties (Chantre et al. 2002; Koshy and Lester 1994; Loo-Dinkins and Gonzalez 1991; Loo-Dinkins et al. 1991; St.Clair 1994; Vargas-Hernandez and Adams 1991). The interior variety is amongst the most important commercial species in the interior of British Columbia (Nigh et al. 2004) and has the potential to produce high quality veneer (Hesterman and Gorman 1992) and lumber (Wagner et al. 2002). Recent advances in molecular marker technology have aided in the research of the structure, organization and evolution of genomes, and facilitate the creation of linkage maps and identification of marker-trait associations (Cervera et al. 2000). Several of these 105 techniques utilize the advantages of polymerase chain reaction (PCR) detection. One such technique, amplified fragment length polymorphisms (AFLP), has become a superior alternative to other PCR based techniques, such as random-amplified polymorphic DNA (RAPD), as it is more reproducible within and between labs and can produce more polymorphic markers at an equal cost (Cabrita et al. 2001; Cervera et al. 2000; Jones et al. 1997; Vos et al. 1995). The rapid production of many polymorphic loci for a given set of PCR reactions makes AFLPs ideal for saturating linkage maps. As such, several linkage maps have been developed using AFLPs for tree species (Cervera et al. 2001; Chagne et al. 2002; Remington et al. 1999; Scalfi et al. 2004; Travis et al. 1998; Wu et al. 2000; Yin et al. 2003; Zhang et al. 2004). Although, to date, there is no AFLP linkage map for Douglas-fir, several linkage maps have been developed utilizing RAPDs and restriction fragment length polymorphisms (RFLPs) (Jermstad et al. 1998; Krutovskii et al. 1998). Linkage maps are a preliminary tool for analyzing the organization of genomes, and can be used as a base map for QTL (quantitative trait locus) identification in forest trees (Jermstad et al. 2001a; Jermstad et al. 2001b; Markussen et al. 2003; Sewell et al. 2000; Sewell et al. 2002), candidate gene mapping (Brown et al. 2003) and comparative mapping between species (Chagne et al. 2003; Krutovsky et al. 2004). Jermstad et al (2001a, 2001b and 2003) developed and employed a linkage map for Douglas-fir to identify QTLs for adaptive traits including bud flush, fall and spring cold hardiness, and growth initiation and cessation. Furthermore, the same map was used for comparative genome mapping with loblolly pine and other conifer species (Krutovsky et al. 2004). Conifer genomes are very large and it is unlikely that they will be sequenced in the near 106 future (Krutovsky et al. 2004). Consequently, this realization has provoked the pursuit of alternative methods for exploring and understanding the relationships between species, and among genes conferring control over traits of interest. Traditional linkage analyses in trees employ marker information from the offspring of a cross to develop linkage maps for the two parents. Subsequently, the maps are then aligned using common markers between parent maps (Chagne et al. 2002; Jermstad et al. 1998; Wu et al. 2000). Although codominant markers are ideal for map alignment, dominant intercross markers can be used for the same purpose. One of the criticisms of linkage mapping is the limitation of the analysis to a single individual or pair of individuals (for a sex-averaged map), which has implications for applications of derived maps. A linkage map has been created for loblolly pine (Pinus taeda L.) integrating segregation data from two outbred, three-generation pedigrees using intercross markers to align maps (Sewell et al. 1999). The resulting consensus linkage map consisted of 357 markers (258 RFLPs, 67 RAPDs and 12 isozymes) and covered ~1359cM over 20 linkage groups. Twelve linkage groups were integrated with map information from all four mapping populations (all four parents), two were specific to their QTL pedigree and six were specific to single-populations (Sewell et al. 1999). The current study presents a consensus AFLP linkage map for coastal Douglas-fir using eight full-sib families from the British Columbia Ministry of Forests second generation progeny test program employing a joint likelihood function (Hu et al. 2004) to calculate average recombination rates and LOD scores across families. This technique facilitates the development of linkage maps for multiple full-sib families using dominant markers and has implications for tree improvement programs, which utilize large 107 numbers of outbred pedigrees. Furthermore, it can be used to integrate genetic information from a large number of genetic backgrounds, which will aid in a more comprehensive analysis of genome organization and QTL effects within a population. Materials and Methods Plant Material Plant material was collected from eight full-sib families from the British Columbia Ministry of Forests second generation progeny test program for coastal Douglas-fir. The families were sampled from four sites in south-western British Columbia. Ten individuals per family were sampled randomly within each site (total of 320 trees). The trees were planted in 1977 in row plots of four trees replicated in four blocks per site. In May of 2004, branches with newly flushing buds were cut and collected from the upper part of the live crown of each individual. Buds were immediately removed from the branches, placed into cryo vials, and frozen in a vapour tank for transport. In the lab, the frozen bud material was stored at -80°C. DNA Isolation DNA was isolated from bud material using a cetyl trimethyl ammonium bromide (CTAB) procedure adapted from (Doyle and Doyle 1987). In short, frozen buds were 108 ground by mortar and pestle under liquid nitrogen. To each sample, 400uL of concentrated buffer base (0.2M hydroxymethylaminoethane [Tris], 0.04M ethylenediaminetetraacetic acid [EDTA] and 2.8M NaCl at pH 8.3) and 400uL of 0.1M CTAB were added and briefly vortexed. The buffer-bud mixture was then incubated at 65°C for one hour with routine shaking every 10 minutes. Samples were spun at 10,000 rpm for 2 minutes in a 4°C centrifuge and the supernatant collected. RNase (10ug) was added to each sample and incubated at 37°C for 30-45 minutes, after which 700uL of chloroform:isoamyl alcohol (24:1) was added and rotamixed for an additional 30 minutes. The mixture was spun for 10 minutes at 10,000 rpm in a 4°C centrifuge and the supernatant collected. Cold isopropanol (400pL) was added to each tube, mixed gently, and placed at -20°C for 30 minutes. The tubes were then spun for 30 minutes at 10,000 rpm in a 4°C centrifuge and the supernatant discarded. The pellet was washed twice with 200uL of ice-cold 70% ethanol and the DNA dried in a speedvac for 5 minutes, re-suspended in lOOuL of sterile water and incubated at 65°C for one hour. DNA from each sample was quantified on a spectrophotometer (Pharmacia Biotech Ultrospec 3000) at 260nm and 280nm, diluted to lOOng/uL and stored at -20°C. AFLP Template Preparation and Reactions The restriction-ligation (RL) protocol employed was adapted from Vos et al. (1995). Restriction double-digests were completed using EcoKUMsel and PstVMsel. Previously, Paglia and Morgante (1998) report that AFLP profiles generated using EcoRI have a low signal to noise ratio and limited number of discernable polymorphic bands in 109 conifers. As a result, both EcoRI and PstI AFLP profdes were generated and compared for polymorphisms and ease of scoring. EcoRI produced an abundance of polymorphic loci, whereas Pst\ was rather monomorphic (data not shown). As a result, EcoRI was used as the frequent cutter to generate AFLP profdes for this study investigating Douglas-fir. Approximately lOOOng of DNA were digested in a lOuL solution of 5 x R L buffer (50mM Tris-HAc pH7.5, 50mM MgAc, 250mM KAc and 25mM DTT), 8.04U of EcoRI, 6.7U of Msel and incubated for 2-3 hrs at 37°C. A 2.5uL solution containing lOmM ATP, 5 x R L buffer, 5pmol/uL of EcoRI adapter, 50pmol/uL of Msel adapter and 0.5U of T4 DNA ligase was added to lOuL of the digested DNA product and incubated for 3 hours at 37°C. After ligation, the RL product was diluted 1:10 with dH^O and stored at -20°C. The adapter sequences employed are as follows: EcoRI Top strand 5' AACGACGACTGCGTACC 3' Bottom strand 3' CTGCTGACGCATGGTTAA 5' Msel Top strand 5' GACGATGAGTCCTGAG 3' Bottom strand 3' TACTCAGGACTCAT 5' Amplification of the DNA template was adapted from Remington et al (1999). Pre-amplification was carried out using EcoRI (E+AC) and Msel (M+CC) primers: EcoRI 5' GACTGCGTACCAATTC 3' Msel 5' GATGAGTCCTGAGTAA 3' A 7.5uL solution containing 15ng of E+AC, 15ng of M+CC, 2mM dNTP mix (contains equal volumes of 0.4mM dATP, dCTP, dGTP and dTTP), lOxPCR buffer (Roche Diagnostics Corp.) and 0.6U Taq polymerase (Roche Molecular Systems) was added to 110 2.5uL of RL product. The thermocycler conditions for PCR amplifications were as follows: 1 min denaturation at 94°C, 28 cycles of 30 sec at 94°C, 30 sec at 60°C and 60 sec at 72°C followed by a 5 min extension at 72°C. The pre-amplification product was diluted 1:40 with dH 20 prior to final amplification and stored at -20°C. Final selective amplification was conducted using 10 primer pair combinations of E+3 and M+4 selective nucleotides (selected from 92 screened primer combinations). A final solution of lOuL containing 2.5uL of pre-amplification product, 0.25pmol of M13 labelled primer, 2.52ng of E+3 tailed primer, 2.52ng of M+4 primer, 2mM dNTP mix, lOxPCR buffer (Roche Diagnostics Corp.), and 0.6U Taq polymerase was placed in a thermocycler for final amplification. The Ml3 labelled primer (LiCor) and final amplification primers have the following sequences: M13 labelled primer 5' CACGACGTTGTAAAACGAC 3' EcoRI 5' CACGACGTTGTAAAACGACGACTGCGTACCAATTC 3' Msel 5' GATGAGTCCTGAGTAA 3' PCR amplifications were carried out as follows: 1 min denaturation at 94°C, 3 cycles of 30 sec at 94°C, 30 sec at 65°C and 60 sec at 72°C, 12 cycles of 30 sec at 94°C, 30 sec at 65°C (-0.7°C/cycle) and 1 min at 72°C, 23 cycles of 30 sec at 94°C, 30 sec at 56°C and 60 sec at 72°C and a 5 min extension step at 72°C. After final amplification, 3 uL of formamide loading buffer was added to each sample and stored in aluminium foil at -20°C. I l l Detection and Scoring of AFLP Fragments The AFLP products were resolved using a LiCor 4200 autosequencer with 25cm plates. The gel was prepared with a 30mL solution containing 7% Long Ranger polyacrylamide (FMC BioProducts), 7M urea and 5><TBE (0.45M Tris, 0.45M boric acid and 0.01M EDTA). Ammonium persulfate (200uL) was added to the urea solution and filtered. TEMED (N,N,N,N - tetramethyl-ethelenediamine; 15uL) was added to the filtered solution to begin the solidification process. Forty-eight lane gels were used with luL of sample in each well. Each gel contained one family with parents replicated three times (46 individuals) and two LRD-labelled molecular-weight markers (LiCor; size standard LRDye™) as standards. Two primer sets were loaded per gel, each containing a different Ml3 primer channel (700nm or 800nm). Electrophoresis was carried out using lxTBE running buffer under the following parameters: motor speed of 4, 2000V, 35mA 70W and 50°C plate temperature. Each gel was run for four hours and images saved as TIFF files. A total of 92 primer combinations were screened using 12 individuals per primer set (three individuals from each family selected from Adam River, Gold River and Squamish River). Five primer combinations were screened on each gel image separated by IRD-labelled standards. All E+3/M+4 and some E+3/M+5 combinations with an E+AC and M+CC pre-amplification were screened (Table 3.1). Primer pairs were first screened based on the number of polymorphic loci within families and, finally, on the ease of scoring. Ten primer pairs were selected for final analysis. 112 Table 3.1: Number of polymorphic loci generated by each primer combination during primer screening. Eco + Mse + ACA ACC ACG ACT Total CCAA 19 0 20 2 41 CCAC 20 5 14 6 45 CCAG 38 28 34 18 118 CCAT 0 1 2 9 12 CCCA 4 1 17* 17 39 CCCC 26 15 25 13 79 CCCG 31 19 28* 51 129 CCCT 10 0 0 2 12 CCGA 35 9 22* 8 74 CCGC 32 37* 9 9 87 CCGG 48* 39 20* 24 131 CCGT 52* 41 22* 7 122 CCTA 8 7 3 8 26 CCTC 18 9 23* 14 64 CCTG 48 17 30* 20 115 CCTT 24 1 24 6 55 CCAGA 26 18 28 6 78 CCAGC 34 39 24 15 112 CCAGG 15 16 20 19 70 CCAGT 17 10 28 31 86 CCCAA 15 30 0 21 66 CCGAA 17 16 19 25 77 CCGGA 32 24 28 23 107 Total 569 382 440 354 ' 1745 * primer combinations selected for final analysis Reproducibility tests were performed using 10 individuals replicated four times on each gel with the same primer combination. Two primer pairs were used and the data analyzed for missing or additional bands within replicated individuals for a given locus. Scoring was completed by eye using the Saga (Generation 2) software program. Loci that segregated in a 1:1 ratio (backcross markers) and 3:1 ratio (intercross markers) were scored for each family. If a locus was detected in one family, the same locus was assessed and scored, regardless of segregation, in every family. Patterns of segregation were easily visible on the gel image and the existence and segregation pattern (backcross, 113 intercross, no segregation or not existent) of each locus was recorded from the image. The data was recorded as present (+) or absent (-) by the computer software which ultimately generated reports for each family. The segregation assessment recorded from the gel image was used to extract backcross markers from the data for each family. The markers were imported into Joinmap (Van Ooijen and Voorrips 2001) and checked for segregation distortion. Loci with severe segregation distortion were reviewed by eye and removed if deemed unreliable. Linkage Analysis Linkage analysis was conducted on backcross markers. Average recombination rates and LOD scores for all pairwise combinations of markers across families were calculated by Dr. Kermit Ritland (in-house written program in Fortran 95) using a joint likelihood function (Hu et al. 2004). All pairwise data with LOD > 0 were input into Joinmap (Van Ooijen and Voorrips 2001) as a pairwise data fde for linkage mapping. Grouping was carried out from LOD 2 to 10 with a step of one LOD. Groups with three or more markers came together at LOD of 2 and 3. All groups with > 3 markers were retained for mapping. Maps were generated using the Kosambi mapping function with recombination rates < 0.45 and LOD > 0 with a ripple performed after adding each locus and a jump threshold of 5. In some cases, due to the lack of adequate REC and LOD information for some groups, complete maps could not be created. In these cases, markers within these groups were split based on linkage and two or more maps were generated. For these situations, linkage between smaller groups and larger groups was 114 assessed. When there was evidence that smaller groups were linked within larger groups, the smaller groups were removed from further analyses. A third round of marker addition was permitted, however, only maps from the first and second rounds were accepted as reliable map orders. Linkage groups with > 4 loci were retained as major linkage groups to represent the Douglas-fir genome. Total map distance was calculated by adding the total map distance covered by each linkage group. The average distance between markers was calculated by dividing the total map distance by the number of intervals in the map. Marker Distribution and Family Contributions AFLP markers should resemble a random sample from the genome (Vos et al. 1995). The distribution of AFLP markers throughout the linkage map was tested using three methods. The first method (Remington et al. 1999) examined marker clustering by testing each linkage group for the expected number of markers (ki)'. A J = mGj [3.1] where m is the total number of markers and Gj is the map distance of each linkage group adjusted for chromosome ends as Gi = M; + 2s (M; = map distance between terminal markers and s = average marker spacing for the linkage map). The observed number of markers per group can then be tested using the cumulative Poisson distribution with parameter \ . Since this is a two-tailed test, p-values < 0.025 are significant at a = 0.05. 115 The next two methods were adopted from Cervera et al. (2001). The second method uses a Pearson correlation to estimate the relationship between linkage group length and marker number. The Corr procedure of SAS (SAS Version 9.1) was used to estimate the correlation coefficient and associated p-value. The final method employs an estimate of the coefficient of dispersion (CD) to confirm that the observed frequency distribution of markers follows a Poisson distribution (Cervera et al. 2001). Each linkage group was divided into lOcM segments. If a segment at the end of a linkage group was < 7.5cM, the interval was disregarded. The number of markers in each interval was then counted. The CD was calculated as the ratio of the variance over the mean. A value greater than one indicates clustering, while a value less than one suggests a lower number of markers than expected in any given lOcM interval. Family contribution to the linkage map was also tested using a contingency table with rows equal to the number of families (8), and columns equal to the number of linkage groups (19). The number of markers provided by each family to each linkage group was then determined. In order for marker information to contribute to the linkage data of a linkage group, two markers must be present from a common family. Therefore, if a family only contributed one marker to a linkage group, it was disregarded. Using this test, we can assess: i) if family and linkage group are independent, ii) if family contribution to each linkage group is the same as expected, and iii) if individual family contribution is the same as expected over the entire linkage map. 116 Results AFLP Polymorphisms Primer screening revealed substantial polymorphic loci for E+3/M+4 primer combinations (Table 3.1). Combinations of markers with E+ACG, E+ACC and E+ACA produced the most readily scorable images and were employed to produce profiles for the final analysis. Profiles generated using primer combinations with E+ACT were generally smeared and difficult to score. E+3/M+5 primer combinations also produced substantial numbers of polymorphic loci, however, their profiles were difficult to score and were not used for the final analyses. For the four Eco primers used, E+ACA generated the most polymorphic loci followed by E+ACG, while both E+ACC and E+ACT produced the lowest number of polymorphic loci. Of the M+4 primers, those with guanine at their 3' end generally produced the most polymorphic loci. In total there were 1745 polymorphic loci detected after the initial screening. Ten primer combinations were then chosen to produce loci for linkage mapping based on the number of polymorphic loci and profile clarity. On average, primer combinations using E+ACA produced more polymorphic loci (Table 3.2), and were substantially easier to score. For each primer combination, there were two to four times more backcross markers present than intercross markers. Some families produced more polymorphic markers than others (Table 3.3). For example, families 2 and 38 provided 117 Table 3.2: Backcross, intercross and the total number of markers generated by each primer combination in the final AFLP analysis. Eco Mse Backcross Intercross Total ACA CCGG 145 39 184 CCGT 167 54 221 ACC CCGC 85 24 109 ACG CCCA 100 31 131 CCCG 53 12 65 CCGA 79 29 108 CCGG 109 44 153 CCGT 104 35 139 CCTC 108 36 144 CCTG 72 36 108 the most polymorphic loci, which may reflect profile clarity for these families. Similarly, families 2, 38, 92 and 151 produced the cleanest profiles, while families 7, 26, 62 and 75 produced profiles that were more difficult to score. Loci that were difficult to score, or for which segregation was obscure, were usually eliminated from further analysis. Map Results A total of 531 markers that segregate in a 1:1 ratio from eight full-sib families were used to calculated average pairwise LOD and REC data. The initial grouping with Joinmap was able to group 244 markers at LOD thresholds of 2 and 3. The final map contained 120 markers mapped on 19 linkage groups (Figure 3.1). On average, there were 118 Table 3.3: Backcross, intercross and total number of markers generated for each full-sib family. Family Backcross Intercross Total 2 164 36 200 7 116 43 159 26 125 49 174 38 147 44 191 62 139 46 185 75 115 29 144 92 109 49 158 151 107 44 151 6.3 markers per group with an average of 9.3cM between markers. The total map distance covered by the 19 linkage groups was 938.6cM. Another 63 markers were mapped in groups of three on 21 linkage groups that covered a distance of 505cM with an average distance of 12.0cM between markers. A total of 61 markers that were initially grouped were left unmapped. Marker Distribution and Family Contributions There was no evidence of marker clustering. The cumulative Poisson test suggests that markers on linkage groups approximate a Poisson distribution (Table 3.4). There is a significant positive correlation between marker number and marker size (0.57; p = 0.0108) and the coefficient of dispersion (0.91) suggests lower marker numbers than expected in some lOcM segments (Table 3.5). The test for family contribution to the linkage map suggests that families and linkage groups are not independent (p < 0.0001). That is, some families contribute more 119 - ACACCOG_0145 0.0-O 6.9-7.8-- ACACCGG_0075 -ACACCGT_0136 17.2 20.9 23.1 23.2 28.4" ACGCCGG_0297 ACGCCTC_0229 ACGCCTC_0249 ACACCGG_0296 - ACGCCGG_0296 46.9" 50.1 -64.1 -67.9" 12.1 -18.2 -21.6" 24.7" 26.8" 31.6" 33.1 " 35.6" -ACCCCGC_0176 - ACGCCGG_0245 - A C G C C G A J H 8 8 -ACGCCTG_0376 - A C C C C G C J H 5 3 2 - A C A C C G T J 1 7 8 - ACGCCGT_0286 -ACGCCGT_0178 - ACACCGTJJ285 - A C A C C G T J 2 8 7 - ACGCCGG_0325 - ACACCGG_0326 - ACACCGG.0327 11.9-18.4-27.3" -ACACCGG_0245 3 5 6 --ACCCCGC_0144 54.3-54.7" 15.4" 20.5-22.3-23.4-36.7-38.1 -o.o-24.1" 31.3--ACGCCTC_0290 -ACACCGG_0185 - A C A C C G T J H 3 1 -ACGCCGA_0383 -ACACCGT_0401 -ACACCGG_0289 -ACGCCGG_0288 -ACGCCGA_0528 -ACGCCGG_0191 -ACGCCCA_0569 - ACACCGG_0191 - ACGCCCG_0445 - ACACCGG_0258 -ACACCGT_0235 -ACGCCCA_0652 - ACGCCGT.0466 - A C G C C T C _ 0 2 6 5 5 - ACGCCGT_0556 - A C A C C G T _ 0 5 5 6 o.o-4.5" 55.0-- ACGCCGA_0277 0.0 -- ACGCCCA_0223 -ACGCCCA_0127 - ACGCCTC_0306 -ACGCCTG_0115 - ACGCCCA_0408 - ACGCCCA.0235 15.5-16.8" 46.3-47.6" 49.8" 58.2" 63.6" 76.8" 87.1" o.o-15.8" 19.2 -- ACGCCGT_0635 0.0" 8 - ACGCCCA_0255 -ACCCCGC_0217 - ACGCCCG.0255 -ACGCCTC_0143 25.1 " 30.4-0.0-- ACGCCTC_0297 3 7 . 0 ^ N ^ A C A C C G T J 3 9 1 37.1 ACGCCGT_0392 32.4-35.2-- ACGCCCA_0252 - ACACCGG.0237 - ACACCGT_0597 - ACGCCCA_0063 -ACGCCTC_0356 -ACACCGT_0101 -ACGCCCG_0418 -ACACCGG_0153 -ACGCCGG_0153 - ACGCCTG_0302 10 - ACGCCTC_0268 -ACACCGT_0371 - ACGCCGT_0371 26.9" 27.8" 42.2 -ACGCCCA_0439 44.7 -ACGCCTC_0146 - ACGCCCA_0343 47.1 -50.0" 51.9" 55.5" 61.2-o.o-15.2-16.5-55.5" 0.0-3.6-4.0" - ACGCCGA_0370 11 o.o-7.6" !5.5--ACACCGT.0115 -ACGCCCG_0127 -ACGCCCA_0614 32.7-32.8" - A C A C C G G - A C G C C G G . - A C G C C G G . - A C A C C G G . - A C G C C C A , - A C G C C G A. 0316 0315 .0338 0339 ,0060 0294 - ACGCCCAJJ2I3 0.0" -ACGCCTC_0133 10.1 -- ACGCCGG_0090 14 - ACCCCGC.0277 - ACGCCCA_0345 3 4 6 -39.9" -ACACCGG_03I0 -ACGCCGG_0311 17 -ACGCCCG_0524 - ACGCCGA_0072 50.1 — A C G C C G T _ 0 5 9 5 0.0 -0.0 - ~ 5 B x ' ACACCGT_0609 0.1 -nP' ACGCCGT_0608 12 13.3--ACACCGG_0138 - ACGCCGG_0238 -ACGCCGA_0299 -ACCCCGC.0131 - ACCCCGC_0327 - ACGCCGG_0408 - ACACCGG_0407 - ACGCCCA_0237 19.8" 15 13 - A C C C C G C J M 3 0 29.1" 35.1" -ACGCCTG_0085 44.3" -ACGCCTC_0160 0.0" -ACGCCTC_0591 16 16 8 • -ACGCCTG_0154 25.2--ACACCGG_0181 -ACGCCGG_0180 -ACGCCTC_0189 -ACGCCGA_0121 -ACGCCTC_0144 -ACGCCGT_0194 - ACACCGT_0195 18 -ACACCGG_0352 -ACGCCGA_0066 - ACGCCGG_0407 -ACACCGG_0408 19 - A C C C C G C J M 6 9 Figure 3.1: AFLP linkage map for integrated segregation data from eight full-sib families of coastal Douglas-fir. Marker names are the E+3/M+4 primer combination and size. Map distances are in cM estimated by the Kosambi mapping function. Table 3.4: Marker density per linkage group. The expected number of markers (XJ is equal to the observed number of markers (mj if the Poisson two-tailed p-value is > 0.025. L G Mi Q h P-Value (cM) (cM) 1 14 67.9 78.3 7.04 0.994 2 9 35.6 44.5 4.00 0.992 3 8 72.0 92.6 8.32 0.547 4 7 38.1 50.7 4.56 0.908 5 4 31.3 52.2 4.70 0.495 6 7 107.7 143.6 12.91 0.056 7 4 37.1 61.9 5.56 0.348 8 11 87.1 104.6 9.40 0.762 9 5 30.4 45.5 4.09 0.770 10 4 35.2 58.6 5.27 0.394 11 11 61.2 73.4 6.60 0.963 12 5 55.5 83.2 7.48 0.243 13 4 15.7 26.2 2.36 0.909 14 6 50.1 70.1 6.31 0.557 15 5 44.5 66.7 6.00 0.446 16 4 35.1 58.4 5.25 0.397 17 4 39.9 66.5 5.98 0.288 18 4 .44.3 73.8 6.63 0.209 19 4 50.2 83.6 7.52 0.131 Sum 120 938.6 1334.4 120 LG = linkage group m, = number of markers M; = actual linkage group length G; = inferred linkage group length Xj = expected number of markers P-Value = Poisson two-tailed p-value or less to linkage data than expected in some linkage groups (Table 3.6). Families 7, 26, 62 and 75 occur throughout the linkage map as expected under the chi squared distribution, while families 2, 38, 92 and 151 occur more than expected in some linkage groups. Major contributions to the individual chi squared statistics occur from linkage 121 Table 3.5: Frequency data and results of the coefficient of dispersion analysis for marker distribution. Frequency Class Frequency Markers Proportion of (Markers/Interval) Markers 0 21 0 0 1 44 44 0.44 2 16 32 0.32 3 7 21 0.21 4 1 4 0.04 Total 89 101 Mean 1.13 StdDev 1.04 CD* 0.91 Markers cluster if CD > 1 groups, which are made of markers derived solely from one of these four families. Individual linkage groups may contain markers provided by several families, however, in most cases, the linkage contribution is biased to a small number of families, often one or two. For example, linkage group 7 is composed of markers derived from family 2, whereas linkage group 1 contains markers from six of the eight families, but is biased with more markers being contributed by families 2, 62 and 75 (Table 3.6). 122 Table 3.6: Observed and expected (in brackets) number of markers contributing to linkage from each family. Linkage Group Family 1* 2 3 4* 5 6 7* 8* 9* 10* 11 12* 13* 14* 15* 16* 17* 18* 19* 2* 8 0 0 7 3 0 4 0 5 0 4 0 0 6 5 0 4 0 0 (6.4) (4.0) (3.4) (3.0) (13) (4.5) (0.9) (5.1) (1.5) (1.3) (4.9) (1.9) (0.9) (1.7) (1.5) (0.9) (0.9) (0.9) (1.3) 7 3 0 0 2 0 2 0 2 2 2 0 0 0 2 2 0 0 0 0 (2.4) (1.5) (1.3) (1.1) (0.5) (1.7) (0.3) (1.9) (0.6) (0.5) (1.8) (0.7) (0.3) (0.6) (0.6) (0.3) (0.3) (0.3) (0.5) 26 2 4 4 3 3 3 0 2 0 0 2 0 0 0 0 0 0 0 2 (3.5) (2.2) (1.9) (1.6) (0.7) (2-4) (0.5) (2.8) (0.8) (0.7) (2.7) (1.0) (0.5) (0.9) (0.8) (0.5) (0.5) (0.5) (0.7) 38* 5 5 5 0 0 4 0 3 0 0 4 4 4 0 0 4 0 0 0 (5.3) (3.3) (2.8) (2.5) (1.1) (3.7) (0.7) (4.2) (1.2) (1.1) (4.0) (1.6) (0.7) (1.4) (1.2) (0.7) (0.7) (0.7) (1.1) 62 7 2 2 2 0 2 0 6 0 0 2 0 0 0 0 0 0 0 0 (3.2) (2.0) (1.7) (1.5) (0.6) (2.2) (0.4) (2.6) (0.7) (0.6) (2.4) (1.0) (0.4) (0.9) (0.7) (0.4) (0.4) (0.4) (0.6) 75 5 2 0 0 0 0 0 4 0 0 0 0 0 0 0 0 0 0 0 (1.5) (1.0) (0.8) (0.7) (0.3) (1.1) (0.2) (1.2) (0.4) (0.3) (1.2) (0.5) (0.2) (0.4) (0.4) (0.2) (0.2) (0.2) (0.3) 92* 0 6 5 0 0 7 0 0 0 4 7 5 0 0 0 0 0 4 0 (5.3) (3.3) (2.8) (2.5) (1.1) (3.7) (0.7) (4.2) (1.2) (1.1) (4.0) (1.6) (0.7) (1.4) (1.2) (0.7) (0.7) (0.7) (1.1) 151* 0 0 0 0 0 3 0 7 0 0 4 0 0 0 0 0 0 0 4 (2.5) (1.3) (1.6) (1.2) (1.9) (2.0) (1.8) (0.8) (0.6) (0.7) (0.6) (0.3) (0.3) (0.5) (0.3) (0.3) (0.5) (0.5) (0.3) * significantly different from expected at a = 0.05. Discussion Printer Screening and Segregation AFLP markers are very reproducible and the technique is capable of producing large numbers of polymorphic loci per PCR reaction. Primer combinations of E+3/M+4 with an E+2/M+2 pre-amplification gave the best results. The frequent cutter Pstl was used to compare profdes and polymorphisms with EcoRI, and found to lack polymorphic loci in Douglas-fir. Although Paglia and Morgante (1998) suggest that Pstl is less prone to cutting within highly repetitive regions of the conifer genome, this restriction enzyme did not generate profiles of the same quality as EcoRI in the present study. Since conifer genomes contain large amounts of repetitive DNA, Pstl may result in profiles with higher signal to noise (Paglia and Morgante 1998). This analysis was more concerned with the ease of scoring than the number of polymorphic bands produced. Since the signal to noise ratio between Pstl and EcoRI profiles was similar, priority was given to maximizing the number of polymorphic markers produced per reaction. All primer combinations given the pre-amplification of E+AC/M+CC were screened for polymorphisms and ease of scoring (Table 3.1). Primer combinations using E+ACA produced the most polymorphic loci and were the easiest to score. Msel primers with guanine as the 3' selective nucleotide gave the best results. This is consistent with Remington et al. (1999) who report optimal results from primer combinations with one CpG unit in the selective region of EcoRI or Msel primers and Paglia and Morgante (1998) who report more polymorphic loci in Pstl profiles with CpG units within the 124 selective region of final amplification primers. However, our results suggest that primer combinations with E+ACG produce significant polymorphisms but are difficult to score. The best combinations for ease of scoring and polymorphism production were E+ACA with M+CCGG and M+CCGT (Table 3.2). Several E+3/M+5 primer combinations were screened but produced profiles that had darker backgrounds and were difficult to discern. Mismatches can occur when three or more selective nucleotides are added to the 3' end of a primer resulting in inconsistent results and genotyping error. The quality of DNA for AFLP generation is very important. AFLPs were attempted using one year old Douglas-fir needles. Using old tissue, we were unable to generate useable AFLP profiles (data not shown). This was most likely due to DNA quality and the presence of extractives, and not to DNA quantity. Newly flushing bud material was optimal for isolating clean DNA. The presence of contaminants in the DNA can affect the restriction digest and PCR amplifications necessary to generate AFLP profiles. DNA quality between families appeared to be variable (Table 3.3), even when using freshly flushing buds. Fore example, families 2 and 38 consistently provided the cleanest AFLP profiles and were easy to score. Consequently, they also contributed the most polymorphic markers to the linkage analysis, which likely reflects the number of easily scorable markers rather than heterozygosity. 125 Map Comparisons Many linkage maps have been created for tree species using AFLP markers. Two of the most comprehensive maps were generated for loblolly pine and maritime pine (Pinuspinaster Ait.) (Chagne et al. 2002; Remington et al. 1999). Both maps resulted in 12 linkage groups which correspond to the number of haploid chromosomes in pine. Several other maps have been generated for conifers (Travis et al. 1998; Yin et al. 2003) and hardwoods (Cervera et al. 2001; Scalfi et al. 2004; Wu et al. 2000; Zhang et al. 2004) using AFLPs. Generally, the conifer maps result in greater than 12 linkage groups, more than expected for pines. This is consistent with the map generated in the present study. Nineteen major linkage groups were generated for Douglas-fir, which is more than the expected 13 haploid chromosomes. It seems that an excessive number of AFLP markers must be generated in order to accurately depict the actual chromosomes. However, although a saturated map will more closely approximate the actual size and shape of the chromosomes, a map with lower density and more linkage groups has the potential to cover the same genome distance. Travis et al. (1998) report that 10 primer combinations are enough to cover 72% of the genome at a density of lOcM between markers, and that 10-20 primer combinations are optimal for QTL analysis. Hu et al (2004) suggest that 50 individuals from ten crosses is sufficient to generate linkage maps using the joint likelihood method. In this study, we used 10 primer combinations from eight families with 40 individuals per family to generate a linkage map for future QTL analysis. Although our map resulted in more than the expected number of linkage groups, it is expected to cover an adequate amount of the genome for QTL mapping. 126 One of the most recent linkage maps generated for coastal Douglas-fir employed both dominant (RAPDs) and codominant markers (RFLPs) (Jermstad et al. 1998), and has been the basis for QTL analysis of adaptive traits (Jermstad et al. 2003; Jermstad et al. 2001a; Jermstad et al. 2001b) and comparative mapping with loblolly pine (Krutovsky et al. 2004). The map presented in this paper is of similar size and density to the one generated by Jermstad et al. (1998), who created a linkage map with 141 markers distributed on 17 major linkage groups covering a distance of l,062cM and a density of 7.5cM between markers. The present AFLP map is composed of 120 markers on 19 linkage groups covering 938.6cM with an average of 9.3cM between markers. The advantage of this linkage map over those generated in other studies is the integration of segregation data from eight full-sib families. The final map is a synthesis of data from each of the families and represents a more diverse part of the population. Rather than relying on a single cross or family to generate a linkage map, using this technique, a more significant component of the genetic variation within the population can be assessed and mapped, generating a more representative illustration of the organization of the genome of a population. However, the family contribution data in Table 3.6 suggests that several linkage groups are biased for specific families. These results have implications for potential applications of the map. The family specific linkage groups are expected to merge with other groups given more AFLP markers and individuals to give a more accurate depiction of the Douglas-fir genome. 127 Unmapped Loci A large number of markers were left unmapped in this analysis. The linkage map was generated using segregation data from eight full-sib families; some markers did not segregate or appear in more than one family, while others occur in multiple families. In order to calculate pairwise data for two markers, the markers must segregate in a common family (for example, each marker must segregate in family 2 and 7 in order to calculate average pairwise LOD and REC statistics). In order for Joinmap to order markers in a group, each marker must show linkage to at least two other markers. In some instances, a marker within a linkage group segregated in multiple families and showed significant linkage to markers that occurred in only one family. If markers that flank the multiple-family marker do not occur in a common family with the markers that occur in a single family, single family markers were left unmapped. This occurred in many instances and can account for the large number of markers left unmapped after grouping. It may also explain the large number of loci left ungrouped after calculating pairwise statistics. It has been suggested that increasing the number of markers may improve map density and reduce the number of linkage groups by bringing individual groups together. Jermstad et al. (1998) attempted to identify linkage among groups by adding more markers, but found that no improvements occurred. In the current study, it may be more advantageous to add more families in order to improve map density. Adding more families should increase the number of loci that segregate in multiple families, and will increase the chance that linked markers will segregate in common families. Given 128 adequate marker segregation in all families, this technique would provide a true average linkage map for the study population. The linkage map presented in this paper was initially grouped at LOD thresholds of 2 and 3 which are not uncommon for trees (Cervera et al. 2001; Chagne et al. 2002; Jermstad et al. 1998; Scalfi et al. 2004; Travis et al. 1998; Wu et al. 2000). To improve the confidence of the map by increasing the minimum LOD threshold for grouping, more offspring would be required per family. After grouping, all available pairwise data was needed to order markers onto linkage groups. This meant using all pairwise data with LOD values greater than or equal to zero and with a recombination rate of 0.45 or less. Low LOD values indicate that two markers are either far apart on a chromosome or are not linked. These markers may be located on a common linkage group, but at distances to allow significant recombination. The LOD and REC data, although sometimes low, was necessary to place some single-family markers on the final map. Applications Linkage maps are a necessary precursor to QTL analyses, comparative genomics and candidate gene mapping. Generally, a specific set of parents are chosen to represent the genome of a particular species or population, and can serve as a backbone for comparing QTL activity in different genetic backgrounds. The use of a single family to represent a population or species is, however, limiting. Use of these maps in QTL analyses, comparative genomics and candidate gene mapping presents a very limited 129 picture of the genome of the population and eliminates variation that may occur for QTLs within the species. By integrating multiple families, a more comprehensive map can be generated and used to represent a larger portion of the population. This has the potential to confer a more accurate picture of the number and effect of QTLs acting in the population. This type of analysis has implications for the tree improvement sector. Tree improvement programs are concerned with the improvement of tree characteristics such as volume and wood quality for a group of selected individuals that will become the parents of the next breeding population. Testing these parents requires the assessment of progeny resulting in large scale, full-sib family field trials. By adopting this type of analysis, improvement programs can create "population" linkage maps to assess QTL effects in the progeny of selected trees. 130 References Aubry, C. A., Adams, W.T., and Fahey, T.D. 1998. Determination of relative economic weights for multitrait selection in coastal Douglas-fir. Canadian Journal of Forest Research. 28(8): 1164-1170. Brown, G.R., Bassoni, D.L., Gill, G.P., Fontana, J.R., Wheeler, N.C, Megraw, R.A., Davis, M.F., Sewell, M.M., Tuskan, G.A., and Neale, D.B. 2003. Identification of quantitative trait loci influencing wood property traits in loblolly pine (Pinus taeda L.). III. QTL verification and candidate gene mapping. Genetics. 164 (4): 1537-1546. Cabrita, L.F., Aksoy, U., Hepaksoy, S., and Leitao, J.M. 2001. Suitability of isozyme, RAPD and AFLP markers to assess genetic differences and relatedness among fig (Ficus carica L.) clones. Scientia Horticulturae. 87 (4): 261-273. Cervera, M.T., Remington, D., Frigerio, J.M., Storme, V., Ivens, B., Boerjan, W., and Plomion, C. 2000. Improved AFLP analysis of tree species. Canadian Journal of Forest Research. 30 (10): 1608-1616. Cervera, M.T., Storme, V., Ivens, B., Gusmao, J., Liu, B.H., Hostyn, V., Van Slycken, J., Van Montagu, M., and Boerjan, W. 2001. Dense genetic linkage maps of three Populus species (Populus deltoides, P. nigra and P. trichocarpa) based on AFLP and microsatellite markers. Genetics. 158 (2): 787-809. Chagne, D., Brown, G., Lalanne, C , Madur, D., Pot, D., Neale, D., and Plomion, C. 2003. Comparative genome and QTL mapping between maritime and loblolly pines. Molecular Breeding. 12 (3): 185-195. Chagne, D., Lalanne, C , Madur, D., Kumar, S., Frigerio, J.M., Krier, C , Decroocq, S., Savoure, A., Bou-Dagher-Kharrat, M., Bertocchi, E., Brach, J., and Plomion, C. 2002. A high density genetic map of maritime pine based on AFLPs. Annals of Forest Science. 59 (5-6): 627-636. Chantre, G., Rozenberg, P., Baonza, V., Macchioni, N., Le Turcq, A., Rueff, M., Petit-Conil, M., and Heois, B. 2002. Genetic selection within Douglas fir (Pseudotsuga menziesii) in Europe for papermaking uses. Annals of Forest Science. 59 (5-6): 583-593. 131 Doyle, J.J., and Doyle, J.L. 1987. A rapid DNA isolation procedure for small quantities of fresh tissue. Phytochemical Bulletin. 19: 11-15. Gartner, B.L., North, E.M., Johnson, G.R., and Singleton, R. 2002. Effects of live crown on vertical patterns of wood density and growth in Douglas-fir. Canadian Journal of Forest Research. 32 (3): 439-447. Hesterman, N.D., and Gorman, T.M. 1992. Mechanical-properties of laminated veneer lumber made from interior Douglas-fir and lodgepole pine. Forest Products Journal. 42(11-12): 69-73. Hu, X.-S., Goodwillie, C , and Ritland, K.M. 2004. Joining genetic linkage maps using a joint likelihood function. Theoretical and Applied Genetics. 109: 996-1004. Jermstad, K.D., Bassoni, D.L., Jech, K.S., Ritchie, G.A., Wheeler, N.C., and Neale, D.B. 2003. Mapping of quantitative trait loci controlling adaptive traits in coastal Douglas fir. III. Quantitative trait loci-by-environment interactions. Genetics. 165 (3): 1489-1506. Jermstad, K.D., Bassoni, D.L., Jech, K.S., Wheeler, N.C., and Neale, D.B. 2001a. Mapping of quantitative trait loci controlling adaptive traits in coastal Douglas-fir. I. Timing of vegetative bud flush. Theoretical and Applied Genetics. 102 (8): 1142-1151. Jermstad, K.D., Bassoni, D.L., Wheeler, N.C., Anekonda, T.S., Aitken, S.N., Adams, W.T., and Neale, D.B. 2001b. Mapping of quantitative trait loci controlling adaptive traits in coastal Douglas-fir. II. Spring and fall cold-hardiness. Theoretical and Applied Genetics. 102: 1152-1158. Jermstad, K.D., Bassoni, D.L., Wheeler, N.C., and Neale, D.B. 1998. A sex-averaged genetic linkage map in coastal Douglas-fir (Pseudotsuga menziesii [Mirb.] Franco var 'menziesii') based on RFLP and RAPD markers. Theoretical and Applied Genetics. 97: 762-770. Jones, C.J., Edwards, K.J., Castaglione, S., Winfield, M.O., Sala, F., van de Wiel, C , Bredemeijer, G., Vosman, B., Matthes, M., Daly, A., Brettschneider, R., Bettini, P., Buiatti, M., Maestri, E., Malcevschi!, A., Marmiroli, N., Aert, R., Volckaert, G., Rueda, J., Linacero, R., Vazquez, A , and Karp, A. 1997. Reproducibility testing of RAPD, AFLP and SSR markers in plants by a network of European laboratories. Molecular Breeding. 3 (5): 381-390. 132 Koshy, M.P., and Lester, D.T. 1994. Genetic-Variation of Wood Shrinkage in a Progeny Test of Coastal Douglas-Fir. Canadian Journal of Forest Research. 24 (8): 1734-1740. Krutovskii, K.V., Vollmer, S.S., Sorensen, F.C., Adams, W.T., Knapp, S.J., and Strauss, S.H. 1998. RAPD genome maps of Douglas-fir. Journal of Heredity. 89 (3): 197-205. Krutovsky, K.V., Troggio, M., Brown, G.R., Jermstad, K.D., and Neale, D.B. 2004. Comparative Mapping in the Pinaceae. Genetics. 168: 447 - 461. Loo-Dinkins, J.A., and Gonzalez, J.S. 1991. Genetic control of wood density profile in young Douglas-Fir. Canadian Journal of Forest Research. 21 (6): 935-939. Loo-Dinkins, J.A., Ying, C.C., and Hamm, E.A. 1991. Stem volume and wood relative density of a nonlocal Douglas-fir provenance in British-Columbia. Silvae Genetica. 40 (1): 29-35. Markussen, T., Fladung, M., Achere, V., Favre, J.M., Faivre-Rampant, P., Aragones, A , Perez, D.D., Harvengt, L., Espinel, S., and Ritter, E. 2003. Identification of QTLs controlling growth, chemical and physical wood property traits in Pinus pinaster (Ait.). Silvae Genetica. 52 (1): 8-15. Nigh, G.D., Ying, C.C., and Qian, H. 2004. Climate and productivity of major conifer species in the interior of British Columbia, Canada. Forest Science. 50 (5): 659-671. Paglia, G., and Morgante, M. 1998. PCR-based multiplex DNA fingerprinting techniques for the analysis of conifer genomes. Molecular Breeding. 4: 173-177. Remington, D.L., Whetten, R.W., Liu, B.-H, and O'Malley, D.M. 1999. Construction of an AFLP genetic map with nearly complete genome coverage in Pinus taeda. Theoretical and Applied Genetics. 98: 1279-1292. SAS. Version 9.1. SAS System for Windows. Copyright (c) 2002 - 2003 SAS Institute Inc. SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc., Cary, NC, USA. 133 Scalfi, M., Troggio, M., Piovani, P., Leonardi, S., Magnaschi, G., Vendramin, G.G., and Menozzi, P. 2004. A RAPD, AFLP and SSR linkage map, and QTL analysis in European beech (Fagus sylvatica L.). Theoretical and Applied Genetics. 108 (3): 433-441. Sewell, M.M., Bassoni, D.L., Megraw, R.A, Wheeler, N.C., and Neale, D.B. 2000. Identification of QTLs influencing wood property traits in loblolly pine (Pinus taedaL.). I. Physical wood properties. Theoretical and Applied Genetics. 101 (8): 1273-1281. Sewell, M.M., Davis, M.F., Tuskan, G.A, Wheeler, N.C., Elam, C.C., Bassoni, D.L., and Neale, D.B. 2002. Identification of QTLs influencing wood property traits in loblolly pine (Pinus taeda L.). LI. Chemical wood properties. Theoretical and Applied Genetics. 104 (2-3): 214-222. Sewell, M.M., Sherman, B.K., and Neale, D.B. 1999. A consensus map for loblolly pine (Pinus taeda L.). I. Construction and integration of individual linkage maps from two outbred three-generation pedigrees. Genetics. 151 (1): 321-330. St.Clair, J.B. 1994. Genetic variation in tree structure and its relation to size in Douglas-fir. I. Biomass partitioning, foliage efficiency, stem form, and wood density. Canadian Journal of Forest Research. 24 (6): 1226-1235. Travis, S.E., Ritland, K., Whitham, T.G., and Keim, P. 1998. A genetic linkage map of Pinyon pine (Pinus edulis) based on amplified fragment length polymorphisms. Theoretical and Applied Genetics. 97 (5-6): 871-880. USD A. 2000. Wood handbook: wood as an engineering material. University Press of the Pacific, Madison, Wisconsin. 463 pgs. Van Ooijen, J.W., and Voorrips, R.E. 2001. JoinMap 3.0, software for the calculation of genetic linkage maps. Plant Research International, Wageningen, the Netherlands. Vargas-Hernandez, J., and Adams, W.T. 1991. Genetic variation of wood density components in young coastal Douglas-fir - Implications for tree breeding. Canadian Journal of Forest Research. 21 (12): 1801-1807. Vos, P., Hogers, R., Bleeker, M., Reijans, M., Vandelee, T., Homes, M., Frijters, A , Pot, J., Peleman, J., Kuiper, M., and Zabeau, M. 1995. AFLP - a new technique for DNA-fingerprinting. Nucleic Acids Research. 23 (21): 4407-4414. 134 Wagner, F.G., Gorman, T.M., Pratt, K.L., and Keegan, C.E. 2002. Warp, moe, and grade of structural lumber curve sawn from small-diameter Douglas-fir logs. Forest Products Journal. 52 (1): 27-31. Wu, R.L., Han, Y.F., Hu, J.J., Fang, J.J., Li, L., Li, M.L., and Zeng, Z.B. 2000. An integrated genetic map of Populus deltoides based on amplified fragment length polymorphisms. Theoretical and Applied Genetics. 100 (8): 1249-1256. Yin, T.M., Wang, X.R., Andersson, B., and Lerceteau-Kohler, E. 2003. Nearly complete genetic maps of Pinus sylvestris L. (Scots pine) constructed by AFLP marker analysis in a full-sib family. Theoretical and Applied Genetics. 106 (6): 1075-1083. Zhang, D., Zhang, Z., Yang, K., and Li, B. 2004. Genetic mapping in (Populus tomentosa x Populus bolleand) and P. tomentosa Carr. using AFLP markers. Theoretical and Applied Genetics. 108 (4): 657-662. 135 Chapter 4 Identification of Quantitative Trait Loci for Wood Quality and Growth Traits for Eight Full-Sib Coastal Douglas-fir Families Introduction Coastal Douglas-fir {Pseudotsuga menziesii (Mirb.) Franco var menziesii) is the most intensively managed tree species on the west coast of North America (Aubry et al. 1998), as its wood is extremely valuable as a construction material due to its superior strength properties (USDA 2000). Much is known of the heritability and correlations of commercially important quantitative traits for Douglas-fir. For example, wood density and its components are generally under moderate to strong genetic control with heritabilities estimated between 0.2 - 0.5 (St.Clair 1994; Vargas-Hernandez and Adams 1991), which is consistent with other species (Hannrup et al. 2000; Hannrup et al. 1998; Ivkovich 1996; Ivkovich et al. 2002a; Loo et al. 1985; Louzada and Fonseca 2002; Nicholls et al. 1980; Talbert et al. 1983; Yanchuk and Kiss 1993). Wood density is arguably the most important wood attribute, as it contributes significantly to overall wood strength and is often negatively correlated with growth traits (St.Clair 1994; Vargas-Hernandez and Adams 1991). The whole-tree dollar value of Douglas-fir, however, has been shown to be most influenced by volume (Aubry et al. 1998). Volume and growth traits are under moderate (h2 = 0.17 - 0.32) genetic control in conifers (Ericsson and Fries 2004; Hannrup et al. 2000; Hannrup et al. 1998; Ivkovich 1996; St.Clair 1994; Yanchuk 136 and Kiss 1993) and have been the focus of tree improvement programs (Aubry et al. 1998; Vargas-Hernandez and Adams 1991). Fibre properties, such as fibre length and cell wall thickness, are more important to paper quality than solid wood products and thus have implications for the pulp and paper industry (Chantre et al. 2002). Fibre length and coarseness interact to provide bonding between fibres and improve paper strength (Seth and Kingsland 1990), but are under little to moderate (h2 = 0.10 - 0.37) genetic control (Ivkovich 1996; Loo et al. 1985; Pot et al. 2002). Wood strength and stiffness is highly associated with microfibril angle (Cramer et al. 2005; Evans and Ilic 2001), which is under variable genetic control (h2 = 0.17 - 0.51; H 2 = 0.11 - 0.72) (Donaldson and Burdon 1995; Hannrup et al. 2004; Ivkovich et al. 2002b; Myszewski et al. 2004). The chemical constituents of wood, including lignin, cellulose and hemicellulose, influence the overall characteristics of the woody material, and significantly impact pulp and paper processing. QTL (quantitative trait locus) analyses have been conducted in several tree species for wood quality and adaptive traits. Quantitative traits of commercial importance including microfibril angle, wood specific gravity and volume percentage of latewood, have been extensively studied in loblolly pine (Pinus taeda L.) (Brown et al. 2003; Neale et al. 2002; Sewell et al. 2000; Sewell et al. 2002; Sewell et al. 1999). QTLs for lignin and cellulose content provided evidence of environmental interactions suggesting a complex pattern of QTL activity (Sewell et al. 2002). Furthermore, QTLs identified for wood quality traits were shown to be stable through time (Brown et al. 2003; Sewell et al. 2000). In several instances, QTLs from different developmental stages and a variety of traits were shown to co-locate on linkage maps (Arcade et al. 2002; Brown et al. 2003; 137 Grattapaglia et al. 1996; Jermstad et al. 2001a; Neale et al. 2002; Sewell et al. 2000; Sewell et al. 2002; Thamarus et al. 2004; Yoshimaru et al. 1998). The co-localization of multiple QTLs for a trait from different developmental stages or years is clear evidence that a QTL exists, and equally important, validates the QTL (Sewell et al. 2000; Sewell et al. 2002). In tree breeding efforts, an obstacle to the use of QTL data for selection has been a lack of QTL verification. However, several QTLs for wood quality and chemistry traits in loblolly pine have been verified using related and unrelated families. In related families, 43% of QTLs were detected in both families while only 16% of QTLs were common to unrelated families (Brown et al. 2003; Neale et al. 2002). These results suggest that different alleles segregate in different families. QTL analysis for adaptive traits have been conducted in Douglas-fir (Jermstad et al. 2003; Jermstad et al. 2001a; Jermstad et al. 2001b; Jermstad et al. 1998; Wheeler et al. 2005). Similar temporal and spatial patterns of QTL activity were identified for timing of vegetative bud flush and spring and fall cold hardiness. To our knowledge, no QTL analysis for wood quality traits has been conducted in Douglas-fir. Initially, QTL mapping was identified as a potential tool for tree improvement programs. Marker assisted selection (MAS) and marker assisted early selection (MAES) have been suggested as applications for QTL analyses. Several studies have been conducted to investigate the potential use of QTL mapping in marker aided selection (Johnson et al. 2000; Strauss et al. 1992; Wu 2002). MAS or MAES may work for selection within families for traits of low heritability, but have limited potential for implementation more broadly. Strauss et al. (1992) suggest that MAS may be justified for three possible scenarios: i) to identify QTLs associated with severe threats to forest health 138 where time is urgent, ii) used in situations where extremely high-value families exist and where a limited number of clones or genotypes will be identified for extensive use, or iii) combine MAS with phenotypic selection in mapped families to improve genetic gains among and within families. It has also been suggested that QTL analyses are more effective than traditional approaches in testing many old hypotheses regarding quantitative traits, including the nature of gene action, the number of genes controlling quantitative traits, the significance of epistasis in hybrid breakdown, and the evolution of adaptive and morphological trait complexes. Genetic linkage and QTL maps are currently being used for comparative mapping between species (Chagne et al. 2003; Krutovsky et al. 2004), and to identify the numbers of QTLs of major effect, and their temporal and spatial interactions (Arcade et al. 2002; Byrne et al. 1997; Jermstad et al. 2001a; Jermstad et al. 2001b; Sewell et al. 2000; Sewell et al. 2002). Additionally, composite maps may act as a base for candidate gene mapping (Brown et al. 2003; Wheeler et al. 2005). Traditional linkage and QTL analyses in forest trees have been conducted in single pedigrees with sex-averaged linkage maps (Chagne et al. 2002; Jermstad et al. 1998; Sewell et al. 1999; Wu et al. 2000) or maps generated for specific individuals using megagametophyte tissue (Remington et al. 1999; Travis et al. 1998). One of the major concerns about the use and implementation of such QTL data is the lack of representation of multiple genetic backgrounds. There is evidence for QTL-by-environment interactions and evidence that QTL effects are different in unrelated families (Brown et al. 2003; Neale et al. 2002). These results suggest that QTL analyses using several families may be required to better identify the QTLs acting within a population. The current study 139 presents the results of a QTL analysis for wood quality traits using eight full-sib families replicated on four sites for coastal Douglas-fir. Materials and Methods Sample Population The sample population consisted of eight full-sib Douglas-fir families (2, 7, 26, 38, 62, 75, 92, and 151) selected from the British Columbia Ministry of Forests second generation progeny test program. The families were chosen from a total of 15 full-sib families previously characterized for wood properties based on growth and density data. Families with varying growth and density combinations were selected for linkage map construction and QTL analysis (Table 4.1; Appendix 2). The families were sampled from four sites established in 1977 in southwestern British Columbia (two on Vancouver Island and two on the B.C. mainland). Forty individuals were selected from each family (10 individuals per family per site; 320 trees in total) for identification of QTLs affecting growth and wood property traits. The trees were sampled in 2003 at the age of 26. 140 Table 4.1: Average height (HT), diameter (DBH), volume (VOL), earlywood density (EWD), latewood density (LWD), latewood proportion (LWP) and average density (AD) of the full-sib Douglas-fir families employed for QTL analysis. Family Trait 2 7 26 38 62 75 92 151 HT (m) 18.10 16.43 17.67 18.61 18.52 18.05 15.97 15.99 DBH (cm) 24.60 • 24.08 26.93 26.50 25.76 25.85 21.89 21.40 VOL (m3) 0.36 0.33 0.40 0.42 0.40 0.38 0.27 0.24 EWD (g/m3) 303.85 312.75 295.83 291.20 306.09 296.38 313.41 290.59 LWD (g/m3) 645.70 620.81 610.50 617.26 638.55 617.93 628.13 609.27 LWP 0.38 0.38 0.34 0.31 0.35 0.32 0.38 0.33 AD (g/m3) 432.06 430.94 403.22 394.42 424.10 400.20 433.70 401.85 Phenotypic Data Growth Traits and Core Sampling Height (HT) and diameter at breast height (DBH) data was collected for each tree. Height was measured using a Vertex instrument (Vertex ILT; Haglof, Sweden) and diameter was estimated at breast height (1.3m) using a diameter tape. Both height and diameter are estimates at 26 years of age. Volume was calculated using Schumacher's equation for Douglas-fir (0.000047966*[DBH1 8 1 3 8 2]*[HT 1 0 4 2 4 2]) for each tree. Bark to bark increment core samples (10mm in diameter) were taken at breast height from each tree in a north to south direction. Fibre Length and Coarseness Fibre length (FL) and fibre coarseness (CS) were measured specifically on wood material extracted from growth rings corresponding to ages 15-17 from the southern 141 portion of the increment cores. Wood samples were macerated in Franklin solution (1:1, 30% hydrogen peroxide:glacial acetic acid) for 48 hours at 70°C. The solution was decanted and the remaining fibrous material was washed under vacuum with de-ionized water until a neutral pH was achieved. Fibres were dried at 50°C overnight, and the moisture content measured to determine fibre mass. Sub-samples were then re-suspended in 10 mL of de-ionized water and fibre properties determined on a Fibre Quality Analyzer (LDA02, OpTest Equipment Inc., Canada). All samples were run in triplicate. Fibre length was measured in millimeters and fibre coarseness as mass per unit length (mg/m). Both traits are averages for ages 15-17. Wood Density Wood density traits were measured on the northern portion of tree increment cores by X-ray densitometry. Cores were cut to 1.67mm thicknesses on a precision pneumatic twin blade saw to expose the radial face for analysis. Earlywood density (EWD), latewood density (LWD), latewood proportion (LWP) and average density (AD) were measured for each ring from pith to bark. QTL analyses were performed on composite traits (average across all rings) and individual rings for all traits. Composite traits represent the average density or latewood proportion at age 26. Microfibril Angle Microfibril angle estimates were generated by X-ray diffraction and light microscopy (Megraw et al. 1998). The 002 diffraction spectra for individual earlywood 142 growth rings from six sample trees were screened for T-value distribution and symmetry on a Bruker D8 Discover X-Ray diffraction unit equipped with an area array detector (GADDS). Wide-angle diffraction was used in the transmission mode, and the measurements were performed with CuKai radiation (K = 1.54A), the X-ray source fit with a 0.5mm collimator, and the scattered photon collected by the GADDS detector. Both the X-ray source and detector were set to theta = 0°. Thirty-two individual growth rings selected from the six sample trees were sectioned using a microtome (0.20um -0.30um) and processed for compound light microscopy (Wang et al. 2001). Wood sections were placed in lmL of 5% (wt/vol) cobalt chloride (C0CI2) and heated to 80°C for 2 hours then floated in a sonicator (47KHz) for a subsequent 2 hours. The sections were mounted on slides and allowed to dry overnight. Differential interference contrast (DIC) microscopy at 400* magnification was used to visualize individual fibres. Images were collected using Qcapture and saved as TIFF files. The angle of microfibrils within individual fibres was measured using the ImageJ software. Sections were poor from three growth rings and were discarded. The remaining 29 sections were used to create a standard curve (R2 = 0.92) using the average T-value from the 002 diffraction arc peaks integrated over 2-theta, and the known angles. Microfibril angle measurements were then estimated for all samples by collecting the 002 diffraction intensity profiles and measuring the T-values for the earlywood portion of the growth ring corresponding to year 17, and comparing them to the best-fit linear relationship generated for known (measured microscopically) microfibril angles and T-values. 143 Wood Chemistry Wood chemical composition was measured using a modified Klason analysis (Huntley et al. 2003) on increment core material from the northern portion of the tree. The wood material from pith to bark was ground with a Wiley mill to pass a 0.4mm screen (40 mesh). The ground wood was extracted in lOOmL of acetone for 8 days (with repeated changes) to remove extractable components and to minimize the formation of "pseudolignin" during Klason analysis. The extracted lignocellulosic material was air-dried to remove the solvent and analyzed for sugar and lignin composition. A 0.2g sample of extracted wood was transferred to a 15mL reaction vial cooled on an ice bath. A 3mL aliquot of 72% (w/w) H 2 S O 4 was added to the sample and thoroughly mixed for one minute. The reaction vial was immediately transferred to a water bath maintained at 20°C, and mixed for one minute every 10 minutes. After 2 hours of hydrolysis, the contents of each test tube were transferred to a 125mL serum bottle using 112mL of de-ionized water to rinse all residue and acid from the reaction vial. The serum bottles (containing 115mL 4% (w/w) H2SO4 plus wood) were sealed and autoclaved at 121°C for 60 minutes. Samples were allowed to cool, and the hydrolysates vacuum-filtered through pre-weighed medium coarseness sintered-glass crucibles. Each sample was washed with 200mL of warm (~50°C) de-ionized water to remove residual acid and sugars, and dried overnight at 105°C. The dry crucibles were re-weighed to gravimetrically determine Klason lignin (acid-insoluble lignin). The filtrate was then analyzed for acid-soluble lignin by absorbance at 205 nm using UV/VIS Spectrometer (Lambda 45, PerkinElmer Instruments Inc., USA) according to TAPPI Useful Method UM250 (Tappi Useful Method 1991). 144 The concentration of sugars (arabinose, galactose, glucose, mannose and xylose) in the hydrolysate were determined using High Performance Anion Exchange Liquid Chromatography (HPLC). The HPLC system (Dionex DX-500, Dionex, USA) was equipped with an ion-exchange PA1 (Dionex) column, a pulsed amperometric detector with a gold electrode, and a Spectra AS50 autoinjector (Spectra-Physics, USA). Prior to injection, samples were filtered through 0.45um HV filters (Millipore, USA). A 20uL volume of sample was loaded containing fucose as an internal standard. The column was equilibrated with 250mM NaOH and eluted with de-ionized water at a flow rate of 1.0 mL/min. Total lignin (TL), arabinose (ARA), galactose (GAL), glucose (GLU), mannose (MAN) and xylose (XYL) content were measured as the proportion of the initial mass of wood used in the analysis. Wood chemistry data represent the whole core chemical content at age 26. Genotypic Data and Map Construction Ten AFLP (amplified fragment length polymorphisms) marker combinations were used to develop a comprehensive linkage map for the eight full-sib families used in this study (see chapter 3 for details). The map was generated by calculating average LOD and recombination rates across families using a joint likelihood function (Hu et al. 2004). The map was generated using Joinmap (Van Ooijen and Voorrips 2001) and contains 120 markers distributed across 19 linkage groups. The total distance covered by the linkage map is 938.6cM with an average of 9.3cM between markers. 145 QTL Analysis QTLs were detected and positioned using the "sib-pair analysis with parents of known genotype" interface of the QTL Express online software package (Seaton et al. 2002). This analysis is based on the sib-pair analysis that was first proposed by Haseman and Elston (1972) and improved upon by Visscher and Hopper (2001). The regression analysis uses the relationship between the identity-by-descent (LBD) probabilities and the squared difference and corrected squared sum of phenotypic values between sibs at each locus. Since site can have a significant effect on trait values, the GLM procedure of SAS (SAS Version 9.1) was used to transform phenotypic data to remove site and block effects using the following linear model: Yijip = p + Fi + Si + FSii + B j (, ) + FBij(i) + Ep(iji) [4.1] where Yyip is the individual phenotypic observation, u, is the overall mean, Fj is the fixed family effect, Si is the random site effect, FSJI is the random family-by-site interaction, Bj(i) is the random block effect nested within sites, FBJJQ is the random family-by-block interaction nested within site and Ep(jji) is the random residual effect. The residuals were used as phenotypic input for QTL Express. The linkage map was scanned for each trait using a chromosome-wide bootstrap analysis with 100 iterations to calculate critical F-values at the 0.05 and 0.01 alpha levels. If a QTL was detected at either significance level, the linkage group was re-scanned for the trait using 1000 iterations to calculate the 146 critical F-statistics. The final analysis was conducted using a 1-QTL model with a lcM step size as described by Visscher and Hopper (2001). The proportion of variation explained by each QTL was calculated using an ANOVA (analysis of variance) to partition phenotypic trait values into their respective components (Table 4.2) based on the following linear model: Yijkip = u + Fj + S, + FS a + GF k ( i ) + SGFik ( i ) + BSm + FBS i j ( l ) + E [4.2] where Y^ip is the individual phenotypic observation, u is the overall mean, Fj is the fixed family effect, Si is the random site effect, FSu is the random family-by-site interaction, GFk(i) is the random genotype nested within family effect (QTL effect), SGFik(i) is the random site by genotype nested within family interaction (site by QTL interaction), BSJQ is the random block nested within site effect, FBSJJQ is the random family-by-block nested within site interaction and EP(jikj) is the random residual effect. The proportion of variation explained by the nearest marker to the QTL position determined by QTL Express was used to estimate QTL effect. QTL effects were estimated as follows: a; = SSCXF) [4.3] SSTotal where a; is the proportion of variation explained by the QTL at the i t h marker, SSG<F) is the sum of squares for genotype nested within family (QTL) and SSjotai is the total sum of squares. 147 Table 4.2: Components of variance and degrees offreedom used to calculated the proportion of variation explained by each QTL for the corresponding marker. df Components of Variance F (f-1) O 2 E + mna 2FBs + bna 2GFS + sbna2MF + mbna2Fs + smbnc2F S (s-1) a2E + fmna2Bs + bna 2GFs + fmbna2s FS (f-l)(s-l) a2E + mna2FBs + bna 2GFs + mbnc2Fs G(F) f(m-l) a2E + bno 2GFS + sbna2GF SG(F) f(s-l)(m-l) O 2 E + bna 2GFs B(S) s(b-l) a 2 E + fmna2Bs FB(S) s(f-l)(b-l) a E + mna2FBs Error sfmn(n-l) 0 2 E Expected Number of Undetected QTLs The number of undetected QTLs was estimated using the analysis proposed by Otto and Jones (2000). This analysis assumes that the frequency distribution of QTL effect size approximates an exponential distribution. If two or more QTLs are detected for a given trait, the expected number of undetected QTLs and their average effect size can be estimated. The number of undetected QTLs (IIQTL) was calculated as: noTL = D [4.4] M - 0 where D is the sum of additive effects of the detected QTLs, M is the average of the additive effects of detected QTLs and 0 is the threshold value for detecting QTLs calculated by satisfying the following equation: 0 = ami„ - 0 - (M- 0/na) + ami„ Exp[-aminnd/(M- 0)]/l-Exp[-amin d/(M- 0)] [4.5] 148 where ami n is the minimum effect of detected QTLs and rid is the number of detected QTLs. Confidence intervals can be calculated by satisfying the following equation for n: -na + (M- 0)nnd/D - ndln[(M- 0)n/D] - %2m/2 = 0 [4.6] where x2i[o.o5] = 3.841 and n is the confidence interval. The average effect of undetected QTLs can be calculated as: Mundetected = M(l-[(l-T)/ l-Exp{-(l-T)/T}]) [4.7] where x = (M- 0)/M. The expected number of QTLs for any trait can be calculated by adding the number of undetected QTLs (IIQTL) with the number of detected QTLs (rid). Results Number and Effect of Detected QTLs for Compound Traits A total of five QTLs were detected for growth traits on three linkage groups (Table 4.3; Figure 4.1). Two QTLs were detected for height that explained 16.1% -17.7%) of the phenotypic variation. Only one QTL was detected for diameter which had 149 Table 4.3: QTLs detected using interval mapping. The F is the F-value for the regression analysis and Fo.os and Fo.oi are critical F-values for significance at the corresponding a levels. Marker is the closest AFLP marker to the QTL position and is used to assess the proportion of phenotypic variation explained by the QTL. P-values are reportedfor the significance of genotype nested within family for the single marker QTL analysis. Group Trait QTL F Fo.05 Fo.oi Marker Marker Effect P-value Position Position (cM) (cM) 1 HT 15 8.03 6.53 13.25 ACGCCGG 0297 17.2 0.161 0.08 2 HT 31 11.02 5.86 12.09 ACGCCGG 0325 31.6 0.177 0.00* VOL 32 6.83 5.88 13.35 ACGCCGG 0325 31.6 0.113 0.00* 4 DBH 38 24.15** 7.88 16.01 ACGCCCA 0652 38.1 0.086 0.04* VOL 38 24.25** 7.08 14.52 ACGCCCA 0652 38.1 0.096 0.06 7 LWD 37 9.16 8.2 21.09 ACACCGT 0391 37 0.040 0.34 12 MFA 13 9.58 7.29 18.38 ACACCGG 0138 15.2 0.036 0.05 13 GLU 0 7.59 4.18 9.43 ACCCCGC 0327 0 0.062 0.08 14 FL 0 16.86 6.9 19.02 ACGCCCA 0213 0 0.074 0.02* CS 50 8.05 6.35 15.31 ACGCCGT 0595 50.1 0.052 0.16 GAL 33 9.96 5.3 12.47 ACGCCCA 0345 32.8 0.119 0.04* LWD 41 12.98 10.67 26.81 ACGCCCA 0345 32.8 0.060 0.10 15 LWD 42 24.38** 8.15 19.06 ACGCCTC 0160 44.5 0.080 0.19 GAL 44 15.64** 5.9 11.11 ACGCCTC 0160 44.5 0.048 0.21 GLU 44 22.59** 6.39 12.65 ACGCCTC 0160 45.5 0.044 0.24 MAN 44 27.37** 4.77 10.44 ACGCCTC 0160 45.5 0.067 0.11 TL 44 24.5** 5.02 9.69 ACGCCTC 0160 44.5 0.050 0.26 16 CS 17 8.18 6.67 13.99 ACGCCTG 0154 16.6 0.074 0.17 17 FL 10 7.66 6.36 17.52 ACGCCGG 0311 10.1 0.036 0.32 ARA 35 13.88 7.05 19.76 ACGCCTC 0189 34.6 0.049 0.22 18 CS 0 9.6 6.53 14.3 ACGCCTC 0144 0 0.146 0.00* FL 0 12.19 7.06 17.9 ACGCCTC 0144 0 0.157 0.02* * significant at a = 0.05 ** significant at a = 0.01 HT = height at age 26; DBH = diameter at breast height age 26; VOL = volume at age 26; LWD = average core latewood density; CS = fibre coarseness for rings 15-17; FL = fibre length for rings 15 - 17; GAL = core galactose content; GLU = core glucose content; MAN = core mannose content; ARA = core arabinose content; TL = core lignin content; MFA = microfibril angle at age 17 an effect of 8.6%. Two QTLs were detected for volume. Each co-locates with QTLs for height and diameter, as expected given that volume is calculated based on diameter and height. The QTL detected on group two co-locates with a QTL for height and explains 11.3%) of the variation, while the volume QTL on group four co-locates with a QTL for diameter and has an effect of 9.6%. The markers associated with QTL positions for 150 growth traits are significant at the 0.05 level for QTL effects except for ACGCCGG_0297 on linkage group one (p = 0.082) and ACGCCCA_0652 on linkage group four (p = 0.057). There were three QTLs detected for each fibre trait (length and coarseness), located on four linkage groups, and one QTL for microfibril angle. The three fibre length QTLs were located on linkage groups 14, 17 and 18 and explain 3.6% - 15.7% of the phenotypic variation. The fibre coarseness QTLs were located on linkage groups 14, 16 and 18 and have effects ranging from 5.2% - 14.6%, while the microfibril angle QTL is located on linkage group 12 and explains 3.6% of the phenotypic variation. Although the fibre property QTLs on group 14 are located at opposite ends of the linkage group, the QTLs on linkage group 18 are located at the same position. Three of the five unique markers associated with fibre property QTLs are not significant for QTL effects as is the marker associated with microfibril angle. Only three QTLs were detected for compound wood density traits, all associated with latewood density. The QTLs had effects ranging from 4.0% - 6.0% of phenotypic variation and were located on three linkage groups. The QTL on group 14 co-locates with individual ring QTLs and the QTL on group 15 is located near a QTL associated with several wood chemistry traits. All of the markers associated with compound latewood density QTLs were not significant for QTL effects. A total of seven QTLs were detected for wood chemistry traits. Only one QTL was detected for lignin content and explained 5.0% of the phenotypic variation. This QTL co-locates with QTLs associated with glucose, galactose and mannose located on group 15, which explain between 4.4% and 6.2% of the phenotypic variation. The 151 markers associated with these QTL are not significant for QTL effects. A QTL for arabinose was also detected on linkage group 17, which explains 4.9% of the variation. A second galactose QTL is located on group 14 with an effect size of 11.9% and a second glucose QTL was detected on group 13 with an effect size of 6.2%. The marker associated with the second galactose QTL is significant for QTL effects (p = 0.042), however, the arabinose and glucose markers are not (p = 0.22 and 0.080, respectively). Number and Effect of Detected QTLs for Ring Density Traits In total, 78 ring density QTLs were detected for average density, latewood density, earlywood density and latewood proportion at the suggestive (0.05) and significant (0.01) level (Table 4.4; Figure 4.1). Twenty-two QTLs were detected for average density. These QTLs explained 0.1% - 14.9% of the phenotypic variation and co-located with other density trait QTLs located on 9 different linkage groups. Seventeen QTLs were detected for earlywood density with effects ranging from 0.1% -6.5%, which are located on seven linkage groups. Twenty-six latewood density QTLs were located on nine linkage groups and explained 0.5% - 13.1% of the phenotypic variation, while thirteen latewood proportion QTLs were detected on nine linkage groups with effect sizes ranging from 0.4% - 10.2% of the variation. The amount of variation explained by these QTLs is low and many of the markers associated with QTLs are not significant for QTL effects. Several QTLs share similar positions on common linkage groups. When two or more QTLs were located within 15cM, they were grouped as a single QTL (Figure 4.1). Ring density clusters with QTLs from all four wood density 152 traits were labeled "DEN". In total, there were 11 QTL ring density clusters containing 67 QTLs (86% of ring density QTLs) and 11 independent ring density QTLs distributed throughout the linkage map (Figure 4.1). Number and Effect of Undetected QTLs for Compound Traits An estimate of the number of undetected QTLs was calculated for compound traits for which two or more QTLs were detected (Table 4.5). Height (21) and volume (12) had the highest number of undetected QTLs of all compound traits analyzed with average effect sizes of 1.6% and 1.7%, respectively. The total number of QTLs affecting height and volume is expected to be 23 and 14, respectively. Fibre property traits had four and five undetected QTLs for a total of seven for fibre length and eight controlling fibre coarseness. The undetected fibre property QTLs had average effect sizes of 1.1% for fibre length and 1.7% for fibre coarseness. The expected number of undetected QTLs controlling core latewood density is six with an average effect size of 1.3%, and the total number of QTLs expected to control latewood density is nine. Galactose has three undetected QTLs and glucose six. Galactose undetected QTLs have an average effect size of 1.4% and glucose 1.2%. Galactose content is expected to be controlled by five QTLs and glucose content eight. 153 ACACCGG_0145 -ACACCGG_0075 -A C A C C G T _ O I 3 6 -ACGCCGGJ)297 ACGCCTC_0229 ACGCCTC_0249 \ n A C A C C G G _ 0 2 % — T l ACGCCGG_0296 -ACACCGG.0245 -ACCCCGC_0144 -A C C C C G C _ 0 1 7 6 -A C G C C G G J K 4 5 -ACGCCGA_02SS -ACGCCTG_0376 -ACCCCGC_0253 -ACACCGT_0I7S -ACGCCGT_0286-ACGCCGT_0I78 -ACACCGT_0285 -ACACCGT_0287 -ACGCCGG_0325 -ACACCGG_0326 -ACACCGG_0327 -Scale 0 ACGCCTC_0290 -ACACCGG_0185 -ACACCGT.013I -ACGCCGA.0383 -ACACCGT_040I -ACACCGG.0289 -ACGCCGG.0288 -ACGCCGA_0528 -ACGCCGG.019I -ACGCCCA_0569 -A C A C C G G J H 9 ! -ACGCCCG_0445 -A C A C C G G J 2 5 8 -A C A C C G T 0235 -ACGCCCA_0652 -ACGCCGT_0466 -ACGCCTC.0265 -5 ACGCCGT_0556 -ACACCGT_0556 -ACGCCGA.0277 -ACGCCCAJJ223 -ACGCCCA_0127 -ACGCCTC_0306 -ACGCCTG_0115-ACGCCCA_0408 -ACGCCCA.0235 -§ I S V A C G C C T C J M 4 3 -S > 2 o ACGCCGT_0635 -ACGCCCA_0255 -ACCCCGC_02J7-8 ACGCCCG_0255 -ACGCCCA.0439 -ACGCCTC_0146 -ACGCCCA.0343 -ACGCCCA.0252 -ACACCGG_0237 -ACACCGT_0397 -ACGCCCA_0063 -ACGCCTC_0356 -A C G C C G A . 0 3 7 0 -13 -r & ACGCCTC_0297.-ACACCGT_039I -ACGCCGT.0392 ACACCGT_OIOI -ACGCCCG_0418-ACACCGG_OI53-ACGCCGG_OI53 -ACGCCTG_0302 -10 ACACCGT.0371 -ACGCCGT_0371 -11 ACACCGT.OI 15 -ACGCCCG_OI27-ACGCCCA_06I4 -A C A C C G G J B I 6 -A C G C C G G J ) 3 I 5 -ACGCCGG_0338 -ACACCGG_0339 -ACGCCCA JX360 -- 35 ACGCCGAJE94 -ACGCCCG_0524 -ACGCCGA_0072 -12 ACACCGG_OI38-ACGCCGG_0238 -ACGCCGA_0299 -ACCCCGC_013! -A C C C C G C J 3 2 7 -ACGCCGG_O408 - . ACACCGG.0407 -13 A C C C C G C _ O I 3 0 -A C G C C C A . 0 2 1 3 -^ ACGCCTCJH33 -O ACGCCGG_0090 -14 ACCCCGC_0277 ACGCCCA_0345 ACGCCGT.0595 -ACACCGT.0609 ACGCCGT_0608 >n ACGCCCA.0237 -15 ACGCCTG_0085 -A C G C C T C _ 0 1 6 0 -ACGCCTC_0591 -16 ACACCGG_018I -ACGCCGG_0180-ACACCGG_0310-ACGCCGG_0311 -17 — u > ACGCCTC_0I89 -ACGCCGA_0121-A C G C C T C _ 0 I 4 4 -18 A C G C C G T _ 0 I 9 4 -A C A C C G T _ O I 9 5 -ACACCGGJJ352 -c —5 "6 D ACGCCGA_0066 -19 ACGCCGG_0407 -ACGCCTG.0154 —H — O _ O ACACCGG.0408 — H ACCCCGC.0169 -Figure 4.1: QTL map for composite and ring density traits for coastal Douglas-fir. QTLs marked "DEN" contain QTLs detected in multiple years from all four wood density traits. The numbers in brackets are the number of QTLs detected within the QTL location. The scale is in cM (Kosambi map function). Table 4.4: Ring density interval mapping QTLs. F is the F-value for the regression analysis and Fo.os and Fo.oi are the critical F-values at the corresponding a value. Marker is the closest marker to the QTL position and "P" is the p-value of the single locus analysis testing genotype nested within family. Year is the year for which each ring corresponds. LG Year Trait QTL F Fo.os Fp.oi Marker Marker Effect P Position Position (cM) (cM) 1 1989 LWD 2 7.19** 1.18 2.18 ACACCGG 0145 0 0.005 0.082 2003 AD 20 . 6.57 5.59 9.4 ACGCCTC 0229 20.9 0.010 0.802 2001 LWP 22 10.24 10.1 16.53 ACGCCTC 0229 20.9 0.070 0.243 1998 EWD 24 4.8 • 4.23 7.41 ACACCGG "0296 23.2 0.051 0.384 1995 AD 28 5.33 4.28 6.91 ACGCCGG 0296 28.4 0.016 0.619 1994 AD 28 6.65 5.39 8.78 ACGCCGG 0296 28.4 0.071 0.106 1995 EWD 28 5.44 3.88 5.97 ACGCCGG 0296 28.4 0.016 0.555 1995 LWD 28 4.04 3.37 6.09 ACGCCGG 0296 28.4 0.073 0.003 1996 LWD 28 8.02 6.91 10.11 ACGCCGG 0296 28.4 0.065 0.053 1996 AD 29 5.69** 3.97 5.63 ACGCCGG 0296 28.4 0.013 0.469 1990 AD 29 17.49 13.22 22.98 ACGCCGG 0296 28.4 0.110 0.053 1994 EWD 29 11.83 10.22 15.95 ACGCCGG 0296 28.4 0.054 0.096 1989 AD 30 16.96 11.16 17.99 ACGCCGG 0296 28.4 0.088 0.016 1989 EWD 35 9.93 8.41 13.37 ACACCGG 0245 34.9 0.015 0.062 1990 EWD 35 22.76 14.53 26.54 ACACCGG 0245 34.9 0.005 0.790 2 1993 AD 27 11.76 6.94 12.31 ACACCGT 0287 26.8 0.052 0.006 1996 EWD 28 5.67 5.04 12.68 ACACCGT 0287 26.8 0.035 0.260 1992 AD 34 19.53** 8.19 17.68 ACACCGG 0326 33.1 0.086 0.015 3 2000 LWD 48 8.79 8.77 21.46 ACACCGG 0289 54.3 0.057 0.050 2002 LWP 69 11.19 7.14 14.26 ACGCCGA 0528 72 0.004 0.762 4 1991 AD 37 10.2 9.51 19.17 ACACCGT 0235 36.7 0.021 0.617 1993 LWD 37 15.08** 6.62 15.01 ACACCGT 0235 36.7 0.013 0.294 1985 LWD 37 22.04** 7.92 21.84 ACACCGT 0235 36.7 0.022 0.602 1992 LWD 37 23.57 13.37 25.41 ACACCGT 0235 36.7 0.012 0.511 1988 LWD 38 9.22 6.85 20.28 ACGCCCA 0652 38.1 0.104 0.035 1991 LWD 38 17.22 13.06 24.91 ACGCCCA 0652 38.1 0.021 0.423 5 1996 AD 7 6.09** 2.61 4.6 ACGCCTC 0265 9.6 0.003 0.847 1999 AD 7 10.48 5.11 12.46 ACGCCTC 0265 9.6 0.129 0.150 2002 AD 7 10.61 5.45 14.13 ACGCCTC 0265 9.6 0.002 0.784 1998 AD 7 13.15** 3.88 8.08 ACGCCTC 0265 9.6 0.013 0.707 1998 EWD 7 3.72 2.91 5.99 ACGCCTC 0265 9.6 0.032 0.501 1995 EWD 7 4.1 2.49 6.18 ACGCCTC 0265 9.6 0.036 0.324 1997 EWD 7 6.65 6.33 15.29 ACGCCTC 0265 9.6 0.025 0.552 1994 EWD 7 20.96** 6.1 13.21 ACGCCTC 0265 9.6 0.001 0.772 1999 LWP 7 13.11** 6.02 11.68 ACGCCTC 0265 9.6 0.100 0.202 1996 LWD 9 6.15 4.19 9.52 ACGCCTC 0265 9.6 0.013 0.638 1990 LWD 24 9.31 7.59 15.3 ACGCCGT 0556 24.1 0.039 0.277 1993 EWD 26 20.24 10.39 21.28 ACGCCGT 0556 24.1 0.026 0.356 1990 LWP 26 5.38 4.41 9.23 ACGCCGT 0556 24.1 0.011 0.652 1990 AD 27 22.77 10.33 25.64 ACGCCGT 0556 24.1 0.043 0.402 1991 EWD 27 8.38 5.47 10.92 ACGCCGT 0556 24.1 0.030 0.473 6 1988 LWD 1 50.1** 8.09 15.7 ACGCCGA 0277 0 0.061 0.316 2002 LWP 107 16 11.04 20.28 ACGCCCA 0235 107.7 0.046 0.170 7 2001 LWD 21 9.68 7.91 19.04 ACGCCTC 0297 21.1 0.035 0.433 8 1998 LWP 56 8.14 6.95 12.16 ACGCCCA 0252 58.2 0.009 0.245 155 Table 4.4 continued LG Year Trait QTL Position (cM) F Fo.os Fo.oi Marker Marker Position (cM) %Var P 9 2001 EWD 0 11.1** 4.99 9.17 ACGCCTC 0356 0 0.041 0.044 1995 LWD 0 4.08 2.74 5.82 ACGCCTC 0356 0 0.020 0.672 1994 LWD 0 5.38 2.82 6.46 ACGCCTC 0356 0 0.021 0.431 1993 LWD 0 10.26** 5.31 9.7 ACGCCTC 0356 0 0.015 0.790 1985 LWD 0 27.32** 5.99 10.77 ACGCCTC 0356 0 0.010 0.695 11 2001 LWP 12 16.64 11.51 19.49 ACGCCGA 0370 0 0.018 0.579 2003 AD 14 7.74 5.81 10.49 ACACCGT 0115 26.9 0.021 0.394 2003 LWP 14 9.4 8.17 14.72 ACACCGT 0115 26.9 0.041 0.068 12 1988 LWP 25 7.76 5.28 10.28 ACGCCGG 0238 16.5 0.013 0.517 14 2003 AD 40 6.04 5.82 11.42 ACGCCCA 0345 32.8 0.011 0.566 2001 EWD 50 7.72 6.62 17.08 ACGCCGT 0595 50.1 0.004 0.715 2003 LWP 50 8.97 8.29 14.41 ACGCCGT 0595 50.1 0.102 0.110 1988 LWP 50 10.47 6.18 13.49 ACGCCGT 0595 50.1 0.020 0.426 16 1998 LWD 17 21.38 12.54 26.36 ACGCCTG 0154 16.6 0.067 0.008 2000 AD 20 8.35 7.55 15.52 ACGCCTG 0154 16.6 0.074 0.422 2000 EWD 20 7.8 5.73 18.42 ACGCCTG 0154 16.6 0.065 0.456 1998 EWD 21 5.38 4.68 9.22 ACGCCTG 0154 16.6 0.023 0.660 2001 LWP 22 15.24 7.41 16.7 ACGCCTG 0154 16.6 0.066 0.342 2003 LWD 23 11.55 9.42 20.62 ACGCCTG 0154 16.6 0.012 0.561 1996 LWP 23 16.52 10.41 19.86 ACGCCTG 0154 16.6 0.014 0.693 1986 LWD 35 17.74 9.92 22.71 ACGCCGG 0180 35.1 0.027 0.602 17 1993 LWD 0 22.25** 3.44 10.06 ACACCGG 0310 0 0.018 0.260 1995 AD 6 5.87 2.46 7.42 ACGCCGG 0311 10.1 0.149 0.117 1991 AD 6 10.67 4.9 13.54 ACGCCGG 0311 10.1 0.004 0.574 1988 LWD 6 6.11 2.59 9.56 ACGCCGG 0311 10.1 0.129 0.072 1992 LWD 6 39.46** 6.84 22.72 ACGCCGG 0311 10.1 0.087 0.163 1995 LWD 11 5.26 1.74 5.75 ACGCCGG 0311 10.1 0.131 0.144 1993 AD 34 13.26 4.74 14.11 ACGCCTC 0189 34.6 0.027 0.116 2001 EWD 34 10.02** 2.96 6.59 ACGCCTC 0189 34.6 0.007 0.696 1994 LWD 34 8.59** •1.72 5.56 ACGCCTC 0189 34.6 0.124 0.104 1985 LWD 34 43.82** 2.89 10.39 ACGCCTC 0189 34.6 0.029 0.322 18 1988 AD 3 3.09 2.59 6.89 ACGCCTC 0144 0 0.001 0.837 ** significant at a = 0.01. EWD = earlywood density LWD = latewood density LWP = latewood proportion AD = average density 156 Table 4.5: Number of undetected QTL, the average effect of undetected QTL and the expected number of QTL for composite traits. "Trait n? 6" D M n Q T L c CI (+/-)" Mulcted" Total Q T L HT 2 0.15 0.34 0.17 21.04 3.49 0.0161 23.04 VOL 2 0.09 0.21 0.10 12.01 1.99 0.0168 14.01 FL 3 0.02 0.27 0.09 4.00 1.00 0.0105 7.00 CS 3 0.04 0.27 0.09 5.10 1.27 0.0165 8.10 LWD 3 0.03 0.18 0.06 6.10 1.52 0.0126 9.10 GAL 2 0.03 0.17 0.08 3.19 0.53 0.0140 5.19 GLU 2 0.04 0.11 0.05 6.23 1.04 0.0121 8.23 " number of detected QTLs " the threshold effect size for detecting QTLs c number of undetected QTLs d 95% confidence intervals for the estimate of undetected QTLs ' the average effect of undetected QTLs Discussion QTLs and QTL Effects Many QTL studies in trees have detected QTLs for growth and wood quality traits. In coastal Douglas-fir, a comprehensive QTL study was conducted on adaptive traits, however, to date, the present study is the only known QTL analysis extensively evaluating commercially important wood quality traits in Douglas-fir. Furthermore, most QTL studies have been conducted using a single pedigree growing on a single site. This study was conducted using multiple, unrelated outbred families replicated on the four sites. The resulting QTL map is a synthesis of genotypic and phenotypic information from all eight families on four sites. This approach to QTL mapping may provide a broader picture of QTL activity within a population rather than a restrictive approach which evaluates a single individual or family. 157 The low number of detected QTLs can be explained by either 1) a low number of sampled individuals per family or 2) conflicting QTL effects across families. The study included 320 offspring, however, at any given marker, linkage can be determined by as few as 40 individuals from a single family and may limit the power of QTL detection. Further, if multiple families contribute to a marker, yet the families differ for QTLs located near the marker, the effects of those QTLs can be lost through the sib-pair regression analysis. Therefore, the QTLs presented here are those of largest effect and greatest importance in trait development within the population. QTLs that occur across time, space and genetic background are the most useful and can be used for marker aided selection, studying the genetic architecture of quantitative traits and positional selection of candidate genes (Wheeler et al. 2005). The number and effects of detected QTLs concurs with previously reported QTL studies for growth and wood quality traits in several species (Arcade et al. 2002; Bradshaw and Stettler 1995; Grattapaglia et al. 1996; Markussen et al. 2003; Sewell et al. 2000; Sewell et al. 2002; Wu 1998; Yoshimaru et al. 1998). Many of these studies report clustering of QTLs for related and unrelated traits. Yoshimaru et al. (1998) report QTL clusters for height, diameter and female fertility, while Bradshaw and Stettler (1995) discovered QTL clusters for growth, branch and leaf area traits. Several QTL clusters were detected in this study for growth, density, fibre properties and wood chemistry traits. These clusters most likely represent pleiotropic effects, but may be evidence for clusters of linked genes (Grattapaglia et al. 1996). The QTLs for height and volume on group two and diameter and volume on group four are likely identification of genes that affect height and diameter with corresponding effects on volume. The QTL for height on 158 linkage group one was not associated with a volume QTL, and as such, may be less important for volume production or the effect on volume too small to detect. Pleiotropic effects are probably responsible for the co-localization of QTLs affecting fibre length and coarseness on linkage group 18. However, the large QTL cluster for wood chemistry traits identified on group 15 cannot be attributed to pleiotropy without also considering a cluster of linked genes. Several QTLs for individual ring density traits co-locate and are likely independent verification of the same gene (Brown et al. 2003; Neale et al. 2002). The only compound wood density trait for which QTLs were detected is latewood density. The lack of QTL detection for multiple compound wood density traits is likely a result of the large degree of variation in gene expression patterns in response to temporal climate variation. Using average values across multiple rings may reduce the power of detecting QTLs. The detection of QTLs for compound traits is most useful since they incorporate variation across multiple years (Neale et al. 2002), however, the use of individual ring traits is superior for studying temporal variation in gene expression and QTL identification. In total, there are 11 QTL clusters for individual ring wood density traits. Many of these clusters are between seasonal traits (earlywood or latewood) and average density (linkage group 2, 4, 11, and 17), and are likely QTL that affect average density via the respective seasonal trait. Other QTL clusters for wood density traits include combinations of latewood density, earlywood density, latewood proportion and average density and reflect genes that have a general effect on wood density throughout the year. There are several other wood density traits that are detected in single years. Neale et al. (2002) 159 suggest that QTLs detected in multiple years are likely verified QTLs, whereas those detected in only a single year have a greater potential to be false positives. QTL clusters, such as on group one (Figure 4.1), may be evidence of genes that are determinants of wood density, whereas those detected in only a few years may be evidence for gene action in response to biotic or abiotic factors (Brown et al. 2003). Several QTLs for wood density co-locate with QTL for other traits. A QTL for height and volume co-locates with a QTL affecting earlywood density and average density on linkage group two, and a QTL for diameter and volume co-locates with a QTL affecting latewood density and average density on linkage group four. The co-location of QTLs for wood density and growth traits is further evidence of pleiotropic effects and may be ideal targets for candidate gene mapping and marker aided breeding. Other interactions with QTLs for wood density occur with coarseness on groups 14 and 16, fibre length on group 17, fibre length and coarseness on group 18, arabinose on group 17 and wood chemistry traits on group 15. These examples may be further evidence of pleiotropy and can help to biologically explain correlations between these traits. However, several traits resulted in unique QTL, such as microfibril angle on group 12 (Figure 4.1), which are likely positions of unique genes with major effect for the respective trait. Expected Number of QTLs QTL studies are known to underestimate the total number of loci involved in trait determination (Beavis 1998; Otto and Jones 2000). Strauss et al. (1992) suggest that one 160 of the important applications of QTL analyses is the identification of the number of QTLs controlling quantitative traits. QTL studies are limited in size and include a restricted number of individuals and markers. As a result, only a small number of the actual number of QTLs can be detected, and consequently the detected QTLs are those with large effect (Beavis 1998). Assuming an exponential distribution of effects, the number of undetected QTLs and their average effect size can be estimated for traits where two or more QTLs have been detected (Otto and Jones 2000). However, depending on the history of each allele (its effect on fitness and the mode of selection acting upon it), an exponential distribution of QTL effects may not be an appropriate assumption and the results may be biased. In certain circumstances, the more flexible gamma distribution may be appropriate and more accurate when estimating the number of undetected QTLs and their effects (Otto and Jones 2000). The current study provides estimates for growth and wood traits based on an exponential distribution of QTL effects due to its theoretical support and relative simplicity (Otto and Jones 2000). Otto and Jones (2000) compared their QTL based estimator of undetected QTL to the Castle-Wright-Zeng estimator. The QTL based estimator was shown to be more robust and provided superior estimates and more accurate confidence limits of undetected QTL. However, the QTL based estimate is most accurate when > 3 QTLs have been detected, and generally overestimate the number of undetected QTLs when based on less than three detected QTLs. The estimate is robust against dominance interactions and opposing additive QTL effects. Growth traits are affected by a large number of QTL of small effect. Growth occurs at a time of prolific gene activity which can affect height and volume growth. Volume is affected by fewer QTL than height suggesting that some height QTLs are 161 negligible when considering their effect on volume. The QTLs detected for height and volume were of rather large effect (>9.5%), whereas the average effect size of undetected QTLs is estimated at 1.6-1.7%. This implies a small number of QTL of large effect and a large number of QTL with very small effect. One of the potential applications of QTL studies is marker aided breeding. Knowledge of the number of QTLs affecting traits of interest within breeding populations and individual families is a valuable tool for modeling QTL effects and stability through time in breeding programs (Alvin Yanchuk, personal communication). The expected number of QTL and their effect sizes can be used in these types of applications. Temporal Stability of QTLs Many QTL studies report detection of QTLs in multiple years. Yoshimaru et al (1998) detected two QTLs for height in Japanese cedar (Cryptomeria japonica D. Don) from two consecutive years (four and five years) at the same location, but a third QTL for year 14 height growth at a separate location. The control of height growth, like any quantitative trait, is complex and varies through time. QTL studies of height growth are snap-shots of the accumulation of gene activity through time. This study estimates QTLs for height at year 26 in Douglas-fir. These height QTLs represent genes of largest effect that control height growth at this particular time, and are not likely candidate QTLs for early growth. However, this type of consecutive analysis is warranted and would require the detection of QTL activity within individual years by evaluating height increments. 162 Density traits vary through time as climate varies and maturation occurs. Neale et al. (2002) report temporal variation in detected QTLs suggesting that some QTLs occur over the duration of growth, whereas others occur in only later years. In the current study, the density QTLs on group five show interesting patterns of temporal variation. The QTLs detected from 24 - 27cM occur in years 1990 - 1993. The second QTL on the same group (7 - 9cM) occurs from 1994 - 2002. Other QTLs seem to be present throughout the duration of the experiment (20 - 35cM on linkage group one), whereas others seem to occur in the early years (0 - 1 lcM on group 17). These patterns of temporal variation are likely a result of maturation. Studies have shown that density traits and their heritabilities vary through time for coastal Douglas-fir and are associated with environmental signals (Vargas-Hernandez and Adams 1992, 1994; Vargas-Hernandez et al. 1994). The pattern of detection of QTLs through time likely reflects both maturation and responses to biotic and abiotic factors. 163 References Arcade, A., Faivre-Rampant, P., Paques, L.E., and Prat, D. 2002. Localisation of genomic regions controlling microdensitometric parameters of wood characteristics in hybrid larches. Annals of Forest Science. 59 (5-6): 607-615. Aubry, C.A., Adams, W.T., and Fahey, T.D. 1998. Determination of relative economic weights for multitrait selection in coastal Douglas-fir. Canadian Journal of Forest Research. 28 (8): 1164-1170. Beavis, WD. 1998. QTL analyses: power, precision and accuracy. In Molecular dissection of complex traits. CRC Press, Raleigh, North Carolina, pp. 145-162. Bradshaw, H.D., and Stettler, R.F. 1995. Molecular genetics of growth and development in Populus. IV. Mapping QTLs with large effects on growth, form, and phenology traits in a forest tree. Genetics. 139: 963-973. Brown, G.R., Bassoni, D.L., Gill, G.P., Fontana, J.R., Wheeler, N.C., Megraw, R.A., Davis, M.F., Sewell, M.M., Tuskan, G. A , and Neale, D.B. 2003. Identification of quantitative trait loci influencing wood property traits in loblolly pine (Pinus taedaL.). IE. QTL verification and candidate gene mapping. Genetics. 164 (4): 1537-1546. Byrne, M., Murrell, J.C., Owen, J.V., Williams, E.R., and Moran, G.F. 1997. Mapping of quantitative trait loci influencing frost tolerance in Eucalyptus nitens. Theoretical and Applied Genetics. 95 (5-6): 975-979. Chagne, D., Brown, G., Lalanne, C , Madur, D., Pot, D., Neale, D., and Plomion, C. 2003. Comparative genome and QTL mapping between maritime and loblolly pines. Molecular Breeding. 12 (3): 185-195. Chagne, D., Lalanne, C , Madur, D., Kumar, S., Frigerio, J.M., Krier, C , Decroocq, S., Savoure, A , Bou-Dagher-Kharrat, M., Bertocchi, E., Brach, J., and Plomion, C. 2002. A high density genetic map of maritime pine based on AFLPs. Annals of Forest Science. 59 (5-6): 627-636. 164 Chantre, G., Rozenberg, P., Baonza, V., Macchioni, N , Le Turcq, A., Rueff, M., Petit-Conil, M., and Heois, B. 2002. Genetic selection within Douglas fir (Pseudotsuga menziesii) in Europe for papermaking uses. Annals of Forest Science. 59 (5-6): 583-593. Cramer, S., Kretschmann, D., Lakes, R., and Schmidt, T. 2005. Earlywood and latewood elastic properties in loblolly pine. Holzforschung. 59: 531-538. Donaldson, L.A., and Burdon, R.D. 1995. Clonal variation and repeatability of microfibril angle in Pinus radiata. New Zealand Journal of Forestry Science. 25 (2): 164-174. Ericsson, T., and Fries, A. 2004. Genetic analysis of fibre size in a full-sib Pinus sylvestrish. progeny test. Scandinavian Journal of Forest Research. 19(1): 7-13. Evans, R., and Ilic, J. 2001. Rapid prediction of wood stiffness from microfibril angle and density. Forest Products Journal. 51 (3): 53-57. Grattapaglia, D., Bertolucci, F.L.G., Penchel, R., and Sederoff, R.R. 1996. Genetic mapping of quantitative trait loci controlling growth and wood quality traits in Eucalyptus grandis using a maternal half-sib family and RAPD markers. Genetics. 144: 1205-1214. Hannrup, B., Cahalan, C , Chantre, G , Grabner, M., Karlsson, B., Le Bayon, I., Jones, G.L., Muller, U., Pereira, H , Rodrigues, J.C., Rosner, S., Rozenberg, P., Wilhelmsson, L., and Wimmer, R. 2004. Genetic parameters of growth and wood quality traits in Picea abies. Scandinavian Journal of Forest Research. 19: 14-29. Hannrup, B., Ekberg, I., and Persson, A. 2000. Genetic correlations among wood, growth capacity and stem traits in Pinus sylvestris. Scandinavian Journal of Forest Research. 15: 161-170. Hannrup, B., Wilhelmsson, L., and Danell, O. 1998. Time trends for genetic parameters of wood density and growth traits in Pinus sylvestris L. Silvae Genetica. 47 (4): 214-219. Haseman, J.K., and Elston, R.C. 1972. Investigation of linkage between a quantitative trait and a marker locus. Behavior Genetics. 2 (1): 3-9. 165 Hu, X.-S., Goodwillie, C , and Ritland, K.M. 2004. Joining genetic linkage maps using a joint likelihood function. Theoretical and Applied Genetics. 109: 996-1004. Huntley, S.K., Ellis, D., Gilbert, M , C , C , and Mansfield, S.D. 2003. Significant increases in pulping efficiency in C4H-F5H-transformed poplars: improved chemical savings and reduced environmental toxins. Journal of Agricultural and Food Chemistry. 51: 6178-6183. Ivkovich, M. 1996. Genetic variation of wood properties in balsam poplar (Populus balsamifera L). Silvae Genetica. 45 (2-3): 119-124. Ivkovich, M., Namkoong, G., and Koshy, M. 2002a. Genetic variation in wood properties of interior spruce. I. Growth, latewood percentage, and wood density. Canadian Journal of Forest Research. 32 (12): 2116-2127. Ivkovich, M., Namkoong, G., and Koshy, M. 2002b. Genetic variation in wood properties of interior spruce. II. Tracheid characteristics. Canadian Journal of Forest Research. 32: 2128-2139. Jermstad, K.D., Bassoni, D.L., Jech, K.S., Ritchie, G.A., Wheeler, N.C., and Neale, D.B. 2003. Mapping of quantitative trait loci controlling adaptive traits in coastal Douglas fir. III. Quantitative trait loci-by-environment interactions. Genetics. 165 (3): 1489-1506. Jermstad, K.D., Bassoni, D.L., Jech, K.S., Wheeler, N.C., and Neale, D.B. 2001a. Mapping of quantitative trait loci controlling adaptive traits in coastal Douglas-fir. I. Timing of vegetative bud flush. Theoretical and Applied Genetics. 102 (8): 1142-1151. Jermstad, K.D., Bassoni, D.L., Wheeler, N.C., Anekonda, T.S., Aitken, S.N., Adams, W.T., and Neale, D.B. 2001b. Mapping of quantitative trait loci controlling adaptive traits in coastal Douglas-fir. II. Spring and fall cold-hardiness. Theoretical and Applied Genetics. 102: 1152-1158. Jermstad, K.D., Bassoni, D.L., Wheeler, N.C., and Neale, D.B. 1998. A sex-averaged genetic linkage map in coastal Douglas-fir (Pseudotsuga menziesii [Mirb.] Franco var 'menziesii') based on RFLP and RAPD markers. Theoretical and Applied Genetics. 97: 762-770. 166 Johnson, G.R., Wheeler, N.C., and Strauss, S.S. 2000. Financial feasibility of marker-aided selection in Douglas-fir. Canadian Journal of Forest Research. 30 (12): 1942-1952. Krutovsky, K.V., Troggio, M., Brown, G.R., Jermstad, K.D., and Neale, D.B. 2004. Comparative Mapping in the Pinaceae. Genetics. 168: 447 - 461. Loo, J.A., Tauer, C.G., and McNew, R.W. 1985. Genetic variation in the time of transition from juvenile to mature wood in loblolly pine (Pinus taeda L). Silvae Genetica. 34 (1): 14-19. Louzada, J., and Fonseca, F.M. A. 2002. The heritability of wood density components in Pinus pinaster Ait. and the implications for tree breeding. Annals of Forest Science. 59 (8): 867-873. Markussen, T., Fladung, M., Achere, V., Favre, J.M., Faivre-Rampant, P., Aragones, A., Perez, D.D., Harvengt, L., Espinel, S., and Ritter, E. 2003. Identification of QTLs controlling growth, chemical and physical wood property traits in Pinus pinaster (Ait.). Silvae Genetica. 52 (1): 8-15. Megraw, R.A., Leaf, G., and Bremer, D. 1998. Longitudinal shrinkage and microfibril angle in loblolly pine. In Microfibril angle in wood. University of Canterbury Press, Christchurch, New Zealand, pp. 27-61. Myszewski, J.H., Bridgwater, F.E., Lowe, W.J., Byram, T.D., and Megraw, R.A. 2004. Genetic variation in the microfibril angle of loblolly pine from two test sites. Southern Journal of Applied Forestry. 28 (4): 196-204. Neale, D.B., Sewell, M.M., and Brown, G.R. 2002. Molecular dissection of the quantitative inheritance of wood property traits in loblolly pine. Annals of Forest Science. 59 (5-6): 595-605. Nicholls, J.W.P., Morris, J.D., and Pederick, L.A. 1980. Heritability estimates of density characteristics in juvenile Pinus radiata wood. Silvae Genetica. 29 (2): 54-61. Otto, S.P., and Jones, CD. 2000. Detecting the undetected: Estimating the total number of loci underlying a quantitative trait. Genetics. 156: 2093-2197. 167 Pot, D., Chantre, G., Rozenberg, P., Rodrigues, J.C., Jones, G.L., Pereira, H., Hannrup, B., Cahalan, C , and Plomion, C. 2002. Genetic control of pulp and timber properties in maritime pine (Pinus pinaster Ait.). Annals of Forest Science. 59: 563-575. Remington, D.L., Whetten, R.W., Liu, B.-H, and O'Malley, D.M. 1999. Construction of an AFLP genetic map with nearly complete genome coverage in Pinus taeda. Theoretical and Applied Genetics. 98: 1279-1292. SAS. Version 9.1. SAS System for Windows. Copyright (c) 2002 - 2003 SAS Institute Inc. SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc., Cary, NC, USA. Seaton, G., Haley, C.S., Knott, S.A., Kearsey, M., and Visscher, P.M. 2002. QTL express: mapping quantitative trait loci in simple and complex pedigrees. Bioinformatics. 18 (2): 339-340. Seth, R.S., and Kingsland, M A . 1990. The reinforcing properties of softwood kraft pulps. Pulp and Paper Canada. 91 (7): 68-72, 74-75. Sewell, M.M., Bassoni, D.L., Megraw, R.A, Wheeler, N.C., and Neale, D.B. 2000. Identification of QTLs influencing wood property traits in loblolly pine (Pinus taeda L.). I. Physical wood properties. Theoretical and Applied Genetics. 101 (8): 1273-1281. Sewell, M.M., Davis, M.F., Tuskan, G.A., Wheeler, N.C, Elam, C.C, Bassoni, D.L., and Neale, D.B. 2002. Identification of QTLs influencing wood property traits in loblolly pine (Pinus taeda L.). JJ. Chemical wood properties. Theoretical and Applied Genetics. 104 (2-3): 214-222. Sewell, M.M., Sherman, B.K., and Neale, D.B. 1999. A consensus map for loblolly pine (Pinus taeda L.). I. Construction and integration of individual linkage maps from two outbred three-generation pedigrees. Genetics. 151 (1): 321-330. St.Clair, J.B. 1994. Genetic variation in tree structure and its relation to size in Douglas-fir. I. Biomass partitioning, foliage efficiency, stem form, and wood density. Canadian Journal of Forest Research. 24 (6): 1226-1235. 168 Strauss, S.H., Lande, R., and Namkoong, G. 1992. Limitations of molecular-marker-aided selection in forest tree breeding. Canadian Journal of Forest Research. 22 (7): 1050-1061. Talbert, J.T., Jett, J.B., and Bryant, R.L. 1983. Inheritance of wood specific gravity in an unimproved loblolly pine population: 20 years of results. Silvae Genetica. 32 (1-2): 33-37. Tappi Useful Method. 1991. UM-250: Acid-soluble lignin in wood and pulp. Thamarus, K., Groom, K., Bradley, A., Raymond, C.A., Schimleck, L.R., Williams, E.R., and Moran, G.F. 2004. Identification of quantitative trait loci for wood and fibre properties in two full-sib pedigrees of Eucalyptus globulus. Theoretical and Applied Genetics. 109 (4): 856-864. Travis, S.E., Ritland, K., Whitham, T.G., and Keim, P. 1998. A genetic linkage map of Pinyon pine (Pinus edulis) based on amplified fragment length polymorphisms. Theoretical and Applied Genetics. 97 (5-6): 871-880. USD A. 2000. Wood handbook: wood as an engineering material. University Press of the Pacific, Madison, Wisconsin. 463 pgs. Van Ooijen, J.W., and Voorrips, R.E. 2001. JoinMap 3.0, software for the calculation of genetic linkage maps. Plant Research International, Wageningen, the Netherlands. Vargas-Hernandez, J., and Adams, W.T. 1991. Genetic variation of wood density components in young coastal Douglas-fir - Implications for tree breeding. Canadian Journal of Forest Research. 21 (12): 1801-1807. Vargas-Hernandez, J., and Adams, W.T. 1992. Age-age correlations and early selection for wood density in young coastal Douglas-fir. Forest Science. 38 (2): 467-478. Vargas-Hernandez, J., and Adams, W.T. 1994. Genetic relationships between wood density components and cambial growth rhythm in young coastal Douglas-fir. Canadian Journal of Forest Research. 24 (9): 1871-1876. Vargas-Hernandez, J., Adams, W.T., and Krahmer, R.L. 1994. Family variation in age trends of wood density traits in young coastal Douglas-fir. Wood and Fiber Science. 26 (2): 229-236. 169 Visscher, P.M., and Hopper, J.L. 2001. Power of regression and maximum likelihood methods to map QTL from sib-pair and DZ twin data. Annals of Human Genetics. 65: 583-601. Wang, H.H., Drummond, J.G., Reath, S.M., Hunt, K., and Watson, P. A. 2001. An improved fibril angle measurement method for wood fibres. Wood Science and Technology. 34: 493-503. Wheeler, N.C., Jermstad, K.D., Krutovsky, K., Aitken, S.N., Howe, G.T., Krakowski, J., and Neale, D.B. 2005. Mapping of quantitative trait loci controlling adaptive traits in coastal Douglas-fir. IV. Cold-hardiness QTL verification and candidate gene mapping. Molecular Breeding. 15 (2): 145-156. Wu, H.X. 2002. Study of early selection in tree breeding. 4. Efficiency of marker-aided early selection (MAES). Silvae Genetica. 51 (5-6): 261-269. Wu, R.L. 1998. Genetic mapping of QTLs affecting tree growth and architecture in Populus: implications for ideotype breeding. Theoretical and Applied Genetics. 96; 447-457. Wu, R.L., Han, Y.F., Hu, J.J., Fang, J.J., Li, L., Li, M.L., and Zeng, Z.B. 2000. An integrated genetic map of Populus deltoides based on amplified fragment length polymorphisms. Theoretical and Applied Genetics. 100 (8): 1249-1256. Yanchuk, A.D., and Kiss, G.K. 1993. Genetic variation in growth and wood specific gravity and its utility in the improvement of interior spruce in British Columbia. Silvae Genetica. 42 (2-3): 141-148. Yoshimaru, H , Ohba, K., Tsurumi, K., Tomaru, N., Murai, M., Mukai, Y., Suyama, Y., Tsumura, Y., Kawahara, T., and Sakamaki, Y. 1998. Detection of quantitative trait loci for juvenile growth, flower bearing and rooting ability based on a linkage map of sugi (Cryptomeria japonica D. Don). Theoretical and Applied Genetics. 97:45-50. 170 Chapter 5 Conclusions This research presents a much needed comprehensive examination of the quantitative genetics of wood quality and growth traits in coastal Douglas-fir. The heritabilities, genetic and phenotypic correlations and QTLs are descriptive of a small population of full-sib coastal Douglas-fir families in south-western British Columbia that have been selected for superior growth traits. The trees sampled represent genetically improved stock grown in natural conditions of the Ministry of Forests second generation progeny test program. The findings suggest that growth traits in coastal Douglas-fir are under moderately strong genetic control with heritability estimates of 0.23 - 0.30. Only a small number of QTLs of relatively large effect were detected for height, diameter and volume at age 26. Height and volume are expected to be controlled by a large number of QTLs (approximately 23 and 14, respectively). The large amount of environmental influence on growth traits may be a result of interactions between many genes that control these traits, and their environment. QTL that occur consistently through time, space and genetic background will be most useful to tree breeders and are ideal targets for candidate gene mapping. Fibre properties are influenced quite heavily by their surrounding environment. The heritability estimate for fibre length is 0.10 and fibre coarseness 0.18. Three QTLs 171 were detected for each fibre trait. The number of expected QTLs for control of fibre properties is approximately 7 - 8. It is likely that some of these QTLs are common to both traits so that the actual number of QTLs affecting fibre traits is lower. The small heritability estimates suggest that QTLs affecting fibre properties are affected by environmental variation, whether seasonal climatic differences (earlywood and latewood differences) or differences in maturation through time. Density traits are under strong genetic control with relatively large heritability estimates (0.21 - 0.54). This suggests consistent within family patterns of QTL activity. The only QTLs detected for compound density traits were for latewood density. Three QTLs were detected for latewood density and it is expected that approximately nine QTLs are responsible for average latewood density at age 26. Latewood is produced in late summer when resources diminish and growth is slowed. With less physiological activity during active latewood production, it is no surprise that it is affected by a smaller number of genes. Earlywood is produced during active growth in spring when resources are abundant, and as such, it is expected to be affected by a larger number of QTLs with small effect, similar to growth traits. The fact that no QTLs were detected for compound earlywood density supports this hypothesis. Wood chemistry traits are under a wide range of genetic control (h2 = 0.16 -0.98). QTLs were detected for lignin, glucose, galactose, mannose and arabinose. The lack of QTL detection for xylose is likely due to control by a large number of QTLs of small effect. Glucose is a prominent cell-wall constituent and is likely influenced by relatively few genes. This is also true of galactose. Glucose is expected to be controlled by 5 and galactose 8 QTLs of small effect. Wood chemical traits are likely influenced by 172 conditions in their surroundings such as patterns of gene expression and carbon allocation. As a result, genes that control one trait are likely to influence others. Many detected QTLs co-located to similar positions on the linkage map. These interactions may represent pleiotropic effects and can be explained by trait correlations. Height and diameter QTLs co-locate with volume in two instances. Given the high correlations between growth traits, co-location of QTLs and pleiotropic effects are expected. These two regions also co-locate with QTL clusters for density traits. The height/volume QTL co-locates with a QTL cluster involving earlywood density, and the diameter/volume QTL co-locates with a QTL cluster involving latewood density. Growth traits are generally negatively correlated with density traits, however, earlywood density and latewood density show no significant genetic or phenotypic correlations with height or diameter for this population. Given the QTL results and assuming pleiotropic effects, genes controlling height growth will influence earlywood density while those affecting diameter will influence latewood density. However, given the low correlations between these traits, the co-location of QTL may indicate the existence of unique genes located near each other in the genome. A QTL cluster for wood chemistry traits involving lignin, glucose, galactose and mannose co-locates with a QTL for compound latewood density. Lignin is negatively correlated with latewood density (-0.54; p = 0.04) as is galactose (-0.17; p = 0.53). Both glucose and mannose have small positive correlations with latewood density, although not significant (0.13 and 0.40, respectively). The co-location of QTL and relatively strong correlations suggest that some, if not all, of these QTL represent few or a single gene 173 with pleiotropic effects. Arabinose co-locates with a QTL cluster for latewood and earlywood density, however, no significant correlations exist between these traits. Several QTLs for fibre length and coarseness co-locate with wood density QTL clusters. Again, there are no significant correlations between wood density traits and fibre properties. The co-location of QTL for density and fibre property traits may be due to pleiotropic effects and may indicate genes with large effect on one trait and small effect on another, or may be evidence for gene clusters. Recommendations for Improvement One of the key components to experimental design is the consideration of sample size. This study used a variety of techniques to explore the quantitative genetics of wood quality and growth traits. The phenotypic correlations and heritabilities are strong and an adequate sample size was used for these estimates. However, genetic correlations are difficult to estimate and require a large number of families. The genetic correlations presented in this thesis show large standard errors and, in some situations, erroneous estimates. To reduce the standard errors and improve estimates, at least 15 more full-sib families should be subject to and added to the analysis. The calculation of linkage statistics for the creation of genetic linkage maps is largely dependent on the number of individuals within families. With more individuals, these statistics become more precise (higher confidence). The number of individuals within families used to calculate linkage statistics in this study was 40. For a stronger map, the number of individuals within families should be increased to between 50 and 174 100. The use of the joint likelihood function to calculate linkage statistics in this analysis allowed the integration of genetic data from eight full-sib families with relatively little marker information. Using conventional methods, the sample sizes and marker information required to integrate eight full-sib families on a single map would be exhausting. However, in order to calculate linkage statistics between markers, the joint likelihood function requires that markers belong to common families. When more than one family is common to two markers, average linkage statistics can be calculated. The ideal situation would be to identify backcross markers that segregate in all eight families. If all markers segregated in all eight families, the resulting linkage map would be a true average map for the eight families studied. This is a difficult task given the amount of genetic diversity for neutral markers in conifers. By increasing the number of AFLP primer combinations used in the analysis (perhaps doubling the number), a more accurate average linkage map could be attained that may better represent the 13 haploid chromosomes of the Douglas-fir genome. QTL analyses are dependent on marker density and the number of individuals within families. The sib-pair analysis used to identify QTLs in this study is the ideal technique for integrating phenotypic and genetic data from multiple full-sib families. The sib-pair analysis was designed for use with multiple, small full-sib families. This study used eight large full-sib families which was sufficient to generate accurate positions of QTLs. The linkage map used in this study was of sufficient density for QTL analysis. Increasing the marker density and improving the family contributions would result in a more accurate QTL analysis for the sample population. 175 Research Significance The methods presented in this thesis are relevant to tree improvement programs that are considering the incorporation of marker aided selection (MAS). One of the major setbacks to the implementation of MAS in tree improvement programs has been the inability to generate linkage maps and perform QTL analyses on multiple outbred families. By using a combination of the joint likelihood approach for linkage mapping and sib-pair analysis for QTL mapping, the integration of data from multiple outbred families is possible. The resulting QTL map would reveal QTL of strong effect that occur across families and locations. These QTL are most important to tree breeders as they will work consistently given altered genetic backgrounds and environmental conditions. Tree breeders are interested in the flow of genes within the breeding population and their effect on traits of interest. In order to model such activities, researchers require an estimate of the number of genes controlling any given trait. The approach outlined in this thesis can be used with the results of any QTL analysis to predict the number and effect size of QTLs affecting traits of interest. In this way, a more accurate number of QTLs can be used in models of gene flow within breeding populations. Future Research Prospects The comparison of maps within and between species is a novel approach to integrating data from different studies. The integration of codominant markers common with markers used in other linkage map and QTL studies will allow comparison of linkage groups and QTL data. The Conifer Comparative Genome Project (CCGP) has 176 made available ESTPs (expressed sequence tagged polymorphisms) for use as anchor loci for comparing studies of conifer genomes. Using this approach, the Douglas-fir QTL map for growth and wood quality traits can be compared to similar studies in other species and may be used for future verification studies. 177 Appendix 1 ANOVA Tables Trait Source DF MS F Prob<F DBH Family 14 150.0416 5.24 <0.0001 Site 3 1902.4705 46.68 <0.0001 Family* Site 42 28.6404 1.22 0.1909 Block(Site) 12 40.7536 1.93 0.0297 Family*Block(Site) 165 23.4808 1.11 0.2038 Error . 363 21.0945 HT Family 14 38.2861 4.09 0.0002 Site 3 341.4325 16.68 0.0001 Family* Site 42 9.3520 1.32 0.1123 Block(Site) 12 20.4681 4.17 <0.0001 Family*Block(Site) 165 7.0801 1.44 0.0024 Error 362 4.9136 VOL Family 14 0.1571 5.03 O.0001 Site 3 1.3253 33.82 O.0001 Family* Site 42 0.0312 1.67 0.0127 Block(Site) 12 0.0392 2.29 0.0083 Family*Block(Site) 165 0.0187 1.09 0.2466 Error 362 0.0171 FL Family 14 0.1398 2.63 0.0079 Site 3 1.6353 19.41 0.0001 Family* Site 42 0.0532 1.28 0.1368 Block(Site) 12 0.0843 2.09 0.0169 Family*Block(Site) 164 0.0414 1.03 0.4113 Error 347 0.0403 Cs Family 14 4.5207E-03 5.35 <0.0001 Site 3 5.5236E-02 26.53 0.0001 Family* Site 42 8.4508E-04 1.02 0.4510 Block(Site) 12 2.0819E-03 2.10 0.0166 Family*Block(Site) 164 8.2973E-04 0.84 0.9043 Error 347 9.9276E-04 178 Appendix J continued: Trait Source DF MS F Prob<F LWP Family 14 0.0832 6.79 <0.0001 Site 3 0.7831 28.72 <0.0001 Family* Site 42 0.0122 1.10 0.3362 Block(Site) 12 0.0273 2.23 0.0101 Family*Block(Site) 165 0.0112 .0.91 0.7424 Error 361 0.0122 LWD Family 14 3710.3303 4.42 <0.0001 Site 3 45503.8544 31.97 <0.0001 Family* Site 42 839.4974 1.02 0.4517 Block(Site) 12 1423.3608 2.11 0.0156 Family*Block(Site) 165 824.6554 1.22 0.0603 Error 361 673.8951 EWD Family 14 2769.7074 13.09 <0.0001 Site 3 7172.6071 11.44 0.0008 Family* Site 42 211.6192 1.04 0.4114 Block(Site) 12 626.8776 3.06 0.0004 Family*Block(Site) 165 202.7432 0.99 0.5252 Error 361 204.9057 AD Family 14 5648.4720 9.63 <0.0001 Site 3 28430.4395 17.89 0.0001 Family* Site 42 586.2844 1.29 0.1315 Block(Site) 12 1588.8255 3.07 0.0004 Family*Block(Site) 165 453.8017 0,88 0.8332 Error 361 517.7905 Ara Family 14 8.3441E-06 2.28 0.0199 Site 3 1.5904E-05 16.90 0.0001 Family* Site 42 3.6583E-06 4.42 <0.0001 Block(Site) 12 9.4131E-07 1.68 0.0690 Family*Block(Site) 165 8.2733E-07 1.48 0.0013 Error 361 5.5988E-07 Gal Family 14 2.1471E-04 4.16 0.0002 Site 3 1.1044E-03 22.94 <0.0001 Family* Site 42 5.1613E-05 0.89 0.6686 Block(Site) 12 4.8139E-05 0.98 0.4650 Family*Block(Site) 165 5.8207E-05 1.19 0.0927 Error 361 4.8997E-05 179 Appendix J continued: Trait Source DF MS F Prob<F Glu Family 14 7.1038E-03 14.58 <0.0001 Site 3 2.5908E-03 17.68 0.0001 Family* Site 42 4.8734E-04 2.41 <0.0001 Block(Site) 12 1.4651E-04 1.00 0.4467 Family*Block(Site) 165 2.0215E-04 1.38 0.0063 Error 361 1.4625E-04 Xyl Family 14 1.7766E-04 6.67 <0.0001 Site 3 2.6582E-04 16.08 0.0002 Family* Site 42 2.6631E-05 2.48 <0.0001 Block(Site) 12 1.6530E-05 2.04 0.0200 Family*Block(Site) 165 1.0741E-05 1.33 0.0145 Error 361 8.0883E-06 Man Family 14 3.6388E-04 3.10 0.0023 Site 3 1.1270E-03 14.61 0.0003 Family* Site 42 1.1747E-04 2.07 0.0006 Block(Site) 12 7.7138E-05 1.72 0.0603 Family*Block(Site) 165 5.6699E-05 1.27 0.0348 Error 361 4.4780E-05 TL Family 14 5.8976E-04 1.25 0.2773 Site 3 2.5405E-03 10.98 0.0009 Family* Site 42 4.7128E-04 3.00 O.0001 Block(Site) 12 2.3130E-04 2.38 0.0057 Family*Block(Site) 165 1.5687E-04 1.62 0.0001 Error 361 9.7090E-05 MFA Family 14 134.3126 4.18 0.0002 Site 3 934.9823 78.05 <0.0001 Family* Site 42- 32.0976 1.11 0.3192 Block(Site) 12 11.9788 0.53 0.8932 Family*Block(Site) 165 28.9777 1.29 0.0257 Error 359 22.4877 180 Appendix 2 Family Means and Standard Deviations 181 Appendix 1.1: Growth trait family means and standard deviations. Parents Diameter (cm) Height (m) Volume (m ) Family Female Male Mean StdDev Mean StdDev Mean StdDev 2 408 358 24.60 6.80 18.10 2.80 0.36 0.19 7 206 210 24.08 6.84 16.43 3.70 0.33 0.18 26 441 344 26.93 6.10 17.67 2.69 0.40 0.20 38 214 129 26.50 6.26 18.61 2.94 0.42 0.19 46 33 227 22.96 4.23 15.91 2.47 0.26 0.11 56 199 292 22.80 5.38 15.92 3.08 0.27 0.14 62 300 62 25.76 6.37 18.52 2.92 0.40 0.20 75 121 439 25.85 6.00 18.05 2.69 0.38 0.19 92 350 231 21.89 6.30 15.97 3.31 0.27 0.17 130 46 426 20.35 5.31 15.80 2.62 0.23 0.13 150 73 48 23.41 5.26 16.96 3.13 0.30 0.15 151 67 452 21.40 5.55 15.99 2.59 0.24 0.13 154 152 70 22.49 5.08 16.73 2.59 0.28 0.13 155 83 32 23.46 5.29 17.17 2.56 0.31 0.15 156 440 101 22.04 5.30 15.89 2.50 0.25 0.13 Appendix 1.2: Fibre trait family means and standard deviations. Parents Fibre Length (mm) Fibre Microfibril Anlge (°) Coarseness (g/m) Family Female Male Mean StdDev Mean StdDev Mean StdDev 2 408 358 2.261 0.254 0.196 0.044 27.81 5.89 7 206 210 2.357 0.238 0.211 0.036 25.01 . 5.78 26 441 344 2.304 0.206 0.185 0.030 26.39 5.62 38 214 129 2.319 0.222 0.202 0.048 28.62 4.58 46 33 227 2.182 0.221 0.176 0.032 28.99 4.69 56 199 292 2.260 0.336 0.196 0.034 29.47 5.94 62 300 62 2.345 0.175 0.209 0.032 25.36 6.56 75 121 439 2.288 0.279 0.190 0.035 25.77 6.15 92 350 231 2.314 0.233 0.201 0.040 27.55 5.96 130 46 426 2.241 0.189 0.184 0.037 29.62 4.74 150 73 48 2.288 0.185 0.199 0.031 29.23 5.46 151 67 452 2.365 0.205 0.207 0.038 23.57 5.62 154 152 70 2.441 0.234 0.195 0.033 27.68 5.76 155 83 32 2.324 0.217 0.179 0.038 27.55 4.14 156 440 101 2.220 0.204 0.183 0.036 24.65 5.16 Appendix 1.3: Density trait family means and standard deviations. Parents Latewood Earlywood Latewood Average Proportion Density (g/m3) Density (g/m3) Denstiy (g/m3) Family Female Male Mean StdDev Mean StdDev Mean StdDev Mean StdDev 2 408 358 0.3769 0.0558 303.85 16.25 645.70 27.13 431.61 53.68 7 206 210 0.3773 0.0588 312.75 14.25 620.81 35.41 432.64 54.37 26 441 344 0.3361 0.0484 295.83 15.37 610.50 36.47 403.98 55.10 38 214 129 0.3098 0.0414 291.20 15.36 617.26 29.49 394.59 50.83 46 33 227 0.3625 0.0549 308.98 13.80 634.13 29.08 426.65 50.45 56 199 292 0.3296 0.0487 286.88 19.17 626.42 31.26 399.20 55.00 62 300 62 0.3504 0.0424 306.09 13.17 638.55 28.05 423.68 53.20 75 121 439 0.3222 0.0446 296.38 17.88 617.93 33.17 400.51 53.66 92 350 231 0.3808 0.0658 313.41 18.57 628.13 26.54 433.75 58.99 130 46 426 0.3590 0.0617 296.27 13.72 626.19 37.23 416.13 58.06 150 73 48 0.3613 0.0594 298.30 16.41 621.69 35.30 418.88 56.49 151 67 452 0.3312 0.0560 290.59 16.54 609.27 33.79 401.48 53.97 154 152 70 0.3570 0.0577 296.75 16.77 614.14 33.94 410.93 51.86 155 83 32 0.3499 0.0477 315.61 16.69 628.25 27.73 424.69 54.89 156 440 101 0.3545 0.0697 294.84 15.47 609.67 34.91 409.41 56.24 Appendix 1.4: Chemistry trait family means and standard deviations. Parents Arabinose Galactose Glucose Xylose Family Female Male Mean StdDev Mean StdDev Mean StdDev Mean StdDev 2 4 0 8 3 5 8 0 . 0 1 3 3 0 . 0 0 1 2 0 . 0 3 2 2 0 . 0 1 1 2 0 . 4 1 1 2 0 . 0 2 0 4 0 . 0 4 4 4 0 . 0 0 4 0 7 2 0 6 2 1 0 0 . 0 1 2 3 0 . 0 0 1 0 0 . 0 3 3 8 0 . 0 0 9 7 0 . 4 1 7 7 0 . 0 1 8 5 0 . 0 4 1 6 0 . 0 0 3 4 2 6 4 4 1 3 4 4 0 . 0 1 2 9 0 . 0 0 1 4 0 . 0 3 1 3 0 . 0 0 6 4 0 . 4 0 7 7 0 . 0 1 2 3 0 . 0 4 9 4 . 0 . 0 0 4 7 3 8 2 1 4 1 2 9 0 . 0 1 2 9 0 . 0 0 1 1 0 . 0 3 0 2 0 . 0 0 7 0 0 . 4 0 7 2 0 . 0 1 6 3 0 . 0 4 5 4 0 . 0 0 3 3 4 6 3 3 2 2 7 0 . 0 1 1 7 OTJOIO 0 . 0 2 9 4 0 . 0 0 6 1 0 . 4 2 5 1 0 . 0 1 0 9 0 . 0 4 6 0 0 . 0 0 3 8 5 6 1 9 9 2 9 2 0 . 0 1 2 2 0 . 0 0 0 8 0 . 0 3 0 7 0 . 0 0 6 9 0 . 4 3 4 7 0 . 0 1 2 3 0 . 0 4 6 0 0 . 0 0 3 8 6 2 3 0 0 6 2 0 . 0 1 2 8 0 . 0 0 1 1 0 . 0 2 7 9 0 . 0 0 4 4 0 . 4 0 8 3 0 . 0 1 2 2 0 . 0 4 3 7 0 . 0 0 2 8 7 5 1 2 1 4 3 9 0 . 0 1 3 1 0 . 0 0 1 2 0 . 0 3 7 4 0 . 0 0 8 1 0 . 4 0 9 4 0 . 0 1 3 1 0 . 0 4 7 8 0 . 0 0 3 0 9 2 3 5 0 2 3 1 0 . 0 1 3 2 0 . 0 0 1 0 0 . 0 2 9 6 0 . 0 0 5 5 0 . 4 1 6 4 0 . 0 1 3 4 0 . 0 4 7 6 0 . 0 0 3 4 1 3 0 4 6 4 2 6 0 . 0 1 2 2 0 . 0 0 0 7 0 . 0 3 1 3 0 . 0 0 4 8 0 . 4 4 3 3 0 . 0 1 1 4 0 . 0 4 5 7 0 . 0 0 3 8 1 5 0 7 3 4 8 0 . 0 1 2 4 0 . 0 0 0 7 0 . 0 3 5 6 0 . 0 0 8 8 0 . 4 3 1 1 0 . 0 1 3 9 0 . 0 5 0 3 0 . 0 0 3 5 151 6 7 4 5 2 0 . 0 1 2 6 0 . 0 0 1 0 0 . 0 3 1 4 0 . 0 0 7 5 0 . 4 0 3 4 0 . 0 1 6 3 0 . 0 4 7 2 0 . 0 0 2 8 1 5 4 1 5 2 7 0 0 . 0 1 1 8 0 . 0 0 0 9 0 . 0 3 1 7 0 . 0 1 0 9 0 . 4 4 3 5 0 . 0 1 8 2 0 . 0 4 4 6 0 . 0 0 3 5 1 5 5 8 3 3 2 0 . 0 1 2 1 0 . 0 0 1 0 0 . 0 3 1 9 0 . 0 0 7 2 0 . 4 3 7 6 0 . 0 1 3 5 0 . 0 4 4 9 0 . 0 0 3 1 1 5 6 4 4 0 1 0 1 0 . 0 1 2 2 0 . 0 0 0 7 0 . 0 3 3 5 0 . 0 0 6 8 0 . 4 3 7 4 0 . 0 1 0 0 0 . 0 4 7 5 0 . 0 0 2 3 Appendix 1.4 continued: Parents Mannose Lignin Family Female Male Mean StdDev Mean StdDev 2 408 358 0.1248 0.0101 0.3004 0.0164 7 206 210 0.1251 0.0099 0.3045 0.0157 26 441 344 0.1200 0.0066 0.3087 0.0128 38 214 129 0.1220 0.0050 0.3070 0.0089 46 33 227 0.1267 0.0070 0.3052 0.0097 56 199 292 0.1252 0.0069 0.3048 0.0096 62 300 62 0.1200 0.0053 0.3106 0.0122 75 121 439 0.1168 0.0051 0.3108 0.0088 92 350 231 0.1223 0.0049 0.3026 0.0123 130 46 426 0.1290 0.0081 0.3006 0.0106 150 73 48 0.1199 0.0082 0.3084 0.0167 151 67 452 0.1179 0.0082 0.3119 0.0156 154 152 70 0.1218 0.0121 0.3040 0.0150 155 83 32 0.1214 0.0092 0.2991 0.0116 156 440 101 0.1207 0.0071 0.3102 0.0127 

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}]}"
                            data-media="{[{embed.selectedMedia}]}"
                            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:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0074997/manifest

Comment

Related Items