Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Genetic variation among populations of Pissodes strobi (white pine weevil) reared from Picea and pinus… Lewis, Kornelia G. 1995

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

Item Metadata

Download

Media
831-ubc_1995-0621.pdf [ 19.69MB ]
Metadata
JSON: 831-1.0075158.json
JSON-LD: 831-1.0075158-ld.json
RDF/XML (Pretty): 831-1.0075158-rdf.xml
RDF/JSON: 831-1.0075158-rdf.json
Turtle: 831-1.0075158-turtle.txt
N-Triples: 831-1.0075158-rdf-ntriples.txt
Original Record: 831-1.0075158-source.json
Full Text
831-1.0075158-fulltext.txt
Citation
831-1.0075158.ris

Full Text

G E N E T I C V A R I A T I O N A M O N G POPULATIONS O F Pissodes strobi (WHITE PINE W E E V I L ) R E A R E D F R O M Picea AND Pinus HOSTS AS INFERRED F R O M RAPD M A R K E R S by K O R N E L I A G. LEWIS B.Sc, University of Victoria, Victoria, British Columbia, 1991 A THESIS SUBMITTED IN 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 O F SCEENCE in T H E F A C U L T Y O F G R A D U A T E STUDD2S (Department of Forest Sciences) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA October 1995 ©Kornelia G. Lewis, 1995 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of f-Qr^Sf SclttXtS The University of British Columbia Vancouver, Canada Date DE-6 (2/88) 11 ABSTRACT The white pine weevil, Pissodes strobi, is an insidious pest of spruce regeneration in British Columbia and has forced forest managers to virtually eliminate the planting of Sitka spruce in coastal B.C. This leader weevil is capable of feeding on a wide variety of conifers and in B.C., P. strobVs preferred hosts are species of spruce. This attack pattern is counter to that observed in eastern Canada where species of pine are the major hosts attacked. The RAPD marker technique was used to examine genetic variation within and among populations of P. strobi across it's range in B.C. This was done to assess population structure and to determine if P. strobi populations are distinct relative to host tree species. Weevils were collected from a total of 12 populations: five Sitka spruce (three of which were from Vancouver Island and two from the mainland), three white spruce and two Engelmann spruce from the B.C. interior and two Jack pine (obtained from Ontario) populations. The bulked DNA technique was initially explored in an attempt to identify genetic markers specific to the Y chromosome (these in essence being haploid markers). Six putative male specific markers were tentatively identified. However, dot-blots failed to confirm the genomic specificity of these markers so the genetic variation was assessed using RAPD markers generated from diploid DNA. From each population DNA was extracted from 30 individuals (15 males; 15 females). Ten different primers were used with each of the 360 DNA extracts and 69 polymorphic RAPD markers were made available for analysis. Since there was no general consensus as to how to analyze RAPD data, markers were used to estimate allele frequencies as well as being treated as simple binary type data (presence or absence of band). Population genetic analyses, based on the allele frequency data, included examining the percentage of polymorphic loci, mean expected Ul heterozygosity, Wright's F-statistics, and cluster analyses (three different methods tested with three different genetic distances/similarities). Multivariate analyses was performed with the binary coded data and included principal component analysis as well as parametric and nonparametric discriminant procedures. Both multivariate and population genetic analyses indicated the following: 1) populations of P.strobi from Sitka spruce on Vancouver Island are distinct from populations obtained from mainland B.C. Sitka spruce; 2) populations of P. strobi collected from Engelmann and white spruce form a distinct complex thus, for management purposes investigators should recognize three groups of P. strobi in B.C. and; 3) P. strobi reared from the eastern host, Jack pine, are most similar to the Engelmann/white spruce complex. TABLE OF CONTENTS Abstract ii Table of Contents iv List of Tables viii List of Figures x Acknowledgments xiv Chapter 1 Literature Review of Pissodes strobi 1 1.1 General Introduction 1 1.2 Taxonomic Review 4 1.2.1 Pissodes strobi 4 1.2.2 Summary of Taxonomic Review 8 1.3 Hosts, LifeCycle and Behavior, and Types of Controls Tested Against Pissodes strobi 9 •1.3.1 Specific Hosts of Pissodes strobi 9 1.3.2 Generalized LifeCycle 11 1.3.3 Control Techniques Tested Against Pissodes strobi 14 1.3.3.1 Introduction 14 1.3.3.2 Mechanical Controls 14 1.3.3.3 History of Insecticides Used Against Pissodes strobi 16 1.3.3.4 Silvicultural Methods Tested as Control Options for Pissodes strobi 20 1.3.3.5 Predators and Parasites as Biological Control Agents for Pissodes strobi 27 1.4 Development of Trees Which are Genetically Resistant to Pissodes strobi for Use in Control Programs 29 1.5 Approaching Pissodes strobi Control Via an Integrated Pest Management System 41 Chapter 2 Systematics and the RAPD Marker Technique 44 2.1 Systematics 44 2.1.1 Introduction and Terminology 44 2.1.2 Theoretical and Philosophical Debates of Molecular Phylogenetics 46 2.1.2.1 Classical-Balance Debate 46 2.1.2.2 Neutral Theory (Molecular Clock Hypothesis) 48 2.1.2.3 Phenetic vs Cladistic Approach to Systematics 50 2.2 R A P D Marker Technique 53 2.2.1 Development of Technique 53 2.2.2 Entomological Studies Utilizing RAPD Markers 55 2.2.3 Procedures for Analyzing RAPD Data 66 2.2.3.1 General Introduction 66 2.2.3.2 Population Genetics Techniques 72 2.2.3.3 Multivariate Techniques 77 2.2.4 Addressing the Dominance Problem of R A P D Markers 80 Chapter 3 The Search for Male Specific RAPD Markers in Pissodes strobi 82 3.1 Methods 82 3.1.1 Sample Design and Collection of Weevils 82 3.1.2 Extraction of Total Genomic D N A 86 3.1.2.1 Bulked D N A Extracted From 20 Individuals 86 3.1.2.2 Bulked D N A Extracted From 100 Individuals 88 3.1.2.3 Bulked D N A Extracted From 12 Individuals 88 3.1.2.4 Quantification of D N A 89 3.1.3 Optimization of RAPD Assay Conditions 89 3.1.3.1 Amplification of RAPD Markers 89 3.1.3.2 Resolution of RAPD Products 90 3.1.4 Screening of Primers to Identify Putative Y-Specific Markers 91 3.1.5 Verification of Y Specificity via the Dot-Blot Procedure 92 3.1.5.1 Preparation of Dotted Membranes 92 VI 3.1.5.2 Pre-Hybridization, Labeling Reaction and Hybridization Protocol 93 3.1.5.3 Preparation of Putative Male Specific Markers for Use as Probes 95 3.2 Results and Discussion 97 3.3 Conclusions and Recommendations for Future Work 121 Chapter 4 Genetic Variation of Pissodes strobi Based on Diploid R A P D Markers 123 4.1 Methods 123 4.1.1 Generation of RAPD Markers Based on D N A Obtained From Diploid Tissue 123 4.1.2 Scoring of Diploid RAPD Markers 128 4.1.3 Data Analysis 129 4.1.3.1 Computer Programs Used 129 4.1.3.2 Considerations in Compiling Data Sets 129 4.1.3.3 Multivariate Techniques 130 4.1.3.3.1 Introduction 130 4.1.3.3.2 Principal Component Analysis 131 4.1.3.3.3 Stepwise Discriminant Analysis 132 4.1.3.3.4 Nonparametric Discriminant (Neighbor) Analysis 133 4.1.3.3.5 Canonical Discriminant Analysis 135 4.1.3.4 Population Genetic Analysis 135 4.1.3.4.1 Introduction 135 4.1.3.4.2 Percentage of Polymorphic Loci and Mean Expected Heterozygosity 136 4.1.3.4.3 Wright'sF-Statistics 137 4.1.3.4.4 Cluster Analysis: Methods and Genetic Distances or Similarities Used 138 4.1.3.5 Additional Analyses 139 4.2 Results 139 4.2.1 Multivariate Approaches 139 vii 4.3 Discussion 172 Chapter 5. Overall Conclusions and Recommendations 189 Bibliography 192 Appendices 208 viii L I S T O F T A B L E S T A B L E P A G E 1. Hopkins(1911) original hierarchial classification of Pissodes strobi and a taxonomic account of the four members in the Pissodes strobi species group (adapted from Boyce et al. 1994). 5 2. North American tree species (native and introduced) recorded as hosts of Pissodes strobi (adapted from Vandersar et al. 1977). 10 3. Summary of types of data analyses conducted on RAPD markers (entomological investigations). 67 4. Populations of Pissodes strobi sampled in RAPD marker study. 84 5. Summary of Pissodes strobi collected, sexed, and frozen in 1992 for use in examining genetic variation with RAPD markers. 98 6. Performance (failed to yield products, gave one band, resulted in multiple amplification products) of primers screened using DNA obtained from 100 bulked weevils and from 20 bulked weevils (each gender). 105 7. Summary of putative Y chromosome specific markers seen using three different extracts of bulked DNA. Presence or absence (in each gender) of markers in DNA from 100 bulked weevils (Figure 7) and DNA generated from 12 weevils (Figure 8) are compared to markers seen in DNA from 20 bulked weevils (Figure 6). 107 8. Scoring (presence) of four putative Y chromosome specific RAPD markers in reactions containing DNA extracted from 12 individual weevils of each gender. 115 9. Grouping of 12 sample populations in RAPD assays and gel electrophoresis (first attempts). 125 10. Summary of RAPD markers scored from each of the 10 random primers used. 140 TABLE 11. RAPD markers which discriminated (stepwise procedure: 69 RAPD markers examined) between male and female Pissodes strobi in each population.. 12. Mean sample size per locus (N), percentage of polymorphic loci and mean expected heterozygosity (H) for each of the 12 sample populations (total data set 60A and 60B). Numbers in brackets are standard errors. 13. Mean percentage of polymorphic loci and mean expected heterozygosities (H) for populations pooled into four groups: S s w a M a n d , S s v a n L . , Se-SwCompiex, and jPine (total data sets 60A and 60B). 14. Variance components and F statistics combined across all loci for total data sets 60A and 60B. 15. Matrix of Nei's unbiased genetic identities (1978) as inferred from RAPD data. Values between populations of P. strobi in 4 different groups, averaged across all pairwise comparisons between populations, shown. 16. Matrix of Nei's unbiased genetic identities (1978) as inferred from RAPD data. Values between populations of P. strobi in 3 different groups, averaged across all pairwise comparisons between populations, shown. 17. Comparison of population genetic statistics obtained from allozyme markers (Phillips and Lanier 1985) and RAPD markers (total data set 60A - monomorphic markers retained). X LIST OF FIGURES FIGURE PAGE 1. Studies dealing with host resistance to Pissodes strobi arranged according to Hanover's (1975) diagrammatic representation of the host-insect complex. Numbers in upper portion correspond to those in lower portion and vice-versa. Location of sites in B.C. (1992) from which white pine weevil infested leader were obtained. 3. Example of inconsistent results obtained when DNA was quantified by absorbance at 260 r|m. Banding profiles produced in the first trial using primers 314 to 320 inclusive. 5. Test of Tag at half the concentration (0.025 U/uL) previously used. 6. Eight primers which tentatively yielded putative Y-specific markers (in primary screening with bulked D N A from 20 weevils) used in one RAPD assay with bulked D N A obtained from 20 weevils. 7. Eight primers which tentatively yielded putative Y-specific markers in initial primer screening with bulked D N A from 20 weevils used in one RAPD assay with bulked DNA obtained from 100 individuals. 8. Eight primers which tentatively yielded putative Y-specific markers in initial primer screening with bulked DNA from 20 weevils used in one RAPD assay with bulked DNA generated by pooling D N A from 12 individual extractions (both genders). 9. Primer 219 used in reactions containing DNA extracted from 12 individual male weevils and 12 individual female weevils. 10. Primer 374 used in reactions containing DNA extracted from 12 individual male weevils and 12 individual female weevils. 33 83 100 101 102 106 109 110 111 112 11. Primer 376 used in reactions containing D N A extracted from 12 individual male weevils and 12 individual female weevils. 114 Products from reactions using retrieved RAPD markers specific to primers 219, 374, and 376 as template DNA visualized on a 2% agarose gel. Electrophoretogram of seven sets of replicated reactions using primer 376. Illustration of autoradiogram obtained from dotted-membrane hybridized with radioactive probe created from the 765 base pair DNA RAPD marker generated with primer 376. Products from a single RAPD assay (46 reaction tubes using primer 322) resolved via electrophoresis on a 1% synergel™/2% agarose gel. RAPD profiles (second attempts with primer 350) from 10 of the 12 sample populations resolved via electrophoresis on a single 1% synergel™/2% agarose gel. Individual scores plotted on Principal Component Axis one and two (total data set 60A - correlation matrix specified). Numbers in brackets indicate percentage of total variation accounted for by each axis. Individual scores plotted on Principal Component Axis one and two (total data set 60B - correlation matrix specified). Numbers in brackets indicate percentage of total variation accounted for by each axis. Comparison of mean error rates between data used to develop classification scheme (2/3 of data) and data used to test the classification (1/3 of data) (series 1-60 loci used in analysis). Comparison of three different partitionings of the data: mean error rates of the test data (1/3 of data set) based on classification into populations (69 loci used in analysis). xii FIGURE PAGE 21. Comparison of different groupings for classification - data set 60B (series 1: K set at 2 and 12). 149 22. Comparison of varying the number of loci used in the analysis and random versus nonrandom selection of loci (K set at 12). 150 23. Plot of canonical variable two versus canonical variable one (individual scores plotted) for total data set 60A. Numbers in brackets indicate percentage of variation (between populations maximized) accounted for by each axis and colored dots indicate group (population) means (red = Ss from Vancouver Island, orange = Ss from mainland B.C., yellow = Sw, green = Se, and brown jPine) 151 24. Plot of canonical variable two versus canonical variable one (individual scores plotted) for total data set 60B. Numbers in brackets indicate percentage of variation (between populations maximized) accounted for by each axis and colored dots indicate group (population) means (red = Ss from Vancouver Island, orange = Ss from mainland B.C., yellow = Sw, green = Se, and brown = jPine). 152 25. Plot of canonical variable two versus canonical variable one (individual scores plotted) for data set 60B, males only. Numbers in brackets indicate percentage of variation (between populations maximized) accounted for by each axis and colored dots indicate group (population) means (red = Ss from Vancouver Island, orange = Ss from mainland B.C., yellow = Sw, green = Se, and brown jPine). 153 26. Plot of canonical variable two versus canonical variable one (individual scores plotted) for data set 60B, females only. Numbers in brackets indicate percentage of variation (between populations maximized) accounted for by each axis and colored dots indicate group (population) means (red = Ss from Vancouver Island, orange = Ss from mainland B.C., yellow = Sw, green = Se, and brown = jPine). 154 27. Distribution of FXY statistics for total data set 60A. 160 FIGURE 28. Distribution of FXY statistics for total data set 60B. 29. Dendrogram showing the phenetic relationships among Pissodes strobi populations based on Nei's unbiased genetic identity and the UPGMA clustering method (total data set 60A). 30. Dendrogram showing the phenetic relationships among Pissodes strobi populations based on Prevosti's genetic distance and the UPGMA clustering method (total data set 60A). 31. Dendrogram showing the phenetic relationships among Pissodes strobi populations based on Nei's unbiased genetic identity and the UPGMA clustering method (total data set 60B). 32. Dendrogram showing the phenetic relationships among Pissodes strobi populations based on Prevosti's genetic distance and the UPGMA clustering method (total data set 60B). 3 3. Correlation between Nei's unbiased genetic identity and Mahalanobis' distance. A: total data set 60A; B: total data set 60B: and Rs = Spearman Rank Order correlation coefficient. 34. Correlation between Prevosti's distance and Mahalanobis' distance. A: total data set 60A; B: total data set 60B: and Rs = Spearman Rank Order correlation coefficient. ACKNOWLEDGMENTS xiv I would like to extend my gratitude to all the people who contributed to the success of this study. Weevil infested leaders were provided by many different individuals and I thank them all. I continually leaned on Rory Mcintosh and Brian Sieben for their computer expertise and they cheerfully came to my aid. Special thanks to Dr. V . LeMay of the Department of Forest Resources Management - without her guidance and support the multivariate procedures would have been overwhelming. Jeff Glaubitz of the Forest Biotechnology Laboratory is gratefully acknowledged for his continual leadership in all the molecular procedures. Dr. A. Wardle is also thanked for her initial contributions in the D N A extraction and R A P D techniques. I thank all the members of my supervisory committee. Specifically, Dr. Rene Alfaro, who obtained funds from G R E E N P L A N , without which this study would not have got started. Dr. Yousry El-Kassaby of Pacific Forest Products was my industrial sponsor for the GREAT award. He is also acknowledged for initially proposing this study and providing statistical advice. A heartfelt thank-you to my co-supervisors Dr. John Carlson and Dr. John McLean. Dr. J. Carlson provided me with space in his overtaxed Biotechnology lab and I thank him. Dr. J. McLean was there for me every step of the way by both encouraging me and always listening to me. I would like to thank all my family and friends for supporting me throughout my university years. Marilyn, my youngest sister, listened to my every complaint - never judging, just supporting. Finally, I wholeheartedly thank my partner Laurie. He has had to endure much as I struggled to complete this work and his support never waivered. I dedicate this Thesis to him. 1 CHAPTER 1: LITERATURE REVIEW OF Pissodes strobi 1.1 General Introduction This thesis contains four chapters. In chapter one, the taxonomic history of Pissodes strobi is presented. The hosts, life cycle and behavior of the white pine weevil, along with control techniques utilized against P. strobi are extensively reviewed to provide background biological information on a weevil which has negatively impacted the forest industry for the past 200 years. For those individuals primarily concerned with the laboratory and statistical procedures used to examine genetic variation in P. strobi, it may not be imperative to review this chapter in its entirety. Rather, those readers may wish to restrict themselves to the general introduction and the sections related to the intraspecific taxonomic review and development of weevil resistant trees (sections 1.1, 1.2 and 1.4). Chapter two deals with concepts in systematics and the development of the RAPD marker technique along with explanations of the statistical procedures used to examine population structure of P. strobi. The search for male specific RAPD markers, via the bulked DNA technique, is presented in chapter three. Chapter four covers the entire genetic variation study based on RAPD markers derived from diploid DNA of weevils collected from Sitka spruce (Picea sitchensis (Bong.) Carr), in coastal British Columbia (B.C.) (5 populations), Engelmann spruce (P. engelmannii Parry) (2 populations) and white spruce (P. glauca (Moench) Voss) (3 populations) from the interior of B.C., as well as from jack pine (Pinus banksiana Lamb.) in Ontario (2 populations). The white pine weevil, Pissodes strobi (Peck) (Coleoptera: Curculionidae), was first described by Peck (1817) from the dominant shoot of the "Weymouth pine" Pinus strobus L. It is 2 the most destructive insect of second-growth stands of spruce (Picea spp.) in B.C. (Alfaro and Borden 1985; VanderSar et al. 1977; Wallace and Sullivan 1985). In the spring, female weevils select suitable host trees and oviposit in the leading shoot of the previous year. Ravenous larval feeding quickly results in the death of the shoot (Wallace and Sullivan 1985) and a minimum of two years growth is lost. Repeated attack by this pest has such devastating impact on trees that the planting of Sitka spruce a species desired for its excellent wood qualities, is heavily curtailed in most areas of coastal B.C. (Alfaro 1994). This Nearctic insect is distributed from coast to coast in North America. Its range extends as far north as the Yukon and Northwest Territories in Canada and as far south as mid-Colorado and northern Georgia in the western and eastern United States (US) respectively (Langor and Sperling 1994). In general, host preferences of P. strobi are quite distinct between populations occurring in western and eastern North America. Although shown to feed on a wide variety of conifer species (Alfaro and Borden 1982), in B.C., populations of P. strobi preferentially attack species of spruce including Sitka spruce, Engelmann spruce, white spruce and interior spruce (P. glauca (Moench) Voss var. allertianan (S. Brown) Sarg.), the natural hybrid of Engelmann and white spruce. In eastern Canada, the white pine weevil principally damages species of pine (Wallace and Sullivan 1985), such as jack pine and eastern white pine (P. strobus L.), from which the common name was derived. Historically, Hopkins (1911) taxonomically differentiated the western and eastern weevil populations based primarily on differential host preference. Three separate species were recognized: two western species (classified from Sitka and Engelmann spruce and named by Hopkins as P. sitchensis and P. engelmanni respectively) and one eastern species (P. strobi, classified from eastern white pine and previously named by Peck). Subsequent cytogenetic and cross breeding experiments (Manna and Smith 1959; Smith 1962; Smith and Sugden 1969), 3 physiological (VanderSar et al. 1977; Alfaro 1988) and protein electrophoretic evidence (Phillips and Lanier 1985) refuted the hypothesis of three separate species. Consequently, western and eastern populations were surmised to be conspecific, with geographic races of the ancestral eastern group recognized in western Canada (Manna and Smith 1959; VanderSar et al. 1977; Phillips and Lanier 1985; Alfaro 1988). Many control options for P. strobi have been investigated but none have proven to be entirely effective (Cozens 1983; Alfaro etal. 1994; de Groot and Helson 1994; Retnakaran and Jobin 1994). The development of weevil resistant trees, a potential environmentally benign control option, is an area of research being actively pursued in B.C. today (Wood 1987; Kiss and Yanchuk 1991; Carlson et al. 1994; King 1994). Development of weevil-resistant lines, for all species of spruce susceptible to weevil attack in B.C., requires an understanding of the genetic variation within and among populations of P. strobi. This is particularly true in light of the fact that P. strobi was initially subdivided into different species based on host preference. Populations of P. strobi reared from three western Picea hosts (Sitka, Engelmann and white) and from one eastern Pinus host (jack pine) were investigated in this study. The principal objective was to estimate the extent of genetic variation within and among populations of P. strobi using the newly developed Randomly Amplified Polymorphic D N A (RAPD) marker technique (Welsh and McClelland 1990; Williams etal. 1990). The specific objectives of my study were to: 1) assess the population structure of Pissodes strobi (do population differences exist and, if so, do populations group according to some factor such as tree host or geography); and 2) investigate the utility (i.e., in revealing population trends) of different statistical procedures when applied to data collected from RAPD markers. 4 1.2 Taxonomic Review 1.2.1 Pissodes strobi Thorough investigations of species belonging to the genus Pissodes began with Hopkins' 1911 study. In evolutionary terms, Hopkins' classified the smaller bodied, shorter and stouter beaked (snout) forms as the more primitive types as compared to longer and slender beaked forms which were regarded as more advanced. Hopkins based the classification of species within Pissodes on a combination of morphological (length of beak, scale patterns of the adults) and physiological (host-tree association) characters. He assigned the names P. sitchensis and P. engelmanni to slender beaked weevils found attacking the terminal shoots (leaders) of Sitka and Engelmann spruce respectively. Slender beaked weevils collected from leaders of eastern white pine were recognized as the separate species, P. strobi, as previously named by W.D. Peck (1817). Pissodes sitchensis, P. engelmanni and P. strobi were placed in the same Division, Subdivision, Section, Subsection and Series (Table 1), and were classified as different species by the presence or absence of borders on the posterior spots of the elytra in addition to their host-tree associations. Overall, Hopkins recognized 30 North American Pissodes species, 23 of which he named (Hopkins 1911). Eleven of these species, along with the hierarchical placement of the seven divisions of the classification system of that day, are indicated in the left two columns in Table 1. In the 1920's three other North American Pissodes species were added to Hopkins list-P. terminalis (characterized by adult coloration and host-use) by Hopping (1920) and P. robustus and P. ochraceus (classified primarily by adult coloration and length of beak ) by VanDyke (1927) - bringing the total number of North American Pissodes species to 33 at that time. o (O to to to II v> o o CO 3_S to oo s S s s a. s s Co a gt o o s o o. ~ » 5 BJ » P3 cr P o > <-• CO 2 3 ^ o 2-OQ S 3. p S 3, a . CD W » o p — £• 3' o co 3 C cu W i-l <B 1 — 1 VO l — £ -^J VO a •-I 3 P 3 P 3 Cu P VO o\ J O o p 3 I — vo o> vo 3 o co o> 3 . CD co Is) CO 3 cn O CD -< a. a> co CO 1 co 01 co cr CO CO a. o o> -t-co to ^ 9 2 c r S S f f i . c° o 3 „ • 3 w o l - 2 cr' 4 - U ) TO 55 as TO 00 H _ H - I— VO to ^ - p • ^3 ^0 !*0 3 c Co TO" TO TO I-I TO 3 • N 3 U W o o I 4 -TO 3 Co' O 5' 0Q 1 1 Co 1 ^  i 8 3. N ?r ST. 3 o 3*12. 5 OQ Cu H p t»r Ol 3 O c o B: VO O N VO VO to o GO 3 p 3 Cu GO c OQ Cu Ol 3 i—i VO C \ VO 00 ~~J II § | Co to U ) 00 o I 4-I I I fir £5 5 ~ I TO o T3 3 TO 3 TO 3 Co Co o Co O 3 _r o P co S - X ) "1 5 _t 3 co g — 5 — CO o ^ ^ a 4- cn TO cr oo i . ~ j vo o 4 - 4 - 4 ^ 3* o> CO 3* O a, CO 3 O c a. S5 Co 3* TO 3 O a a. o co p 3 p I TO p 3 Du TO a P o ET BL O <. 5' 3 co GO 5' CO o C 3' OQ co to CO *o a o S ' CO p 3 Cu CO o 3 3 CO O s» m 3 O 0) o CO 3 o 3 H p a* C<5 © c_ -a TO 5»5 S CO «» <e © 2. 2-_ cr© g i i a» & 3 s» cr S- P — o. o_ O co 3 S W a: © tt, TO © i vo P S Q> P <-•• P © S o 3 « o e s © * e •1 3 3 cr n 2 5 Hopkins' (1911) monograph also gave flail generic descriptions with accurate illustrations of larvae from 14 of the Pissodes spp. Although Hopkins did not compile a larval taxonomical key, he distinguished the different larval species on the presence or absence of eye-spots and the shape of several anatomical parts. Boving (1929), however, was unable to differentiate between the larvae of P. strobi and P. approximates based on Hopkins earlier work. Working with only the mature larval forms from these two species, Boving fine-tuned Hopkins earlier work and subsequently identified the following larval characteristics as being taxonomically discriminating: head length and shape, frons length, inner clypeal setae, setae of the 9th abdominal segment and tissue type surrounding the spiracles. The genus Pissodes underwent a major taxonomic revision in the 1950's and 1960's heavily influenced by the work of S.G. Smith and associates. Results from both karyotyping and cross-breeding experiments were used to evaluate the validity of many of the 33 Pissodes species and several were refuted. Some of the more important studies by this group which dealt directly with the status of P. strobi include Manna and Smith (1959), Smith (1962), Smith and Sugden (1969) and Smith and Takenouchi (1969). In the earliest of these four works, the karyotypes of 12 Pissodes species were examined. Manna and Smith (1959) found that P. engelmanni, P. sitchensis and P. strobi all contained 34 chromosomes (2N; diploid) and furthermore there were no discernible differences in their karyotypes. The species status of these three Pissodes representatives was therefore highly suspect. Evidence provided by cross-breeding experiments (viable progeny being produced from matings between P. engelmanni, P. sitchensis and P. strobi (Smith 1962)) coupled with the uniform cytogenetic patterns, compelled Smith and Sugden (1969) to synonymize P. engelmanni and P. sitchensis under P. strobi with the former two identified as geographic races (ecotypes) of P. strobi Peck. In total, 21 of Hopkins recognized species were consolidated into nine valid species and two chromosomally polymorphic hybrid 7 complexes (Smith and Sugden 1969) from evidence presented in these elegant cytogenetic studies. The taxonomic status of P. strobi and its ecotypes was further examined by VanderSar et al. (1977) using forced- and choice-feeding bioassays. In the forced-feeding bioassays (data collected on number of feeding punctures) weevils reared from Sitka spruce, Engelmann spruce and eastern white pine failed to discriminate among the three hosts. In the choice-feeding bioassays, however, clear differences were observed: weevils reared from Sitka spruce fed equally well on all three hosts; weevils obtained from Engelmann spruce preferred both eastern white pine and Engelmann spruce over Sitka spruce; and weevils collected from eastern white pine fed significantly more on eastern white pine than on the two western spruce species. These feeding preferences led VanderSar etal. (1977) to support Smith and Sugden's (1969) contention of one P. strobi species. Furthermore, the pattern of increased adaptation from east to west suggested that P. strobi had originated in the east and migrated west. Alfaro (1988) also concurred with the hypothesis of western geographic weevil races based on results obtained from no-choice oviposition and choice-feeding bioassays. In this study, weevils reared from Sitka spruce and Engelmann spruce were demonstrated to be two distinct races of P. strobi with somewhat different nutritional requirements Genetic variation based on allozyme data has also been used to address the validity of P. strobVs status as a single species (Phillips and Lanier 1985). Eight weevil populations reared from Sitka, Engelmann, white or Table Mountain spruce (pure and hybrids) and nine populations reared from jack or eastern white pine were used to assess genetic divergence in P. strobi. Populations were grouped a priori into three regions (divisions) based on geographic locality: Pacific Coast (ecotype P. sitchensis), Rocky Mountains (ecotype P. engelmanni) and Northeast (ancestral P. strobi). As indicated, these regions were basically defined to delineate P. strobi and its two 8 western ecotypes. Phenetic cluster analysis, based on Nei's genetic identity (I) (Nei 1972), failed to group populations according to geographic region, and overall, mean genetic identity was high (I = 0.963). These results suggested little genetic divergence between the eastern and western populations, once again lending support to the hypothesis of one P. strobi species. Additionally, the belief that P. strobi had originated in the east and spread west was further supported by the trends exhibited in mean heterozygosity and percentage of polymorphic loci (i.e., increased genetic variation from west to east). Thus karyotyping, feeding- and oviposition-bioassays and allozyme markers all support the contention that the western and eastern groups constitute a single P. strobi species with distinct differences existing between the western and eastern groups. 1.2.2 Summary of Taxonomic Review Taxonomy in the study of Pissodes, is an exciting and challenging area. Four Pissodes species (P. strobi, P. nemorensis, P. schwarzii and P. terminalis) have been informally grouped into the P. strobi species group (Smith and Takenouchi 1969). The taxonomic review pertaining to these species is presented in Appendix I. This review illustrates the complexities and anomalies displayed in members belonging to the genus Pissodes. The historical taxonomic progression of all members belonging to the P. strobi species group is summarized in Table 1. In terms of intraspecific variation, thus far, three powerful techniques have all demonstrated that P. strobi's status as a single species holds true. Differential host preference is believed to be an inherited characteristic (Alfaro 1988) and, thus, the western populations are considered ecotypes of the ancestral eastern population. Having been established as a single species from evidence drawn from cytogenetics, ecology/physiology^ehavior and allozymes, I wished to apply the RAPD marker technique (Welsh and McClelland 1990; Williams et al. 1990) 9 to reexamine population structure within P. strobi. Differences in genetic makeup could be a direct reflection of behavioral differences; therefore, at this point, I present a brief overview of P. strobi's specific hosts, lifecycle and behavior. A literature review pertaining to the various control strategies applied against the terminal weevil is also presented. 1.3 Hosts, LifeCycle and Behavior, and Types of Controls Tested Against Pissodes strobi 1.3.1 Specific Hosts of Pissodes strobi In North America, although all native and introduced spruce and pine (Pinus spp.) are susceptible to attack by the white pine weevil (Belyea and Sullivan 1956), seven pine species and nine spruce species are most commonly attacked (VanderSar et al. 1977). Smith and Sugden (1969) also identified 14 of these host trees as breeding sites for P. strobi, merely omitting the two natural hybrids of spruce which occur in western Canada (i.e., hybrid between P . engelmannii and P. glauca and hybrid between P. sitchensis and P. glauca (Table 2)). Table 2 lists the Latin and common names of the 16 host species given by VanderSar et al. (1977). In eastern North America native and exotic pines are most at risk, but P . strobi also attacks several spruce species, such as Norway spruce (P. abies (L.) Karst. (Belyea and Sullivan 1956; Godwin and ODell 1967; VanderSar etal. 1977; Wallace and Sullivan 1985). The top four eastern hosts, listed in descending order, as stated by Godwin and ODell (1967), are eastern white pine , Norway spruce (P. abies (L.) Karst.), jack pine, and Scots pine (P. sylvestris L.). Belyea and Sullivan (1956) indicated white spruce (P. glauca (Moench) Voss), black spruce (P. mariana (Mill) BSP) and red pine (P. resinosa Ait.) as being among the least preferred eastern hosts. Plank and Gerhold (1965) concurred with these attack preferences and showed eastern white pine to be a highly preferred host, jack pine to be an intermediate host and red pine to be non-host. 10 Table 2. North American tree species (native and introduced) recorded as hosts of Pissodes strobi (adapted from Vandersar et al. 1977). Host Species Common Name Pinus strobus L. eastern white pine P. sylvestris L. Scots pine P. banksiana Lamb. jack pine P. resinosa Ait. red pine P. rigida Mill. pitch pine P. pungens Michx. table mountain pine P. contorta Dougl. Lodgepole pine Picea abies (L.) Karst. Norway spruce P. glauca (Moench) Voss white spruce P. marianana (Mill.) B.S.P. black spruce P. pungens Engelm. blue spruce P. rubens Sarg. red spruce P. sitchensis (Bong.) Carr. Sitka spruce P. engelmannii Parry Engelmann spruce P. glauca (Moench) Voss B.C. Interior spruce var. allertiana (S. Brown) Sarg.a Picea x lutzii Littleb Lutz spruce a = Natural hybrid between P. engelmannii and P. glauca b = Natural hybrid between P. sitchensis and P. glauca 11 Phillips and Lanier (1983b) have also indicated that eastern white pine is the preferred native host and that few (if any) adult P. strobi emerged from white spruce leaders when attacked. These rankings are somewhat different from recently reported Quebec studies, wherein Norway spruce was assigned the ranking of number one host (Hamel et al. 1994). More importantly, white spruce was also indicated as a highly preferred species (Trudel et al. 1994; Hamel et al. 1994). The preference suitability of white spruce in eastern Canada, therefore, appears variable. Counter to the pattern in the east, in western North America and specifically in B.C., P. strobi clearly attacks native spruce over pines (VanderSar et al. 1977; VanderSar 1978; Alfaro and Borden 1985; Alfaro 1988). P. strobi has been reported on Lodgepole pine (P. contorta Dougl.) in western Canada, but, in general, this tree species is considered a non-host (Alfaro 1988). Eastern and western white pine, both highly susceptible in eastern North America, are also not at risk and have shown overall resistance to attack in B.C. (Plank and Gerhold 1965; Alfaro and Borden 1985). Alfaro (1988) demonstrated that western weevil populations obtained from spruce trees (Engelmann and Sitka) can be successfully induced to oviposit in pine leaders. However, weevils reared in Lodgepole pine still selected spruce over pine in choice-feeding experiments. Thus, Alfaro (1988) concluded that the differential preference for spruce in the west is an inherited characteristic and hence, under genetic control. Tree species most at risk to attack by P. strobi in B.C. include Sitka spruce, Engelmann spruce and white spruce as well as hybrids of these tree species (Alfaro and Borden 1985). 1.3.2 Generalized LifeCycle Weevils emerge in the spring from their overwintering sites when daily air temperatures reach approximately 20° Celsius (C) (therefore, microhabitat 6° C or greater (Wallace and 12 Sullivan 1985)) (Sullivan 1960; Gara et al. 1971; McMullen 1976a,b). Initial activity is concerned with seeking out and feeding on suitable host trees followed (closely) by mating and egg laying (= ovipositing). Alfaro and Borden (1985) indicated a three step procedure whereby P. strobi seek out suitable sites for feeding and ovipositing: step 1) finding the host; step 2) examining the host; and step 3) sustained feeding or acceptance for ovipositing. Visual cues appear to play an integral part in seeking out suitable hosts, as VanderSar and Borden (1977c) have indicated. Weevils orient to vertical silhouettes in the laboratory, with the taller and wider (i.e., > 3.0 cm in diameter) silhouettes preferred (VanderSar and Borden 1977a,c). After initial random movement, weevils move towards the vertical silhouettes of the year-old leaders of host trees. Weevils respond positively to phototaxis (Sullivan 1959; Droska et al. 1983; VanderSar and Borden 1977b) and thigmotaxis (Harris et al. unpublished), negatively to geotaxis (VanderSar and Borden 1977b) and begin conducting taste tests in the uppermost portion of the tree (i.e., on the underside of lateral branches and on the year-old leader) (step 2). Final rejection or acceptance of a leader for sustained feeding and ovipositing depends, in part, on a complex mixture of volatiles and non-volatiles, present in the bark and needles, which act as either feeding stimulants or deterrents (VanderSar and Borden 1977a; Alfaro etal. 1980). Bark thickness (Kriebel 1954; Sullivan 1961) and ability to circumvent resin ducts (Stroh and Gerhold 1965; Tomlin and Borden 1994a,b) also influence the white pine weevil's behavior concerning host selection. In general, weevils prefer to oviposit in, and aggregate on, the longer and thicker leaders (Harman and Kulman 1967) which possess relatively thick bark (1.8 to 2.2 mm (Sullivan 1961)) and shallow cortical resin ducts (Wallace and Sullivan 1985). Oviposition is primarily influenced by changes in temperature and humidity (Belyea and Sullivan 1956; Sullivan 1959,1960,1961). Maximum oviposition occurs when bark temperatures are between 25° and 29° C (with relative humidity 20 - 55%) and ceases when bark temperature exceeds 35° C (Sullivan 1960; McMullen 1976a,b). Egg laying proceeds 13 for approximately five to six weeks (Belyea and Sullivan 1956; Silver 1968; Wallace and Sullivan 1985) with single female weevils capable of depositing up to 200 eggs in one leader (Belyea and Sullivan 1956). At present, it is believed that any individual female weevil will probably only oviposit in one leader per year although leaders may contain eggs from two or more female weevils. Eggs hatch in about two weeks (Belyea and Sullivan 1956) and first instar larvae immediately begin feeding within their protective environment (Silver 1968). The larvae consume phloem and cambium tissue as they mine downwards in the leader and soon form what is known as a feeding ring (Belyea and Sullivan 1956; Silver 1968). Trees attacked by P. strobi often exude copious quantities of resin and it has been suggested that P. strobi larvae form feeding rings as defense against drowning in resin (Wallace and Sullivan 1985). Although the production of resin may be primarily associated with phellogen restoration after wounding (Silver 1968), it definitely serves to pitch out or drown P. strobi eggs and larvae (Belyea and Sullivan 1956). A chemical factor, believed to be associated with the head capsules of the weevil larvae, has been shown to cause accelerated crystallization of the resin acids. When larvae are present in sufficient masses, such as in feeding rings, effective quantities of the chemical appears to be produced and hence, resin flow is significantly reduced (Wallace and Sullivan 1985). The rate of larval feeding is influenced by bark temperatures; higher temperatures stimulate feeding, thus inducing faster movement down the stem (Sullivan 1960). Survival of larvae is dependent on a number of factors, such as leader size and speed in forming and joining the feeding ring (Sullivan 1960). Intraspecific competition is reduced in leaders of larger diameters and survival is enhanced. Larvae feed for approximately five to six weeks and thereafter the fourth instar larvae (although a fifth instar has also been occasionally observed (Harman 1970)) construct chip cocoons in the pith or wood of the now dead leader. Pupation occurs within the chip cocoon and 14 adults complete their development within two weeks (Belyea and Sullivan 1956; Silver 1968). The adults chew their way out of the leader, generally emerging in August and September. Dispersion in fall is limited with activity primarily consisting of maturation feeding and little (if any) fall mating or ovipositing occurs (Overhulser and Gara 1975). 1.3.3 Control Techniques Tested Against Pissodes strobi 1.3.3.1. Introduction The first control options were used against P. strobi in North America more than 175 years ago (Wallace and Sullivan 1985; de Groot and Helson 1994) and additional controls have been vigorously sought since the turn of the century (Belyea and Sullivan 1956). This is due to the fact that even at the time of naming in 1817, P. strobi was recognized as a pest of noteworthy economic importance. In the 1800's P. strobi damage directly impacted the ship building industry, since tall, sturdy, straight white pine logs were needed for ship masts (MacAloney 1932). Clearly, the crooked, deformed stems produced after P. strobi damage yielded few, if any, logs suitable for this purpose. 1.3.3.2 Mechanical Controls Peck (1817) himself suggested the first control option to be used against P. strobi, namely clipping and burning of the infested leaders. This method, while effective in destroying the vile weevil larvae, also destroyed the beneficial insects associated with P. strobi. Most predators and parasites of the weevil must overwinter in the leader to complete their development (Hulme et al. 1986). Hopkins (1911) recommended that clipped leaders, rather than being immediately burned, 15 should be placed in buckets or barrels covered with wire screening. The wire covered containers were left to overwinter in the stand and the leaders destroyed the following year after the emergence of the invaluable predators and parasites. The size of the wire screening was to be small enough to contain the adult weevils, but large enough to allow the predators and parasites, which on the whole are much smaller than the adult weevil, to escape. Peirson (1922) stated that ordinary wire fly screening fulfilled these requirements and MacAloney (1932) indicated that fourteen-mesh wire screening would also suffice. Throughout the years, frustrated with the lack of control options, foresters and scientists have repeatedly reassessed the feasibility of utilizing leader clipping in some fashion as a viable control option (Peirson 1922; MacAloney 1932; Marty and Mott 1964; Wood and McMullen 1983; McMullen et al. 1987; Rankin and Lewis 1994). Hulme et al. (1987) further refined the dimensions of the wire screening required to contain all the adult weevils while simultaneously allowing for the maximum escape of predators and parasites. Screen with openings 1.7 mm square were determined to be optimum. Wire screens with 1.7 mm square openings were tested in a five year leader clipping trial conducted in a geographically isolated 38.5 ha weevil infested plantation in the interior of B.C. (Rankin and Lewis 1994). Leader clipping in conjunction with parasite augmentation was investigated in this study. Parasite augmentation was accomplished by clipping additional infested leaders from adjacent plantations. Once clipped, these leaders, along with the infested leaders clipped in the 38.5 ha stand, were placed in wire covered, five gallon pails and left to overwinter in the stand. Heavy costs were incurred in this project and leader clipping was deemed to be of limited value (Rankin and Lewis 1994). Time and time again researchers have stated that it is the labor intensity resulting in extreme costs which has forced them to eliminate leader clipping as a control treatment (Peirson 1922; MacAloney 1932; Belyea and Sullivan 1956; Marty and Mott 1964; Smith and McLean 16 1993; Rankin and Lewis 1994). Leader clipping costs of one dollar per acre per year were considered prohibitive in 1932 as were costs of 250 dollars/ha per year in 1994 (Rankin and Lewis 1994). Additionally, McMullen et al. (1987) calculated in a computer simulation model that unacceptable levels of damage may still occur even given 10 successive years of clipping (starting at year 8 or year 10) which significantly reduced weevil populations. Most studies do agree, however, that leader clipping can be effective if thorough and initiated early enough (i.e., when weevil damage is first observed), but should only be implemented in small, geographically isolated, high value stands such as seed orchards (Pierson 1922; MacAloney 1932; Hulme et al. 1986; Wood and McMullen 1983). Other labor intensive methods which have also been eliminated as viable, cost effective control options, but which may be utilized in seed orchards, include the use of sticky bands (tanglefoot) and the hand picking of weevils from the trees (Pierson 1922; MacAloney 1932). 1.3.3.3 History of Insecticides Used Against Pissodes strobi Insecticides have been evaluated for their utility in P. strobi control programs since the late 1800's. Paris Green, a formulation derived from copper aceto-arsenite, was the first of many different chemicals tested (de Groot and Helson 1994). Representatives from six of the eight different categories of insecticides listed in the pesticide applicators handbook (classified according to their chemical makeup) (Miller and Craig 1980) have been applied against the weevil. These include formulations belonging to inorganic compounds, organochlorine compounds, organophosphorous compounds, carbamate compounds, synthetic pyrethroids and insect growth regulators, de Groot and Helson (1994) have presented an excellent review of the use of chemical insecticides for the control of the white pine weevil. 17 From the late 19th century to the early 20th century various inorganic compounds, sometimes in conjunction with fish or whale oil, were investigated as potential control agents against P. strobi. Two commonly applied inorganics which showed mixed results in different trials were lime-sulfur and lead arsenate. MacAloney (1932) reported that both of these insecticides were effective weevil repellents. Peirson (1922) supported these conclusions for lime-sulfur, but found lead arsenate to be less than adequate. Conversely, Graham in 1916 (Peirson 1922) advocated neither of these two inorganic chemicals for use in weevil control programs. In any event, chemical control was not considered a cost effective treatment in the early 1900's, due mostly to the high cost of labor, and was not utilized on a large scale basis (MacAloney 1932). Powerful new chemicals to combat P. strobi became available with the development of organochlorines. In field trials conducted in the late 1940's all four organochlorines tested (DDT (dichloro diphenyl-frichloroethane), chlordane, toxaphene, and BHC (benzene hexachloride)) were demonstrated to be effective against the white pine weevil (de Groot and Helson 1994). Lead arsenate was also reevaluated at this time and was, in fact, advocated as the preferred chemical, since it was visible on the leader for at least four months and it showed no contact action on the parasites. BHC, while effective, was decidedly not preferred since individuals found this to be an extremely disagreeable substance to work with and workers adamantly voiced their concerns. By 1956 both lead arsenate and DDT were recommended by Belyea and Sullivan as cost-effective control agents. Applications were to be applied in the spring using sprayers attached to either helicopters or fixed-wing aircraft. Aerial spraying of DDT to protect eastern white pine from P. strobi damage was carried out in Ontario from 1961 to 1973 (de Groot and Helson 1994). Marty and Mott (1964) and Silver (1968) were also staunch supporters of using chemical insecticides against the white pine weevil. 18 The use of organochlorines was followed (closely) by the development and testing of organophosphates and carbamate compounds. Organophosphates and carbamates, on the whole, are considered less noxious than organochlorines, since they exhibit much shorter residual activity. Malathion, Guthion® (azinphos-ethyl), dimethoate, Bidrin® (dicrotophos) and Metasystox® (oxydemeton-methyl) are just a few of the organophosphates tried in the mid 1960's to the early 1970's. The latter compound was noted as being particularly versatile since effective control was obtained by either spraying it directly onto the foliage (even after oviposition) or injecting it into the tree's trunk (de Groot and Helson 1994). Two carbamate compounds tested during this period were carbaryl and mexacarbate. On the whole, these insecticides were less effective than either organophosphates or organochlorines (which were now also being applied as granular systemic formulations). Ultimately, the active use of synthetic organic chemicals applied as either granular or foliar insecticides against the weevil ended when concerns regarding negative environmental impacts became too great. While the United States continued to spray Lindane (a potent organochlorine applied as a foliar spray) up until 1986, Canada had discontinued testing granular and foliar applications of organochlorines, organophosphates and carbamates sometime in the late 1970's. The focus of investigations shifted to more user-friendly, environmentally benign, chemical formulations (biological control agents). Two such groups tested were insect growth regulators and synthetic pyrethroids. Insect growth regulators mimic the action of hormones normally found in the insect. When applied externally they interrupt normal insect development and the insect dies without becoming a reproducing adult (Miller and Craig 1980). Retnakaran and Jobin (1994) recently indicated that the insect growth regulator diflubenzuron achieved near perfect control (> 99%) in a 15-year-old Scots pine (Pinus sylvestris L.) plantation when applied as a foliar spray. To achieve such outstanding results, they stated that thorough coverage of the leader and 19 application very early in the spring (prior to the snow melting) was absolutely essential. Based on these encouraging results Retnakaran and Jobin (1994) have now recommended diflubenzuron as an effective white pine weevil control option available for use in Canada. Various pyrethroids including Pounce®, Pydrin®, permithrin, esfenvlarate and fluvalinate have also been tested as potential control agents for P. strobi (de Groot and Helson 1993). Permithrin may prove to be another viable control option, as de Groot and Helson (1993) have demonstrated. They showed a significant reduction in weevil damage in a high-value jack pine plantation when Permithrin was sprayed on leaders at the rate of 70 to 140 g(AI)/ha. Work continues in this area, since at present Permithrin is not yet registered for control of the white pine weevil although it is registered in Canada for a number of other forest insects (de Groot and Helson 1993). Along with investigating different types of insecticides work has also continued on different methods of application, in particular injecting chemicals directly into the tree (Fraser and Heppner 1993). Synthetic organic chemicals are considered applicable for this closed system method since negative environmental impacts have not been demonstrated with this procedure. Significant control over two years was shown, when implants containing two different organophosphates were tested in a 10-year-old Sitka spruce plantation on Nootka Island, B.C. (Fraser and Szeto 1994). Fraser and Szeto recommended that future investigations center on improving chemical formulations so that delivery of the insecticide into the tree can be improved. They have also suggested the initiation of larger operational trials with time and motion studies since the cost benefits of implants have not been established. It is hoped that chemical formulations injected directly into the tree may prove to be another environmentally friendly control option available to forest managers. Given the current public attitude towards the use of synthetic, contact insecticides, it is doubtful that these agents will ever again be applied using ground or aerial spray equipment on an 20 operational basis to curtail weevil activity (de Groot and Helson 1994). They may perhaps be utilized in small high-value plantations, but the focus has now shifted away from chemical insecticides towards biological control agents, such as insect growth regulators and synthetic pyrethroids. Indeed, de Groot and Helson (1994), have now discouraged the use of chemical insecticides for overall control of the white pine weevil except as a very limited part of an integrated pest management system. 1.3.3.4 Silvicultural Methods Tested as Control Options for Pissodes strobi. The establishment of monocultures of species susceptible to P. strobi creates a highly vulnerable stand. Infestation can become so heavy and damage so great that total rehabilitation of the plantation is required. Monocultures of trees susceptible to weevil attack are perhaps feasible if, as reported by Stiell (1979) and Alfaro and Omule (1990), an initial dense planting of the seedlings is undertaken followed by a precommercial thinning to release the unattacked trees. This strategy arose from the general observation that attack levels in low density stands were much more severe than in extremely dense stands (MacAloney 1932). Even given relatively high attack rates in dense stands of eastern white pine weevil damage has been observed to be reduced since, due to competition, the trees produce relatively straight stems. In British Columbia, Alfaro and Omule (1990) examined the effects of three different spacings of Sitka spruce and concluded that the more closely spaced regime (2.74m x 2.74 m) sustained a lower intensity of attack than the more open plantations. They stated that by planting Sitka spruce at this spacing and precommercially thinning at age 25 a first log of good quality should be achieved. This recommendation must now be evaluated at an operational level accompanied by a cost-benefit 21 analysis to ensure that close planting followed by thinning is indeed an effective, viable control option against P. strobi. The recognition that eastern white pine growing in mixture with other species sustained considerably less damage than open-grown monocultures of the pine was noted as early as the 1800's by Peck (Sullivan 1961). This awareness invoked great enthusiasm among foresters at the turn of the 20th century, since it was believed that by establishing eastern white pine in a mixed regime with other hardwoods or softwoods all the problems associated with P. strobi could be successfully overcome. Peirson (1922) suggested that eastern white pine should be grown with a hardwood component which over-topped the pine by two feet when the pine was below six feet. When the pine was greater than six feet the over-topping was to be a minimum of three feet. Peirson (1922) recommended that larger-leafed trees such as oak or maple were the tree species of choice to be mixed with the pine, since they provided greater protection than the smaller-leafed hardwoods such as birch or poplar. He indicated that the best results could be achieved if the pine was evenly scattered throughout the stand and did not constitute more than 20-25 % of the total number of dominant trees. Belyea (1923) reported on the results of establishing mixed softwood stands. A 40 % reduction in weevil attack was noted when white pine and scotch pine were planted in alternate rows (spacing approximately two metres by two metres). Overall intermixing of scotch pine and white pine was unsuccessful in reducing weevil attack. MacAloney (1932) related the reduction in weevil attack in mixed stands to the shading of the pines from the sun. P. strobi had been labeled as a sun-loving creature with oviposition and feeding generally taking place in the sun. These activities were hypothesized to have been adversely effected by shade. Additionally, MacAloney (1932) indicated that in mixed stands a protective barrier against weevil flight was established. Both Peirson (1922) and MacAloney (1932) were optimistic that the weevil could be entirely controlled through silvicultural means and hence planting of eastern white 22 pine as a monoculture was thereafter abandoned. MacAloney (1932) stated that foresters should determine which tree species to intermix with eastern white pine based on ecological characteristics specific to each site. This began 30 years of silvicultural trials in an attempt to minimize weevil damage (Belyea and Sullivan 1956). In 1956 Belyea and Sullivan reviewed the studies in which eastern white pine had been planted in mixtures with other species. Based on their evaluation of these trials they were less than enthusiastic concerning the progress made in weevil management. Unsatisfactory results of growing eastern white pine either intermixed with red pine or under rapidly growing aspen or other hardwoods was reported (i.e., due to very poor survival of the white pine from competition). The overall conclusion was that successful control had not been achieved in the different trials because mixed plantations were established without understanding how the conditions in a mixed stand directly or indirectly curtailed weevil activity. Belyea and Sullivan (1956) undertook the challenge of collecting some of this crucial biological information by initiating a trial in 1951 at the Petawawa Forest Experimental Station. It was soon learned that adult feeding, mating and oviposition was primarily influenced by bark temperature of the leader and by atmospheric moisture. Even given this knowledge they adamantly stated that more facts pertaining to the physical and biological requirements of the insect needed to be established so that the exact mechanism behind a reduction in weevil activity in mixed stands could be ascertained. Belyea and Sullivan (1956) also reinforced the need to examine site specific requirements when selecting a nurse crop. Thus, the optimistic attitude in 1922 of definitely curtailing weevil activity by avoiding monocultures of white pine had shifted to. "a satisfactory solution to the problem of growing white pine successfully in a mixture cannot be expected in the immediate future" (Belyea and Sullivan 1956). 23 Sullivan (1961) continued along Belyea and Sullivan's (1956) track and reported on investigations in which the behavior and biology of the weevil was intensely examined in both exposed and shaded stands of white pine. In this study attributes pertaining to the trees, as well as microenvironmental conditions, in each of the two scenarios, were related to the weevil's behavior and needs. Many invaluable pieces of information were gathered from Sullivan's investigation. Three important conclusions from Sullivan's work are itemized below: 1. Shading reduced radiant heating which in turn decreased bark temperature and increased moisture in the stand. Reduced radiant heating not only decreased overall weevil activity but also resulted in eastern white pine leaders with reduced vigor and smaller diameters. These smaller diameter stems did not provide the adequate bark thickness required for successful deposition of eggs (i.e., thinner than 0.8 mm) and hence oviposition was further inhibited. 2. The reduced light entering the stand also affected the pattern of oviposition; instead of being highly concentrated in the upper 10 cm of the leader oviposition was spread haphazardly over the entire length of the leader (and often over the entire stem). It was hypothesized that positive phototaxis no longer occurred (which served to more or less hold the weevil to the upper most portion of the stem). 3. Since oviposition was significantly reduced fewer larvae were produced in the leader and hence the formation of a feeding ring was not observed. Unable to join a protective feeding ring individual larvae succumbed to host tree defenses by being drowned in pitch. Harman and Kulman (1967) and Droska in 1982 (Wallace and Sullivan 1985) collected additional information pertaining to weevil biology and behavior in shaded white pine stands. These studies showed that higher mortality occurred within shaded stands during overwintering and that less migration into heavily shaded stands was observed in the spring. Logan (1962) reported on growing eastern white pine beneath an overstory of aspen 30 to 40 feet tall. He indicated that the presence of large aspen did not appear to seriously impede the growth of the pine seedlings and that the smaller leader diameters produced would reduce susceptibility to P. strobi. Berry and Stiell (1976) were among the first investigators to explore the use of strip cutting. Up until this time studies concerning the effects of shading concentrated on underplanting and shelterwood methods. Strip cutting was surmised to be an applicable strategy in regenerating natural stands or in stand conversion. Berry and Stiell (1976) wished to ascertain the exact degree of shade required to limit weevil attack while simultaneously allowing for optimum tree growth. Four treatments in which direct insolation in North-South oriented strips was reduced by 0, 25, 50, or 75 % were tested. The width of the clear cut in relation to stand height was used to control the amount of light. Treatments were established in two stands: one a mixture of hardwoods and softwoods, the other a mixture of only deciduous species. Results indicated that reduction of 25 % in light had promise for controlling weevil damage in softwood or mixed wood stands. The reduction in damage, however, was obtained at the cost of reduced tree height. Pine planted on strips in the mixed deciduous stand did not receive adequate shade protection under any of the treatments, since the weevils mated and oviposited before the hardwoods leafed out in the spring. The study was continued until 1982 and the long term results relayed in 1985 by Stiell and Berry more or less concurred with the preliminary results reported in 1976. Studies investigating the effects of overstory shading on P. strobi attack to spruces in British Columbia have added to the knowledge gained from the eastern white pine studies. McMullen (1976a) indicated that shading reduced heat accumulation below the threshold for successful brood development. VanderSar and Borden (1977c) showed that the weevil's visual response to the host leader silhouette, essential in the selection of host trees, was disrupted by overstory shade. Trials to determine the feasibility of utilizing different types of shade regimes to suppress weevil damage in spruce trees have now been established in British Columbia (Cozens 1987; 25 McLean 1989, 1994). At the University of British Columbia Malcolm Knapp Research Forest two plots of Sitka spruce were planted in 1980 to test the effects of a red alder (Alnus rubra Bong.) overstory on the incidence of P. strobi. In one plot Sitka spruce were planted after the deciduous overstory had been entirely removed (open-grown/no shade) and in the other plot spruce seedlings were established under 3-year-old natural red alder regeneration (the plot had been strip cleared such that alternate rows of alder running in a north-south orientation remained). Infestation rates tabulated 13 years after establishment indicated that attack was significantly greater in the open-grown regime (McLean 1994). In terms of growth, however, both leader lengths and radial dimensions were significantly reduced on the Sitka spruce growing under the red alder overstory (McLean 1989,1992). A red alder side-shading trial with east west oriented rows was subsequently established at the Malcolm Knapp Research Forest in the hopes that limited shading would allow for better leader growth while simultaneously conferring protection against weevil attack (McLean 1992,1994). The trees were laid out such that four rows of Sitka spruce were flanked between rows of naturally regenerated red alder. This setup thus shades the spruce from direct insolation but allows reflected light from the sky to enter the stand (McLean 1989). In 1985 Cozens (1987) had previously established a similar type of side-shading trial in the Prince George Forest Region with interior spruce seedlings (P. glauca x engelmannii) and a natural deciduous overstory of trembling aspen (Populus tremuloides Michx.) and western white birch (Betula papyri/era Marsh, var. commutata). Three treatment plots in each of three plantations (blocks) were established. Treatments included overstory shade (untreated controls with an intact deciduous overstory), side shade and no shade (all the deciduous overstory removed). Taylor and Cozens (1994) reported that differences in attack rate between treatments were not observed in the immediate three years following treatment but thereafter averaged 21.3%, 14.8% and 15.1% for the no shade, side shade and overstory shade respectively. Thus an 26 average of 6% reduction in weevil attack was shown for the treatments with treatment effects significant at an a level of 0.01. Differences in annual height increment between the no shade and side shade were observed for the first four years after treatment (i.e., height in the side shade treatment was significantly less). However, five years after establishment average annual height increment was not significantly different (a = 0.05) between the side shade and no shade treatments. Annual height increment was significantly less for the overstory shade treatment relative to the other two treatments throughout the study period, as in agreement with McLean's (1989,1992) findings. Thus, Taylor and Cozens (1994) have recommended the use of side shade in areas of high weevil hazard to reduce weevil attack rates. Recently, Taylor et al. (1994) evaluated the effects of overstory shading on white pine weevil damage in eight interior white spruce stands. Sites were selected to cover a wide range of both weevil attack intensity and percentage of crown cover. In this study a highly significant logistic model, which related percentage of weevil attack to tree height and diameter, number of past attacks and percent crown cover, was developed. Attack intensities decreased with increasing crown cover. Additionally, a significant relationship which predicted leader length from percent total crown cover, tree height and diameter was also reported. Not surprisingly, leader growth was negatively correlated to crown cover. Therefore, increasing crown cover, while decreasing levels of attack, adversely affected growth rates. At present it is not known how leader length reduction translates into reductions in merchantable volume. Taylor et al. (1994) concluded that this relationship needs to be established so that sound economic decisions pertaining to the optimization of overstory effects can be made e.g., what amount of overstory results in the greatest reduction in weevil attack without an unacceptable decrease in merchantable volume? 27 1.3.3.5 Predators and Parasites as Biological Control Agents for Pissodes strobi Biological control methods provide an environmentally acceptable avenue for controlling economically destructive forest pests. These techniques make use of the insect's "natural enemy complex" which include the entomopathogens such as viruses, bacteria, fungi, microsporidia and of course the insect's predators and parasitoids. Although the emphasis of using insect predators and parasitoids as control agents has been on those for introduced pests, successful employment of this strategy has also been seen with native pests (Pimentel 1963). Many different insect predators and parasitoids have been reported for P. strobi but in monocultures planted after clearcutting they fail to keep weevil populations at economically acceptable levels. Thus, increasing effort is being put into actively seeking ways in which to enhance the effectiveness of the these insect predators and parasitoids in P. strobi population control. Stevenson (1967) identified 13 hymenopteran parasites and one dipteran predator associated with P. strobi developing in Engelmann spruce leaders. Alfaro et al. (1985) examined the insect's associated with weevil's in Sitka spruce leaders and distinguished 14 species of hymenoptera, three species of dipteran and one species of psocopteran. The three principle natural control agents named by Stevenson (1967) were Dolichomitus terebrans nubilipennis (Viereck), Eurytoma pissodis Girault and Lonchaea corticis Taylor; E. pissodes and L. corticis are also recognized as important enemies of P. strobi developing within Sitka spruce (Hulme et al. 1986). Both D. terebrans nubilipennis and E. pissodis are ectoparasitoids belonging to the order Hymenoptera (Stevenson 1967). To complete development these two insects generally overwinter in the leader; D. terebrans nubilipennis as prepupae and E. pissodis as mature larvae. Lonchaea corticis is a dipteran predator of P . strobi (Stevenson 1967; Alfaro and Borden 1980; Alfaro et al. 1985; Hulme 1989,1990). This predator principally consumes P. strobi pupae 28 (Hulme 1989,1990; Hulme and Harris 1989) but prepupae and weak and unhealthy P. strobi larvae may also be attacked (Alfaro and Borden 1980; Hulme 1989). As with the two ectoparasites, mature larvae of L. corticis overwinter within the leader and they continue feeding on any P. strobi pupae which remain in situ (Hulme 1989; Hulme and Harris 1989). Technical feasibility is a critical parameter for any biological control method. Augmentation of natural enemies of forest pests can occasionally be accomplished by artificial mass laboratory rearing, but for P. strobi the focus has been on obtaining the insect associates directly from infected P. strobi leaders. Exploiting cold-hardiness (Hulme et al. 1986) and adult girth (Hulme et al. 1987) are two avenues which have been explored for separating P. strobi from its insect associates for use in biological control attempts. Hulme et al. (1986) demonstrated that most of P. strobVs predators and parasitoids could survive temperatures of -26° C. Conversely, few P. strobi in leaders of Sitka spruce survived at this low temperatures. Consequently, they recommended a one day cold treatment of clipped infested leaders at -26° C so that the insect associates would then be available for augmentation of natural populations. The use of adult girth to separate P. strobi's insect associates has previously been discussed under mechanical control options. Separation is accomplished since P. strobi adults cannot escape through 1.7 mm square openings while the insect associates can. This procedure may indeed be useful, but extreme care must be taken to ensure that all infected leaders in a plantation are clipped prior to weevil emergence. Allodorus crassigaster (Provencher) is an egg-larval parasitic braconid associated with P. strobi in western coastal regions of North America (Hulme 1994). Unlike the previous three natural enemies of P. strobi discussed above, this insect does not occur throughout the range of P. strobi and does not overwinter in the leader (Alfaro et al. 1985). Scientists at the Pacific Forestry Centre in Victoria have recently evaluated this insect for its potential in the biological control of the white pine weevil (Hulme 1994). Results appear promising since the adult braconid is specific to Pissodes weevils, has excellent searching abilities and the female exhibits very high fecundity. This latter quality makes it an excellent candidate if one could mass produce these insects in laboratory settings. Hulme (1994) has indicated that the cooler coastal areas of B.C. would most benefit from release of^4. crassigaster since in this region an increase in the parasite population should occur for more than one year with a single release (i.e., long-term, self-sustaining pest regulation should occur (inoculation) (Hulme 1992)); this is not the case for areas in which A. crassigaster is normally absent (i.e., releases would have to be made each year (inundation) (Hulme 1992)). Presently, the International Institute of Biological Control is investigating close relatives of A. crassigaster in Europe that may be utilized on a more broad basis in P. strobi control programs. It is, however, doubtful that these European forms will ever be utilized since this could prove to be dangerous to the native predators and parasites. None of P. strobi's natural enemies are as yet available in large numbers for biological control operational trials (Alfaro et al. 1995). 1.4 Development of Trees Which are Genetically Resistant to Pissodes strobi for Use in Control Programs Selection and breeding of trees for insect resistance provide a possible avenue for minimizing the extreme damage caused by P. strobi. The first step in this process requires that trees exist in nature which show resistance to insect attack (Hanover 1975). Several pine and spruce trees have long been recognized as being putatively resistant to P. strobi attack (Wright and Gabriel 1959; Marty and Mott 1964; Silver 1968; Alfaro and Wegwitz 1994). Once potential resistant trees have been identified it is essential to demonstrate the heritability of resistance to achieve genetic gains (in terms of reduced attack) in tree improvement programs. The evaluation 30 of heritability begins by establishing replicated test plots in which vegetative or seed propagations of candidate and susceptible trees are grown. Hanover (1975) has outlined the steps needed to rigorously demonstrate heritable resistance of trees to insects. In simple terms, total phenotypic variance is the sum of three variances: environmental variance, additive variance and dominance variance. The last two variances account for the genetic variation. Heritability is the proportion of the total phenotypic variance of a trait accounted for by the genetic variance of the trait (Solbrig and Solbrig 1979). When selection is based on individuals the genetic variance can be partitioned into its different components. The quantification and partitioning of heritability is a critical step in determining what level of selection to utilize in order to maximize genetic gains in tree improvement programs. Narrow-sense heritability is a measure of how an individual's phenotype will predict the performance of its progeny while broad-sense heritability indicates how an individual's phenotype will predict the performance of its clones (King 1994). Broad-sense heritability encompasses both the additive and the dominance genetic variance. Dominance genetic variance arises from the interaction of alleles and is that portion of the genotypic value not expressed in the breeding value (King 1994). Narrow-sense heritability isolates the additive genetic variance, which determines resemblance between relatives, and thus is the measure most important to tree breeders. If the genetic variance in a trait is largely non-additive, genetic gains are optimized by utilizing clonal options instead of seed options (King 1994). Mean heritabilities can be calculated for different levels of selection such as family selection. The detection, testing, propagation and ultimately the evaluation of the level of resistance should result in a gene pool of weevil-resistant trees for use in control programs. The steps outlined by Hanover (1975) have more or less been followed in trials conducted in British Columbia in which replicated test plots containing vegetative or seed propagations of 31 candidate trees for both Sitka spruce (Ying and Ebata 1994) and interior spruce (the complex of white spruce, Engelmann spruce and their hybrid swarms) (Kiss and Yanchuk 1991; King 1994) have been established. A moderate genetic basis for resistance to weevil attack was demonstrated for interior spruce in Kiss and Yanchuk's (1991) study. As in most other weevil resistance trials, they used the amount of weevil damage as the variable to test for genetic differences between tree families and tree individuals. Family mean heritability was estimated at 0.77 ± 0.11 and heritability on an individual tree basis at 0.18 ± 0.03. King (1994) partitioned this individual heritability into the additive and dominance components and indicated that virtually all the genetic variation in the individual was additive. Kiss and Yanchuk (1991) plotted mean percent damage versus mean 10-year family height and revealed a trend of decreasing attack with increasing mean family height. These results are counter to those reported by Alfaro et al. (1993) for Sitka spruce where tall families were observed to be generally more attacked. Under the supervision of Chen Ying, of the B.C. Ministry of Forests (BCMoF), 43 different Sitka spruce provenances are being tested for weevil resistance at 14 different locations in B.C. (Ying and Ebata 1994). Coastal test sites were established from 1973 to 1975 and locations span from Dragon Lake in northern B.C., to Head Bay situated on Vancouver Island. Seed sources (provenances) were collected from "plus" trees located on the east and west coasts of Vancouver Island, the lower mainland, the Queen Charlotte Islands, the Skeena River area and from the Alaska Panhandle (Wood 1987). Assessment at one of the sites (Sayward trial) revealed significant variation in terms of weevil attack among the provenances (Wood 1987). Therefore, a clonal trial near Fair Harbour (Vancouver Island) was established in 1984 to test the repeatability (i.e., to eliminate the effects of site) of provenance differences in weevil attack (Ying and Ebata 1994). Eight different provenances representing the range of provenance variation in weevil attack evidenced at the Sayward site were selected for testing at the Fair Harbour clonal trial (Ying and Ebata 1994). 32 Ying and Ebata (1994) have since concluded that weevil resistant Sitka spruce provenances do indeed exist and they have recommended which specific seed sources should be utilized in different coastal regions in B.C. They indicated that by simply using seeds from resistant provenances (now available) in high weevil-hazard zones, and no other level of selection such as clonal or family selection, levels of attack should be reduced by at least one half. The search and successful identification of resistant phenotypes, would be greatly aided if physiological traits linked directly or indirectly to weevil resistance could be identified (i.e., the elucidation of resistant mechanisms). To identify the tree characteristics directly associated with or indirectly correlated to resistance a complete understanding of the interrelationships between the host and insect (in this context) is required. This is by no means a simple task and demands a multidisciplinary approach by geneticists, physiologists and entomologists, as evidenced at the White Pine Weevil Symposium held in Richmond, British Columbia in 1994. Hanover (1975) presented an excellent diagrammatic representation of host-insect interactions as related to tree-resistance to an insect. He identified/revealed specific aspects of an insect's feeding and oviposition behavior which could be adversely influenced by tree resistance. Altered insect behavior potentially bestows host resistance since the possibility of successful utilization of the tree (by the insect) is reduced (Hanover 1975). Hanover's schematic representation of these host-insect interrelationships is presented in Figure 1. The four categories of physiological tree characteristics defined by Hanover (1975) (and clearly shown in Figure 1) as underlying the mechanisms of resistance are: 1) morphology and anatomy of the host; 2) chemical repellents produced by the host; 3) chemical attractants produced by the host; and 4) nutritional status of the host. Hanover (1975) emphasized the fact that actual host resistance mechanisms are probably a combination of traits either within the host itself or between the host and insect. to c fB O (/) -I sr o-n n' =r <* o o. 2. ft> V" £». 5' S in O ST 3 3 " •a o » » c 5." 3 § ST « w> »£• O 3 S •a TJ fB TJ o 5" 3 o o 3 2 <g rt- O O ~t ft) et-_,. o 2. w o » 3 O < fD "1 fB -a -a tn Q. Q. •< 5' 2 "« fP 3^ < 3 •5 3 » 2. ft> T J fB ft to < o. a. CD o ft 13 VO <! - J O" O co co O CO Os - 00 VO 0\ 3 •2 S VO 1 - 1 3 & ft o o o 3. a* c cr. - J C\ t_h to P— o 3 CO o ~% cc 3 O f -1 f> be "n i — o 3 a 3 f Q n> to o Q L ^ 3 d/'n Q. <» f/'Oi /7/0 JO/ 3 3 3 3 CO < to c » £ 2! | CO 3 05 ^ VO 0? Ui 3 vo p. CO o »-l D. ft 3 vo —) a* a o cn p» re vo 00 > e? o 3 D-CO o a ft 3 vo 00 Ln l - 1 1—I < o S £3 e x g 2 o Cu *" a ft re I-I ~ » as cn j, 5 » S. S. 3 vo O * \ \ t 3 C7> m Q " — • O s o C L m / CO m o H CD o M CL. ft 3 vo - J ~-J 03 3 & ft o o o 3 s. cr c l ' r r 1 2. cn cn i—> re vo o vo V O ee 34 In Figure 1 pertinent P. strobi studies which addressed weevil resistance (directly or indirectly) are indicated. Both eastern and western investigations, pertaining to pines and spruces respectively, are cited. References for genetic investigations (tree and insect) and tree heritability studies (as previously discussed) are shown in Figure 1. In the extreme right and left hand columns in the upper portion of the diagram noteworthy studies which contributed to general knowledge of P. strobVs feeding and oviposition behavior are listed. These studies did not attempt to directly correlate tree characteristics to weevil resistance but rather, focused on understanding the physical and visual cues involved in P. strobi host selection (both feeding and oviposition). Understanding the mechanisms by which P. strobi selects individual trees for feeding and oviposition could also provide an avenue for screening candidate resistant trees for propagation (Wood 1987). Thus, references contributing to this knowledge were included in Figure 1. This literature was previously reviewed (section 1.3.2) and will not be further detailed. Citations in the lower portion of Figure 1 (numbers 1 to 16 assigned) were itemized by the type(s) of tree characteristic examined in each study. These studies were again referenced in the upper portion of Figure 1 (numbers 1 to 16) beside the type of weevil behavior apparently altered by the tree resistant characteristic under exploration. The literature cited in the lower portion of Figure 1 (omitting previously discussed heritability studies) will now be reviewed. The morphological and anatomical category defined by Hanover (1975) encompasses all the developmental changes in the physical characteristics of the tree. Kriebel (1954), working with eastern white pine, was the first to relate P. strobi attack intensity to bark thickness. He showed that the thinner the bark the less frequently the tree was attacked and surmised that oviposition activity was negatively affected by this trait. The number, size and distribution of resin ducts is another characteristic which has received considerable attention with respect to weevil-resistant trees (Stroh and Gerhold 1965; Plank and Gerhold 1965; Tomlin and Borden 35 1994a). Stroh and Gerhold (1965) related the number and distribution of the outside and inside cortical resin ducts, as well as bark thickness to weevil attack. They indicated that P. strobi feeding on eastern white pine preferred thicker barked leaders and that feeding was also affected by the distribution of the resin ducts. Specifically, weevils feeding in the cortex of the leaders were observed to avoid resin ducts and ceased feeding if they could not circumvent the ducts. Thus, Stroh and Gerhold (1965) suggested that the selection and breeding of trees with shallower and abundant resin ducts could be a means of producing weevil resistant trees. Plank and Gerhold (1965) conducted a laboratory investigation in which four pine species (eastern and western white pine, jack pine and red pine) were examined for characteristics which could be linked to weevil resistance. Among the 10 morphological characteristics they attempted to correlate with weevil feeding were leader diameter, bark thickness and number and depth of resin ducts. Choice feeding bioassays revealed no clear correlation with any of these characteristics and furthermore, western white pine, known to be resistant to weevil attack under natural conditions, was accepted by the weevil's as a suitable food source. These authors stated, however, that western white pine exhibited larger and a greater number of outer resin ducts than evidenced in the eastern white pine and that perhaps this accounted for resistance in the field. Plank and Gerhold (1965) hypothesized that results were less than revealing and contradictory to those seen in the field perhaps due to the artificial conditions of their test situation (i.e., resin pressure significantly reduced in severed leaders). Tomlin and Borden (1994a) working with Sitka spruce also demonstrated that the number of outside resin ducts was significantly greater in resistant trees relative to susceptible trees. This observation has since been confirmed for interior spruce as well (Alfaro pers. comm.1). 1 Research Scientist, Canadian Forestry Service, Pacific Forestry Centre 36 Under the category of chemical repellents Hanover (1975) included repellents which act either prior to or after host-tree contact (i.e., volatiles as pre-contact repellents and non-volatiles which repel after contact or ingestion). Chemical attractants enter into the context of host resistance from the stance that the host produces insufficient quantities of the attractant thus avoiding initial attack. Studies which attempted to correlate chemical composition to weevil resistance often encompassed mechanisms of both chemical repellents and chemical attractants and hence will be presented concurrently. Resistance mechanisms addressing the nutritional status of the tree are extremely difficult to identify since nutritional status is perhaps inseparable from developmental changes in host anatomy or the presence of chemical repellents and attractants as Hanover (1975) himself prudently stated. This type of resistance mechanism was thus not considered when summarizing the research under the different physiological characteristics of the host. In reference to studies cited under chemical repellents and chemical attractants (Figure 1) Santamour and Zinkel (1976), Bridgen etal. (1979), and Wilkinson (1979) examined pines in eastern North America. Al l other weevil chemical repellent and attractant investigations referenced were undertaken in British Columbia and will be the primary focus of the next section. Western redcedar (Thujaplicata Donn ex D. Don in Lamb.) (has been shown to be an unacceptable P. strobi food source possibly due to volatile chemicals present in the foliage (VanderSar and Borden 1977a). Conversely, the resistance of Lodgepole pine (Alfaro and Borden 1982) and western white pine (VanderSar 1978; Alfaro and Borden 1982) has been surmised to be due to sub-optimal qualities or quantities of feeding stimulants. Alfaro and Borden (1980) reported that weevil feeding stimulants were present in a wide variety of conifers (33 native and exotic species studied). Alfaro and Borden (1980) suggested that weevil stimulants were restricted to conifers (Alfaro and Borden 1985) since they were absent in other types of 37 species tested (e.g., angiosperms). Both VanderSar and Borden (1977a) and Alfaro and Borden (1982) concurred that Sitka spruce was highly susceptible to weevil attack possibly due to the presence of chemical feeding stimulants in the leader. A complex mix of volatiles and non-volatiles found in the phloem and on the surface of the bark and needles (all associated with the leader) were identified as feeding stimulants and deterrents in Sitka spruce in an investigation conducted by Alfaro et al. (1980). Feeding stimulants were also shown to be present in the xylem of leaders but absent in xylem collected elsewhere from the tree (Alfaro and Borden 1985). Specifically, the monoterpenes a-pinene, 13-pinene and B-myrcene were postulated as acting synergistically with the non-volatiles present in the bark. These compounds apparently enhanced feeding while piperitone showed a feeding deterrent effect. More importantly, (+) camphor and limonene (in addition to other terpenes) showed mixed results acting as feeding stimulants at low concentration and feeding deterrents above a specific threshold concentration. At present, the general consensus is that monoterpenes are under strong genetic control (Hanover cites 12 studies which have demonstrated this genetic relationship) and thus they offer hope in terms of tree improvement programs if found to be somehow associated with weevil resistance. Indeed, Alfaro (1988) suggested that weevil resistant trees could possibly be produced by manipulating the array of feeding stimulants present in host bark by crossing Sitka spruce with other less susceptible Picea spp. The studies discussed in the preceding paragraph focused upon host and non-host trees of P. strobi in attempting to reveal weevil resistant mechanisms. Brooks et al. (1987) were among the first British Columbia investigators to specifically compare chemical profiles between Sitka spruce putatively identified as either weevil susceptible (60 trees) or resistant (38 trees). Unfortunately, in general, no definitive relationship between the monoterpene profiles (foliage samples) and weevil resistance was demonstrated. Although consistent differences in isoamyl 38 isovalerate and isopentenyl isovalerate (cortex samples) were shown between resistant and susceptible trees, the degree of overlapping concentrations was considered to be too great to facilitate accurate screening of resistant chemotypes based on this factor alone. Recently, Manville etal. (1994) also examined chemical profiles between resistant and susceptible spruce trees. Unlike Brooks et al. (1987), which restricted examination to monoterpenes, chemical analysis of leaf tissues included examining the entire terpene content (mono- sesqui- and diterpenes). Putative resistant and susceptible trees from both Sitka (Fair Harbour trial (Ying and Ebata 1994)) and interior (Kiss and Yanchuk 1991) spruce trees were sampled. Several important preliminary results emerged from this study such as: 1) terpene profiles found in Sitka spruce were shown to be under strict genetic control, unaffected by both environment or weevil attack; 2) expression of terpenes in Sitka spruce was strongly influenced by contributions from the male parent; and 3) the genetic variability within a family was observed to be as large as the genetic variability within populations. Based on this latter result, Manville et al. (1994) indicated that the selection of potential tree candidates for seed orchards should be conducted at the individual not family level. To facilitate individual tree selection two preliminary multivariate models based on factor analyses were presented which separated resistant from susceptible types. The multivariate models, based on six bark or foliar terpenes, correctly classified 15 out of 16 clone trees taken from the BCMoF interior Prince George Nursery at Red Rock as either susceptible or resistant. In eastern white pine Bridgen et al. (1979) revealed no relationship between oleoresin chemical composition (monoterpenes and resin acids) and viscosity factors and weevil resistance. At that time Bridgen et al. (1979) concluded that the chemical and physical characteristics they focused on could not be successfully used to predict weevil resistance in eastern white pine. 39 Research is currently being conducted at Simon Fraser University in an attempt to develop a multicomponent resistance index for Sitka spruce (Tomlin and Borden 1994b). With this approach, trees are evaluated for resistance based on the density of outer resin ducts, the amount of foliar isovalerates, the total cortical resin acid content and whether a significant feeding deterrency is demonstrated in paired-twig bioassays. Results from initial testing of the index using 11 different clones (with a range of observed attack intensities) established at the Fair Harbour provenance trial (Ying and Ebata 1994) were somewhat disappointing. Work continues on fine-tuning the index by exploring other possible indicators of resistance which may be incorporated into the index and by character weighting to reflect the relative importance of each putative resistance criteria (Tomlin and Borden 1994b). Having discussed pertinent literature related to examining tree characteristics putatively associated with weevil resistance a brief discussion pertaining to the different approach taken by Sahota et al. (1994) seems warranted. Their general tactic was to examine the behavior and physiological processes of ovarian development of mature and immature weevils (in terms of the presence of yokey eggs) caged on resistant and susceptible Sitka spruce leaders. A 10 step procedure/mechanism explaining the processes which lead to successful oviposition on a leader and ultimately brood development was outlined. They proposed that feeding on susceptible leaders caused physiological changes in the weevil which in turn altered the weevil's behavior from moving/feeding to moving/oviposition. Once moving/oviposition behavior occurred the authors suggested that the weevil became "domiciled" on the leader. Furthermore, Sahota et al. (1994) hypothesized that feeding on resistant leaders prevented the necessary physiological changes from taking place so that change to moving/oviposition behavior was effectively blocked or delayed. Thus, the weevil would continue with its moving/feeding pattern and ultimately move on to more susceptible leaders. At present, these authors have not attempted to identify any 40 specific host factors which could account for inhibition of the moving/oviposition behavior. If, in fact, this mechanism proves to be sound it could lead to another procedure whereby trees could be screened for P. strobi resistance. Any method which can quickly and accurately predict the susceptibility to weevil attack is worthy of exploration. Remarkable progress in somatic embryogensis (Roberts 1994) has now provided an avenue for mass propagation of resistant trees. Modern technology has also provided an alternative to searching for physiological traits linked to resistance, namely genetic markers linked to resistance. Carlson et al. (1994) are attempting to identify RAPD markers linked to weevil resistance. Bulk Segregant Analysis (BSA) (general technique thoroughly discussed in Chapter Three) was used with D N A obtained from resistant and susceptible interior spruce clones (Kiss and Yanchuk 1991). Three RAPD markers linked to weevil resistance have tentatively been identified. Work continues in this area with the long term goal of converting R A P D markers to "SCARS" (Sequence Characterized Amplified Regions (Michelmore et al. 1991)) markers. SCARS markers should provide a quick, reliable and simple tool for screening individual trees for weevil resistance (Carlson et al. 1994). Tree breeders have for sometime understood that trees do not always perform the same in different physical environments. Similarly, when developing trees resistant to insect attack they must also consider the confounding effects of the insect's genotype, as shown in Figure 1. Ideally, of course, tree resistance mechanisms should not only be genetically codetermined in the insect and host but would also be stable and applicable even given pressures from different weevil genotypes or environments (Hanover 1975). Leaf oil terpenes are believed to be under strong genetic control and not largely influenced by ecological factors on the whole (Manville et al. 1994), but this may not be the case for other tree characteristics linked to resistance. Thus, it is imperative to know the degree of genetic variation present in populations of P. strobi throughout 41 British Columbia. The RAPD marker procedure will help to determine whether different genotypes of weevils must be recognized and considered in the development of weevil resistant spruce trees in B.C. It is hoped that tree-breeders have learned from the mistakes of their counterparts in plant-breeding and will ultimately employ strategies which will result in durable resistance which will not break down over time (Namkoong 1994). Establishing the extent of genetic variation present in P. strobi populations throughout B.C. should contribute to the successful completion of this long term goal. 1.5 Approaching Pissodes strobi Control Via an Integrated Pest Management System Integrated Pest Management (IPM), figuratively speaking, can be viewed as an holistic approach to pest management. In this system the ecological interaction of the pest with the host is stressed (Retnakaran et al. 1982) so that the ultimate goal is not eradication of the noxious insect but reduction in pest damage to economically acceptable levels (Alfaro et al. 1994,1995). This management regime makes use of a combination of control methods with emphasis on protection of the environment. Use of synthetic contact chemicals is only utilized as a last resort, under imperative conditions. IPM was first implemented for destructive agricultural pests beginning with the cotton boll weevil in the 1930's. Professor Dwight Isley developed the system which utilized desirable plant strains, close weevil population monitoring, enhancement of the natural enemy complex, sanitation and contact insecticides to reduce cotton boll weevil damage (Retnakaran et al. 1982). Scientists enthusiastically embraced the concept of IPM in the early 1970's and today it is the strategy of choice to combat pests. An example of a highly successful IPM system in a forestry setting is the program developed for the Douglas-fir tussock moth, Orgyiapseudotsugata (McDunnough) (Shepherd and Otvos 1986). 42 Dixon and Houseweart (1982) were among the first investigators to suggest developing an IPM program for P. strobi. Biological control, resistant trees and destruction of overwintering sites for the weevils were among the potential control methods considered. Recently, Alfaro et al. (1994,1995) outlined a detailed IPM system for white pine weevil in British Columbia incorporating several control options. Tactics were described for both established and future spruce plantations (Alfaro et al. 1994,1995). These two strategies rely on knowing the exact cost-benefits associated with each possible control option so that sound decisions can be made pertaining to which combination of control to implement in each different ecoregion in B.C. The IPM strategy developed for protection of new plantations depends on first assessing each site for the risk of weevil damage. Thus, areas that are at extreme risk should never have monocultures of spruce planted and the deciduous portion, or alternate species, would be encouraged to be the primary component of the stand. High risk sites would be planted with genotype mixtures of resistant trees, susceptible trees and with non-host trees. Other management options include employing dense stocking, pruning to improve form, and insecticide injections to crop trees (Alfaro et al. 1994,1995). Sites specified as being at risk, but only mildly so, would primarily employ silvicultural tactics. Forest sites in B.C. can be delineated as high or low weevil hazard based on the accumulated number of degree days (dd) above a developmental threshold needed for brood development (McMullen 1976a,b). If the climate is such that the required number of degree days is not accumulated then brood development is highly unlikely and these sites are classified as low hazard areas. Pissodes strobi climate-based hazard rating is possible throughout B.C. since the accumulated degree days needed for weevil development in both coastal (888 dd (McMullen 1976a)) and interior (785 dd (McMullen 1976b)) environments have been determined. Heppner and Wood (1984) established weevil hazard zones for Vancouver Island and the coastal mainland of B.C. Sieben (1991) later assessed the weevil hazard for much of the Cariboo and 4 3 Prince George Forest Regions. Sieben (1995) is continuing with this work for the Nelson Forest Region and for the MacKenzie River Basin in northern B.C. Trial plantations in which the IPM system is being tested have recently been initiated in limited areas in B.C. Alfaro et al. (1994,1995) have stressed the need to establish additional trials in different ecoregions so that a sound IPM program for minimizing P. strobi damage will be established throughout B.C. 44 CHAPTER 2: SYSTEMATICS AND THE RAPD MARKER TECHNIQUE 2.1 Systematics 2.1.1 Introduction and Terminology The branch of biology known as systematics deals with the detection, description and explanation of diversity in the biological world (Moritz and Hillis 1990). In systematics, man-made systems of classification express degrees of similarity between organisms (Jefferey 1977). While the establishment of systematic groups, known as taxa, is the process of classification, the process of nomenclature is the allocation of names to the taxa (Jefferey 1977). The first formalized hierarchical system of nomenclature based on the description and categorization of biological diversity was developed by Linnaeus (Moritz and Hillis 1990). I have adopted the opinion held by Moritz and Hillis (1990) that views systematics as encompassing the study of both intraspecific and interspecific diversity. Historically, hierarchical classification systems did not encompass evolutionary theory (Moritz and Hillis 1990). Not until the days of Darwin was it proposed that classification should be based on phylogenetic relationships. Today, L i and Graur (1991) have defined systematics as "taxonomy and phylogenetics" with phylogenetics defined as "the reconstruction of the evolutionary history of a group of taxa or genes" and taxonomy as "the principles and procedures according to which species are named and assigned to taxonomic groups". Futuyma (1986) viewed systematics from more or less the same perspective and defined it as "the study of the historical evolutionary and genetic relationships among organisms, and of their phenotypic similarities and differences". 45 The Dictionary of Biology (Martin 1985) identifies four different fields of taxonomy based on the type of information or approach used in the classification procedure: 1) classical taxonomy utilizing morphological and anatomical features; 2) biochemical taxonomy based on similarities in the structure of proteins and nucleic acids; 3) cytotaxonomy involving the comparison of the size, shape, and number of somatic chromosomes and; 4) numerical taxonomy which employs mathematical procedures to quantitatively assess similarities and differences between taxonomic groups. The Harper Collins Dictionary of Biology (Hale and Margham 1991) defined biochemical taxonomy from a slightly different stance and labeled it experimental taxonomy which encompasses the determination of genetic relationships. Since genetic relationships are often determined from molecular markers the term molecular systematics has arisen. Moritz and Hillis (1990) have identified three primary applications of molecular systematics: examining population structure (e.g., determining geographic variation, mating systems, and heterozygosity), identification of species boundaries and estimation of phylogenies. The term molecular evolution is aptly applied to phylogenetic relationships derived from molecular markers (Li and Graur 1991). The field of molecular evolution relies on the collaboration of two quite distinct disciplines: molecular biology which furnishes the empirical data and population genetics which provides the theoretical framework from which an understanding of the evolutionary processes may be gained (Li and Graur 1991). Appendix II lists noteworthy molecular markers traditionally used to study genetic relationships among organisms. Pertinent references for each laboratory technique, a brief outline of the general procedure and the type of data collected, the most applicable level of hierarchy to address and various advantages and concerns of each molecular marker are presented. Consciously omitted 46 from Appendix II was information pertaining to the RAPD marker technique, which will be discussed at length in part two of this chapter. Although molecular systematics has been employed in this study primarily to examine population structure in P. strobi, some of the more important theoretical debates which have tended to overshadow and confuse the area of molecular phylogenetics will now be discussed. In essence, these issues have diverted attention away from the application of molecules as highly informative genetic markers (Avise 1994). Additionally, I felt that a discussion on these theoretical debates was relevant since I grappled with many of these concepts and came to realize they were extremely pertinent in understanding theoretical population genetics as a whole. Understanding these concepts was also critical for thoroughly comprehending the population genetic models used herein to analyze my RAPD data (to be discussed). 2.1.2 Theoretical and Philosophical Debates of Molecular Phylogenetics 2.1.2.1 Classical-Balance Debate Evolution, in itself, is a highly complex and controversial concept. Dobzhansky, in 1937, defined evolution as "a change in genetic composition through time" (Avise 1994). Changes in genetic composition can arise from alterations in the frequencies of alleles or genotypes (Futuyma 1986). The four fundamental forces which can operate to bring about changes in allele frequencies are mutation, genetic drift, gene flow and natural selection (Hartl and Clark 1989; Futuyma 1986). Nonrandom mating, by itself, may not result in a change in allele frequencies but could affect the frequency of genotypes in a population (i.e., it often results in a reduction in the number of heterozygotes) (Futuyma 1986). Population geneticists attempt to measure genetic variation, often with the objective of gaining insight into how (and which of) these forces may 47 have operated to bring about the changes observed. A prerequisite to this central challenge is that genetic variation must exist; however the magnitude of genetic variation present in natural populations has been hotly debated. Herein lies the basis for the historical classical versus balance debate which dominated systematics prior to the mid-1960's (Avise 1994). Supporters of the "classical school" alleged that genetic variability in most species was low, while proponents of the "balance view" maintained that genetic variation was high. Classicists adamantly refuted the contention that most loci were polymorphic and individuals typically heterozygous at a large fraction of genes. They viewed natural selection as a cleansing agent, purging the genome of mutational variation. Recombination was seen as an inconsequential process. Furthermore, genetic variation was discerned as a tremendous burden to a population ultimately resulting in diminished fitness which could lead to population extinction (Avise 1994). Since genetic variation was believed to be almost nonexistent any genetic variation among populations was seen as being of profound importance. Conversely, the "balance school" viewed genetic variation as the "norm" and adaptively relevant. Natural selection was seen as favoring genetic polymorphisms to such an extent that the balance school believed that heterozygosis among gene loci (for individuals or a population) tended towards 100 percent. The extensive phenotypic variation observed in most natural populations, the genetic basis for many morphological variants and the genetic response of animals and plants to artificial selection all indirectly supported the balance hypothesis. Not until genetic variability studies were based on multilocus protein electrophoresis was it possible to directly estimate what fraction of genes in an individual were actually heterozygous. Although crude protein electrophoresis techniques were available in the 1930's, an electrophoretic revolution truly began in 1966 (Avise 1994). Results from hundreds of studies, conducted over a 20 year span, using both plant and animal samples, revealed levels of genetic 48 variation that were high but also somewhat variable among species. Estimates of population heterozygosity typically ranged from 0.00 -0.20 and estimates of the proportion of polymorphic loci from 0.00 - 0.80 (Avise 1994). Protein electrophoresis data provided critical information on genetic variation but was not wholeheartedly embraced as the ultimate methodology for addressing phytogeny. Rather, for approximately 20 years after the protein-electrophoretic revolution began, empirical population geneticists often focused their research goals on several non-phylogenetic issues. These included investigating the extent of genetic variation which remained undetected using this technique, the sampling effects of the procedure, and how protein electrophoretic variation related to fitness (Avise 1994). 2.1.2.2 Neutral Theory (Molecular Clock Hypothesis) The clear demonstration, via protein electrophoresis, of abundant genetic variation in natural populations did not entirely eliminate the "classical school" of thinking. Rather, neoclassicists, or neutralists, emerged. Neutralists are embroiled in the neutralist-selectionist controversy. They contend that molecular variants are selectively neutral (or nearly so) (Moritz and Hillis 1990). Neutralists have been viewed as neoclassicists because both neutralists and classicists have maintained that balancing selection (intermediate phenotype most fit) does not function in maintaining polymorphism and that directional selection (extreme phenotype most fit) serves to cleanse the genome of deleterious alleles (Avise 1994). Unlike classicists, however, original proponents of the neutrality theory did not refute the existence of high molecular variability but argued that this variation reflects the transient passage of neutral alleles (Futuyma 1986). Thus, molecular variants are seen as confirming no differential fitness effects on their carriers and hence, selection is discounted as a factor in population genetic models. Neutrality 49 theory proclaims that polymorphism levels are upheld at relatively high levels due to a balance between mutational input and random allelic extinction via genetic drift (Avise 1994). Genetic drift refers to random changes in the frequencies of two or more alleles due to sampling variation of gametes from generation to generation (Futuyma 1986; Avise 1994). It refers to shifts in frequencies of pre-existing alleles and can be viewed as a special case of sampling error which is inversely related to sample size (Avise 1994). In reference to the origin and substitution of new alleles, neutrality theory proclaims that each nucleotide is equally likely to mutate to any other (Haiti and Clark 1989). Furthermore, the rate of neutral evolution has been shown to be entirely independent of population size, being equal to the mutational rate to neutral alleles (Avise 1994). Hence, supporters of neutrality theory have maintained that protein and D N A sequences have diverged among species at a constant rate (Futuyma 1986). These fundamental assumptions have provided the foundation for the somewhat controversial "molecular clock" hypothesis. In essence, protein and D N A sequences which have diverged among species at a constant rate provide a "molecular clock" by which the time since common ancestry can be estimated (Futuyma 1986). To calibrate the "molecular clock" investigators must have abundant fossil evidence from an organism so that an estimate of the number of base substitutions per millennium (on average) can be obtained (i.e., mutational rate). This estimate is then extrapolated to different organisms that have an inadequate fossil record but are assumed to exhibit identical mutational rates (Futuyma 1986). As well as being paramount to the neutral theory of evolution and to estimating time since divergence, constancy of rates is also an assumption of various methods for estimating phylogeny (Moritz and Hillis 1990). The selection-neutrality controversy has by no means been laid to rest. Avise (1994) has indicated that the debate rages on for two fundamental reasons. First, both selectionists and neutralists are able to vehemently support their positions given any set of observations. Second, 50 the exact definition of natural selection is by no means universal or precise. Thus, there is no general consensus whether to include such phenomenon as "meiotic drive" or "selfish genes" as agents of natural selection. In any event, Avise (1994) argues that this debate is perhaps only critical for applications such as phylogeny reconstruction and is more or less inconsequential for clonal identification or parentage assessment. 2.1.2.3 Phenetic vs Cladistic Approach to Systematics There are two contending schools of thought which differ widely in their approach to addressing molecular or other data in a systematic context. These two general schools of systematics are phenetics and cladistics. Pheneticists maintain that classification should be based on quantitative measures of overall (phenetic) similarity (or its converse, distance). Members of this school advocate gathering information on as many characteristics as possible for use in classification (Futuyma 1986; Avise 1994). Pheneticists are also known as numerical taxonomists since they have developed elaborate numerical methods for grouping species on the basis of overall similarity (Futuyma 1986). The groupings of the operational taxonomic units (OTUs) based on numerical taxonomic methods are portrayed in diagrams aptly named phenograms (Futuyma 1986). Although OTUs are most commonly species or higher taxa they may also be well-isolated conspecific populations, individuals or nonrecombined alleles of a gene (Avise 1994). Phenograms may also represent phylogenetic trees if the following two objectives are met: 1) the organisms are grouped according to genealogical ties; and 2) the time of divergence between organisms is estimated. This latter statement (as discussed above) assumes constant rates of evolution between lineages (Futuyma 1986). If differential rates of evolution were to have 51 occurred, the phenogram would not represent the true phylogenetic tree (Futuyma 1986). Phylogenetic trees are depicted with both nodes (taxonomic units) and branches (pathways connecting branches). They can be constructed as either rooted or unrooted trees. In general, if information exists such that the common ancestor of all OTUs under investigation is known (or somehow deduced), this node serves as the root. From the root, unique pathways lead to every other node and the direction of each path corresponds to evolutionary time (Li and Graur 1991). Unrooted trees, also referred to as networks, serve only to specify the relationships among the OTUs and do not define the evolutionary path (Li and Graur 1991). Advocates of the cladistic school follow many of the principles laid out by the famous German entomologist Willi Hennig (died 1979) (Futuyma 1986; Avise 1994). Cladists do not believe that classification should be based on overall similarity (Avise 1994). Rather, cladists classify organisms on the basis of the historical sequences by which they have diverged from common ancestors (Futuyma 1986). In this type of classification it is the branching (cladistic) relationships among species that are expressed, regardless of the degree of similarity or difference among the species (Futuyma 1986). Hence, cladists do not focus on accumulated change within lineages (anagenesis) and so are not concerned with the branch lengths of the phylogenetic trees (Avise 1994). Stated from a speciation perspective, anagenesis refers to gradual speciation wherein directional change within a lineage occurs. The World Book Dictionary (Barnhart and Barnhart 1991) defines anagenesis as "progressive evolutionary change within a species". True speciation refers to the splitting of lineages giving rise to two distinct lineages. The splitting of lineages into different species is called cladogenesis (Solbrig and Solbrig 1979). The challenge to cladists is to correctly distinguish ancestral characters (plesiomorphs) from advanced characters (apomorphs) since they attempt to distinguish clades (the set of species descended from the ancestral species (Futuyma 1986)) on the basis of synapomorphs (shared 52 derived character states). Inherent in the preceding statement is the need for cladists to differentiate symplesiomorphic characters (ancestral character shared by several species) from synapomorphic characters (Avise 1994). Cladists have used various criteria to suggest primitiveness for a character, such as presence in fossil records, commonness among an array of taxa, early appearance in ontogeny and presence in an outgroup (Avise 1994). It is the latter criterion which has appeared to yield the most reliable results and is thus the criteria most commonly used by cladists today (Avise 1994). Cladists depict clades in diagrams termed cladograms, often utilizing only one true synapomorphy in the classification procedure. It is the use of only one character to define genealogical relationships that has been the focal point of intense criticism from pheneticists. Most researchers do agree that shared-derived traits are a solid basis for clade delineation (Avise 1994) but issues such as the selection of characters and the correct identification of synapomorphies cause concerns. One of the primary reasons for addressing the cladistic-phenetic debate in this thesis is that data collected from molecular markers can be of two distinctly different forms (i.e., in terms of systematics and specifically phylogenetic analysis): distance data, where differences are measured as a single quantitative variable, and character data, where differences are measured as a series of discrete variables (Moritz and Hillis 1990). While character state data can be converted to distance data the reverse statement does not hold (Moritz and Hillis 1990). In so far as phylogenetic investigations are concerned, strict cladistic approaches cannot be applied to distance data (Avise 1994). Furthermore, Avise (1994) stated that even given molecular character state data, such as that obtained from protein electrophoresis, strict cladistic analysis is not particularly relevant due to the high risk of homoplasy (similar but of different origin) at the level of individual electromorphs. As previously explained it was imperative to have a full grasp of the debates surrounding molecular phylogenetics so that sound decisions pertaining to data analysis could be made. Texts such as Avise (1994) and Hillis and Mortiz (1990) clarified many of these issues and presented the arguments in excellent chronological accounts. Both of these books are highly recommended as required reading for intermediate population genetic courses which encompass the use of molecular markers in systematics. 2.2 RAPD Marker Technique 2.2.1 Development of Technique The RAPD marker procedure (Williams et al. 1990), a recently evolved spin-off of the Polymerase Chain Reaction (PCR) method, is proving to be a powerful tool in molecular taxonomy (Paran et al. 1991; Chapco etal. 1992; Perring etal. 1993) and population genetics (Chalmers et al. 1992; Chapco et al. 1992; Perring et al. 1993) and in genome mapping (Carlson et al. 1991; Martin et al. 1991; Tulsieram et al. 1992; Welsh et al. 1991; Reiter et al. 1992; Hunt and Page 1995). Basic reaction ingredients in D N A PCR amplification include: 1) template D N A extracted from the organism under investigation; 2) a thermostable D N A polymerase, such as that isolated from the thermophillic bacterium Thermus aquaticus (enzyme named Taq) (Saiki et al. 1988), used to link together the deoxynucleotide triphosphates (dNTPs); 3) each of the four dNTPs (dATP, dCTP, dGTP, dTTP - adenine, cytosine, guanine and thymine respectively) needed for in vitro D N A construction; 4) M g 2 + , critical for Taq activity, D N A synthesis and annealment of primer to template DNA; 5) reaction buffer for maintaining homeostasis; and 6) primers which target a specific segment of DNA. PCR amplification proceeds through a series of steps, the first step being denaturation of the double-stranded template DNA. Denaturation is achieved by applying a high temperature condition (e.g., 94° C). Primer annealing follows 54 denaturation: this step is conducted at a specific temperature in the presence of excess primers to avoid incomplete amplification. DNA synthesis commences, after primer annealment, generally at a temperature of 72° C. These three steps constitute one PCR cycle and the entire PCR amplification procedure may encompass as many as 35 cycles. Products of the first round of amplification serve as templates for the second round of amplification. Thus, amplification proceeds exponentially after the second round with the major PCR product being a double-stranded segment of DNA whose length is defined by the distance between the two primers employed (Escote-Carlson 1991). Products of DNA amplification are resolved by gel electrophoresis and visualized after staining with an agent that binds to DNA, such as ethidium bromide (EtBr). The general RAPD technique, technically invented first by Welsh and McClelland (1990) but also developed independently and more or less simultaneously by Williams et al. (1990), results in the production of multiple amplification products from PCR at low stringency annealing (for either the first two cycles or all the cycles) with a single primer of arbitrary sequence. These conditions are counter to the standard PCR method, wherein annealing is done at high stringency with a pair of oligonucleotide primers (10 to 28 nucleotides in length (Escote-Carlson 1991)) designed to target one specific segment of DNA (Innis et al. 1990 ) (all other PCR ingredients discussed above remain the same). Welsh and McClelland's (1990) procedure differed from Williams et al. (1990) in that the former group employed arbitrary primers 20 and 34 nucleotides in length (45 and 75 % GC content (G = guanine, C = cytosine)) while the latter group used random primers 10 nucleotides in length (GC content maintained at or above 50%). Temperature profiles and number of cycles used during the amplification procedure also differed between the two investigative teams: Welsh and McClelland used two cycles of low stringency annealing (40° C) followed by 30 cycles of high stringency annealing (60°) while Williams employed low 55 stringency annealing (36° C) throughout the entire amplification procedure (45 cycles). Welsh and McClelland entitled their technique arbitrarily primed PCR (AP-PCR) and Williams originated the R A P D acronym. I have used the RAPD terminology throughout my thesis since the conditions used in this study were most similar to those reported by Williams et al. (1990). R A P D marker studies commonly employ a primer 10-nucleotides in length constructed with an enriched GC content (Rafalski et al. 1991). Random primers as short as five nucleotides in length, however, have been demonstrated to yield complex, informative D N A banding patterns (when coupled with resolution by polyacrylamide gel electrophoresis followed by silver staining) (Caetano-Anolles et al. 1991) and can also be used in RAPD investigations. R A P D amplification generally yields a different genomic fingerprint for each primer (Chapco et al. 1992; Perring et al. 1993) and for each genotype present in a population (Carlson et al. 1991; Chapco et al. 1992; Perring et al. 1993). This unique system thus facilitates the rapid elucidation of a multitude of genetic markers present in an organism or population (Hedrick 1992). Polymorphic RAPD markers, visualized as the presence (or absence) of bands (Rafalski et al. 1991), are similar to, and can be used in much the same fashion (e.g., constructing genetic maps, investigating genetic variation), as other genetic polymorphisms (e.g., Restriction Fragment Length Polymorphisms = RFLPs). 2.2.2 Entomological Studies Utilizing RAPD Markers Entomological applications of RAPD markers have included determining paternity (Fondrk etal. 1993), establishing linkage maps (Hunt and Page 1995), studying egg laying behavior (Apostol et al. 1993; Vanlerberghe-Masutti 1994), testing for the presence of parasitoids (Black et al. 1992; Vanlerberghe-Masutti 1994), and investigating interspecific (Black et al. 1992; 56 Chapco et al. 1992; Kambhampati et al. 1992; Welsh et al. 1992; Landry et al. 1993; Perring et al. 1993) and intraspecific (Ballinger-Crabtree etal. 1992; Puterka etal. 1993; Dillwith and Soos 1994; Edwards and Hoy 1994; Lu and Rank 1995) genetic variation. Fondrk et al. (1993) demonstrated that RAPD markers could be used to successfully determine the paternity of worker honeybees (Apis mellifera). Apis mellifera make excellent RAPD subjects since they exhibit haplodiploidy i.e., females are diploid, males are haploid. Segregating alleles can be identified since reproductive females can be analyzed along with their haploid sons (Fondrk et al. 1993). Haplodiploidy thus provides powerful avenues for circumventing the problem of dominance in RAPD markers. In Fondrk et al. (1993), protein electrophoresis was first used to positively establish parentage of Fl worker progeny; offspring had been produced from crossing a queen honeybee homozygous for two enzymes (malate dehyrogenase and esterase) with four male drones heterozygous for the same two enzymes. Since the four male drones all displayed a unique combination of alleles, fathership was unequivocally determined from the allelic variants exhibited by each offspring. Extracts obtained from the worker heads were used to conduct protein electrophoresis and total genomic DNA extracted from the workers thorax and abdomen was used for RAPD analysis. Twenty RAPD markers derived from eight, 10 base oligonucleotides (GC enriched) were tested to see if identification of fathers adhered to that/paternity established from the allozyme markers. All thirty-two worker progeny (eight from each of the four different fathers) were classified correctly in cluster analysis (demonstrating the effectiveness of RAPD markers in paternity analysis). Hunt and Page (1995) used RAPD markers to construct a linkage map and to estimate total genome size and recombination rates in A. mellifera. One thousand 10-nucleotide primers were initially screened. One hundred and thirty-two of these primers were used to generate 365 informative RAPD markers. Twenty-six linkage groups, which encompassed 3110 centiMorgans 57 (cM), were identified. Since honeybees exhibit only 16 chromosomes, 10 gaps were indicated on the map. The large size of the map compelled Hunt and Page (1995) to conclude that honeybees exhibit very high recombination rates. Complete genome size in the honeybee was estimated to be approximately 3450 cM. Apostol etal. (1993) developed an elegant statistical procedure for estimating the number of full-sibling families at an oviposition site of Aedes aegypti using RAPD markers. The procedure basically entailed establishing mathematical models for determining the expected proportion of matches (i.e., shared presence and absence of RAPD bands) among families (Mp). The validity of the approach was tested using established full-sibling families; sample size consisted of 15 individuals each from 10 different mosquito families. Thus, the actual number of families was known and compared to the number predicted from cluster analysis. Forty RAPD markers generated from five, 10-nucleotide primers (60-90 % GC content) were used in the analysis. To apply the technique, the frequencies of these RAPD alleles had to first be estimated. Restricting the analysis to allele frequencies (dominant allele) between <0.1 and >0.6 yielded a discriminating M p value of 0.77. The M p value of 0.77 was then used as a cutoff point in cluster analysis for estimating the number of families. A near perfect linear relationship (r2 = 0.99) was shown in regression analysis of the actual number of families versus the predicted number of families. Apostol et al. (1993) recommended this technique for investigating egg laying behavior in any species in which clusters of individuals are suspected to consist of mixtures of full siblings. Vanlerberghe-Masutti (1994) used RAPD markers to investigate egg laying behavior of a Trichogramma species which parasitizes Lepidopteran egg clusters. In this study four parasitized egg clusters were examined. RAPD analysis done on these egg clusters generated a total of 30 different markers. Results obtained from both multifactorial and parsimony analysis indicated that 58 several different female Trichogramma parasitized each egg cluster. Thus, Vanlerberghe-Masutti (1994) concluded that very little competition existed among female Trichogramma of this species. Vanlerberghe-Masutti (1994) also reported on the use of RAPD markers for investigating parasitization in aphids. She revealed that RAPD markers specific to the parasite were evident from aphid D N A extract approximately four days after infection. These results are similar to Black et al. (1992) who also utilized the RAPD marker technique to detect the presence of an endoparastic wasp (Diaeretiella rapae (Mcintosh)) within Russian wheat aphid (Diuraphis noxia (Mordvilko)) bodies. RAPD analysis detected the presence of this parasitoid no sooner than six days after parasitization (Black et al. 1992). R A P D markers have been used in numerous entomological studies to examine interspecific genetic variation. Black et al. (1992) showed significant levels of R A P D marker variation among four species of aphids: the greenbug (Schizaphis graminum (Rondani)), the Russian wheat aphid, the pea aphid (Acyrthosiphon pisum (Harris)) and the brown ambrosia aphid (Uroleucon ambrosiae (Thomas)). Morphological and biochemical polymorphisms had been shown to be rare between these species (Black et al. 1992). Large amounts of R A P D marker variation were produced using four 10-nucleotide primers (three having 60% GC content, the other 40% GC content). Substantial genetic variation was detected among individuals in each species, within and among biotypes, among populations and among color morphs. Black et al. (1992) also demonstrated the reproducibility of the RAPD marker technique using 10 sister nymphs produced through apomictic parthenogenesis; RAPD amplification profiles were identical for all 10 sisters, as theoretically expected. Chapco et al. (1992) examined the population genetics of several melanopline and oedipodine (two subfamilies of Acrididae) grasshopper populations using the R A P D marker system. Fourteen different grasshopper populations were used in this study: 11 populations 59 belonging to the subfamily melanoplinae, three populations representative of the subfamily oedipodinae. A total of twenty-four, nine-nucleotide primers (GC content ranging from 55 to 67%) were employed in four separate experiments. Chapco et al. (1992) showed that mean percent similarity (i.e., percentage of shared bands) was 51.2% within species and 35.0% between species or genera (overall). Chapco et al. (1992) stated that sample size was inadequate in their study (i.e., mostly one individual per population) to precisely estimate polymorphism levels and establish phylogenetic relationships. Kambhampati et al. (1992) used RAPD markers to investigate genetic relationships among five species of mosquitoes belonging to two subgroups of Aedes scutellaris (three species belonging to subgroup Ae. scutellaris, two species belonging to subgroup Ae. albopictus). They wished to determine the utility of the RAPD marker technique for differentiating species and populations, for identifying unknown species, and for reconstructing phylogeny. Two 10-nucleotide primers (60% GC) were utilized with D N A extracted from, usually, four individuals from each population. Diagnostic RAPD profiles were generated for each of the five species examined. The establishment of conserved D N A fragment profiles was invaluable since differentiation of species within each of the two subgroups proved to be extremely difficult morphologically (Kambhampati et al. 1992). In terms of differentiating species, all individuals, including several unknowns, were classified or grouped correctly at the species level in nonparametric discriminant and cluster analysis respectively. Furthermore, the multivariate analysis was able to differentiate individuals to the population level. Cluster analysis did not, however, reflect previously established ancestral relationships among the two subgroups. Kambhampati et al. (1992) suggested that phylogeny may not have been accurately reconstructed possibly due to employing species-specific D N A fragments as opposed to lineage-specific D N A fragments. 60 Two entomological investigations which used RAPD markers to aid in identifying new species were published by Landry et al. (1993) and Perring et al. (1993). Landry et al. (1993) examined genetic variation among five biotypes of parasitic wasps belonging to the genus Anaphes. Phylogenetic analysis using maximum parsimony was done with 57 R A P D markers derived from 13 10-nucleotide primers (60 to 70% GC). To test for stability of the tree shape, phylogenetic analysis was repeated (several times) excluding five randomly selected loci. Al l cladograms were virtually identical and indicated two major clades. Landry et al. (1993) concluded that three of the biotypes belonged to the species Anaphes sp.nov., a new species not yet described, and the other two biotypes to Anaphes sordidatus (Girault). Between species genetic distances ranged from 0.63 to 0.81. Landry et al. (1993) also demonstrated that genetic variation was much higher in Anaphes sp.nov. relative to Anaphes sordidatus with RAPD markers. These results were as biologically expected since with Anaphes sordidatus up to six insects can emerge from each parasitized host egg. This is counter to Anaphes sp.nov. where one insect commonly emerges from each host egg. Mating among the microhymenoptera occurs immediately after adult eclosion. Males emerge first and wait on the host egg shell for emergence of their mates. Thus, the probability of inbreeding is much greater in Anaphes sp.nov. (Landry et al. 1993). A study conducted by Perring et al. (1993) used results from protein electrophoresis, the RAPD marker technique, and cross-breeding experiments to identify a new species of whitefly. Sample size for the allozyme and RAPD marker portion consisted of 10 individuals each from 17 different populations; whiteflies were coded as "type A " for the "cotton strain" and "type B " for the "poinsettia strain". Fourteen different enzymes were evaluated, eight of which showed allelic polymorphisms. Fixed allelic differences between type A and type B whiteflies were observed at three of the eight polymorphic allozymes. Nei's genetic distance between type A and type B was 61 reported to be 0.24. Seven different 15- or 16-nucleotide primers (all at least 50% GC content) were used in RAPD analysis. Different genomic profiles between type A and type B whiteflies were revealed for all seven RAPD primers. Populations of the same type showed 8.0 to 100% similarity in RAPD banding profiles, but less that 10% similarity was seen between types (Perring et al. 1993). Reproductive isolation was demonstrated since no offspring were produced between types A and B (even though courtship behavior was observed). The value of Nei's genetic distance, the low similarity in RAPD markers and the lack of viable offspring all compelled Perring et al. (1993) to conclude that the type B whitefly was indeed a different species than type A. They suggested that this whitefly be referred to as the silverleaf whitefly. Welsh et al. (1992) examined genetic variation among 30 worldwide isolates of the corkscrew shaped bacteria, Borrelia burgdorferi, using RAPD markers. This study has been included under entomological investigations since the spirochete is the causative agent of Lyme disease: Borrelia burgdorferi are transmitted from animals to humans by the bite of a hard-bodied tick (i.e., species belonging to the Ixodes ricinus complex) (Anderson and Magnarelli 1994). Although ticks are not insects, most entomology texts, such as Borror et al. (1981), include descriptions of these common Arachnids. Samples consisted of 14 European, 14 North American and two Japanese B. burgdorferi isolates. Fifty-four RAPD markers generated from three primers (18, 20, and 30 nucleotides in length: GC content ranging from 39 to 50%) were used in phylogenetic analysis. Three main groups and two unique strains were identified. Group I encompassed two European and 13 North American isolates, group II consisted of nine European and 1 Japanese isolate and group III contained three European isolates. Hence, isolates from groups II and III were unique to Eurasia and it appeared that the North American strain originated from European isolates. Genetic distance were estimated to be 0.307 (between groups I and III), 0.387 (between groups I and II), and 0.268 (between groups II and III). These 62 distances indicated divergence near the level of distinct species (Welsh et al. 1992). Genetic distances within each group ranged from 0.043 to 0.050. Welsh et al. (1992) were confident that RAPD analysis delineated the isolates into correct groupings, since the same three groups were identified independently via protein electrophoresis, DNA-DNA hybridization and rRNA gene restriction patterns. The RAPD marker system is perhaps most robust for, and best utilized in, studies examining differentiation among conspecific populations (Kambhampati et al. 1992). Puterka et al. (1993) collected samples from 36 worldwide populations of the Russian wheat aphid. Aphids were obtained from wheat, barley, wheatgrass and oat and using both enzyme electrophoresis and the RAPD marker system examined genetic variation within this species. Sample sizes ranged from 20 to 28 individuals per population for the allozyme investigation. Average expected heterozygosity, based on 20 isozyme loci, was 4.9 per cent for the 31 world populations examined. Allelic frequencies of three polymorphic allozyme markers were used to calculate Nei's genetic distances among populations. Cluster analysis performed on these distances revealed five distinct groups (genotypes). Virtually no intrapopulation differentiation was revealed with RAPD markers (or for that matter with the allozyme markers) and hence banding patterns from only one individual per population were used for RAPD analysis. Sixty-nine polymorphic (interpopulation) RAPD bands were scored from seven 10-nucleotide RAPD primers (primarily 60-70% GC content). Genetic distances (1-M) from the RAPD data were derived from the fraction of bands matching (M) (presence and absence) between two aphids species. Cluster analysis utilizing RAPD genetic distances yielded similar results to those obtained from the allozyme analysis. Unlike the allozyme markers, however, all populations were differentiated using RAPD markers. Puterka et al. (1993) concluded that the RAPD system was superior to allozymes for revealing the true level of genetic variation in aphids (i.e., allozymes traditionally 63 underestimated genetic variation). No correlation was found between geographic distances and RAPD genetic distances. This led Puterka et al. (1993) to surmise that clones were probably established randomly through commerce rather than through migration. Dillwith and Soos (1994) have also used the RAPD marker system to examine intrapopulation genetic variation in aphids. Their report at the Annual Meeting of the Entomological Society of America in 1994, however, focused on an elaborate and sound scoring system which they developed for RAPD markers. Results pertaining to the genetic variation they observed among aphid populations have not yet been published. One of the best and most comprehensive RAPD marker studies to date was conducted by Ballinger-Crabtree et al. (1992). Eleven different geographic populations of mosquitoes, belonging to two subspecies of Aedes aegypti (seven populations from subspecies formosus, four populations from subspecies aegypti), were examined. Ten individuals (five males; five females) were sampled from each population. Forty RAPD primers, each 10-nucleotides in length, were initially screened. Nonparametric discriminant analysis identified 16 RAPD markers (produced from three of the primers; 70% GC content) as being informative for discriminating amongst the geographic populations. Individuals were identified 100% correctly between subspecies and 89% correctly among the geographic populations. Unknowns, included in the multivariate analysis, were classified correctly into subspecies but were not classified correctly into the geographic populations from which they arose. Genetic distances, calculated from both presence/absence data (two different measures tested: Similarity Index (SI) and percent match) and allele frequency data (Nei's unbiased genetic distance), were used for three separate cluster analyses. Dendrograms were created utilizing two different clustering methods: UPGMA (Unweighted Pair-Group Method Using Arithmetic Averages) and single-linkage. All three dendrograms indicated the same two major branches. These branches corresponded to the separation of the Ae. 64 aegypti aegypti and Ae. aegypti formosus subspecies populations. Unknown groups (also included in the analysis) were correctly identified according to subspecies by all clustering analyses performed. Correct designation of unknowns at the population level, however, only occurred when clustering via single-linkage was used on SI derived genetic distances. Ballinger-Crabtree etal. (1992) also demonstrated the heritability of RAPD markers in mosquitoes: RAPD profiles from a family of mosquitoes indicated that all fragments present in the Fl progeny (six males, four females) were observed in one or both of the parents. Edwards and Hoy (1994) also utilized RAPD data to determine allele frequencies in populations of Hawaiian Diglyphus begini (Hymenopteran parasitoid of leafminers). In the absence of chemical pesticides Edwards indicated that D. begini does an excellent job of controlling leafmining pests of tomato plants. It was Edwards and Hoy's (1994) contention that agricultural practices could lead to disruptions of D. begini populations in individual fields which could cause local instability. Thus, their primary objectives were to examine population structure and investigate gene flow in populations of this introduced parasite species. This information was deemed critical for making proper management decisions. Individuals were collected (from tomato plants) from four different sampling sites located on three of the Hawaiian Islands. Three collections were undertaken from site A and site B, one collection from site C, and five collections from site D. RAPD analysis was done using DNA extracted from 10 individuals (all males, thus haploid tissue) from each collection. Analysis was conducted on 23 RAPD bands generated from six different primers (40 primers screened). Allele frequencies were used to calculate Nei's unbiased genetic distances; subsequent cluster analysis (UPGMA) yielded a dendrogram which indicated groupings corresponding somewhat to geographic location. Multidimension scaling did not, however, reflect geographic distances. Hence, Edwards and Hoy (1994) concluded that population differentiation could not be attributed to geographic distance 65 alone. Results concerning gene flow were only briefly mentioned but were determined by calculating FS T. To my knowledge, Edwards and Hoy's (1994) findings have not yet been published. In a recent RAPD study conducted by Lu and Rank (1995) 131 polymorphic and 31 monomorphic bands were used to estimate population genetic parameters in the alfalfa leafcutting bee (Megachile rotundatd). Five geographically isolated populations ofM. rotundata were examined in the study. Four of these populations (coded a, m, c and r) were collected from sites in Saskatchewan. The fifth population (coded f) originated from 1000 cells imported from France in 1982 and subsequently raised in isolation in Saskatoon. DNA was extracted from five haploid male bees in each population; thus, total sample size was 25. Sixteen 10-nucleotide primers (50-80 % GC content) were used to generate the 162 RAPD bands. Estimates of population parameters indicated that population f had the highest average between-population heterozygosity, nucleotide divergence and Nei's genetic distance with the other four populations. Population f also showed the smallest within-population heterozygosity and nucleotide divergence. These results were not surprising given that the French population was known to have originated from a narrow genetic base. Interestingly, average within-population heterozygosity (0.3305) was 10-fold higher than that previously seen at the protein level (0.026). Mean nucleotide divergence (1.0 with a standard error approximately 0.2), however, was in the range formerly established using RFLP analysis of mt-DNA (0.26 - 1.01). Cluster analysis revealed that populations r and m were the most similar (Nei's distance = 0.02), with population c joining this group at a distance of 0.03. Clustering appeared to be highly correlated with geographic latitude. Lu and Rank concluded (1995) that RAPDs were valuable for estimating population genetic variation in haploid M. rotundata even given a small sample size of five 66 individuals. They suggested that the technique was also applicable to the evaluation of divergence in diploid populations if a sufficient (i.e., much larger than five) sample size was used. 2.2.3 Procedures for Analyzing RAPD Data 2.2.3.1 General Introduction RAPD data are visualized as the presence or absence of bands on electrophoretograms. Bands can be scored as discrete characters (binary coded; character state data) and the data analyzed as such. Analyses are also possible using various quantitative measures computed from the binary coded RAPD data. At present, there is no general consensus regarding statistical procedures for analyzing RAPD data even when the primary objective has been to determine genetic relationships among populations (at various taxonomic levels) of an organism (Table 3). Types of RAPD data analyses undertaken in entomological studies to date are summarized in Table 3. Analyses have included various multivariate techniques such as multifactorial scaling (Edwards and Hoy 1994) and nonparametric discriminant analysis (Neighbor) (Ballinger-Crabtree etal. 1992; Kambhampati etal. 1992). Estimated heterozygosity and nucleotide divergence (Lu and Rank 1995) are among the population genetic parameters computed from RAPD data. Wright's F-statistic (1951,1978) approach has also been utilized with the RAPD marker technique (Edwards and Hoy 1994). Cluster analysis has been perhaps the most widely applied procedure for analyzing RAPD data. In terms of the 15 different entomological studies summarized in Table 3, seven studies utilized the UPGMA clustering algorithm and one study used the Kitsch (+) clustering algorithm. Cluster analysis was done on a number of different similarity or distance measures. The similarity or distance measures were computed from either the number of matching bands (presence only or t-1 03 3 a. re VO VO Is o « f ° 3 2 re 3 •* a* £. o" 3 <B. 3 - . 3 . N CD O 3* 0> 3 o a. g: (4; o ^§ ft o •5 cn z O 03 O ^—' s ~ P . p3 > > Z o 3 CD PS I-3* w 3 63 o 3* 03 •o o o re vo vo CO 3 3 S 03 D. cn | J " I i S t .§ &. cn cr o* • a a m S S X n « ES (t (t 3 3 3 | . g. g <B 3* 2. 5' F> 03 5- » CD 3 3 CD CD CD h0 >fl W " 1 ^ 3 era ?0 Z o 3 CD > o ' %• 3 * M 3 ' i-i > a | CD B- £ o S » 3 Q3 CD Is 03 O re 03 a. 5 w era f? 2 <B"8 a. s. c. 5" 3 5 >-• an vo 5» vo 03 2. CD CD cr. cn o i 5 3 CD z O 03 03 w o" 3 & ° 03 3 f ? CD Cu P <; CD > » >< M l i e 2 o 2 <8 cr 3 S. CD e-f=r- 2 O 3 > C S CD I CO I a n a E l Q. g o 3 ! 3 ^  §. S a s CD 3 a l i i 5 <"> o ft S Q as cn 7 ^ 0Q o co CO CO < o. CD Cu cn 3 / cr. ET. era © 3 L9 o 3 2 >-h g Q, rt 5 rt rf rt 3 K 2 in 6 o a. s TO rt 3 rt 3 rt. B* (15 VO VO to T l rt 3. a VO vo t-1 o rt s & rt tn 3. rt 03 as r: cn rt s P I rt OQ rt 3 m X CO CO 3 0 13 ere 1 a &g-5. 3 rt sr 8"S' I < ia P. .—* o ia P. Not Si Not * s cn oTl CO •-•> rt rt o. 1 har o co o o rt •5 cn P rt a. rt 5 O l-H 5 11 3 3 rt rt cn O* 5" w x a* co o g CD 0s 3 §. £ £ J 8" g*' B rt o g > o H PS BT ST 5*" O o B 5" B rt CL rt a t/i I a it O 03 S i l l £ a g rt. o 55 rt B rt a l i i 1- UJ CO 2 ^  *" OO o cn £»< rt rt C a s. So? B > < O* rt P-B? TO g CO 5 oo l-H C/3 89 S 3 vo E2 03 s Cu co o o tf TO » CD ^ I % a* M c s 09 -< cn 5: sr 03 ft sr CO ET o -r re VO VO to T3 ' B O w m S ft cn a w 03 P rt CT S» o C o. °» m" P Cu cn M O 03 M> ES o. tf Z o Z o EJ rt S5 E) 03 ? I •*"" » >— 00 vo 2 vo 7 to n I tf CD tf 03 C/3 05 H> ?r cn 0Q •9 3 5. 03 ^ S3 & "8 3 B a 3- E. £T I I a l a 1 5?.<5' s» o sri ss « 5 a- 03 3 e 5- ° rt S3 ,— CA § c » cn S-o H 03 ft" w O o B 5" B rt rt' o & 3 S. g o> 05 t o P C m 3 GO a rt sr a & I "as. a a 69 3 05 VO VO c>,2 3 m gl§ B S « g 5' & 5" O ft —> i-l 3 IJ a 2 ft CO CO f-i "2. 5" O ft 2. co o: P rt a. p ft 3 01 01 -co <-»• £ 0> 05 5 3 3 g o o O o 8 S ! | §-1 —^ crq cr ft ft •g • l i s . s. I i o n " » S3 0 o tp o >-t> o >o £ . 3 £ B» w. & §» § cro ft p rt cr. o o 3 f? <8 o' 05 ft a. s ct- P* • j2. • ~- " 2 M a o rt cn 3. p oT o" < K 05 rt t i a ft § i f S - i | f . | I cr 05 o p CL re VO VO U) 1-1 cr §* ^ § f ,ft 13 '< 05 ST s cn I ? 0= ft D. 05 N SL w 3 o. ol ft 3 ft ffl I* 0) 3 3 3 3 t t . ft o. ft I o 3 0Q ft / f l ft ^ l -K CQ CA 9 cf. * 3 3 > I H ft" o o 8 B rt O. s ft o 5-ft CO c:2Q a l l CO 3. 3 OQ ft D. cn 3,' 05 3 OL CD T3 O o O T3 C o ft T3 3 Cu If i -S3" VO , vo OQ tf 2 a a & §* « 8 g © a a-2 " f t ii & & o 5i 3 Fe Cu o H w ft o 3 5" B fe s I c/3 a i r CA PS C73 III a e a. 3 OQ rt p. <B § p CA IZ, 72 presence and absence) or from allele frequencies (determined from each population under investigation). Phylogenetic analysis using parsimony (PAUP) was also employed in two of the interspecific entomological investigations using RAPD markers. In this study, R A P D data were considered as both discrete (binary coded) and continuous variables (allele frequencies). Analyses were conducted with the objective of examining population structure within P. strobi. Background information pertaining to the population genetic procedures and multivariate techniques used are briefly introduced below. 2.2.3.2 Population Genetics Techniques The determination of allele frequencies is fundamental to population genetics. When R A P D markers are generated from haploid tissue each of the two alleles assumed to be present at a single locus (particular size of D N A fragment) are directly observed. This makes the computation of allele frequencies straightforward; the frequency of the dominant allele (p) is the sum of the bands observed at one locus divided by the total number of individuals sampled in the population. The frequency of the recessive allele (q) can be computed by summing the number of individuals not exhibiting the marker (absent band) and dividing by the sample size. It can also be obtained by simply computing q = 1-p, since by definition the sum of the allele frequencies is always one (Li and Graur 1991). If RAPD markers are derived from diploid tissue allele frequencies are no longer directly observed and they must be estimated from q2. Lynch and Milligan (1994) formulated mathematical equations needed to estimate p and q from q2. Their method incorporated a factor to correct for small sample size since they suggested that estimating p and q directly from q 2 would yield a downwardly biased estimate of q. It has been suggested that anything less than 1000 individuals per population should be considered a small sample when 73 estimating allele frequencies from RAPD markers produced from diploid DNA. When 1000 individuals per population are used the probability of observing the rare recessive allele (i.e., q at a very low frequency) is greatly increased. A simple measure often used as a descriptor of genetic variation is the percentage of polymorphic loci. L i and Graur (1991) defined genetic polymorphism as "the coexistence of two or more alleles at a locus". When the frequency of the most abundant allele at a locus approaches 100% it is highly unlikely that any other allele will be observed in a sample unless the sample size is very large (Li and Graur 1991). For this reason most investigators set a cutoff point and consider loci to be polymorphic only when the frequency of the most common allele is less than 95 or 99% (exact limit specified by each author) (Wright 1978). Monomorphic loci are those in which virtually all individuals in a population exhibit the same allele at a locus (Li and Graur 1991). Heterozygosity is another simple descriptor which characterizes the extent of genetic variation. Since each heterozygote carries different alleles, heterozygosity directly reflects genetic variation (Li and Graur 1991). With co-dominant molecular markers heterozygosity in a population can be calculated either as the mean frequency of heterozygotes over all loci (observed heterozygosity! or as the mean frequency of heterozygotes expected in a population in Hardy-Weinberg equilibrium (expected heterozygosity or gene diversity) (Li and Graur 1991). Dominant molecular markers, such as those derived from the RAPD marker technique, restrict calculation to only expected heterozygosity since genotypes are not directly seen (i.e., the heterozygotes cannot be distinguished from the homozygote dominant individuals). Expected heterozygosity (h), under Hardy-Weinberg equilibrium, for any single locus and any single population is defined for a two allele system as h = 2q(l - q) (Lynch and Milligan 1994). Mean expected heterozygosity (H) is simply the average of the h values over all loci studied. Gene 74 diversity is equivalent to H and can be interpreted as the probability that two randomly chosen genes from a population are different. This measure is applicable to any haploid, diploid, or polyploid organism whether they exhibit random mating, selfing, or asexual reproduction (Nei 1987). In this study, mean expected heterozygosity and gene diversity have been used interchangeably. When more than one population is examined it is possible to partition total gene diversity into its within- and between-population components ( H w and H B respectively) (Lynch and Milligan 1994). Examining these components allows inferences to be made concerning population subdivisions (Lynch and Milligan 1994). Alternatively, Wright's F-statistics (1951,1978) may be used to measure the extent of population subdivision. Analyses that attempt to quantify population subdivision are based on the premise that the populations sampled are representative of the species and therefore share a common evolutionary history. The focus is on determining the extent to which different populations within the species have differentiated over the time since the ancestral population (Weir 1990). Population subdivision entails an inbreeding-like effect and results in a reduction in the number of heterozygotes and an increase in the number of homozygotes. Quantifying the reduction in heterozygosity is hence appropriate for examining population subdivision. To compute Wright's F-statistics (1951,1978) different levels of heterozygosity must first be calculated or estimated, such as heterozygosity pertaining to individuals (H:), populations (demes) (HD), subpopulations (H s) and the total population (HT). H can be viewed as the average heterozygosity of all the genes in an individual. In studies utilizing dominant molecular markers this measure cannot be computed since heterozygous genotypes are not observed. H D and H s represent the level of heterozygosity that would be found in a deme or subpopulation respectively if the deme or subpopulation were undergoing nonrandom mating. H T represents what the heterozygosity would be if all the populations were pooled together and 75 mated randomly (Haiti and Clark 1989). In the absence of population subdivision allele frequencies are identical in all populations, hence, H r = H D = H s = H T . If population subdivision has occurred, the Wahlund effect ensures that H T would always be greater than H s . Wright (1951,1978) used the principles discussed above to formulate his F-statistics. He interpreted his F-statistics as the correlation between specified classes of gametes, relative to a specified total population. Note the terminology used since pairs of genes may be related by being within individuals, between individuals within demes, between demes within subpopulations, etc. (Weir 1990) (the correlations which can be computed depend on the sampling design and the molecular marker used). The F-statistics considered in this study included F S T , F D T and F D S . F S T is an estimator which measures the amount of differentiation among subpopulations, relative to the limiting amount under complete fixation, F D T measures this among demes and F D S measures the amount of differentiation among demes within subpopulations. (Wright 1978). Allele frequencies are also used to calculate genetic distances. Genetic distances express the genetic difference between two populations as a single number (Weir 1990). Genetic distances (or their converse similarity) may be utilized in a number of different fashions: they may be computed simply as a data reduction procedure or as a means of comparing pairs of extant populations. They may also be employed for reconstructing the evolutionary history for a group of populations. Each one of these applications require different distance measures and require genetic models, particularly when inferring phylogenies (Weir 1990). Many different genetic distances have been formulated and utilized in molecular marker studies. Wright (1978) thoroughly evaluated five different genetic distances for their effectiveness in establishing the pattern of relationships among four human populations. He explored Prevosti's distance, Wright's modified Prevosti distance, Roger's distance and the arc and chordal distances proposed by Cavalli-Sforza and Edwards. Conspicuously absent from Wright's comparison was Nei's 76 genetic distance (1972), the most widely used measure to date for inferring phytogenies (Weir 1990). Nei's unbiased genetic identity (I) (1978) is a commonly used similarity measure. As Wright (1978) so elegantly stated, phylogenetic trees cannot be used to reconstruct the evolutionary history in the case of differentiation among populations, between which there is no reproductive isolation. In this case, it is the network of genetic distances that has real meaning and cluster analysis can be performed to indicate which groups are most similar. Wright (1978) concluded that the genetic distance measures most applicable for revealing groupings among intraspecific populations were those established by Prevosti (Wright 1978) and Cavalli-Sforza and Edwards (arc distance, but modified by Wright (1978)). The basic principle behind cluster analysis is to utilize a clustering method upon a matrix of resemblance coefficients (distances or similarities) (Romesburg 1984) to ultimately produce a diagram (tree or dendrogram) which shows at a glance the degrees of similarity between all pairs of objects (in our case OTUs or populations). One of the most popular clustering methods employed in population genetic analysis has been UPGMA (Sokal and Michener 1958). Avise (1994) described UPGMA as "a simple and appealing example of the phenetic approach to data summary". As evidenced in Table 3 most investigators performing cluster analysis upon genetic distance matrices used the UPGMA method. Other noteworthy clustering methods include the Single Linkage Clustering Method (SLINK) and the Complete Linkage Clustering Method (CLINK) as thoroughly explained by Romesburg (1984). Table 3 shows that PAUP has also been utilized with RAPD data. PAUP is a computer program which has often been employed to reconstruct evolutionary trees and hence infer phylogenies among interspecific populations. Maximum parsimony, the method utilized in PAUP, is an approach which attempts to mimic the principles followed by cladists (Avise 1994). 77 2.2.3.3 Multivariate Techniques Principal components analysis (PCA) is a multivariate procedure which is often employed in a descriptive fashion. It is a technique which may be used to summarize associations in a large set of observed variables (such as loci) and hence correlations amongst these variables can be examined (Tabachnik and Fidell 1983). The overall objective of P C A is to transform an original large set of variables into a smaller set of linear combinations (= Z; 's) (Dillon and Goldstein 1984) while maintaining as much of the original information (i.e., total variation) as possible. These linear combinations, also called principal components or P C A axes, are uncorrelated with each other. Furthermore, the P C A axes are extracted so that the first axis ( Z , ) displays the largest amount of variation, the second axis ( Z 2 ) displays the second largest amount of variation and so on (Manly 1992). Although it is possible to extract as many P C A axes as there are original variables often the general goal is to obtain a parsimonious description of the structure underlying a set of multivariate data. This is achieved when a few principal components account for most of the total variation displayed in the original data set. The amount of variation captured or explained by each axis is indicated in its eigenvalue. The eigenvalues for each axis are simply the diagonal elements in the sample covariance or correlation matrix used in the procedure. When different scales of measurement are used in collecting data on the variables it is customary to first standardize the variables to have zero means and unit variances. In this case computations are performed on the correlation matrix (Manly 1992). Eigenvectors provide the coefficients for the principle components; by examining these coefficients (also called loading scores) the investigator is able to examine associations amongst the original variables (Dillon and Goldstein 1984). Additionally, by obtaining two (or three) dimensional graphs (of individual scores plotted on Z i vs Z 2 ) overall structure in the data can be visualized. 78 Cladists have long advocated the use of PC A in taxonomy for finding structural relationships among specimens (individuals sampled) without a priori subdivision of the samples into discrete populations (Wiley 1981). More recently, in a workshop which addressed problems associated with analyzing data obtained from dominant RAPD markers, PCA was suggested as an appropriate technique for examining patterns of differentiation among (sub)-populations (Gilesey 1981). 1994). Principal component analysis can also be used as a prelude to other multivariate techniques such as multiple linear regression (MLR) or the general linear model (GLM) procedure (Dillon and Goldstein 1984). That is, it serves as a data reduction technique so that a few uncorrected variables (Zi's), free of multicollinearity, can be subsequently used as independent variables in MLR or other multivariate procedures. Discriminant analyses are multivariate procedures which do require {a priori) subdivision of the samples into discrete and identifiable populations (Eisenbeis and Avery 1972). These techniques have proven to be invaluable in systematics as Wiley (1981) and other cladists have indicated. Discriminant procedures have shown to be particularly useful in establishing the membership of an (unknown) individual collected from one of several distinguishable groups. In a strict sense, discriminant analysis does not classify but rather identifies an observation as belonging to a particular group (Reyment et al. 1984). These procedures have also proven their worth in delineating species boundaries (Wiley 1981) as Tizado and Nieto Nafria (1994) recently demonstrated. Results from canonical discriminant analysis conducted on 15 quantitative characters indicated that the group of aphids under study, (considered to represent a single species), actually belonged to two distinct species. Discriminant analysis techniques can be broadly grouped into two categories: parametric procedures, where the predictor variables are assumed to follow an approximate multivariate 79 normal distribution within each class, and nonparametric procedures, where no assumptions concerning the distribution of the predictor variables (within each class) have been made (Dillon and Goldstein 1984). Canonical discriminant analysis is a parametric procedure. In this technique, the interrelationships between a number of populations are simultaneously examined, with the end in view of representing these interrelationships graphically in perhaps two or three dimensions (Reyment et al. 1984). This technique involves computing new axes (= canonical discriminant functions or canonical variates or canonical components) which are linear combinations of the original variables (in my case loci). Maximum separation of the groups is obtained when the F ratio of between-groups to within-groups variation is as large as possible (Manly 1992). Successive discriminant functions are generated under the constraint that there is no correlation between each new axis and any previous axis within groups (i.e., the residual between-groups to within-groups variability maximized in successive axes (Dillon and Goldstein 1984)). The number of canonical axes possible for any one analysis is one less than the number of a priori groups designated or number of variables used, which ever is less (Wiley 1981). Nearest Neighbor analysis is a nonparametric discriminant procedure for classifying observations into one of several classes. This technique identifies observations as belonging to a particular group by computing distances (Mahalanobis) to the 'k-number' of nearest neighbors. The number of neighbors (k) to be examined is an option in the procedure, specified by the individual conducting the analysis. Placement of individuals into a specific group depends on both the surrounding neighbors and on prior probabilities specified for each group. Prior probabilities (the odds of an observation belonging to a particular group) are generally set at being equal unless the sample size for each group is an estimate of the chance of being in that group. Table 3 gives examples where nearest neighbor analysis has been conducted on RAPD data in previous 80 entomological investigations. Khamphampati et al. (1992) successfully utilized the procedure with presence/absence RAPD data to examine differentiation among five species of mosquitoes and among four conspecific populations of one of these species (Aedes albopictus). Nearest neighbor analysis was used by Ballinger-Crabtree et al. (1992) to identify a subset of RAPD fragments useful for discriminating among 11 geographic populations of Aedes aegypti. 2.2.4. Addressing the Dominance Problem of RAPD Markers The inherent problem of dominance (inability to distinguish heterozygotes from dominant homozygotes) in the RAPD system does not allow the direct measurement of allele frequencies and actual heterozygosities when performed with D N A extracted from diploid tissue. Genetic diversity statistics (based on allele frequencies) can be estimated from information based on R A P D markers generated .with diploid tissue (as previously discussed), but not without addressing various concerns (Lynch and Milligan 1994). Of particular note, is the lower level of both accuracy and precision in the estimation of F S T (Lynch and Milligan 1994). Conifers provide a simple solution to the measurement of allele frequencies by supplying haploid tissue in the megagametophyte of the seeds (Bartels 1971). Animal taxa which base sex determination on the X Y chromosome system (presence or absence of the Y chromosome), also provide an avenue for directly observing allele frequency. Although gametes are not always easily obtained from many animal taxa, if genetic markers specific to the Y chromosome could be identified in diploid tissue, a direct measurement of allele frequencies would be obtained. This assumes that only two alleles are present at any one loci (Lynch and Milligan 1994). P. strobi is representative of an insect in which male sex determination results from the presence of the Y chromosome (Smith and Takenouchi 1969). The objective of the study 81 outlined in chapter three was to identify male specific RAPD markers in P. strobi and thus overcome one of the major weakness of the RAPD marker technique (i.e., allele frequencies could be observed, not estimated, for subsequent use in the investigation of genetic variation within P. strobi). To aid in this endeavor, the bulked DNA technique, which has shown to detect markers in specific genomic regions (Michelmore et al. 1991), was explored and will be discussed in chapter three. 82 CHAPTER 3: THE SEARCH FOR MALE SPECIFIC RAPD MARKERS TN Pissodes strobi 3.1 METHODS 3.1.1 Sample Design and Collection of Weevils White pine weevil infested leaders were obtained from 10 spruce plantations in B.C. Clipping of leaders was done from mid-July to early-August, 1992. Sample sites ranged from Northern B.C. (Kitimat) to Vancouver Island. Sample locations included five Sitka spruce (three from Vancouver Island, two from mainland B.C.), three white spruce and two Engelmann spruce plantations. Figure 2 indicates the general locations in B.C. from which weevil infested leaders were obtained. As indicated in Table 4, two separate samples were acquired from the plantation near Kitimat, B.C. (extreme Northwestern site). These samples were treated as separate populations. Hence, 11 white pine weevil populations were acquired from spruce hosts in B.C. The assigned population names shown in Table 4 will be used hereafter (either full name or shortened prior to the underscore). In general, the following criteria for selecting and sampling within each B.C. plantation was used. (1) The primary component of each stand was to be Picea sp. (2) If two plantations were to be sampled from one area, the sites were to be at least 50 km apart. (3) Current (1992) weevil attack intensity was to be moderate to high (at least 5-15% of each stand was to have shown 1992 weevil attack). (4) A minimum of 40 weevil infested leaders were to be cut. (5) Only one weevil attacked leader was to be clipped from each selected tree. (6) The leaders were to be collected evenly throughout the stand (i.e. not obtained from one small portion of the plantation). In nine of the 10 B.C. sampling sites, the minimum of 40 weevil infested leaders was 83 TAIGA PLAINS NORTHERN BOREAL MOUNTAINS BOREAL PLAINS • = Sitka spruce Van. Is. • = Sitka spruce Mainland = White spruce • = Engelmann spruce • = Jack pine (Ontario) SUB-BOREAL INTERIOR SOUTH CENTRAL INTERIOR INTERIOR MOUNTAIN SOUTH . INTERIOR Figure 2. Location of sites in B.C. (1992) from which white pine weevil infested leaders were obtained. 84 Table 4. Populations of Pissodes strobi sampled in RAPD marker study. City or Town Close to Sample Site3 Assigned Name Tree Host Personnel Leaders Obtained Fromb Gold River S V AN 1 GOLDRIVER Ss P.F.P. Gold River SVAN2NGOLDRIVER Ss FIDS Tofino SVAN3TOFINO Ss FIDS Kitimat-1 SMIKITIMATI Ss MoF Kitimat-2 SMIIKITIMATII Ss MoF Vancouver SMIII_RESFOR Ss K.Lewis McBride Wl_MCBRIDE-26 Sw MoF McBride W2_MCBRTDE-4&5 Sw MoF Prince George W3PRLNCEGEORGE Sw FIDS Golden E1GOLDEN-2 Se MoF Golden E2GOLDEN-119 Se MoF Gogama -1991 Pl_GOGAMAl JP E. Tomlin Gogama-1992 P2GOGAMA2 JP Applejohn a: First 10 locations sites in B.C., last two located in Ontario b: P.F.P. = Pacific Forest Products; FIDS = Forest Insect and Disease Survey; E.Tomlin of Simon Fraser University; Applejohn associated with Forestry Canada 85 achieved, the exception being a white spruce plantation wherein only 20 infested leaders were clipped. This area was not originally targeted as a sample site, but a sample was supplied and therefore utilized. Upon arrival at U.B.C., each leader was trimmed of lateral branchlets, placed in an individual rearing tube and maintained at room temperature. The 0.6 m long, 5 cm diameter cardboard tubes were sealed by taping thin, expandable, cardboard caps to each end. A 4 mL, glass insect collecting vial was attached to one of the caps by cutting a circular hole (smaller than diameter of vial cap) in the cardboard and "punching" the vial through. Glass vials were checked daily for emerged weevils and emerged parasites. Additionally, all rearing tubes were opened weekly and weevils which remained inside the tube collected. Collection of reared weevils (and parasites) occurred from late-July to mid-September. Following collection, individual weevils were placed in labeled, 1.5 mL Eppendorf (Epp) tubes. Epp tubes were placed at 4° C for one to three nights (to allow weevils to empty gut contents). Thereafter, the gender of each live weevil was determined by the method of Harman and Kulman (1966). Individual weevils were transferred to sterile, labeled, Kontes 1.5 mL microtubes (transfer done in a Laminar flowhood to avoid aerosol contamination). Tubes were immersed in liquid nitrogen (for quick freezing of weevils) and stored at - 80° C until needed for DNA extraction. In addition to obtaining P. strobi from 11 spruce tree populations, weevils were also collected from jack pine. Weevils from P2_GOGAMA2 (Table 4) were obtained from 30 infested leaders clipped in early August, 1992, from a jack pine stand near Gogama, Ontario (due North of Sudbury). Pissodes strobi in these leaders were reared and treated as above. Weevils from P1_G0GAMA1 (Table 4) had been collected as live adults, from the same stand near Gogama, in May, 1992 (weevils, therefore, from the 1991, or previous, generation). These insects had been 86 used in a study at Simon Fraser University, B.C. and supplied to us as live weevils (grouped in one jar). The eastern samples were treated as separate populations since they represented sampling from two consecutive years. 3.1.2. Extraction of Total Genomic DNA Prior to sample collection in the summer of 1992, a colony of P. strobi was maintained at the University of British Columbia (UBC) under the supervision of Dr. A. Wardle. Weevils in this colony had been collected in 1990 and 1991 from Sitka spruce trees, either from the U B C Malcolm Knapp Research Forest, Maple Ridge, B.C., or from sites on Vancouver Island. Insect specimens from this colony were initially used to perfect the bulk D N A extraction technique (addressed below). Furthermore, aliquots from these bulk D N A extractions were used in RAPD assays to screen for primers which produced markers putatively specific to the Y chromosome. 3.1.2.1 Bulked DNA Extracted From 20 Individuals Bulk D N A extractions were done using both 20 individuals (bulk-20) and 100 individuals (bulk-100) of each gender. To prevent contamination with aerosols possibly containing RAPD products (presumed to be present in the general laboratory area), all extractions were done in a Laminar Flowhood. The following protocol was followed for extraction of D N A from 20 bulked individuals. This method was done separately for each gender. Twenty live P. strobi were placed in a ceramic mortar, quick frozen in liquid nitrogen and ground to a fine powder with a ceramic pestle (ensuring tissue remained frozen throughout). The homogenate was transferred to a sterile, 30 mL corex tube and gently suspended in 4 mL buffer 87 (0.2 M Tris/ 0.04M EDTA/ 2.8 M NaCl, pH = 8.3) containing 1% (w:v) dissolved polyvinylpyrrolidone (PVP) and 0.2% 13-Mercaptoethanol (v:v). The tube was spun at 10,000 rpm (JA-20 rotor) for five minutes and the yellow, merky supernatant discarded. The pellet was resuspended in 8 mL of 2X CTAB buffer (IX CTAB = 0.05 M Tris/ 0.01 M EDTA/ 0.7 M NaCl, pH = 8.3/ 0.1% B-Mercaptoethanol (v:v)/ 0.025 M CTAB (cetyltrimethyl ammonium bromide)) containing 1% polyvinylpolypyrrolidone (PVPP) and placed at 65° C for one hour. During this period, the tube was gently inverted every 15 minutes to aid in the release of DNA (trapped in solid matter). Following incubation, the tube was spun at high speed (12,000 rpm - JA-20 rotor)) for 10 minutes and the clear supernatant (approximately 7 mL) decanted and saved. The pellet was resuspended in an additional 4 mL of 2X CTAB buffer, incubated an additional 10 minutes at 65° C and centrifuged at 12,000 rpm (JA-20 rotor) for 10 minutes. The supernatant (approximately 4 mL) was decanted and added to the first. Ribonuclease (sigma RNase-A, boiled to inactivate deoxyribonuclease) (5u.g/insect) was added to the supernatant and the tube placed at 37° C for one half hour. Total genomic DNA was extracted with an equal volume of chloroform:isoamyl alcohol (24:1). DNA was precipitated with ice-cold isopropanol (equal volume) followed by cold incubation at -20° C. DNA was collected after a 30 minute high speed spin (12,000 rpm - JA-20 rotor). The DNA pellet was washed twice with 4 mL of 70% ethanol (EtOH), air dried in a Laminar flowhood and dissolved in 2 mL TE (10 mM Tris/1 mM EDTA, pH = 8.0). To ensure DNA resuspension, tubes were placed at 65° C for one hour followed by placement at room temperature overnight (tubes covered with tinfoil). Thereafter, extracts were stored at either -20° C or 4° C (small aliquots of stock solution). This procedure followed that of Boyce et al. (1989) with the exception of the PVP, PVPP and RNase-A treatments which were introduced by J. Glaubitz (Forest Biotechnology laboratory at UBC, supervised by Dr. J. Carlson). 88 3.1.2.2 Bulked DNA Extracted From 100 Individuals Bulk D N A extraction from 100 individuals followed the same procedure as outlined above (solution amounts increased 2.5 X relative to 20 weevils). Extractions were done after grinding 50 weevils at one time (repeated twice for each gender). After total genomic D N A was extracted and dissolved in TE, extracts were pooled and thoroughly mixed. 3.1.2.3 Bulked DNA Generated From 12 Individuals Towards the latter part of this study, bulked D N A was generated by pooling equal aliquots (same concentrations) from 12 individual (bulk-12) D N A extractions (done for each gender). Individual weevils had been collected from leaders clipped in 1992 at the Malcolm Knapp Research Forest (Sitka spruce). As with previously mentioned D N A extractions, isolation of total genomic D N A from single weevils was adapted from the methodology of Boyce et al. (1989). Grinding of individual weevils was done in Kontes 1.5 mL, sterile microtubes, with tubes either submerged in liquid nitrogen or placed in dry ice. A sterile Reusable Pellet Pestle Mixer R , designed to fit the Kontes tubes, was used for crushing. Prior to use, pestles had been treated with a 10% solution of sodium hypochlorite (bleach) to ensure decontamination of unwanted D N A (Prince and Andrus 1992) Crushing of the tissue was not as extensive (exoskeleton was not thoroughly ground) as in the bulk D N A extractions, therefore, neither PVP or PVPP were needed (both serve to bind and hence eliminate phenolic compounds). As in the bulk D N A extractions, 5 ug of RNase-A was used for each insect. Solution amounts differed from the bulk D N A extraction as follows: 1) 89 homogenate was initially suspended in 500 uL of 2X CTAB buffer; 2) the pellet was resuspended in 300 uuL of fresh 2X CTAB buffer following the one hour incubation at 65° C and; 3) the D N A pellet was washed twice with 200 uL of 70% EtOH, dried and resuspended in 100 uL of TE. 3.1.2.4 Quantification of DNA Quality and quantity of D N A extracts were initially determined by absorbance at 260 -nm and 280 t|m. D N A dilutions based on absorbance at 260 r|m were used for all R A P D assays described in Chapter Three. For use in the dot-blot procedure (section 3.1.5), D N A was quantified by comparing staining intensities of D N A samples to lambda D N A standards run simultaneously on a 1% agarose gel. To visualize D N A fragments, gels were stained in a 1 ug/mL solution of ethidium bromide for 30 minutes, thoroughly rinsed in 2 liters of distilled water and photographed on a U V transilluminator. This method allowed for examination of both general fragment size and the presence of R N A in each D N A extract. 3.1.3 Optimization of RAPD Assay Conditions 3.1.3.1 Amplification of RAPD Markers To optimize the RAPD assay conditions for weevil template DNA, various concentrations of DNA, magnesium and enzyme were investigated. Total reaction volume was also varied, as was amount of mineral oil overlay. Optimum amplification (and most cost effective) of RAPD markers was achieved in a total reaction volume of 12.5 u.L, containing 0.3125 Units (U) Taq 90 polymerase enzyme (Perkin-Elmer Canada), I X Perkin-Elmer Reaction Buffer II, 25 rig (based on spectrophotometer readings) weevil template DNA, 200 u M of each dNTP (Perkin-Elmer Canada), 0.3 u M oligonucleotide (10 base) primer and 2.5 mM M g C l 2 . The reaction, containing only the primer and template DNA, was overlaid with 20 uL mineral oil and subjected to an initial denaturation step at 94° C for seven minutes. After addition of the premix (mixture excluding primer and template DNA) samples were amplified in a Perkin-Elmer Cetus D N A thermal cycler model 480. A denaturation step at 94° C for two minutes, was followed by 45 cycles of: 94° C for two minutes, 36° C for one minute and 72° C for two minutes. Final extension was done for 10 minutes at 72° C, followed by a 4° C soak until recovery. 3.1.3.2 Resolution of RAPD Products Optimum resolution of amplification products was achieved by loading the entire sample onto a 1% Synergel™/ 2% agarose gel. A 0.5X solution of phosphate-buffered Tris-EDTA (0.5X solution: 0.04 M Tris base, 0.004 M EDTA, pH adjusted to 8.0 with phosphoric acid) was used in forming the gel and as a running buffer in electrophoresis. Molecular D N A markers (100 base pair ladder) were used in the extreme left and right hand lanes (upper and lower rank of each gel). Loading into adjacent lanes, on each gel, was done using the following pattern: products generated from bulked female D N A ran next to products generated from bulked male D N A which in turn ran next to negative controls (layout repeated twice for each primer). This design allowed for quick recognition of both putative male specific bands and spurious amplification products. The reproducibility of each primer specific reaction was also easily verified using this loading pattern. Once loaded, samples were rapidly moved from the well by applying high voltage (at 91 least 0.26 volts (V)/cm2) for 10 minutes. Gels underwent electrophoresis, on average, for five hours at 100 V (0.20 V/cm2). Staining of gels was done in a dilute (0.5 ug/mL) solution of ethidium bromide for 30 minutes. Gels were destained for 20-30 minutes in distilled water and photographed using Polaroid type 67 film. 3.1.4 Screening of Primers to Identify Putative Y-Specific Markers Screening of primers to identify Y-specific markers was initially performed using bulked D N A obtained from 100 weevils (each gender). A negative control, reaction containing each ingredient except the template weevil DNA, was included for each primer. Positive controls (primers known to amplify weevil bulk DNA) were used in each RAPD assay (48 reactions prepared for each assay). Each primer was employed twice in negative controls and twice in reactions containing bulked male D N A and bulked female DNA. A total of 70 different primers were screened using bulked D N A extracted from 100 individuals. RAPD amplification assays were repeated with 45 of these primers using template D N A which had been obtained from 20 bulked weevils. An additional 59 primers were tested using aliquots of the latter D N A extraction. Eight primers (219, 291, 365, 374, 376, 383, 386 and 387), which appeared to target regions specific to the Y chromosome in the initial screening trial, were used in three separate RAPD assays. These trials employed bulked D N A extracted from 100 individuals, 20 individuals and bulked D N A generated by pooling extracts from 12 individual extractions. Three primers (219, 374, and 376) which produced the most distinct (clear, sharp band on electrophoretogram) putative Y-specific markers were used in amplification reactions with template D N A extracted from 12 female and 12 male individual weevils (i.e., not bulk DNA). 1 92 Weevils used for individual DNA extractions were collected in 1992 from infested leaders clipped at the Malcolm Knapp Research Forest. 3.1.5 Verification of Y Specificity via the Dot-Blot Procedure 3.1.5.1 Preparation of Dotted Membranes Total genomic DNA was extracted from 96, individually frozen weevils (collected summer of 1992), for use in the dot-blot procedure. Extractions were done on 48 individuals of each gender. Twenty-four weevils from each of the following four populations were selected: SMIIIRESFOR (UBC) and S VAN 1 GOLDRIVER (both Sitka spruce sites), W2MCBRIDE-4&5 (white spruce) and E2GOLDEN-119 (Engelmann spruce). With the exception of the UBC samples, one male and one female weevil (theoretically full-sibs) were chosen from 12 different leaders in each population. As previously indicated, DNA was quantified via comparison to Lambda DNA standards run on a 1% agarose gel. Ninety-one (46 males, 45 females) of the 96 extractions yielded a minimum of 200 rig of DNA needed to prepare four membranes. Four separate dot blots were prepared using Hybond™ - N + (nylon) membranes. A separate membrane was prepared for each of four different hybridization probes. Probes were generated from DNA fragments of the four most distinct, putative Y-specific RAPD markers. Preparation of probes will be outlined in detail following explanation of the entire dot-blot procedure. Nylon membranes were cut 9 cm by 12 cm, to fit into the Bio-Dot SF microfiltration apparatus (fit with BIO-Dot module). The critical parameters used in preparing each dot blot are described below. The nylon membrane was thoroughly wetted in distilled water for a minimum of 20 minutes. Prior to placement in the Bio-Dot apparatus, excess water was removed from the 93 membrane by blotting between two pieces of filter paper. Five hundred uL of distilled water was placed into each of the 96 wells and drawn to the membrane by applying a gentle vacuum. This step served to rehydrate the nylon membrane. Ninety-one DNA samples, containing either 50 r\g (14 of the 91 samples) or 100 r\g of weevil DNA, had been denatured by suspending DNA in a 0.4 M NaOH/10 mM EDTA solution and placing tubes at 100° C for 10 minutes. Following heat treatment, samples were stored on ice until loading (sample loading volume was 300 uL). On each membrane, a positive control (2.5 r|g of denatured probe) and a negative control (100 r|g of denatured herring sperm DNA) was included. Wells containing neither DNA samples, nor controls were filled with 300 uL of distilled water (3 of the 96 wells). DNA extracts from male weevils were loaded in wells on the upper half of the membrane and DNA extracts from female DNA in the lower half of the wells. Following loading of DNA samples, 500 uL of 0.4 M NaOH was added to each well, and drawn onto the membrane by applying a full vacuum. The Bio-Dot SF apparatus was then disassembled and the blotted membrane removed. The membrane was rinsed in 200 mL of 2X SSC (IX SSC =150 mM NaCl, 15 mM trisodium citrate) air dried overnight and vacuum baked at 80° C for approximately 30 minutes. 3.1.5.2 Pre-Hybridization, Labeling Reaction and Hybridization Protocol Pre-hybridization was done at 65° C, for a minimum of three hours, in a hybridization oven equipped with a VS-250 rotor. For each membrane, 12.5 mL of pre-hybridization solution, containing 6X SSC, 0.5% (w:v) SDS (sodium dodecyl sulfate), 5X Denhardt's solution (IX = 0.02% bovine serum albumin,.0.02% P V P , 0.02% ficol) and 100 ug/mL denatured herring sperm 94 D N A was used. Herring sperm D N A had been denatured by heat (100° C for 10 minutes). Membranes were placed, bound D N A side up, in 38 x 300 mm (with screw cap) thick walled, borosilicate glass tubes. Following addition of pre-hybridization solution, air bubbles between the membrane and glass surface were removed with a 2 mL sterile pipette. Each of the four probes was labeled with P 3 2-dCTP using ingredients from a Gibco BRL™-Random Primer D N A labeling kit. To denature probe DNA, a sterile 1.5 mL Epp tube containing 50 r|g of probe D N A (diluted to 23 uL with distilled water) was placed at 100° C for five minutes and then placed on ice. Two uL of each dATP, dGTP, and dTTP solution were added to the denatured probe DNA, as was 15 uL of Random Primer Buffer Mixture. Following addition of five uL of P 3 2-dCTP (approximately 50 uCi) and 1 uL of Klenow fragment, the reaction was gently but thoroughly mixed, the tube briefly centrifuged and then placed in a 25° C water bath for a minimum of two hours. After incubation, five uL of Stop Buffer, 10 uL of tracking dye and 35 uL of TE was added to the labeling reaction. Unincorporated nucleotides were removed by running the reaction mix through a sephadex G-50 column. The column had been prepared by placing the outer portion (adapter) of a 1 mL syringe in a 15 mL plastic tube. The bottom of the adapter had been plugged with sterile steel wool. A Pasteur pipette was used to fill the adapter with sephadex G-50 solution, ensuring that no air bubbles were trapped in the column. The column was spun in a megafuge (2150 rotor) for four minutes at approximately 14,500 rpm, topped up with sephadex G-50 and respun. Two washes, using 100 uL of TE each, followed. A one uL aliquot of the labeling reaction was removed (after the mixture had run through the column) for quantification of incorporated P 3 2 -dCTP. The labeled probe was denatured for five minutes at 100° C, placed on ice for five minutes 95 and then carefully added to the pre-hybridization solution (solutions thoroughly mixed prior to touching membrane). Hybridization proceeded overnight at 65° C. Unhybridized and poorly bound probe was removed from each membrane with a series of five wash steps. Two washes, each using 2X SSC were first performed. Next, a 0.2X SSC/ 0.5% SDS solution was used. Finally, two washes with 0. IX SSC/0.1% SDS, were done. The first two washes were done for 15 minutes and the latter three for 30 minutes. All washes used 250 mL of solution and were done at 65° C in the hybridization oven. 3.1.5.3 Preparation of Putative Male Specific Markers for Use as Probes DNA fragments, obtained from the four most distinct putative Y-specific bands, were used to prepare the hybridization probes. Primary RAPD amplification was first done on template bulked DNA (male and female) with primers 219, 374 and 376, to generate the four putative Y-specific markers (one DNA marker each from primers 219 (990 bp) and 374 (590 bp) and two markers from primer 376 (650 bp and 765 bp)). Duplicates of each reaction were done. Products were resolved and visualized as previously discussed (2% agarose/1% Synergel™, ethidium bromide staining). The putative Y-specific bands were cut from the gel, under UV transillumination, using new razor blades to excise each gel band. Extreme care was taken to ensure each slice contained only the band of interest (gloves and blades were frequently changed as well). DNA was recovered from each agarose slice by centrifugation through blotting paper. This method involved fitting a 1.5 mL Epp tube with a small, cup-shaped piece of blotting paper into which the agarose slice was placed. The tube had been punctured at the bottom and was mounted into a second (not punctured) tube. A one minute high speed spin was used to drive the 96 electrophoresis buffer and D N A fragment into the intact tube. The dehydrated agarose remained in the blotting paper. The retrieved buffer/DNA solution was diluted with 600 uL of TE. To generate 50 r|g of D N A (of each putative Y-specific marker) required for use as hybridization probes, secondary amplifications were done. Five uL of the retrieved DNA/buffer solution was used as the source of template D N A in the secondary amplifications. For each D N A marker, secondary amplifications of both 20 and 25 cycles (separate reactions) were investigated. Aside from total reaction volume in the secondary amplifications being 25 uL, concentrations of each ingredient in the R A P D assay were as previously described. A five uL aliquot of each final product was run on a horizontal, 1% agarose gel also loaded with Lambda D N A standards and D N A markers (100 bp ladder). This step was performed to quantify the D N A produced in the secondary amplifications and to ensure that only the D N A marker of interest had been amplified. Difficulties were encountered in isolating and reamplifying the 765 bp D N A fragment associated with primer 376. Neither secondary amplification (20 or 25 cycles) yielded 50 r\g of the pure fragment required for use as a probe. Therefore, the reaction subjected to 25 cycles underwent an additional 10 cycles in the thermocycler (after addition of fresh Taq). A pure product of sufficient yield was not evident when a five ul aliquot of the final product was visualized on an ethidium bromide stained, 2% agarose gel. Initial amplification, resolution, retrieval and reamplification were repeated using primer 376. Sufficient yield of the 765 bp D N A marker was still not achieved. Finally, primary amplification was done on seven replicated reactions (primer 376, bulked DNA), entire products run on a 1% Synergel™/2% agarose gel and the 765 bp fragment retrieved from each successful reaction. Following retrieval through blotting 97 paper, liquid from the agarose slices was pooled and evaporated to 23 uL in a speed vacuum. The entire 23 uL was then used as the source of template D N A in the labeling reaction. Appendix III outlines the rationale for each step undertaken in the search for putative male specific RAPD markers. Appendices IV and V illustrate the procedures described above in the isolation and generation of the hybridization probes. 3.2 R E S U L T S AND DISCUSSION Weevils which emerge from different leaders (in any one site) should, at most, be half-sibs. This assumes that each female only oviposits in one leader. If oviposition does continue in additional leaders, it seems plausible that the female would select leaders located in proximity to each other. They could, in fact, be on the same tree if the tree had been previously attacked by weevils. This hypothesis is supported by the contention that P. strobi are poor flyers (Cozens 1983). To avoid the probability of clipping leaders which could have been oviposited by the same female (hence double sampling of families) only one infested leader was taken from each tree. Additionally, samples were collected evenly throughout the stand. A summary of the weevils reared, sexed and frozen from each sampling location is given in Table 5. Leaders collected from the Malcolm Knapp Research Forest showed very poor adult emergence; weevils were collected from only 12 of the 40 leaders clipped. Although the W3_PRINCEGEORGE sample had weevils emerge from only 19 leaders, only 20 leaders were originally obtained from this plantation (Table 5). With the exception of the W3_PPJNCEGEORGE and SMIII_RESFOR, adult weevils were collected and subsequently frozen from a minimum of 26 leaders and a maximum of 61 leaders per site (B.C. sites, Table 5). Based on sample size alone, it appears that sufficient weevil families have been gathered (with the 98 Table 5. Summary of Pissodes strobi collected, sexed and frozen in 1992 for use in examining genetic variation with RAPD markers. Sample Number Leaders Number Number Source3 Leaders Minimum Males Females Clipped 1 Weevil Frozen Frozen Emerged SVAN1 42 31 97 69 SVAN2 40 26 85 61 SVAN3 41 38 93 84 SMI 40 28 77 83 SMII 40 33 109 59 SMIII 41 12 30 33 Wl 40 30 52 54 W2 40 28 65 52 W3 20 19 48 38 El 50 30 37 33 E2 55 36 100 71 PI NA b NA 100 100 P2 30 30 92 80 a: Refer to Table 4 for total assigned population names, b: NA= Not Applicable, since live adults obtained. 99 exception of SMIII_RESFOR) so that genetic variation estimates which are statistically sound can be computed following collection of data. Figure 3 shows extracts from 10 different D N A extractions, diluted to 10 r)g/uL (based on absorbance at 260 r|m) and run on a 1% agarose gel. Absorbance at 260 -nm overestimated D N A yield, relative to the Lambda D N A standards, by at least four to eight times (DNA was not even visualized in at least three of the lanes loaded with sample DNA). Impurities in the D N A extracts, also absorbing at 260 rjm, could account for both the inconsistency in the dilutions (some extracts were less pure than others) and the general over estimation of yield. If indeed foreign substances were present in the D N A extracts, they did not, in general, appear to inhibit RAPD amplification. Various problems in standard PCR reactions attributed to contamination of the D N A with PCR inhibitors, have been shown to be eliminated when dilute (1 r|g/uL) samples of D N A were used in each reaction (Sommer et al. 1992). It is suggested that the concentration range of D N A used in R A P D amplification in this study was 0.20 - 0.50 r)g/uL, as quantified by comparison to Lambda D N A standards, and not 2 r|g/uL as indicated by absorbance at 260 -nm. For consistency, all future D N A quantification was done by comparing to Lambda D N A standards and RAPD amplification done with 0.50 r\g/\xL. R A P D amplification products using primers 314 to 320 inclusive are seen in Figure 4. Reactions were done in a total volume of 25 uL with 0.05 U Taq/[iL and 1.9 mM M g 2 + . In an attempt to reduce costs, a subsequent trial using half this concentration of Taq (0.025 U/uL) was done with primers 314, 315 and 319 (Figure 5). To optimize the RAPD amplification technique at this lower concentration of Taq, five different concentrations of M g 2 + (1.0 mM, 1.5 mM, 1.9 mM. 2.5 mM and 3.0 mM) were tested. Figure 5 clearly indicates that M g 2 + concentrations of 100 1 2 3 4 5 6 7 8 9 10 11 12 13 14 \ Figure 3. Example of inconsistent results obtained when DNA was quantified by absorbance at 260 r\m. Lambda D N A standards are shown in the first four lanes (20 r)g/4 uL, 10 rig/4 uL, 5 r|g/4 uL, and 2.5 r|g/4 uL), followed by 10 samples obtained from different D N A extractions (all diluted to 10 r\g/4 uL based on absorbance at 260 n m ) 101 L F M C F M C F M C F M C F M C F M C F M C F M C L 314 315 316 317 L F M C F M C F M C F M C F M C F M C F M C F M C L Figure 4. Banding profiles produced in the first trial using primers 314 to 320 inclusive. Each primer (indicated below profiles) was used twice with bulked female D N A (F) and bulked male D N A (M) (DNA extracted from 20 weevils), as well as in two negative controls (C). Reactions with primer 304 were used as positive controls (previously shown to produce multiple markers). Lambda D N A Hindll l markers (L) are shown in the extreme right and left hand lanes (each rank). 1.9 mM M g 2 + and 0.05 U Taq/]iL was used in a total reaction volume of 25 uL. 102 Figure 5. Test of Taq at half the concentration (0.025 U/uJL) previously used. Three primers (314, 315 and 319) were tested at five different M g 2 + concentrations (1.0 mM. 1.5 mM, 1.9 mM, 2.5 mM and 3.0 mM). Duplicates of each reaction (ran in adjacent lanes) as well as negative controls (every third lane) were done. Bulked male D N A from 20 weevils was used in each reaction (except negative controls). 103 1.9 mM or less, do not yield 100% reproducible, distinct banding patterns in reactions containing 0.025 U Taql\xL. Al l three primers did however, work well at M g 2 + concentrations of 2.5 mM. The number of markers seen in the 2.5 mM reactions in Figure 5 is only slightly reduced relative to the corresponding reactions seen in Figure 4. Of particular note, is the reduction in the number of bands seen in the negative controls with primer 319 and the slight reduction in background noise (caused by spurious amplification products) (Figure 5). In traditional PCR reactions, decreasing the amount of enzyme in each reaction has been shown to increase specificity while lowering reaction costs (Sommer et al. 1992). The evidence presented in Figure 5 suggests that this trend appears to hold true for RAPD amplification as well. Primer 319 produced amplification products in three out of four negative controls seen in Figure 5 and both negative controls seen in Figure 4. It is highly unlikely that the bands in the negative controls are a result of exogenous D N A being introduced by the experimenter, since a strict set of procedures (Kwok and Higuchi 1989) for avoiding D N A contamination were rigorously employed throughout this study. Amplification products in negative controls could be accounted for if the Taq enzyme (solution) itself contained minute quantities of D N A not eliminated during purification of the enzyme (J. Carlson pers. comm.2 ). If used in combination with specific primers, amplification could occur in negative controls. Conversely, products in the negative controls could simply be a function of the primer (i.e., primer-dimers, primer-trimers etc. amplified). For both of these reasons alone, negative controls for each primer must always be used. The utilization of negative controls should eliminate the possibility of scoring false positives (Kwok and Higuchi 1989). 2 Associate Professor, University of British Columbia, Forest Sciences Department, Faculty of Forestry, Biotechnology Laboratory. 104 A subsequent trial with primers 314 to 320 was done to investigate a reduction in total reaction volume by one half (thereby further reducing costs). Based on the previous trial employing 0.025 U Taql\\L (Figure 5) the 12.5 uL reactions were done with 2.5 mM M g 2 + . The results (not shown, but virtually identical to the corresponding reactions seen in Figure 5) suggested that highly reproducible banding patterns could be produced with weevil D N A in a total reaction volume of 12.5 uL with 0.025 U Taql\xL and 2.5 mM M g 2 + . A l l additional primer screening was done in a total reaction volume of 12.5 uL. Almost 90% (62) of the 70 primers screened using bulked D N A extracted from 100 individuals gave multiple RAPD markers (Table 6). While three of these 70 primers failed to produce amplification products, six of the primers produced either only one band or non-specific amplification (seen as a smear on the stained gel). Forty-five of the primers screened above were also used with aliquots of bulked D N A from 20 weevils. Results (i.e., failure to amplify, multiple markers produced etc.) from the 104 primers screened, using D N A extracts from 20 bulked individuals, followed the same general trend as seen using bulked D N A from 100 weevils (Table 6). Of the 104 primers screened using bulked D N A extracted from 20 individuals, eight primers (listed in methods) were tentatively identified which appeared to yield amplification products specific to the Y chromosome. The electrophoretogram obtained using these eight primers in one RAPD assay with bulked D N A from 20 weevils, is shown in Figure 6. Banding profiles were identical for male and female bulked D N A with primers 376 and 387 (no male specific markers evident), but six of the eight primers generated the identical putative Y-specific markers as seen in the original screening trials (not shown). The fragment size, based on migration distance of the 100 bp D N A markers (graph not shown), of each putative Y-specific marker is presented in Table 7. RAPD assays, using these eight primers were repeated with 105 Table 6. Performance (failed to yield products, gave one band, resulted in multiple amplification products) of primers screened using DNA obtained from 100 bulked weevils and from 20 bulked weevils (each gender). Bulk DNA Bulk DNA from 100 from 20 weevils weevils Number of Primers which failed 3 8 to yield amplification products Number of Primers which gave 5 6 either only one marker or spurious amplification products Number of Primers which gave 62 90 multiple amplification products (= multiple genetic markers) Total Primers Screened 70 104 (45 same as bulk DNA from 100) 106 L B F M C F M C B F M C F M C B F M C F M C B F M C F M C L V 219 291 365 374 L B F M C F M C B F M C F M C B F M C F M C B F M C F M C U JJ PI Pt Pf 376 383 386 387 Figure 6. Eight primers which tentatively yielded putative Y-specific markers (in primary screenings with bulked DNA from 20 weevils), used in one RAPD assay with bulked DNA obtained from 20 weevils. Each primer (indicated below RAPD profiles) was used twice with both bulked male (M) and bulked female (F) DNA. Arrows indicate the position of six possible male specific bands, C indicates negative control reactions, B indicates blank lanes, and L indicates size markers (100 basepair ladder). 107 Table 7. Summary of putative Y chromosome specific markers seen using three different extracts of bulked DNA. Presence or absence (in each gender) of markers in DNA from 100 bulked weevils (Figure 7) and DNA generated from 12 weevils (Figure 8) are compared to markers seen in DNA from 20 bulked weevils (Figure 6). Primer Approximate Possible Presence or Absence Length (base in males and females pairs) of putative male marker (Bulked DNA-20) Bulked DNA-100 Bulked DNA-12 219 990 possibly present in both genders faintly present in both genders 291 780 possibly present in both genders absent in both genders 365 950 appears absent in both genders inconclusive 374 590 present in both genders faintly present in both genders 376 None present — 2 markers present only in males 383 520 absent in both genders appears absent in both genders 386 1050 absent in both genders perhaps present in both genders 387 None present — — 108 bulked D N A obtained from 100 and 12 weevils (separate assays). The results from these amplifications are shown in Figure 7 and Figure 8 respectively. The scoring (presence or absence) for each putative Y-specific band seen in Figure 7 and Figure 8 (relative to Y-specific bands seen in Figure 6) is presented in Table 7. Interpretation of the upper rank of the gel shown in Figure 7 was difficult due to smearing of one the D N A markers (caused by undissolved agarose). With respect to bulked D N A from 100 weevils, three of the bands were possibly present in both genders (primers 219, 291 and 374), and three of the bands appeared to be absent in both genders (primer 365, 383 and 386). With respect to D N A generated by pooling extracts from 12 individuals, three of the markers of interest appeared to be faintly present in both genders and two of the markers absent in both genders. Primer 365 gave inconclusive results. Interestingly, primer 376 did appear to target two unique putative Y-specific markers (Figure 8) of approximately 650 and 765 base pairs in length. However, if one carefully examines Figure 6, the 650 bp locus may very well be present in one of the lanes showing products amplified from female D N A using primer 376. The results presented in Table 7 suggest that the putative Y specificity appeared to be refuted for four of the markers and not confirmed in the other two markers. This assumes that homologous bands from the different gels were correctly matched and that marker alleles from different loci never co-migrate to the same position on a gel. Primers 219, 374 and 376 were selected for use with individual (not bulked) D N A extracts. These primers were chosen since they produced the most distinct, putative Y-specific markers (refer to Figure 6 and Figure 8). Both putative Y-specific markers generated by primers 219 and 374 appeared to be present in reactions using female D N A (individuals) (Figure 9: primer 219, Figure 10: primer 374). Scoring for presence or absence of the marker of interest was extremely difficult in both Figure 9 and Figure 10, due to the fuzzy nature of the 109 Figure 7. Eight primers which tentatively yielded putative Y-specific markers in initial primer screening with bulked DNA from 20 weevils, used in one RAPD assay with bulked DNA obtained from 100 individuals. Loading of products (M for male DNA, F for female DNA, C for negative controls) and size markers (L) into lanes is identical to that indicated in Figure 6. Arrows indicate the position of the six possible male specific bands identified in Figure 6 (used as reference points in scoring). 110 Figure 8. Eight primers which tentatively yielded putative Y-specific markers in initial primer screening with bulked DNA from 20 weevils, used in one RAPD from 12 individual extractions (both genders). Loading of products (M for male DNA, F for female DNA, C for negative controls) into lanes is identical to that indicated in Figure 6. Arrows indicate the position of the six possible male specific bands identified in Figure 6 (used as reference points in scoring). Crosses indicate the position of two putative male specific bands (primer 376) not seen in Figure 7 or Figure 6. 112 L B F M C B B F M C B L B U L K E D INDIVIDUAL M A L E S B U L K E D L B F M C B B F M C B L B U L K E D INDIVIDUAL F E M A L E S B U L K E D Figure 10. Primer 374 used in reactions containing DNA extracted from 12 individual male and 12 individual female weevils. Loading of products into sample wells was done in the same order as indicated in Figure 9. Arrows indicate the position of the putative Y-specific marker of interest in each of the reactions done with bulk DNA. Molecular size markers (100 bp ladder) indicated with an L , blank lanes indicated by a B, reactions from bulked male D N A with an M , reactions from bulked female DNA with an F and negative controls done for the bulked D N A reactions with a C. I l l Figure 9. Primer 219 used in reactions containing DNA extracted from 12 individual male and 12 individual female weevils. Products from male D N A are shown in the upper rank of the gel and products from female D N A in the lower rank of the gel. Reactions done with bulked male (M) and bulked female D N A (F) flanked each set of reactions done with D N A extracted from individuals. Arrows indicate the position of the putative Y-specific marker of interest in each of the reactions done with bulk DNA. Molecular size markers (100 bp ladder) indicated with an L , blank lanes indicated by a B and negative controls done for the bulk D N A reactions with a C. 112 L B F M C B B F M C B L B U L K E D INDIVIDUAL M A L E S B U L K E D L B F M C B B F M C B L B U L K E D INDIVIDUAL F E M A L E S B U L K E D Figure 10. Primer 374 used in reactions containing DNA extracted from 12 individual male and 12 individual female weevils. Loading of products into sample wells was done in the same order as indicated in Figure 9. Arrows indicate the position of the putative Y-specific marker of interest in each of the reactions done with bulk DNA. Molecular size markers (100 bp ladder) indicated with an L , blank lanes indicated by a B, reactions from bulked male D N A with an M , reactions from bulked female D N A with an F and negative controls done for the bulked D N A reactions with a C. 113 electrophoretograms. The lack of clarity could possibly be attributed to either improper focusing of the camera or to the longer than normal period of destaining (destained 75 instead of 20-30 minutes). The author has witnessed diffusion of D N A bands, in R A P D amplification products, when destaining was carried out for an extended period of time. Additionally, scoring of bands on these electrophoretograms would have been easier if: 1) reactions containing male D N A had been loaded in lanes next to lanes containing products from female D N A ; and 2) an additional set of positive controls (bulked D N A reactions) had been loaded in the center lanes (each rank). These considerations were followed for primer 376 as clearly shown in Figure 11. Highly resolved, discrete, easily scored bands were produced in this assay. The two putative Y-specific markers associated with primer 376 did not appear to be amplified in any of the reactions containing female D N A (Figure 11). This is not surprising given that the bulked D N A used, which originally targeted these loci, was generated by pooling equal aliquots of D N A from these exact individual extractions. Table 8 summarizes the scoring of putative male specific bands seen in Figures 9 through 11. The four most distinct, putative Y-specific markers (listed in Table 8) were also chosen for use as hybridization probes. Secondary amplification was not done for the traditional 45 cycles (as in the primary amplification) in an attempt to achieve maximum yield of target D N A while minimizing non-specific amplification. Figure 12 indicates that secondary amplification resulted in a sufficient quantity of D N A (for use as probes) in three of the four reactions which went for 25 cycles; products in lanes 17, 18 and 19 determined to be 10 r|g/uL based on comparison to standards. Therefore, D N A contained in these reactions were used as the source of template D N A in the preparation of probes. The results obtained from the reactions which went for 20 cycles (Figure 12) illustrate the various problems which can occur when using bands excised from agarose as the source of template D N A in RAPD amplification (i.e., failed reactions (lane 13), 114 B C F M M F M F F M M F M F F M M F M F C B B U L K E D I N D I V I D U A L B U L K E D I N D I V I D U A L B U L K E D I N D I V I D U A L B C F M M F M F F M M F M F F M M F M F C B * B U L K E D I N D I V I D U A L B U L K E D I N D I V I D U A L B U L K E D I N D I V I D U A L Figure 11. Primer 376 used in reactions containing D N A extracted from 12 individual male and 12 individual female weevils. Reactions done with female DNA(F) ran next to reactions done with male D N A (M). As shown, reactions with bulked D N A were loaded in the right and left hand lanes, flanking the products produced with D N A from individuals. An additional set of reactions with bulked D N A was also loaded in the center lanes of each rank on the gel. Arrows indicate the position of the two putative Y-specific markers of interest in two of the sets of bulk D N A controls. Blank lanes indicated with an M and negative controls with a C. 115 Table 8. Scoring (presence) of four putative Y chromosome specific RAPD markers in reactions containing DNA extracted from 12 individual weevils of each gender. Size of Number of successful- Number of Individuals marker Marker Amplifications (out of 12) appears to be present in (base pairs) Male Female Male Female Primer DNA DNA DNA DNA 219 990 11 10 3 somewhat 4 clearly (Figure 9) 5 faintly 2 faintly 374 590 12 9 4 possible 2 possible (Figure 10) 376 650 12 11 3 0 (Figure 11) 765 12 11 10 0 116 M B 2 3 4 5 6 7 8 9 10 B 12 13 14 15 B 1 7 18 19 2 0 B M D N A S T A N D A R D S 20 C Y C L E S 2 5 C Y C L E S Figure 12. Products from reactions using retrieved RAPD markers specific to primers 219, 374 and 376 as template DNA, visualized on a 2% agarose gel. Eight lambda D N A standards (range: 10.0-0.125 r)g/uL) were loaded into lanes 2-9 followed by reactions which underwent secondary amplification for 20 cycles and 25 cycles respectively (fragment from 219 lanes 12 and 17, fragment from 374 lanes 13 and 18 and fragments from 376 lanes 14 and 19 (larger fragment) and 15 and 20 (smaller fragment)). D N A size markers (100 bp ladder) indicated with an M . 117 poor yield (lane 12) and apparent amplification of non-target bands (lane 14 indicates two bands present; lane 15 indicates the reaction targeted the 650 bp marker not the intended 765 bp marker)). The reaction meant to target the 765 bp fragment associated with primer 376 (5 uL aliquot of reaction seen in lane 20, Figure 12), underwent an additional 10 cycle amplification in an attempt to increase D N A yield. A five uL aliquot of this final amplification product visualized on an ethidium bromide stained agarose gel, revealed that four distinct markers (765 bp fragment and three smaller sized fragments) now appeared to be present (results not shown). Initially, I hypothesized that the three additional bands were a result of obtaining an agarose slice which did not only contain D N A from the 765 bp fragment. That is, minute traces of D N A from three other R A P D markers seen with primer 376 (and also present in reactions with female D N A (Figure 8)), were transferred along with the marker of interest. The additional three markers could, however, have resulted from the presence of secondary primer binding sites on the 765 bp fragment. If this were true, the smaller markers were merely sections of the 765 bp marker and D N A contained in this reaction would have sufficed for preparation of hybridization probe. Based on the former argument, I chose not to use D N A from this reaction as a putative male specific probe. Figure 13 indicates the superior resolution and highly reproducible results that can be achieved running amplification products on a 2% Synergel™/1% agarose gel for six hours and 45 minutes at 100 volts. The six, 765 bp fragments seen on this electrophoretogram were carefully removed, D N A retrieved through blotting paper and used directly as template D N A to produce radioactive, labeled probe. Traditionally, D N A to be used in labeling reactions are retrieved from low melting point agarose. To increase migration distance between the selected marker and the next migrating band (hence increasing the likelihood of isolating only the target DNA) the system described herein was used. Ingredients present in the TPE buffer (particularly phosphorous since 118 L F B M B F B M B F B M B F B M B F B M B F B M B F B M B L Figure 13. Electrophoretogram of seven sets of replicated reactions using primer 376. Reactions using bulked male D N A indicated with an M , those based on bulked female D N A with an F, blank lanes indicated with a B and 100 bp ladder with an L. Stars indicate the two putative Y-specific markers. 119 the Taq enzyme must recognize the 3'-OH end of the phosphate group to join nucleotides) may have somewhat inhibited incorporation of P 3 2-dCTP in the labeling reaction (low scintillation counts were observed) but a sufficiently "hot" probe was obtained. Al l four autoradiograms revealed that probes bound both to male and female DNA. Figure 14 illustrates the autoradiogram obtained from the 765 bp probe associated with primer 376. Development was done after four days exposure. In general, this autoradiogram is representative of the other three. As clearly shown in Figure 14, the probe bound preferentially (reflected in intensity of fluorescence) to the positive control (lower right hand corner) and not at all to the negative control (extreme right, fourth row). Appendix VI gives the D N A loading pattern (sample site and leader number from which each weevil used emerged) for each membrane prepared, as well as indicating which samples used either 50 r\g or 100 r|g of DNA. Intensity of hybridization did not appear to be correlated to either amount of D N A loaded or to the gender of the weevil from which the D N A was obtained (compare information in Figure 14 to that presented in Appendix VI); overall fluorescence was not less in areas loaded with only 50 rjg of D N A and not less in areas with D N A from female weevils. Hybridization of probes to D N A obtained from both genders does not necessarily refute the male specificity of the RAPD markers. These results could be explained if either: 1) each marker/probe encompassed a repetitive sequence of DNA, common to all members of a species; or if 2) the labeling reaction was done with template D N A also consisting of spurious amplification products (i.e., not solely the marker of interest - even though only one band evident in Figure 12 (J. Glaubitz pers. comm.3)). The former hypothesis could be tested by determining the nucleotide sequence of each marker and constructing PCR probes based on the sequences at 3 Ph.D. Candidate. UBC, Forest Sciences Dept., Faculty of Forestry, Biotechnology Laboratory. 120 ••••••• Figure 14. Illustration of autoradiogram obtained from dotted-membrane hybridized with radioactive probe created from the 765 bp DNA RAPD marker generated with primer 376. Samples from male D N A were loaded into the upper four rows and samples from female D N A in the lower four rows. Source of D N A as follows: rows one and five from W2_MCBRIDE-4&5, rows two and six from S V A N 1 _GOLDRI VER, rows three and seven from E 2 G O L D E N - 1 1 9 and rows four and eight from SMIIIRESFOR. The 96th well (lower right hand corner) was loaded with denatured probe (positive control). The 48th well (last well in the 4th row) contained denatured herring sperm D N A (negative control). 121 or near the end of each primer site (J. Carlson pers. comm.2). The latter hypothesis could be tested if traditional cloning methods were used to generate sufficient D N A for use as probes, instead of the secondary amplification used herein. Selection of cloned fragments should result in only the fragment of interest being used to prepare hybridization probes. 3.3 CONCLUSIONS AND RECOMMENDATIONS FOR FUTURE WORK The results of this study clearly demonstrate that differences in genetic makeup between bulked D N A samples are more easily revealed when bulked D N A is obtained from fewer (rather than more) individuals. While Figure 6 (bulked D N A from 20) reveals six markers unique to the bulked male D N A samples, banding profiles are identical between bulked male and female D N A obtained from 100 individuals (Figure 7) (both assays done with same primers). Also, the sensitivity in detecting differences in genetic make-up is illustrated in Figure 8; a marker (650 bp fragment with primer 376) present in only three of the 12 individual males (Figure 11) used to generate the bulked D N A sample is sharply revealed. It is suggested that future studies using the bulked D N A technique (in conjunction with RAPD amplification) to reveal differences in genomic composition use bulked D N A samples generated from between 10 to 20 individuals. It is recommended that the search for male specific RAPD markers in Pissodes strobi should no longer be pursued since: 1) after screening 129 unique primers only four sharply defined putative male specific markers were seen (primers 219, 374 and 376); 2) when tested with D N A extracted from individual weevils, only two of these markers were shown to be absent in genetic profiles based on female D N A and; 3) the dot-blot procedure failed to confirm the male specificity of the markers of interest. Rather than continue to search for male specific markers, or use the putative male specific markers already identified, easy to score R A P D markers (less 122 intensely stained markers ignored) based on diploid tissue will be used. Primer 376 will be used to generate markers, however, the 650 and 765 bp fragments will be treated as diploid markers. Although analysis based on markers from diploid tissue is more difficult, I would rather base genetic variation on estimated allele frequencies from a large number of loci rather than 'measured' allele frequencies from a small number of putative male specific loci. Allele frequencies estimated from a large number of loci does somewhat overcome the problem of dominance since the statistical power of the test is increased (Hedrick 1992). Additionally, as Lynch and Milligan (1994) stated the accuracy of the F S T estimate greatly increases if the assumption of random mating within subpopulations is met. Random mating is probably not unreasonable to assume in Pissodes strobi and does, in fact, hold true for many natural populations of animals (Hartl and Clark 1989). I suggest that this is a much more conservative and sound approach and will be used to examine genetic variation within and among populations of Pissodes strobi. 123 CHAPTER 4: GENETIC VARIATION OF Pissodes strobi BASED ON DD?LOID RAPD MARKERS This chapter describes the approach taken to estimate genetic variation within and among populations of P. strobi from diploid RAPD markers. The extensive statistical analysis which was performed is explained in detail as a guide to future investigators conducting similar types of studies. 4.1 METHODS 4.1.1 Generation of RAPD Markers Based on DNA Obtained From Diploid Tissue Ten primers, which had produced clear, highly reproducible, easy to score banding patterns were selected to generate diploid RAPD markers. DNA extraction (individuals) and quantification (comparison with Lambda DNA standards), RAPD assay conditions (omitting the initial seven minute denaturation at 95° C) and resolution of amplification products were as outlined in Chapter 3. Only 12 of the 13 populations listed in Table 5 of chapter three were utilized; individual weevils from SMIII_RESFOR were not used in this portion of the study due to poor weevil emergence from the leaders (i.e., only 12 families of weevils would have been represented). Total genomic DNA was extracted from each of 30 individuals (15 male, 15 female) from each of the 12 populations, therefore, total sample size (N) was 360. The number of different weevil families examined from each population was maximized, since, wherever possible, the 30 individuals used had emerged from 30 different leaders (i.e., as previously discussed in chapters one and three, individuals from different leaders were assumed to have represented different weevil families). 124 The ten primers were initially used once with each of the 360 DNA extracts. A single RAPD assay utilized only one of the ten primers and consisted of 46 reactions: 30 reactions using the DNA extracted from individuals collected from one population, 15 reactions employing DNA obtained from individuals from another population, and one reaction lacking DNA (negative control). Amplification products obtained from individual RAPD assays (46 reactions) were resolved on a single gel. Table 9 indicates the pairing of populations used in the initial RAPD assays and subsequent resolution of products. Figure 15 shows the electrophoretogram obtained from the RAPD assay using primer 322 with DNA from 30 individuals from population Wl_MCBRJX>E-26 (white spruce) and 15 male individuals from SMII_KITIMATII (Sitka spruce). One further attempt was made to generate products from reactions which had failed the first time through. DNA quantifications were reevaluated and fresh dilutions made for each of the initial failed reactions. Single RAPD assays still only utilized a single primer, but DNA from more than just two populations were used (i.e., number of failed first time reactions varied for each population and each primer, thus the number of populations used in a single RAPD assay varied). Positive controls (DNA which had produced banding profiles the first time round) for each population were also included in this second attempt. Setup of second round RAPD assays and subsequent resolution of products was such that valid comparisons of individual markers between the populations could be made. Figure 16 illustrates this point; primer 350 was used with DNA from 10 of the 12 populations. Additionally, positive controls ensured the reproducibility of banding patterns once again (banding profiles from first and second runs were compared). If reactions failed the second time through, bands were scored as missing values in the data set (indicated by a dot in the spread sheet). 125 Table 9. Grouping of 12 sample populations in RAPD assays and gel electrophoresis (first attempts). Groupings S VAN 1 GOLDRIVER and 15 females SVAN3TOFINO SVAN2NGOLDRIVER and 15 females SMI_KITIMATI Wl_MCBRJX>E-26 and 15 males SlVHIKITIMATn W3 PRINCEGEORGE and 15 males W2_MCBRJX>E-4&5 E1GOLDEN-2 and 15 females SMII_KITIMATII E2GOLDEN-119 and 15 males SVAN3_TOFINO Pl_GOGAMAl and 15 males SMI_KITIMATI P2GOGAMA2 and 15 females W2_MCBRTDE-4«&5 126 ;ure 15. Products from a single RAPD assay (46 reaction tubes using primer 322) resolved via electrophoresis on a 1% synergel™/2% agarose gel. The first 30 numbered lanes contain reactions with D N A extracted from W 2 M C B R I D E - 2 6 (male D N A lanes 1-15) and lanes 31-45 show profiles based on D N A extracted from 15 male weevils from SMII_KITIMATII. Negative control indicated with a C and 100 basepair ladders with an M . 127 Figure 16. RAPD profiles (second attempts with primer 350) from 10 of the 12 sample populations resolved via electrophoresis on a single 1% synergeI™/2% agarose gel. D N A used in each reaction obtained from: lanes 1-5 E l _ G O L D E N - 2 , lanes 6-11 SMII_KITIMATII, lanes 13-15 S V A N 1 GOLDRIVER, lanes 16-18 SVAN3_TOFINO (females), lanes 20-25 E2_GOLDEN- l 19, lanes 26-27 SVAN3_TOFINO (males) , lanes 28-30 P2_GOGAMA2, lane 32 W 2 M C B R I D E -4&5, lanes 34-40 S V A N 2 N G O L D R I V E R , lanes 42-44 S M I K I T I M A T I (females), lanes 46-48 SMI_KITIMATI (males), and lanes 50-54 Wl_MCBRTDE-26. Molecular markers indicated with an M and blank lanes with a B 128 4.1.2 Scoring of Diploid RAPD Markers To score RAPD markers as presence and absence data the following assumptions were made: 1) each scoreable band per primer (different molecular weights) represented an independent locus and thus was treated as a two allele system; 2) marker alleles from different loci did not comigrate to the same position on a gel; and 3) banding patterns from gels were interpreted in a completely unambiguous manner (i.e., the investigator was unbiased and accurate in matching bands from different lanes within and between gels). Throughout the remainder of this thesis the terms RAPD markers, RAPD bands, RAPD loci and RAPD DNA fragments have been used interchangeably even though they are not all truly synonymous. Monomorphic and polymorphic RAPD markers were visually scored once from the electrophoretograms (photographs) by the author. Only relatively easy to score (i.e., usually bright and discrete), highly reproducible bands were generally considered. Presence of a band was scored as a one and absence as a zero. Photographs were grouped by primer and all possible markers for one primer scored prior to scoring markers generated from a different primer. As previously discussed, photographs indicating profiles generated from the second round reactions were used as controls to score bands between gels. The approximate size of each DNA fragment scored from each primer is given in Table 10 in the results section of chapter 4 (page 140). Notes to indicate confidence in scoring each particular band were also made. Table 10 indicates the loci which were scored with very high confidence (superscripted with a 1), high confidence 129 (superscripted with an 2) and medium to low confidence (no superscript). These rankings were used as the basis to partition loci into different data sets for subsequent analysis (to be discussed). 4.1.3 Data Analysis 4.1.3.1 Computer Programs Used Presence and absence data were entered into an EXCEL spreadsheet. Both multivariate and population genetics statistical techniques were explored. Multivariate procedures were performed using SAS (SAS Institute 1989). All multivariate procedures performed were done using the presence/absence data. Data sets from the EXCEL spreadsheets were saved as ASCII files and converted to SAS data sets by a FORTRAN program compiled in MTS by the author. These data sets were subsequently transferred to UNIX, and submitted to SAS procedures for analyses. The computer program BIOSYS-1 (Swofford and Selander 1981) was used for the population genetics analysis. Allele frequencies, correcting for small sample size (Lynch and Milligan 1994), were computed in an EXCEL spreadsheet. BIOSYS-1 data files were created in WORD for Windows from ASCII files generated from the EXCEL programs. The computer programs SIGMAPLOT and SIGMASTAT (Jandell 1994) along with EXCEL were utilized to investigate specific correlations and to generate frequency histograms. 4.1.3.2 Considerations in Compiling Data Sets SAS effectively deals with missing values by eliminating that observation from the analysis (observation = individual (one of 360)), therefore, five bands with too many missing values (Table 10, superscripted with an M) were discounted prior to analysis. This effectively reduced the total 130 data set to information on 73 bands. Markers which were monomorphic overall (Table 10, superscripted with a C), while of interest in the population genetic approach, were of concern in the multivariate procedures (i.e., matrix approaches singularity, therefore, determinant approaches zero and inversion of matrix difficult or not possible (Manly 1992)). Additionally, BIOSYSL, the large version of BIOSYS-1, in its unmodified form, is able to analyze data from a maximum of 60 loci. Given these three considerations, two different data sets, entitled 60A and 60B, were used for much of the analysis. While both sets utilized 60 loci, elimination of loci from the total data set of 73 loci was as follows for each set: 60A: eliminated- first 13 loci which were scored with medium to low confidence (Table 10)-monomorphic markers, therefore, retained. 60B: eliminated- four loci monomorphic across all populations (Table 10, superscripted with a C), three loci which were monomorphic when data was sorted by gender (Table 10, superscripted with a G), five loci which were scored with medium to low confidence-monomorphic markers, therefore, eliminated. Additional data sets explored will be indicated where applicable. 4.1.3.3 Multivariate Techniques 4.1.3.3.1 Introduction As previously indicated, all multivariate techniques were done based on the discrete, binary coded data. Parametric procedures included stepwise discriminant analysis (PROC STEPDISC, SAS Institute 1989) and canonical discriminant analysis (PROC CANDISC, SAS 131 Institute 1989). Principal component analysis (PROC PRINCOMP, SAS Institute 1989) and nonparametric discriminant analysis (PROC NEIGHBOR, SAS Institute 1989) were also performed. Variables measured as presence or absence are nominal-scaled data. They are discrete variables not continuous and their underlying distribution may not be multivariate normal (Dillon and Goldstein 1984). Thus, significance tests and levels indicated from the parametric procedures may not be valid and 'significance' will be reported within quotation marks. Multivariate statistical procedures were considered very useful, even given the violation of assumptions, for the following two reasons: 1) the multidimensionality of the data was retained (Strauss et al. 1992), thus the spread of individuals within a population (and overall population structure) could be examined in two or three dimensional plots; and 2) for comparative purposes with results obtained from analysis based on allele frequencies (i.e., traditional population genetics approach). 4.1.3.3.2 Principal Component Analysis Principal component analysis was done as an exploratory technique to address the extent of correlation amongst the variables and to examine overall structure in the data. R-type PCA (correlations amongst variables examined) was carried out on the correlation matrices of both the 60A and 60B data sets. Plots of principal axis two versus principal axis one (codes assigned for each population and single observations plotted) were obtained and examined for subtle associations often missed by other traditional statistical procedures. Although different scales of measurement were not used on the variables, the most common reason in PCA for standardization of the variables to zero means and unit variances, the correlation matrix was specified to ease interpretation of eigenvalues (i.e., eigenvalues greater than one deemed 'significant' with 132 correlation matrix; 'significance' level of eigenvalues from co variance matrix must be calculated from output (Dillon and Goldstein 1984)). An additional PCA, employed as a data reduction technique, was performed as a prelude to analysis of variance (general linear model procedure -PROC GLM). In this application, the data set consisted of 69 loci with the four monomorphic loci not considered (these four markers were superscripted with a C in Table 10). To test for 'significant' differences among the means of the 12 populations, PCA variables with eigenvalues greater than one were retained for use in the GLM procedure. Loading scores of the original variables (i.e., loci) from each of the 'significant' PCA variables, indicated from the GLM procedure, were then examined. Ten loci with loadings greater than 0.30 were identified and grouped to form an additional data set tested in nonparametric discriminant analysis (discussed below). 4.1.3.3.3 Stepwise Discriminant Analysis Stepwise discriminant analysis was also employed in an exploratory fashion for two reasons: first, to test for differences between males and females within each population (i.e., obtain number of loci which 'significantly' discriminate between males and females for each population as an indicator of overall gender difference) and second, to obtain a subset of loci which 'significantly' discriminated among the 12 populations. Sixty-nine loci, mentioned previously under PCA, were indicated as being eligible for examination in both stepwise discriminant procedures. Selection criterion was based on stringent entry and exit levels set at 'a= 0.05'. Parametric discriminant analyses require that the data follow multivariate normal distributions, thus results were interpreted cautiously and further tested in the Neighbor procedure. 133 4.1.3.3.4 Nonparametric Discriminant (Neighbor) Analysis For all neighbor analysis performed, prior probabilities were set as equal. The 'holdout method' described by Dillon and Goldstein (1984 page 392) was used to generate error rates; the classification criterion was developed from two-thirds of the total data set (i.e., 240 observations), with the remaining one-third used for error rate estimations. To address the reliability of the error rates (i.e., test for strong structure in the data), the data were partitioned into three series and neighbor analysis performed on each of these three series. Partitioning into the two-thirds data sets were as follows: a) series one, the first twenty observations from each population were used; b) series two, the last twenty observations were used and; c) series three, the middle twenty observations were used. While technically not a (completely) random selection, numbering of leaders, and hence ordering of observations within a population, was entirely random (i.e., leaders were taken out of the collecting bag at random and numbered sequentially). Neighbor analysis was initially performed on the three series varying k from 2 to 12. Different k's were tested to address the variability in results obtained by specifying different number of nearest neighbors to examine when classifying an observation. The neighbor analyses discussed above utilized information from 69 variables (previously discussed under PCA). Neighbor analysis was primarily done, however, to obtain misclassification rates from different groupings of the populations (i.e., to examine population structure) and to test different subsets of loci in terms of discrimination power. To address the first objective, separate neighbor analyses were done with observations classified according to the following: 134 1) by population, considering SMIKITIMATI and SMHKITIMATII a single population and Pl_GOGAMAl and P2_GOGAMA2 a single population, thus 10 groups defined; 2) by four groups, with populations pooled according to tree host (i.e., Ss, Se, Sw, and jPine groups) - this partitioning of the populations will be henceforth referred to as 'tree species' grouping; 3) by five groups, as in 2) but populations from Ss on Vancouver Island (Ssvanis.) and Ss from mainland B.C. (SsMainiand) separated; 4) by four groups defined, populations collected from Se combined with those from Sw (Se-Swcompiex), Sitka populations still separated, and jPine- this partitioning will hereafter be referred to as the 'four group' scenario and finally; 5) three groups, Ss (all), Se-SwCompiex, and jPine. Comparison of mean error rates for these five different partitionings of the populations was done for k set at 2 and 12 (total data set 60B employed). To test different subsets of loci and to examine the increase in discrimination power when additional loci were included in the analysis, data sets consisting of 10, 22, 36, 45 and 69 loci were used. Both random and nonrandom selection of these number of loci were tested. Random selection was done by assigning each marker a number from 1 to 69 and then using the random number table in Zar (1984 p. 653). Criterion for nonrandom selection of loci was as follows: 1) 10 loci indicated from PCA followed by GLM analysis; 2) 22 loci scored with very high confidence (superscripted with a 1 in Table 10); 3) 36 'significant' loci identified in the stepwise discriminant procedure; 4) 45 loci scored with very high and high confidence (superscripted with a 1 or 2 in Table 10); 135 5) 69 loci, the full complement, omitting monomorphic loci (as indicated with superscript C in Table 10 and previously discussed in PCA section). Mean error rates between the five different data sets and the random versus nonrandom selection of loci were compared for the 'four group' scenario. 4.1.3.3.5 Canonical Discriminant Analysis Since the assumption of multivariate normality was violated, tests to establish significant differences between the mean values for any pair of groups and tests to indicate whether the canonical discriminant functions varied significantly from group to group were not conducted. Rather, canonical discriminant analysis was employed to examine within group variation, to view basic associations between groups in two dimensions and to obtain Mahalanobis distances among all pairwise comparisons between populations (i.e., 66 different Mahalanobis distances). Nonparametric correlation analyses were subsequently conducted using these Mahalanobis distances and various genetic distances (to be discussed). Canonical discriminant analysis was conducted on both 60A and 60B total data sets, and on both of these data sets when they were sorted by gender (i.e., males and females separately). 4.1.3.4 Population Genetic Analysis 4.1.3.4.1 Introduction The approach taken by Lynch and Milligan (1994) was applied (correcting for small sample size but 'pruning' of loci was not done) and allele frequencies for each population were 136 computed under the assumption of Hardy-Weinberg equilibrium (i.e., F I S = 0). The frequency of the null allele (= q) was inferred from q2, with q2 being defined as the proportion of N sampled individuals that did not exhibit the marker. An asymptotically unbiased estimator of q was estimated from the following equation presented by Lynch and Milligan (1994): Let x=q2 and x1/2= the frequency of q then, q = x,/2/(l - Var(x)/8x2) where, Var(x) = x(l-x)/N With q in hand, the frequency of the marker allele (p) was calculated using p = 1 - q. Population genetic statistics computed using BIO SYS-1 included percentage of polymorphic loci, mean expected heterozygosities, and Wright's F-statistics (1978). Cluster analysis, using three different clustering methods on three different genetic distance or similarity matrices was also done. Data sets 60A and 60B were explored separately in all population genetics procedures. Gender sorted data sets were examined in the cluster analyses. 4.1.3.4.2 Percentage of Polymorphic Loci and Mean Expected Heterozygosity A locus was defined as being polymorphic if the frequency of the most common allele did not exceed 0.95 (Wright 1978). Mean expected heterozygosity for each population was obtained from BIOSYS-1. Mean percentage polymorphism and mean expected heterozygosities for the 'four group' scenario were calculated manually. 137 4.1.3.4.3 Wright's F-Statistics The degree of population subdivision, based on expected heterozygosities, was examined using Wright's F-statistics (1978) approach. In this study, as previously mentioned, Hardy-Weinberg equilibrium was assumed (within each population; hence F B = 0 and F D T considered to be the same measure as Frr (Wright 1978)), thus the F-statistics were applied to measure an inbreeding-like component (encompassing sampling drift). Since gene frequencies were used, assortative mating was not considered. Furthermore, natural selection processes were not addressed since alleles were assumed to be selectively neutral. Two different hierarchial arrangements of the populations were explored and pertinent F-statistics generated in BIOSYS-1 for each of the two hierarchies. In the first hierarchy, groups of populations (demes = D) collected from different tree species were defined as unique subpopulations (= S) (i.e., the 'tree species' scenario as previously defined). In keeping with Wright (1978) and Phillips and Lanier (1985), the following subscripts were applied: T = hypothetical total population (i.e., combined all 12 populations into one single population); D = demes, and refers to each of the 12 different panmictic populations and; Si = subpopulations, referring to the populations pooled according to tree species. Subpopulations in the second hierarchy tested, consisted of weevil populations pooled as follows: Ss from Vancouver Island (SsVan.is), Ss from mainland B.C. (SsMainiand), Se-SwCompieX and jPine (i.e., the 'four group' scenario as previously explained). Thus, the only change in terminology was the use of S2, referring to subpopulations based on these 'four groups' as 138 opposed to being based on 'tree species'. The hierarchial F-statistics obtained from the BIOSYS output, for each loci and summed over all loci, were F^T (where i = 1 or 2), FD T, and F D S (i >. Negative FST's were calculated in BIOSYS due to limitations in the computer algorithm (Swofford 1989) and these negative values incorporated into BIOSYS' computation of mean FS T. To correct for this underestimation, FST's indicated as being negative were assigned values of zero and adjusted FST's manually (calculator) computed. These latter values were used in subsequent analyses. To display genetic variation based on individual loci, frequency distributions, based on the three different F-statistics obtained from each locus, were also constructed. Four discrete categories were defined following Wright (1978): 0.00 - 0.049 (little differentiation), 0.05 - 0.149 (moderately differentiated), 0.150 - 0.249 ( highly differentiated), and greater than 0.249 (extremely differentiated). 4.1.3.4.4 Cluster Analysis: Methods and Genetic Distances or Similarities Used Hierarchial phenetic cluster analysis was done to explore relationships between populations. Dendrograms were produced using three different clustering methods based on three different resemblance coefficients (i.e., genetic distances or similarities) in BIOSYS-1 (Swofford and Selander 1981). Clustering methods specified included UPGMA, single-linkage and the Distance Wagner procedure. With this latter technique none of the 12 populations was classified as an outgroup. The tree was hence rooted at the midpoint of the longest path with the cophenetic correlation coefficient used to determine the addition sequence. All three clustering methods were employed on resemblance matrices containing either Nei's unbiased genetic 139 identities, Prevosti's genetic distances or Cavalli-Sforza and Edward's genetic distances. The following equations define each of these distances: Prevosti's genetic distance = D = 0.5 Z I qx - qy I (half the sum of the absolute difference between the allelic frequencies of the two populations, x and y) Nei's genetic identity = Ixy = J x / (JJ y) 1 / 2 where Jxy = Z(qxqy) and Jx= Z(qx2) and Jy= SXq/ ) (between populations x and y) Cavalli-Sforza and Edwards Distance = D = (2/TC) COS'1 E(qxqy)1/2 4.1.3.5 Additional Analyses Correlations between Mahalanobis distances and two genetic resemblance coefficients (Nei's unbiased genetic identities and Prevosti's genetic distances) were performed to aid in comparing results between the multivariate procedures and the population genetics technique. Correlation coefficients were computed in SIGMASTAT and plots obtained in SIGMAPLOT. 4.2 RESULTS 4.2.1 Multivariate Approaches A total of 78 different RAPD markers were initially scored from the DNA fragments generated by the 10 random primers (Table 10) (mean of 7.8 bands per primer). Twenty-two of these markers were scored with very high confidence (superscripted with a 1, Table 10) and an additional 23 with high confidence (Table 10, superscript 2). The descending order of primers cr P P £ P g iSg* ta Q cn U > t O U > U > U > U > U > U > U ) U > O i — ' 0 0 U > U > t O U i » J O - O . 4 a . v o t O i > J O N t O O O N U i 4 i . T3 P O £ . 0 - 0 o 5. a . tn V>J " • 1— ~ J n II f" N 3 P 3 P O J § 3 cr © o & •0 O p 3 * n sr^ ° CT> < *1 •! (i a C/» ^ cr FT. tr 2 00 73. a . o oo rt P rt O Pi o-j? C/ K J * - • ^ II " 00 _ 2 3 5 P °- 5-a> ! T rt - • 1-1 ^cn H ° | 3 R a-cn tt. 1 ? cr' 3 P ^ & 3 a —. °> cn CS on O OO o £ II c 3 o 3 o 3 o •"1 &r. o 3 P rt 00 cr o o rt cn 2 o 5 > O o o H O o o o Q H O Q 0 > > > O o 8 8 o o O O > o § 8 H 1—j > O H > o o o Q O O N 0 0 0 0 0 0 0 0 ON -j O N a-f P , 1 a-h a-i a-h P ..!.. a-h a-f a-g a-f o U l O 2 VO O o U l o VO . M © , Ov & U i © NJ VO U i K ) O O -~ "I © ^ U l U l VO O „o 00 o ^to VO O O to o U l o 00 o © 00 o o to VO o o to to o o o o 00 o o to o o O to o 4*. ON U i o o o 8 9. ON ~ J o © o »^ ON £ 00 © o „~ vo U l © VO 00 o Lft U l U l W 171 IO 0 0 0 VO O U i U i U i U l 00 o o to ^ H -U i vo 00 U l U l O to VO o ^tO VO 00 o to O -1 — © O o 4s. O o o ON o ON 00 © U l o 00 00 o © to o o © © o o to © © U i U i © 0 O N U i o „ >o ON - J o 10 o U i © © U l to © U i VO o ON o 1 — 1 to to -VO ON o © © to © o U i U i o 0 o © to © to U i o © 00 © © VO © to to o to o to to to o to to © to to U l o to VO o o o 3. 3 a 2 P o o cr. CD 2 3 £r cn 2 9 > R cn T3 wo q g'.a. § o cT> •+> o-o. a. CTpj (M O O cr , 3 ft H as cr e 3 3 as •3 S2 O 3 as ?r f6 cn f» O •t re a o 3 n as rt h-' O •t as 9 O. o 3 3 c cn rt a 141 indicated in Table 10 reflects the order in which the bands were scored (i.e., primer 374 first, primer 304 last). The smallest two DNA fragments generated from primer 333 (450 and 480 bp) and the smallest three from primer 304 (1050, 1190, and 1250 bp) (those in Table 10 superscripted with M) were not used in any analysis since these bands were faint in some of the RAPD reactions (specific to certain populations) and were often scored as missing values. Overall structure in the data, as revealed in the plots of PCA axis 2 versus PCA axis 1, indicated that the Sitka spruce populations were somewhat distinct from the other weevil populations (Figure 17). This trend was shown when monomorphic loci were either included (Figure 17) or discarded (Figure 18) from the analysis. Additionally, while most of the jack pine observations were located in a distinct region of their own, some of the jack pine observations intermixed in the space primarily consisting of Engelmann and white spruce representatives (Figures 17 and 18). Eigenvalues for the first 23 PCA axes were greater than one for both 60A and 60B data sets (Appendix VII); none of these axes accounted for greater that 8.06% of the total variation. Twenty-five PCA axes had eigenvalues greater than one when 69 markers were used in the analysis (Appendix VIII). When these 25 variables were tested for differences amongst the populations in the GLM procedure, 10 of them were indicated as being 'significant' at or near the 0.05 level (Analysis of Variance table from GLM output given in Appendix VIII). Examination of loading scores from each of these 'significant' PCA variables showed 10 markers with coefficients greater than 0.30 (output not shown). All 10 of these markers (374-a, 305a, 305-f, 376-b, 376-f, 322-d, 322-g, 333-c, 382-g, 219-d) had been originally scored with very high confidence (Table 10) and they represented bands from seven of the 10 primers. These 10 loci hence formed the data set tested in nearest-neighbor analysis. S 1 do' c 5! co t o re O. ro 03 CD ro — -» > UJ CD > U) 03 ro ro J» c_ u> CO o cr ro o a a o CH Jt> r> W II II II 1  II a i rt 7? 3 rt rt 1 <J H » ET.<w w w R- rt » » 1 RT 3 v. v> JS 5 •O » 1 i 3 2 C B S 3 rt rt rt „ rt rt rt ^ C S » rt rt _ w1 » 3 a l 3 B.| • 3 3 a (JO C "I rt 00 2 cr 5. rt Q. 5" IT rt w» rt CO 3 rt O . C rt" O S 3 rt hjj 2 3 to ro CD 2 T3 2- EL cro O rt o 17 ^ o § i-t- 3 rt 3 . r t .99 5' jr o 3 rt » 3 a o 3 89 rt rt O c - 1 rt o ? -cr rt S* rt 3 " » a. W3 rt rt 0\ o cd rt o rt o 3 3 •a s rt ro > > ro > a) ) > - » > ro > > -» ro > —» to I ro u> >• > > CD > I ro oo o ro ro ro ro ro ro —' ui c_ cr u> cr cr C cr cr cr o> d n> cr a a n a n a. a n a o o cr o a a o n a © a o o a o n © ro o o. •» CO II II II to » g; 5 * c« 3 5 >a » rt ^ B l-s B 3 " " rt « » "S * i rt rt to rt rt S9 o. cr rt II II rt S-* S3 » Vi Vi •a "a s c rt rt rt rt • 3 _ Sr1 sr 5-5 144 Stepwise discriminant analysis indicated that on the whole very few RAPD bands were useful in 'significantly' discriminating between the males and females in each population (Table 11). Differences between the genders ranged from zero markers in the W1MCBRIDE-26 population to 16 markers in the SVANl_GOLDRIVER population (mean of 4.6 discriminating markers in 11 of the 12 populations). As shown in Table 11, differences between the genders were not explored for SVAN3_NGOLDRIVER population since missing values were present in each of the female individuals in this population. Thus, two complete classes did not exist when 69 markers were specified and so the population was discarded from the procedure. Overall differences between the genders appeared to be minimal based on the stepwise procedure. However, gender sorted data sets were explored in the canonical discriminant and cluster analyses to thoroughly examine potential differences between the genders. Stepwise discriminant analysis to test for differences among the populations revealed that 36 of the 69 markers were 'significant' discriminators. Additionally, once a variable was entered into the model it was never removed. The number of observations available for analysis from each population and the stepwise selection summary (short version) is presented in Appendix IX. Estimated mean error rates from nearest-neighbor analysis were highest with k set at 2 (k varied from 2 to 12). Varying k between 3 and 12 did not yield substantially different error rates. Thus, on the whole, selection of k did not appear to dramatically affect the misclassification rates (Figure 19). Figure 19 clearly shows that mean error rates are considerably less for the data used to develop the classification relative to data used to test the classification scheme. Subsequent figures pertaining to the nearest-neighbor analysis were generated using only the test data mean error rates. All three series tested displayed more or less the same trend of estimated mean error rates with series 1 yielding slightly higher misclassification rates (Figure 20). Given the trends shown in Figure 19 and Figure 20 subsequent neighbor analysis was conducted only on series one 145 Table 11. RAPD markers which discriminated (stepwise procedure: 69 RAPD markers examined) between male and female Pissodes strobi in each population. Population Sample Size3 Males Females RAPD Markersb S VAN 1 GOLDRI VER 13 9 376-d, 382-h, 374-b, 219-h, 382-d, 305-c, 336-c, 374-d, 333-i, 333-e, 350-e, 374-c, 350-f, 374-f, 336-e 376-b S V AN2NGOLDRIVER 10 11 322-h, 219-d, 336f SVAN3_TOFINO 0 0 SMIKITIMATI 7 12 305-a, 350-c, 305-b, 382-f, 322-i, 374-c SlVni_KITIMATII 12 12 374-d, 333d Wl_MCBRIDE-26 7 9 None W2MCBRIDE-4&5 12 8 322-j, 322-i, 374-e, 322-b, 304-f W3PRINCEGEORGE 10 12 322-c E1GOLDEN-2 15 11 376-d, 305-a E2GOLDEN-119 10 10 219-g, 333-d, 305-d, 336-f, 333-i Pl_GOGAMAl 14 10 376-d, 336-g, 304-d P2GOGAMA2 9 10 304-e, 322-e, 350-f, 374-d, 374-f, 376-e, 305-d, 305-c, 333-c, 376-e, 350-d, 382-d a: Due to missing values in data set entire 15 observations (= individuals) per gender could not be used in analysis b: Entry and Staying level set at a = 0.05 146 CM 00 CO CO CM O O O O O O O O O O O O C D O O S t O l O ^ f O N r suonejndod oj Suipjoooe uoi}Boijissep -(O/0) ssjm JOJJS UBST/J s 1 53 • i 1 CD o -»—> -o So g •a <U ta & CD o I— •o <u S C5 es "O C es es o B c 08 u C vi es a . > *S -a fi « .5 es 3 T> « 3 S « « ©\ S 8 1 '= 2 ^ cu eS S o S <*• rL © w e s o o .2 « u es c o .5 U w C7\ CU 3 OX) Mean error rates (%) - classification according to population 148 using k set at 2 and 12 (i.e., employing series one with k set at 2 since this was the worst case scenario in terms of misclassification rates). Misclassification rates were very high when individuals were classified according to the population from which they were collected; in general, 40% of the observations were incorrectly classified (Figure 20 and Figure 21). Classification of individuals improved when populations were grouped (Figure 21). With k set at two or 12, misclassification was still relatively large when populations were grouped according to 'tree species', but when populations were pooled according to the 'four group' scenario mean error rates fell substantially (misclassification approximately only 13% (Figure 21)). Furthermore, combining the Sitka spruce populations (three groups) did not reduce the error rates extensively and separating the Se and Sw populations (five groups) increased the error rates relative to the 'four group' scenario. Due to these results (and to address objective number one stated on page 3), subsequent analysis pertaining to grouping populations a priori was restricted to the 'tree species' grouping and the 'four group' scenario. Discrimination power tended to increase as the number of markers used in the analysis was increased (Figure 22). Random selection of markers always yielded higher error rates than nonrandom selection of markers. Interestingly, mean error rates using the 36 markers identified in stepwise discriminant analysis as 'significant' discriminators yielded virtually the same mean error rates as when the full complement of 69 markers were used (Figure 22: approximately 14% overall misclassification rate). Inclusion of monomorphic loci did not appear to affect the trends shown in the plots of the canonical discriminant analysis (compare Figure 23 with Figure 24). The general associations (between the populations) exhibited in the combined data, sets were also viewed in the gender sorted data sets (Figure 25 and Figure 26; only 60B output given, results from 60A identical and oro c •t rt n o •a Ss SS* o 3 Mean error rates (%) of test data 3 rt 3 rt oro o s "5. 5' ore 0 »t «_ ST c« s rt rt-5' 3 1 a rt s» rt rt ON o cd rt MM* rt Cfl rt rt-as rt as 3 a - - - i | 0 ( O W W 4 - * U l O l O ) o c n o c r i o o i o c n o c n o c n o POPULATION Q o ft S' p CfQ i-t O -g OP TREE SPECIES 5 GROUPS 4 GROUPS 3 GROUPS 6*1 3 e r e ' E 3 s> r* or HB •1 — as o « •"* if s rt - * i B 2 i i r 3 S S" « — ft as s I s r » (TO 9 rt ^ a ° S 2 l l as wi R r> 0 -S 3 2. D 1 J ft to = 5 II 1 1 • — i f 11 II rt "> a. O w a % 3 S 1 i 1 EI 3 i f ? E -. S. 3 S s r § 8 » a s jo £ II | on a i I 3 rt °- oro 9 2 o c 1 ? « 3 ? II © -o C 5 £ rt ft V o 3 3 rt • 3 wi • a. Wl !7» t/3 - +-0 3 < -3 rt © C < rt 1 E 3 p. © g 5i 3 M rt CO • o > I—I CO u W J I n § 1 °" rt £ 3 fl\ sr ©• o * o • © as £ 1 3 I a 3 9 2. -? 1 » ? 3 = ' • i c d c d > ^ ^ H B t * CD Ul > CO B J 1 ro ui > ro ^ • • T c_ > (S ^ " • S CD co e_ ro ui — > CO c ui ro c CD -> > > > - 4 Ul cr ro ui ro co -> a cr cr a • O CD > r > bl II II II 1 1 4 m rt- rt -a rt r S ES rt - B S 3 II * -a 8. 8 « II II CO CO rt-* rt-" • W F rt_ • as Wl Wl •a -a 2 2 rt rt < 2 5 ES. 3 _ 6? 5" e I 3 a 1ST II =/ (A a i i 3 rt a TO °" 9 9 2 O 3 ? l <— -a 3 =' E T" O • 3 e/s >—^  •J rt a. O < as 3 5 o 3 3 rt -. TO 3 3 rt .U 3- Hg si O rs *• ir © » •*» » S 3 5 °- 2 B Bf* as rt B as rt — 1 ^ g I © 8 < 3 I 3 2. 3 se v> ft rs 0 as 3 3 ^ © ? ft 1 < 3 I 1 i as 2 B S — • rt CA S3 r- a CL as © 3 era rt Vi a? © 3 a £ I ? ? £. * 5* 3" a | a PZ & n * » • K" w - o> n *f « * n 3 os — a o I ft p * © II © 2 3*1 f Ji & I 1 3 i£ 3 » 3 3 35s oo S 3" 3 f l -I • 3 S rs «« 3 s JT 3 S rt 2 a > J» > CD u u c_ > - . J> -» M t\) cr u p CO UJ o- S ts cr cr cr cr cr — » cr cr a o o CO a a. » a a a _ a. n a a o a. a Z9\ II § tfl £• * 2. - rt TO CL 3 a re O — c« i re " rt as ae l « 2 8 S ^ • | 51 E m re *" 5* s re TO 3 rt se 3 TO* cn * 2 2 ® 3 « re c« a; 3 § cr =. g e. 2. • 2. =• a» I? I ? re o 1 < rt cn 2 3 3 cn » rt u re a 2,1 < n as B i i " 2 W. re o 3 re 1 8 9 Eg cr •» re © re re 3 2 8 3 -a 9 E i I i g 11. Q, « ' o 2 i 3 CL TO 9 rt » 8 t/5 o cn e 3 CL < El 3 as_ cn rt O 3 o E l o rt CL o CL as as cn rt B* cr 2. M! 2 « s as a g. n a. 2 cn v< Be rt 3 1 ON 0 p 1 as o 3 «5T o o o 01 c_ 3 '— to • m i» j * P- cr rt £3 3 * S rt 3 L _ . rt n » H rx c/i a £ 3,' Sf IT TO rt 3 as 3 3 cn • 3 g 3 rt rt cn cn " 3 " 3 3 3 3 3 rt o re re 3 2. • ~_ &1 5" II s CO =-* 2. - « etc a . 3 a n o • wi H. 9 rt » so a n cr 2 ° 3 ^ rt Q T* 5" 3 • 3 us rt a 3 rt rt 95 3 O or 5. | | I I I z sr. « I I rt o 1 < rt « rt a? rt I 3 tf] J § 2, 5. < B •» 99 h 3 cr cr • ST 3 rt ©, 3 3 1 i i 3 1' 8 1 % 11 R' rt rt o. « n 2 I E rt 3 99 rt «• 2 s 63 cr r m S 3 i s wi 3 CO wi I* < 95 3 rt O B < rt ^ sr 3 CL e a 95 s era rt CO wi f 3 3 Ef 3_' B 3 a o 00 CO CD C7) CO OC M —1 CD CO a i CD -j 0 Ql CT c c_ c_ c_ rj CT a o '— 1° 0 ?• a" ll II II (*• 5 '« B 69" •= 8 I , wi Wl 811 T 8 «• « rt rt rt 3 rt 99 Jr 3 IS. 3 &1 ST 3 °-a. 155 not presented). However, the distance between the class (populations) means was generally greater in the gender sorted data relative to the combined data. Plots of canonical variable two versus canonical variable one showed that the Sitka spruce populations from Vancouver Island were closely associated with each other and were very distinct from the Sitka spruce samples collected from mainland B.C. Class means for the two Sitka spruce samples obtained from the same stand near Kitimat (SMIKITIMATI and SMIIKITIMATII) were almost identical. For data sets 60A and 60B (genders combined), observations collected from Engelmann spruce intermixed with observations collected from white spruce (Figure 23 and Figure 24). The two samples collected over two years from the same jack pine stand displayed different class means but were closest to each other relative to all the other population means. Also, a few of the individual observations belonging to the jack pine populations appeared within the region primarily consisting of Engelmann and white spruce observations after plotting (Figure 23 and Figure 24). Thus, the general trends seen in the PCA plots were even more pronounced in the plots of canonical variable two versus canonical variable one. Additionally, the first two axes accounted for a good portion of the ratio of between-class to within-class variation: 56.09 and 59.54 % from data sets 60A and 60B respectively. 4.2.2 Population Genetic Techniques Table 12 shows the percentage of polymorphic loci and mean expected heterozygosity (H) for all 12 sample populations (total data sets 60A and 60B). When monomorphic loci were included in the analysis, percent polymorphism ranged from a high of 76.7 % (P2_GOGAMA2) to a low of 58.3 % (SMIIKITIMATII) and H ranged from 0.269 (P2GOGAMA2) to 0.216 (SMII_PRINCERUPERTII). The percentage of polymorphic loci was very similar between the cr P c> CD 0 0 CD p cn cn 3 o pa 3 2,2 t-y CD O. 2 ^ 2. VO^S — i 3 00 o cr o' cr CD CD c CD CS o v < o r-f cr CD 3 o cn O o 0 3 CD* 01 O r-t-CD X o CD CD Cu VO U l to <! <5 <5 tyo < - • < < H t> 5* w W M ^ jS, 00 OO t/3 < < < vo to VO to vo to VO to VO to VO to VO to VO to VO to VO to 00 to 00 to VO u i -o 00 VO -o U l -o 00 vo 00 ON to to to to to to to to VO VO VO VO VO VO VO VO U l - J O N 00 <1 u i o 00 O 00 ON -o O N •o U l O N O N - J U l -o —) U l O N O N U l 00 O N O O N O O N O N U l -O o - J o -o o -o lo O O -O o 00 U l 00 U l -4 O N oo o - J oo oo U l - J O N O N O N —1 - J -o o o - J o Lo o -o lo Lo - J Lo Lo o o o © o © o © o © o © to to to to to to to to to to to to ON U l U l O N U l 4s. 4S. p—» to o -ts. VO U l U l -o © Ov O N •o 4s. U l o © o © © © © © © o o o '© o © © o © o o © o o © to to to to to to to to to to to to to JS. O N 4>. U l 4s. -ts. - J 00 U l O N ON o o o o o o © o o o o o to to to to to to to to to to to to 00 oo 00 00 - J 00 - J -ts. U l -o - j vo Ov O N O N 4s. 4s. U l U) © VO '—' o © © o © o o © © o o o © © © 'o o © o © © © © o to to to to to to to to to to to to o to -ts. u> to to -o 00 -ts. U l -ts. o c ft o 3 O N © > ON © O N O O y " > 3 o O N ^ f . © ef. W o r o o O N © > H to a S 3 re •g-B 73 ta •M 3.12. je re c n O N 3 re C« 1? 2 ON O W ae o a. = p* » 5! c n v—• re » » S3 ^3 S TO O. re ON o o •*» fed " 3 > ^ © 3 £ re 3 " 3 s as s re 3 5 c n re q 3 re re c n I X s » 3 re 5 8-re 2. 3 2 TO 3 c n w re ae o re to 991 157 two Kitimat samples but H varied somewhat. Standard errors for mean expected heterozygosity were quite low. Pooling populations according to the 'four group' criterion resulted in the SsMakiand group exhibiting the lowest and the jPine group the highest overall mean percent polymorphism and H (Table 13). In terms of differentiating populations according to geography, a general trend of decreasing variation (based on these two statistics) from east to west was clearly seen (Table 13). The overall summary of the variance components and Wright's F-statistics (1978) combined across all 60 loci is shown in Table 14. Statistics are presented for subpopulations defined by the 'four group' scenario (S2) and for subpopulations defined according to 'tree species' ( S i ) (60A and 60B total data sets). The greatest differentiation was among populations with FDT's = 0.136 when monomorphic markers were included and FDT's = 0.129 when monomorphic markers were discarded from the analysis. Thus, 13.6% or 12.9% of the total variation was due to differences in allele frequencies among populations. These statistics are identical for both hierarchies tested. Subpopulations, as defined by 'four group's, yielded a greater F S T value than subpopulations defined by 'tree species'. On the whole, pooling populations according to 'four group's increased differentiation (among subpopulations relative to total variation) by approximately 2.0 % (compare adjusted F S 2 T values to ¥ s l T values in Table 14). Accordingly, the values for F D S ] were smaller than the values for FD S 2. This statistic compares demes within subpopulations ( S i and S2) and thus indicates the level of differentiation of demes within each subpopulation. Variance components followed the same trends as those shown for the F-statistics. The F-statistic trends discussed above are visualized in Figure 27 (60A) and Figure 28 (60B). In Figure 27C the general trend of increasing differentiation when subpopulations are defined according to the 'four group' scenario as opposed to the 'tree species' scenario is o a* P to hd "° 3*' c a 5 3 cn 00 T3 O C &' o 3 00 1 00 T3 § O c gf o 3 cn OO cn —) VO 4^ O to - J ON 13 s? T3 g <-»• 5' 3 cn - J ON u i Ul i— 1 to NO bo 0 to I—1 to Ul ON 0 0 0 0 to to to to ON Ul to to to to NO I — 1 s—N ^ — N 0 0 0 0 0 0 0 0 1—» 0 0 t—* to Ul NO 0 0 0 0 to to to to 00 00 ON U) 00 I — 1 ON NO ^ — ^ ^ — N / — 0 0 0 0 0 0 0 0 0 0 U l 00 w o c ON O > ON o 3 t-1 0s-0 T l 0 . O cr 3 0 B: 0 ON O > o 3 ON O td 8SI 159 Table 14. Variance components and F-statistics combined across all loci for total data sets 60A and 60B. Comparison8 Variance Component FXY X Y 60A 60B 60A 60B Demes- s, 1.696 1.789 0.104 0.099 Demes- s2 1.219 1.216 0.077 0.070 Demes- Total] 2.289 2.408 0.136 0.129 Demes- Total2 2.292 2.410 0.136 0.129 S,- • Total, 0.594 0.619 0.035 0.033 s2- Total2 1.073 1.194 0.064 0.064 Adj: S, Total, 0.787 0.819 0.049 0.050 Adj: S2 Total2 1.167 1.283 0.069 0.071 a: Subscripts as follows - 1 = hierarchy defined by tree species; 2 = hierarchy defined by four grOUpS: SSMainland, SSvanls.j Se-SWcomplex, jPine. A. F X y - Denies - Tree Species (Si) or Groups (S2) 0.00 - 0.049 0.05 - 0.149 0.15 - 0.25 > 0.25 Categories of F statistics C. FXY = Tree species (Si) or Groups (S2) - Total 0.00 - 0.049 0.05 -0.149 0.15 -0.25 > 0.25 Categories of F statistics Figure 27. Distribution of F X Y statistics for total data set 60A. A. F X Y = Demes - Tree species ( S O or Groups (S 2 ) 0.00 -0.049 0.05 - 0.149 0.15 - 0.25 > 0.25 Categories of F statistics B. FXY = Demes - Total 0.00- 0.049 0.05 - 0.149 0.15 - 0.25 > 0.25 Category of F statistics C. F X Y = Tree species ( S i ) or Groups ( S 2 ) - Total 0.00 - 0.049 0.05 - 0.149 0.15 - 0.25 > 0.25 Categories of F statistics ure 28. Distribution of F X Y statistics for total data set 60B. 162 reflected in the number of loci which showed FST's greater than 0.15. In the 'tree species' hierarchy only 4 loci exhibited FST's greater than 0.15. In the 'four group' hierarchy, 8 loci exhibited FST's greater than 0.15. Conversely, the number of loci showing F D S values greater than 0.15 fell from 17 loci (S,) to 12 loci (S2) (Figure 27A). When monomorphic markers were removed from the analysis the same general trends were seen (Figure 28). Goodness of fit statistics (cophenetic correlation coefficients) for the three different clustering methods on the different resemblance coefficients are shown in Appendix X. Nei's genetic measures could not be utilized with the Distance Wagner tree building algorithm since Nei's measures are nonmetric, as clearly stated in the BIOSYS-1 manual (Swofford 1989). The UPGMA clustering method almost always performed better than the single-linkage method since cophenetic correlations were higher. The Distance Wagner procedure appeared to perform marginally better than UPGMA and showed the highest cophenetic correlations coefficients with 8 of the 12 data sets this procedure was used on. In terms of resemblance coefficients, Nei's I generally yielded the lowest cophenetic correlations. Cavalli-Sforza and Edward's D and Prevosti's D performed similarly overall but in certain analyses the Cavalli-Sforza and Edwards distance yielded slightly higher cophenetic correlation coefficients (e.g., total data set 60A, UPGMA) and in other situations Prevosti's performed better (e.g., total data set 60A, Distance Wagner procedure) (Appendix X). Even given the trends based on the goodness of fit statistics basically only two different trees were seen. The relationships amongst the populations revealed in these two basic trees are shown in Figure 29 and Figure 30. Figure 29 shows the dendrogram derived from UPGMA clustering on Nei's unbiased genetic identity matrix (total data set 60A). This tree shows the Sitka spruce populations from Vancouver Island as being quite different from the Sitka spruce populations obtained from mainland, B.C. Additionally, the jack pine populations clustered together and the populations taken from Engelmann and white spruce 163 .90 Similarity .92 .93 .95 .97 .98 1.00 f + + + + + + + + + + E l _ G O L D E N - 2 W l _ M C B R I D E - 2 6 W 2 _ M C B R I D E - 4 & 5 E 2 _ G O L D E N - 1 1 9 W 3 _ P R I N C E G E O R G E P 2 _ G O G A M A 2 Pl_GOGAMAl S V A N I G O L D R T V E R S V AN 2 _ N G O L D R T V E R S V A N 3 _ T O F I N O SMII_KrTTMATn S M I KITEVIATI .90 .92 .93 .95 .97 .98 1.00 Cophenetic correlation coefficient = .852 Figure 29. Dendrogram showing the phenetic relationships among Pissodes strobi populations based on Nei's unbiased genetic identity and the U P G M A clustering method (total data set 6 0 A ) . 164 Distance -20 .17 .13 .10 .07 .03 .00 + — + — + — + — + — + — + — + — + — + — + — + — + El_GOLDEN-2 E2_GOLDEN-119 Wl_MCBRIDE-26 W2_MCBRIDE-4&5 W3_PRINCEGEORGE P2_GOGAMA2 Pl_GOGAMAl SMII_KnTMATn SMI_KITIMATI S V AN 1 GOLDRIVER SVAN2_NGOLDRIVER SVAN3 TOFINO .20 .17 .13 .10 .07 .03 .00 Cophenetic correlation coefficient = .902 Figure 30. Dendrogram showing the phenetic relationships among Pissodes strobi populations based on Prevosti's genetic distance and the UPGMA clustering method (total data set 60A). 165 clustered together (intermixed). Similar clustering was obtained when Prevosti's distances were used (UPGMA Figure 30) but the Engelmann and White populations were slightly more distinct since they did not intermix. The same population trends were seen with data set 60B (Figure 31 and Figure 32). In all dendrograms, the two samples obtained from the same stand near Kitimat (SMI_KITIMATI and SMII_KITIMATII) clustered together, but sampling effects were evident since distances were not zero or similarities not one. Appendix X I illustrates how the Distance Wagner procedure and gender sorted data sets yielded the same population clustering as found with Prevosti's distance and the U P G M A method. Differences between the males and the females were once again not evident, i.e., virtually identical dendrograms were produced (Appendix XI). Mean genetic identity values for the 'four group' scenario are presented in Table 15 (total data set 60A). Not unexpectedly, within group comparisons for the jack pine populations and the Sitka spruce mainland populations yielded the highest genetic identities. The Vancouver Island group was as similar to the SsMaij«nd group as to the Se-SwCompicx- Within group comparisons (means) were always higher than between group comparisons. Also, upper limits of between group comparisons only marginally overlapped with the lower limits of within group comparisons. The noted exception to this trend was the comparison between the jPine group and the Se-Swcompicx group; these groups showed considerable overlap in their ranges. Hence, Table 16 was constructed which combined these two groups into one. As in the 'four group' scenario, within group comparisons (means) were substantially higher than between group comparison. Correlations between genetic distances (determined in BIOSYS) and metric distances (Mahalanobis' distances obtained from canonical discriminant analysis) were quite high (Figure 33A,B and Figure 34A,B). Genetic distances failed the test of normality and thus the Spearman rank order correlation coefficients have been reported. Values for these coefficients ranged from 166 .90 Similarity .92 .93 .95 .97 .98 1.00 E l _ G O L D E N - 2 W l _ M C B R I D E - 2 6 W 2 _ M C B R I D E - 4 & 5 E 2 _ G O L D E N - 1 1 9 W 3 _ P R I N C E G E O R G E P 2 _ G O G A M A 2 Pl_GOGAMAl SMI1_KITIMATII S M I _ K I T T M A T I S V A N I G O L D R T V E R S V A N 2 _ N G O L D R T V E R SVAN3 T O F I N O + — .90 .92 .93 .95 .97 .98 1.00 Cophenetic correlation coefficient = 0.910 Figure 31. Dendrogram showing the phenetic relationships among Pissodes strobi populations based on Nei's unbiased genetic identity and the U P G M A clustering method (total data set 60B) . 167 Distance .20 .17 .13 .10 .07 .03 .00 ************************* * «» ft*****************-** M M * * * * * * * * * * * * * * * .20 .17 .13 E l _ G O L D E N - 2 E 2 _ G O L D E N - 1 1 9 W l _ M C B R I D E - 2 6 W 2 _ M C B R I D E - 4 & 5 W 3 _ P R I N C E G E O R G E P 2 _ G O G A M A 2 Pl_GOGAMAl SMII_KITIMATn SM1_KITIMATI SVAN I G O L D R T V E R SVAN2_NGOLDRTVER SVAN3 T O F I N O .10 .07 .03 .00 Cophenetic correlation coefficient = .940 Figure 32. Dendrogram showing the phenetic relationships among Pissodes strobi populations based on Prevosti's genetic distance and the UPGMA clustering method (total data set 60B) . 168 Table 15. Matrix of Nei's unbiased genetic identities (1978) as inferred from RAPD data. Values between populations of P. strobi in 4 different groups, averaged across all pairwise comparisons between populations, shown. Group SSvan.Is. Group8 S Sjyfa i n l a n d Se-SWcomplex jPine SSvan.Is. (3 populations) 0.968 (0.958-0.987) S ^ M a i n l a n d (2 populations) 0.932 (0.912-0.948) 0.977 Se-SWcomplex (5 populations) 0.932 (0.900-0.946) 0.928 (0.919-0.944) 0.973 (0.956-0.991) jPine (2 populations) 0.936 (0.913-0.956) 0.927 (0.915-0.932) 0.960 (0.940-0.981) 0.981 a: Ranges indicated in brackets 169 Table 16. Matrix of Nei's unbiased genetic identities (1978) as inferred from RAPD data. Values between populations of P. strobi in 3 different groups, averaged across all pairwise comparisons between populations, shown. Group' Group SSvan.Is. ^ ^Mainland Sw-Se-jPine .SSvan.Is. (3 populations) 0.968 (0.958-0.987) SSMajnland (2 populations) 0.932 (0.912-0.948) 0.977 Sw-Se-jP (7 populations) 0.933 (0.900-0.956) 0.928 (0.915-0.944) 0.967 (0.940-0.981) a: Ranges indicated in brackets 170 A. B. 1.00 0.98 H y 0.96 H H l -H H § e I—I H 43 O Q w H l -H H g e o l -H H 8 a 00 < oo 0.94 H 0.92 oo a 0.90 0.88 • • R^-0.77 • • •V* • • • • > • • • • 0 20 40 60 80 100 120 140 MAHALANOBIS' DISTANCE 1.00 0.98 0.96 0.94 0.92 0.90 0.88 * • • • ^ = -0.78 • • r r i i i i 0 20 40 60 80 100 120 140 MAHALANOBIS' DISTANCE Figure 33. Correlation between Nei's unbiased genetic identity and Mahalanobis' distance. A: total data set 60A; B: total data set 60B; and R s = Spearman Rank Order correlation coefficient. 171 A. B. u H oo Q U H O oo H O > o H Q u H O oo H oo O > 3 i 1 r 0 20 40 60 80 100 120 140 MAHALANOBIS' DISTANCE 0 20 40 60 80 100 120 140 MAHALANOBIS' DISTANCE Figure 34. Correlation between Prevosti's distance and Mahalanobis' distance. A: total data set 60A; B: total data set 60B; and R s = Spearman Rank Order correlation coefficient. 172 0.82 to 0.77 between Prevosti's distances and Mahalanobis distances based on data set 60B and data set 60A, respectively. 4.3 DISCUSSION The variety of analytical methods applied to the RAPD data, in general, all tended to reveal similar population trends in Pissodes strobi. Results from methods utilizing binary coded data and allele frequency data were consistently in agreement and correlations between metric distances (Mahalanobis') and genetic distances were very strong. Principal component analysis was the first analytical method explored and was ideal as an exploratory aid. Although observations did not form tight, non-overlapping groups, general ordination of the observations was immediately revealed. Observations from Sitka spruce tended to form their own unique cloud, and jack pine observations also tended to group together but were closely associated with the Engelmann and white samples. Since no one PCA axis accounted for greater than 8.05% of the total variation, very little redundancy in the data was demonstrated (i.e., overall the variables were not accounting for the same variation). This may be an indication that the variables were highly uncorrelated and a visual examination of the correlation matrices certainly suggested this. Indeed, as Manly (1992) so prudently stated, very little of the total variation will be accounted for by each axis if the variables are uncorrelated. Correlation amongst the variables is an important issue with genetic data since if variables (alleles) are correlated they may be linked. Ruling out correlation eliminates the confounding factor of linkage and hence linkage disequilibrium which could tend to bias results (Felenstein 1974). PCA trends were virtually identical between data sets 60A and 60B. Thus, even though warnings were 173 given in the SAS log output files, the analysis did not appear to be overly sensitive to the inclusion of monomorphic markers. Allele frequencies are quantitative variables and thus consequently better suited for PCA as stated in the SAS manual (SAS institute 1989), but they could not be utilized herein since only 12 populations (observations in this case) existed. With 60 variables included in the analysis the general 3:1 minimum rule (ratio of number of observations to variables) (Maze pers. comm.4) would have been grossly violated and the resulting output may have been highly suspect. Thus, it was decided to utilize the discrete data for PCA. Social scientists, as well as ecologists, often utilize binary coded data to conduct PCA and have full confidence in the trends revealed (Maze pers. comm.4). Stepwise discriminant analysis indicated that 36 of the 69 RAPD markers were useful for discriminating among the 12 populations. Twenty-three of these 36 loci showed FST's (total data set 60A) values greater than 0.10 (in the moderately differentiated range) in the population genetic analysis (comparison not given). In fact, five of the top seven most significant discriminators were the same loci which gave FST's in the extremely differentiated range (greater than 0.25). These results, along with the highly significant correlations shown between Mahalanobis' distances and genetic distances, provide compelling evidence that the choice of analyzing RAPD data from a binary perspective versus a population genetic perspective is inconsequential if strong structure in the data exists. These results also suggest that performing a parametric procedure on discrete data does not necessarily produce meaningless results. Manly (1992) has indicated that excellent discrimination may be possible on non-normal populations but it simply may not be possible to firmly establish the significance of the results (page 90). Interestingly, in the stepwise discriminant procedure (among populations) variables entered into the model were never removed. This suggests that very little interaction between the variables 4 Professor, UBC, Botany Department, Faculty of Science 174 was present and supports the random nature of the RAPD technique (i.e., in randomly amplifying regions of the genome). Ballinger-Crabtree et al. (1992) utilized the nearest-neighbor procedure (PROC NEIGHBOR IDENTITY) to select a subset of markers which were useful for population discrimination. I am not familiar with the identity option for this procedure and it was not obvious in the SAS manual I used (1989 version). Hence, I did not investigate this procedure. None-the-less this statistical technique may be extremely useful in obtaining a subset of worthwhile discriminators and seems worthy of future exploration. Selection of k appeared to be relatively unimportant in the nearest-neighbor analysis. When k was set at two, slightly higher misclassification rates did occur relative to k set between three and 12 but on the whole differences in mean error rates were minor. Since the selection of k is an option in the SAS program and because I had no prior experience with this procedure I felt it necessary to examine misclassification rates based on different k's. Other authors, such as Khambhampati et al. (1992), did not report on their selection of k perhaps because they viewed it as a relatively inconsequential point. Hand (1982) stated (page 234) that 'the choice of k is usually relatively uncritical'. This conjecture was more-or-less confirmed in this study but it is suggested that future investigators utilizing the nearest-neighbor analysis set k at two since this yields slightly higher error rates and is thus the most stringent condition. Khambhampati et al. (1992) used nearest-neighbor analysis to correctly classify unknown individual mosquitoes into one of five different species. They reported that unknowns were 100% correctly classified. I assume that their initial data set (four individuals from 5 populations) was used to develop the classification rule and unknowns were used to validate this classification scheme. This was not stated and perhaps should have been. Ideally, one wishes to develop the classification criterion with the entire data set and test this classification with an entirely independent, different sample. By so doing one achieves a degree of confidence in the error rates 175 since, as Dillon and Goldstein (1984) indicated, error rates generated from samples used to both define and evaluate the classification rule are optimistically biased. This statement was certainly confirmed herein, as clearly shown in Figure 19; mean error rates for the criterion data set were 30 to 40% lower than rates for the test data. An independent sample was not available to me and hence the 'hold-out method' described by Dillon and Goldstein was employed. Other popular and more robust methods for estimating errors of misclassification include the jackknife and bootstrap methods (Dillon and Goldstein 1984). Unfortunately, neither method is an option in the SAS program and it is simply unfeasible to perform by creating different data sets via spreadsheets. Confidence in nearest-neighbor error rates may also be gained when different partitionings of the data set yield virtually the same results. As shown in Figure 20, this was indeed the case herein and hence conclusions based on the nearest-neighbor analysis should be relatively sound. Different groupings of the populations were explored in the nearest-neighbor analysis as a type of clustering method. The groupings which yielded the lowest overall error rates were assumed to cluster the populations the best. Based on this premise the best assemblage of populations occurred with the 'four group' scenario (Figure 21). Restricting the discussion to k set at two, mean error rates decreased from classification according to population to classification according to the 'four group' scenario. One could argue that as the number of groups in which to place observations for classification decreased so too would the mean error rates decrease (with one group ultimately resulting in 100% of the individuals being correctly classified!). This trend of decreasing error rates with decreasing the number of groups in which to place observations was somewhat evident in Figure 21. However, mean error rates for the 'tree species' grouping were approximately 20% higher than those obtained from the 'four group' scenario (and both of these classified observations into one of four groups: by 'tree species' into either Se, Ss, Sw or jP; by 'four groups' into either Ssvan.is., SsMainiand, Se-Swcompiex or jPine). The 'four group' scenario 176 yielded the lower error rates due to the combining of the Se and Sw weevil populations and separating the Ss Vancouver Island and Ss Mainland weevil populations (compare error rates between classification according to 'tree species', five group and 'four group' scenario in Figure 21). Furthermore, when the Ss weevil populations were combined, mean error rates were not substantially different than for the 'four group' situation. This indicates that observations from S s v a n . i i . had not been erroneously identified as S s M a W a n d samples (or vice versa) and, indeed, visual examination of the output confirmed this finding. The mean error rates for random selection of loci were always greater than for nonrandom selection of loci (Figure 22). This suggests that if the investigator has an a priori reason for utilizing a specific subset of variables the results will probably look better than by simply exploring subsets at random. The trend seen for mean error rates based on nonrandom selection of loci is a reflection of the scoring procedure. Mean error rates were actually quite low when only the 22 variables scored with the highest confidence were included in the procedure. In fact, mean error rates were only 5% higher with the 22 high confidence variables than when the 36 variables identified in the stepwise procedure were used (or for that matter than when the entire complement of 69 variables were used). Thus, if time is of the essence, and the investigators do not have access to a computer facilitated scoring system they may wish to only examine RAPD markers which can be very easily scored (i.e., between one and three markers from each primer used). I offer this as a practical recommendation since visually scoring RAPD markers from photographs can be an extremely difficult, tedious, frustrating and time consuming task. Increasing the number of loci used in the nearest-neighbor analysis did more-or-less increase in discrimination power (Figure 22). This supports the findings from the principal component analysis which indicated that the variables were highly uncorrelated and hence accounted for different measures of variation. Since addition of loci resulted in a decrease in 177 misclassification, useful information was gained when using many opposed to few variables. The exception to this trend was when the 36 loci identified in the stepwise procedure yielded virtually the same low mean error rates as when the full complement of 69 variables were used. These results, once again support the contention that performing parametric procedures on variables which are discrete, statistically speaking, can be very useful. Maxwell (1961) and Claringbold (1958) also held this opinion. Claringbold (1958) showed that parametric multivariate procedures were totally applicable using discrete variables. Maxwell (1961) applied the canonical variate procedure (canonical discriminant analysis) to a psychiatric classification problem involving dichotomous variables. Examination of scores plotted on canonical variates one and two compelled Maxwell (1961) to conclude that patients grouped into three separate categories. He found this technique extremely useful and recommended that canonical variate analysis could be employed when the variables were dichotomous. Even given the contentions of Claringbold (1958) and Maxwell (1961) and the findings discussed above, I did not restrict subsequent analysis to the 36 loci indicated from the stepwise procedure since I wished to err on the side of conservatism. I also did not restrict analysis to merely the 22 loci scored with highest confidence since I went to the trouble of scoring all these markers and error rates did decrease slightly when more loci were included. Nearest-neighbor analysis was initially selected as a statistical procedure for analyzing the RAPD data for the following two reason: 1) Khambhampati et al. (1992) had successfully used the technique to classify individual mosquitoes into one of five species (and also achieved correct classification according to the populations from which they arose); and 2) estimated error rates were viewed as being perhaps more representative of the true error rates (relative to those which would have been generated in parametric discriminant analysis) since assumptions pertaining to the distribution of the variables were not in any sense violated. 178 Plots of individual scores on PCA and canonical discriminant axes allowed for examination of the spread of observations within each population. For this reason alone PCA and canonical discriminant analysis were worth pursuing since within population variation cannot be determined using allele frequency data. In terms of Figure 23 and Figure 24, a cursory examination revealed that, although observations belonging to the P2GOGAMA2 were the most spread out, all populations showed a considerable degree of genetic variation. This indicates that populations are not that highly inbred and thus population sizes are probably quite large. The purpose of canonical discriminant analysis is to maximize between group variation while minimizing within group variation. Hence, the groups/populations are forced as far apart as possible, within the constraints of maximizing the F ratio of between-group to within-group variation. The two samples obtained from the Ss stand near Kitimat B.C. (SMI_KITIMATI and SMII_KITIMATII) thus served as a powerful positive control, not only for the molecular technique but also for the statistical procedures. Even though the canonical discriminant procedure attempted to separate SMIKITIMATI and SMIIKITIMATII populations group means were virtually identical (in all canonical plots displayed) and individual observations were entirely intermingled. These results lend credence to the RAPD marker technique and to the use of dichotomous variables in parametric procedures. Interestingly, the two group means from populations P1GOGAMA1 and P2_GOGAMA2 were quite dissimilar. This was unexpected since the weevils had been collected from the same jack pine stand. Since sampling was undertaken over a two year period, with the first year's collection consisting of adult weevils collected from the plantation and the second year from clipped leaders, it is possible that the P1_G0GAMA1 population was actually a composite collection of weevils 'visiting' this plantation, perhaps for the purpose of mating. Thus, the difference in group means between P1GOGAMA1 and P2GOGAMA2 could be a reflection on 179 sexual recombination rates bringing about changes in genotypes. Conversely, differences in group means between Pl_GOGAMAl and P1GOGAMA2 could simply be a sampling effect. Support for this statement is provided in Figure 25 and Figure 26 which shows the canonical plots produced from gender sorted data sets. Although the basic associations between the group means remained the same as when the combined data sets were used, separation of the group means was more pronounced. The overall ordination of observations depicted in the PCA plots was also reflected in the plots of canonical axes two versus canonical axis one (Figures 23 through 26 inclusive). Group means from the three Vancouver Island populations were very similar and quite different from the group means obtained from the SsMainiand populations. Group means from the Engelmann and white spruce populations were located close to each other. Group means from the jack pine samples were most similar to the Engelmann and white spruce populations but formed a distinct group of their own. These results once again support the 'four group' scenario as being the best clustering of the populations. The average proportion of polymorphic loci is a function of the total number of loci studied. Mean expected heterozygosity computed directly from the allele frequencies, does not depend on the total number of loci studied and could thus be viewed as a better measure of genetic variability within populations since it is unaffected by sampling effects (Li and Graur 1991). Percentage of polymorphic loci and mean expected heterozygosity were always higher for data set 60B relative to data set 60A. Overall means were 67.65 and 75.97 for percentage of polymorphic loci and 0.243 and 0.271 H (data sets 60A and 60B respectively). Trends in percent polymorphism and mean expected heterozygosity between the data sets were as expected, since data set 60B had monomorphic loci removed. Results from data set 60A reflect overall genetic variation better given that variation should take into consideration the extent of monomorphic 180 markers. Upper limits for percentage of polymorphic loci and H were slightly higher than those reported by Avise (1994). He summarized results from numerous allozyme studies and found a range of 0 to 80% for percentage of polymorphic loci and a range of 0.00 to 0.20 for mean expected heterozygosities. Lu and Rank (1995) also showed values higher than those reported by Avise (1994) in their RAPD marker study of five geographically isolated populations of Megachile rotundata (alfalfa leaf cutting bee). Mean heterozygosity within the bee populations was 0.3305. This was 10-fold higher than previously estimated from allozyme markers. The 'four group' scenario showed a clear trend of increasing diversity from west to east (Table 13): SsMainiand exhibiting the lowest H and % polymorphic loci and jPine the highest. These findings support the hypothesis that P. strobi originated in the east and migrated west; genetic diversity is traditionally higher in the ancestral populations with population subdivision resulting in the loss of heterozygosity (Wright 1978). In terms of percentage of polymorphic loci, the trend of increasing variation to the east is in complete agreement with that reported by Phillips and Lanier (1985) in their study on allozyme variation in Pissodes strobi (Table 17). Their estimates for percentage of polymorphic loci ranged from 33.4% to 53.5 % and were hence considerably lower than my findings, but, as indicated above, this statistic is a function of the number of loci studied (only 11 allozyme markers were examined). Phillips and Lanier's (1985) estimates for mean expected heterozygosities did not show the same pattern as that seen for percentage of polymorphic loci (Table 17). Although the eastern populations did show higher values than the Pacific coast populations, the Rocky mountain groups (Engelmann and white spruce populations) showed the highest H values of all. Also, H values were much lower than that obtained from RAPD markers (Table 17) and this supports the contention of Lu and Rank (1995) that genetic variation at the DNA level is traditionally much more prominent than at the protein level. 181 Table 17. Comparison of population genetic statistics obtained from allozyme markers (Phillips and Lanier 1985) and RAPD markers (total data set 60A - monomorphic markers retained). ALLOZYMES RAPD MARKERS Percentage of Polymorphic Loci Pacific Coast 33.4 S S Ma inland SSvan.Is. 59.15 62.23 Rocky Mountains Northeast 40.0 53.5 Se-S\Vcomple: jPine 71.02 75.84 Mean expected Heterozygosity Pacific Coast 0.149 Rocky Mountains Northeast 0.186 0.182 S Sjvlainland Ssv an.Is. Se-SWcomple: jPine Wright's F-statistics (1978) (Means over all loci) 0.098 Demes to Total Demes to Total (Between Populations) Regions to Total 0.025 (Between Regions) Demes Within Regions 0.075 0.221 (SE = 0.014) 0.229 (SE = 0.009) 0.252 (SE = 0.005) 0.262 (SE = 0.012) 0.136 Groups(S2) to Total3 0.071 Demes Within Groups (S2) 0.077 Range and means of Nei's genetic identity (1972) Range 0.899 - 0.998 Mean 0.963 0.895 -0.986 0.940 a: Adjusted value (negative values replaced with zeros and mean recalculated) - unadjusted value = 0.064. 182 A degree of population subdivision was evident as indicated by the F D T values (Table 14). More importantly, differentiation among subpopulations for the 'four group' hierarchy was considerably higher than that of the 'tree species' hierarchy. In fact, F S 2 T estimates were double FSIT estimates, thus placing them in the category of moderate differentiation (0.05 to 0.149) as opposed to virtually no differentiation between the subpopulations (< 0.049). Frequency distributions of the number of loci which fell into each of the four differentiation categories effectively illustrated this point (Figure 27 and Figure 28). Loci were grouped into these four categories based on (somewhat) subjective values defined by Wright (1978). To my knowledge, no other author has presented frequency distributions of F-statistics in this manner probably because F-statistics have been traditionally based on a limited number of allozyme markers. Since 60 loci were examined herein, frequency distributions were deemed to be more informative than simply presenting overall means. Conversely, I did not report FXY's derived from each of the 60 loci studied (or allele frequencies which is quite common in allozyme studies) since this would have been simply overwhelming to the reader and a general principle in science is to effectively summarize the data. The allele frequencies in each of the 12 populations for the seven loci which had F D T values greater than 0.25 (thus indicating extreme differentiation between the populations) are presented in Appendix XII. These seven loci were 376-d, 376-f, 322-g, 336-g, 333-f, 219-i and 219-j. Fxy's from these seven loci are also shown for both hierarchies tested (FST's not adjusted to correct for negative computations however). Examination of frequencies and FST's values for locus 376-d reveals why the 'four group' scenario was better for clustering the populations than clustering according to 'tree species'. Specifically, the frequencies of the dominant allele (p) for locus 376-d approached one in the SsMainiand populations and were lower than 0.5 in the SsVm.i». population. Frequencies of the dominant allele for this locus in the Se and Sw populations were also lower than 0.5. Combining the Se and Sw weevil populations and 183 separating the Ss Mainland and Vancouver Island weevil populations thus resulted in this locus becoming a highly informative marker in terms of differentiation between the subpopulations (i.e., FSIT = - 0.047, F S 2 T = 0.182). This same shift was revealed with markers 336-g, 333-f, and 219-j (compare F S T values in Appendix XII). Comparison of the weevil RAPD allele frequencies also illustrates the following statements by Wright (1978) pertaining to the interpretation of F S T : 'The fixation index is thus not a measure of degree of differentiation in the sense implied in the extreme case by absence of any common allele. It measures differentiation within the total array in the sense of the extent to which the process of fixation has gone toward completion.' So, the dominant allele of locus 376-d appears to be moving towards fixation in the two SsMainiand populations while in the other 10 populations it is the recessive allele which is moving towards complete fixation (Appendix XII). The F-statistic procedure of Wright (1978) effectively captured and measured this fixation differentiation. While F-statistics are useful for examining the degree of population subdivision ultimately it is dendrograms based on resemblance coefficients that show at a glance if and how populations cluster together (Romesburg 1984). When distances or similarities based on genetic data are used as the resemblance coefficients, inferences pertaining to genetic divergence between populations can be made. Three different genetic resemblance coefficients (Nei's unbiased genetic identity, Prevosti's and Cavalli-Sforza and Edward's) were explored with three different clustering methods (UPGMA, single-linkage and the Distance-Wagner procedure) in this study. Regardless of which resemblance coefficient or clustering method was used SMIKITLMATI and SMII_KITIMATII clustered together as did all three of the populations collected from Vancouver Island. Additionally, Pl_GOGAMAl and P2_GOGAMA2 grouped together and were very closely associated with the Se and Sw weevil populations. Approximately half the trees showed the Se and Sw weevil populations grouping separately, while the remainder of the trees showed 184 the Se and Sw weevil populations to intermix. Similar groupings in all cluster analyses support the contention that strong structure exists in the data. This conclusion was also reached by Ballinger-Crabtree et al. (1992) with RAPD data. They utilized the single-linkage and UPGMA clustering methods on distance matrices based on both binary data and allele frequency data (Nei's unbiased genetic distance). All three dendrograms showed that the 11 mosquito populations grouped correctly according to the two subspecies from which they were collected (i.e., two major branches revealed). Identification of unknown groups at the population level, however, only occurred when the single-linkage method was used on the SI matrix. Initially, I too wished to explore cluster analysis based on distances determined from band sharing but, since overall grouping of the 12 populations was quite consistent (and overall trends were virtually identical in all types of analyses undertaken) I suggest that performing additional cluster analysis in this study is not warranted. Clustering of P. strobi populations was not shown in Phillips and Lanier's (1985) allozyme study. They also wished to determine if P. strobi populations would group according to some factor such as geography. Examination of their genetic identities (based on information from 11 allozyme loci as presented in Appendix XIII) showed that ranges for comparisons between regions overlapped with ranges for comparisons within regions. This was definitely not the case with RAPD markers where genetic identities were higher for comparisons within groups as compared to comparisons between groups. Although I did not initially perform analysis using Nei's genetic identity (Nei's unbiased identity first explored) values for this similarity measure derived from my RAPD data are presented in Appendix XIII for comparative purposes. When populations were pooled according to the 'four group' scenario very little overlap between within group comparisons and between group comparisons was seen. The one exception to this was with the comparison between the jPine group and the Se-SecomPiex- Thus, there may be a basis for 185 combining these two groups. When these groups were combined, and the reduced (three group scenario) pairwise comparison matrix computed, the trends shown from the RAPD data in Appendix XIII were directly reflected in the dendrograms produced. It is the value of the genetic distance (or similarity) which determines the level of evolutionary divergence for the organisms under investigation. Perring et al. (1993) indicated that for vertebrates and invertebrates the range of D values (Nei's estimate, 1972) would be 0.00 to 0.05 for comparison between populations or races (hence approximately 1.0 to 0.95 for Nei's I), 0.02 to 0.20 between subspecies, 0.1 to 2.0 between species and more than 1.0 between genera. In terms of entomological investigations, Ayala et al. (1975) recognized and quantified five levels of evolutionary divergence in the Drosophila willistoni group. They reported mean D's (Nei 1972) to be 0.031 ± 0.007 between geographic populations, 0.230 ± 0.016 between subspecies, 0.226 ± 0.033 between semispecies, 0.581 ± 0.039 between sibling species and 1.056 ± 0.068 between non-sibling species. Clearly, Nei's genetic distance estimated in this study did not indicate genetic divergence near the species level. Thus, I support the findings of Phillips and Lanier (1985) and conclude that populations of Pissodes strobi belong to a single species. This conclusion is important from a RAPD perspective, since, as stated by Black (1993) similarly sized fragments amplified between two species may not be homologous. Since genetic distances were indicative of an intraspecific investigation strict cladistic analysis was not applicable herein since each branch on a cladogram indicates the evolution of a new species. Also, when dealing with incipient differentiation within a species it is the degree of the fixation process which is of primary interest (Wright 1978) as reflected in the F-statistics. Although I have concluded that the populations of P. strobi examined herein are conspecific an argument could be made for genetic divergence at the subspecies level. If one combines the jPine group with the Se-SwCompiex, the overall mean D value (Nei's unbiased genetic 186 distance) would be approximately 0.07 (results not shown). This value was similar to that reported by Ballinger-Crabtree et al. (1992) with RAPD markers for the joining of two subspecies ofAedes. Comparison of Nei's values for subspecies divergence and the findings of Ballinger-Crabtree et al. (1992) supports the contention that the SsMaWand group and SsVan.is. group could be subspecies of the ancestral P. strobi (which encompasses the jPine populations and the Se-Swcompiex)- I, however, am not prepared to conclude that we are dealing with different subspecies. In fact, I consider the point somewhat inconsequential since there is no general consensus regarding at what divergence value subspecies should be recognized and hence named (Avise 1994). What is paramount is that RAPD markers have shown that a considerable degree of divergence exists among P. strobi populations. British Columbia populations of P. strobi were seen to group into three clusters in virtually every analysis undertaken. For management purposes it is thus recommended that within B.C. P. strobi be treated as three distinct groups: one group on Vancouver Island, one Sitka spruce group on mainland B.C. and the other group residing on Se and Sw. Populations isolated by distance are known to accumulate genetic differences as they adapt to different environments (Avise and Smith 1977) and so the findings that the interior populations are very different from the coastal populations make perfect sense. Certainly weevils in the interior of B.C. are in quite a different environment than those in coastal B.C. Indeed, genetic differences attributed to adaptation to different environments is reflected in the different development heat sums of the interior and coastal weevil groups (McMullen 1976a). Although developmental heat sums for the interior weevils were calculated based on emergence from white spruce the results herein support the contention that this sum is probably also applicable to weevils attacking Engelmann spruce (assuming all other factors are equal). Genetic divergence may also be reflected in different overwintering behaviors. Gara et al. (1971) found that Sitka 187 spruce weevils (in western Washington) apparently fed on laterals and stems during the winter months (on warm days). This is counter to the behavior in eastern Canada where the white pine weevil undergoes a period of complete dormancy during the winter months (Belyea and Sullivan 1956, Sullivan 1959). Different development heat sums and overwintering behavior does not help to explain the divergence between SsVan.is. and SsMainiand weevil groups. The SsVan.is. is probably a result of a founder effect although one would expect this group to exhibit the least genetic variation. This is not the case; the SsMainiand group exhibits slightly smaller values of H and mean % polymorphism (as shown in Table 13). The slightly higher values for these statistics for the Ssvanis (although perhaps not significantly different from those for the SsMainiand group) could be explained if quite a large number of weevils initially formed this group. Overall, statistics for the SsMainiand and SsVan.is. are quite similar though, particularly for genetic distances between these two groups and the Se-SwcomPiex- (Table 15). Genetic distances presented in Table 15 suggest that the SSMainland and SSyan.Is. groups diverged at approximately the same time from the Se-SwCompiex- Also, the general conclusion that the Se-Swc0mPiex is most similar to the jPine populations supports the results of Alfaro (1988). In Alfaro's study P. strobi collected from Engelmann and Sitka spruce were forced to oviposit in Lodgepole pine leaders. Although both groups of weevils successfully oviposited in the pine leaders, P. strobi larvae produced from weevils collected from Engelmann spruce were significantly larger than the larvae produced from weevils obtained from Sitka spruce. Alfaro's results make sense in light of the fact that the Se-Swc0mPiex has been shown herein to be genetically very similar to the eastern weevil populations which primarily attack species of pine. Although I do not recognize different subspecies of weevils a clear distinction should be made between weevils attacking Sitka spruce and all other trees hosts. I propose the use of two common names: the Sitka spruce weevil, for populations attacking Sitka spruce and the white pine weevil for all other populations. Additionally, when referring to the Sitka spruce weevil should specify whether the group pertains to Vancouver Island or mainland B.C. 189 CHAPTER 5: OVERALL CONCLUSIONS AND RECOMMENDATIONS RAPD Methodology 1. When using the bulked DNA technique to detect differences in genomic composition it is recommended that only 10 to 20 individuals are used to generate one bulked sample - any more than this could result in failure to detect any differences between the bulked samples. 2. RAPD assay conditions must be optimized for each unique organism (species) under investigation. In particular the exact concentrations of Mg2+, Taq, and DNA must be determined so that highly reproducible results will be achieved. Although DNA quality was not investigated in this study it is my contention (general observation from other people's studies in the RAPD laboratory) that this is also a critical parameter in accomplishing this goal. 3. To avoid observer bias and thereby eliminate measurement error RAPD markers should be scored by at least two different individuals or blind scoring (individual does not initially know which sample is being scored) should be done. Alternatively, scoring could be performed by a computer system if internal DNA standards were placed in each reaction tube prior to electrophoresis. Ideally, two standards would be used and these would be readily distinguished from the RAPD markers of interest. The use of two standards would help account for the 'waving' migration which often occurs during electrophoresis (i.e., adjacent lanes do not always migrate at the exact same rate). 4. Wherever possible DNA should be extracted from haploid tissue so that genotypes rather than phenotypes are observed. 190 Genetic Variation Based on Diploid RAPD Markers 1. Theoretically, genetic diversity estimates would be much more accurate and precise if between 500 and 1000 individuals per population were used. Since this target seems impossible to meet (unless an entirely automated system is used) 30 individuals is the absolute minimum to use -50 to 100 individuals would be better yet. 2. The inclusion of two samples from the same population serves as an excellent control for both the laboratory methodology and for the statistical analysis. It is recommended that similar studies also include this type of control 3. If computer facilitated scoring is not done it is suggested that only two or three RAPD markers should be scored from each primer used. These markers should be the brightest and most discrete bands. A total of 30 of these markers is probably quite sufficient for obtaining sound genetic variation estimates. 4. All of the multivariate procedures performed herein can be used to examine genetic variation, however, if time is of the essence I recommend that principal component analysis be used as a preliminary tool to examine overall structure in the data. This procedure should then be followed by nonparametric discriminant analysis (Identity option should be used to obtain a subset of discriminating fragments followed by estimation of mean error rates using k set at two). 5. If one does not wish to compute allele frequencies from diploid RAPD data genetic distances can be based on the binary coded data (e.g., Similarity Index based on shared bands). Cluster 191 analysis should follow and it appears that various clustering methods can by used. Goodness-of-Fit statistics must be reported. 6. If feasible (i.e., one has access to the software and the software has been developed) standard errors should be computed for the various statistics estimated. Management of Pissodes strobi in B.C. 1. Investigators should recognize three groups of Pissodes strobi in B.C.: one group on Vancouver Island, one group in the interior of B.C. attacking Se and Sw and the third group attacking Ss on the mainland. Future Direction of Research 1. Strictly speaking the RAPD marker technique should be used in a systematic context for only intraspecific investigations. However, it could also be used to investigate the relationships among the Pissodes strobi species group if comigrating RAPD markers were first sequenced and shown to be homologous: ideally, five different RAPD bands from five individuals from each of the four species should be sequenced. 192 BIBLIOGRAPHY Alfaro, R.I. 1988. Laboratory feeding and colonization of non-host Lodgepole pine by two populations of Pissodes strobi (Peck) (Coleoptera: Curculionidae). The Canadian Entomologist 120: 167-173. Alfaro, R.I. 1994. The white pine weevil in British Columbia: Biology and damage. Pp. 7-22 in: The White Pine Weevil: Biology, Damage and Management. FRDA Report 226, R.I. Alfaro, G. Kiss, and R.G. Fraser (eds.). Proceedings of a symposium held January 19-21, 1994 in Richmond, British Columbia. Alfaro, R.I., and J.H. Borden. 1980. Predation by Lonchaea corticis (Diptera: Lonchaeidae) on the white pine weevil, Pissodes strobi (Coleoptera: Curculionidae). The Canadian Entomologist 111: 1259-1270. Alfaro, R.I., and J.H. Borden. 1982. Host selection by the white pine weevil, Pissodes strobi Peck: Feeding bioassays using host and non-host plants. Canadian Journal of Forest Research 12: 64-70. Alfaro, R.I., and J.H. Borden. 1985. Factors determining the feeding of the white pine weevil on its coastal British Columbia host Sitka spruce. Proceedings of the Entomological Society of Ontario 116 (supplement): 63-66. Alfaro, R.I., J.H. Borden, R.G. Fraser, and A. Yanchuk. 1994. An integrated pest management system for the white pine weevil. Pp. 226-238 in: The White Pine Weevil: Biology, Damage and Management. FRDA Report 226, R.I. Alfaro, G. Kiss, and R.G. Fraser (eds.). Proceedings of a symposium held January 19-21, 1994 in Richmond, British Columbia. Alfaro, R.I., J.H. Borden, R.G. Fraser, and A. Yanchuk. 1995. The white pine weevil in British Columbia: Basis for an integrated pest management system. The Forestry Chronicle 71(1): 66-73. Alfaro, R.I., M.A. Hulme, and J.W.E. Harris. 1985. Insects associated with the Sitka spruce weevil, Pissodes strobi (Coleoptera: Curculionidae) in Sitka spruce, Picea sitchensis in British Columbia, Canada. Entomophaga 30(4): 415-418. Alfaro, R.I., M.A. Hulme, and C. Ying. 1993. Variation in attack by Sitka spruce weevil, Pissodes strobi (Peck), within a resistant provenance of Sitka spruce. Journal of Entomological Society of British Columbia 90: 24-30. Alfaro, R.I., and S.A.Y. Omule. 1990. The effect of spacing on Sitka spruce weevil damage to Sitka spruce. Canadian Journal of Forest Research 20: 179-184. Alfaro, R.I., HD. Pierce Jr., J.H. Borden, and A.C. Oehlschlager. 1980. Role of volatile and non-volatile components of Sitka spruce bark as feeding stimulants of Pissodes strobi Peck (Coleoptera: Curculionidae). Canadian Journal of Zoology 58: 626-632. Alfaro, R.I., and E. Wegwitz. 1994. The spruce weevil in the Green Timbers Plantations. Pp. 222-225 in: The White Pine Weevil: Biology, Damage and Management. FRDA Report 226, R.I. Alfaro, G. Kiss, and R.G. Fraser (eds.). Proceedings of a symposium held January 19-21, 1994 in Richmond, British Columbia. 193 Anderson, J.F., and L.A. Magnarelli. 1994. Lyme Disease: A tick-associated disease originally described in Europe, but named after a town in Connecticut. American Entomologist 40(4): 217-227. Apostol, B.L., W.C. Black IV, BR. Miller, P. Reiter, and B.J. Beatty. 1993. Estimation of the number of full sibling families at an oviposition site using RAPD-PCR markers: Application to the mosquito Aedes aegypti. Theoretical and Applied Genetics 86: 991-1000. Avise, J.C. 1994. Molecular Markers, Natural History and Evolution. Chapman & Hall, Inc., New York. 511 pp. Avise, J.C, and M.H. Smith. 1977. Gene frequency comparisons between sunfish (Centrarchidae) populations at various stages of evolutionary divergence. Systematic Zoology 26: 319-335. Ayala, F. J., M.L. Tracey, D. Hedgecock, and R.C. Richmond. 1975. Genetic differentiation during the speciation process in Drosophila. Evolution 28: 576-592. Ballinger-Crabtree, M.E., W.C. Black IV, and BR. Miller. 1992. Use of genetic polymorphisms detected by the random-amplified polymorphic DNA polymerase chain reaction (RAPD-PCR) for differentiation and identification of Aedes aegypti subspecies and populations. American Journal of Tropical Medicine and Hygiene 47(6): 893-901. Barnhart, C.L, and R.K. Barnhart (eds.). 1991. The World Book Dictionary. World Book Inc., Chicago. 2430 pp. Bartels, H. 1971. Genetic control of multiple esterases from needles and macrogametophtes of Picea abies. Planta 99: 283-289. Belyea, H.C. 1923. The control of white pine weevil (Pissodes strobi) by mixed planting. Journal of Forestry 21: 384-390. Belyea, R.M., and CR. Sullivan. 1956. The white pine weevil: A review of current knowledge. Forestry Chronicle 32:58-67. Berry, A.B., and W.M. Stiell. 1976. Control of white pine weevil damage through manipulation of stand climate: Preliminary results. Canadian Department of Environment, Canadian Forestry Service, Petawawa Forest Experimental Station, Internal Report PS-X-61. 8 pp. Black, W.C. IV. 1993. PCR with arbitrary primers: Approach with care. Insect Molecular Biology!. 1-6. Black, W.C. IV, N.M. DuTeau, G.J. Puterka, JR. Nechols, and J.M. Pettorini. 1992. Use of the random amplified polymorphic DNA polymerase chain reaction (RAPD-PCR) to detect DNA polymorphisms in aphids (Homoptera: Aphididae). Bulletin of Entomology Research 82: 151-159. Booth, D.C, and G.N. Lanier. 1974. Evidence of an aggregating pheromone in Pissodes approximatus and P. strobi. Annals of the Entomological Society of America 67: 992-994. 194 Booth, D.C., T.W. Phillips, A. Claesson, R.M. Silverstein, G.N. Lanier, and J.R. West. 1983. Aggregation pheromone components of two species of Pissodes weevils (Coleoptera: Curculionidae). Journal of Chemical Ecology 9: 1-12. Borror, D.J., D.M. De Long, and CA. Triplehorn. 1981. An Introduction to the Study of Insects. Fifth Edition. Saunders College Publishing, Philadelphia. 827 pp. Boving, A.G. 1929. Taxonomic characters for the identification of the mature larvae of Pissodes strobi Peck and Pissodes approximatus Hopkins (Fam. Curculionidae). Proceedings of the Entomological Society of Washington 31 (9): 182-187. Boyce, T.M., M.E. Zwick, and C.F. Aquadro. 1989. Mitochondrial DNA in the bark weevils: Size, structure and heteroplasmy. Genetics 123: 825-836. Boyce, T.M., M.E. Zwick, and C.F. Aquadro. 1994. Mitochondrial DNA in the bark weevils: Phylogeny and evolution in the Pissodes strobi species group (Coleoptera: Curculionidae). Molecular Biology and Evolution 11(2): 183-184. Bridgen, M.R., J.W. Hanover, and R.C. Wilkinson. 1979. Oleoresin characteristics of eastern white pine seed sources and relationship to weevil resistance. Forest Science 25(1): 175-183. Brooks, J.E., J.H. Borden, and H.D. Peirce, Jr. 1987. Foliar and cortical monoterpenes in Sitka spruce: Potential indicators of resistance to the white pine weevil, Pissodes strobi Peck (Coleoptera: Curculionidae). Canadian Journal of Forest Research 17: 740-745. Caetano-Anolles, G., B.J. Bassam, and P.M. Gresshoff. 1991. DNA amplification fingerprinting using very short arbitrary oligonucleotide primers. Bio/Technology 9: 553-557. Carlson, J., Y.-P. Hong, and G. Kiss. 1994. DNA markers associated with weevil resistance in interior spruce. Pp. 158-168 in: The White Pine Weevil. Biology, Damage and Management. FRDA Report No. 226, R.I. Alfaro, G. Kiss, and R.G. Fraser (eds.). Proceedings of a symposium held January 19-21, 1994 in Richmond, British Columbia. Carlson, J.E., L.K. Tulsieram, J.C. Glaubitz, V.W.K. Luk, C. Kauffeldt, and R. Rutledge. 1991. Segregation of random amplified DNA markers in Fl progeny of conifers. Theoretical and Applied Genetics 83: 194-200. Chalmers, K.J., R. Waugh, J.I. Sprent, A.J. Simons, and W. Powell. 1992. Detection of genetic variation between and within populations of Gliricidae sepium and G. maculata using RAPD markers. Heredity 69: 465-472. Chapco, W., N.W. Ashton, R.K.B. Martel, N. Antonishyn, and W.L. Crosby. 1992. A feasibility study in the use of random amplified polymorphic DNA in the population genetics and systematics of grasshoppers. Genome 35: 569-574. Claringbold, P.J. 1958. Multivariate quantal analysis. Journal of the Royal Statistical Society B 20 (2): 398-405. Cozens, R.D. 1983. The spruce weevil, Pissodes strobi Peck (Coleoptera: Curculionidae): A review of its biology, damage and control techniques with reference to the Prince George Timber Supply Area. British Columbia Ministry of Forests, Forest Service Internal Report PM-PG-3. 195 Cozens, R.D. 1987. Silvicultural control of damage by spruce weevil, Pissodes strobi in the central interior of British Columbia. British Columbia Ministry of Forests, Forest Service Internal Report PM-PG-6. 16 pp. de Groot, P., and B.V. Helson. 1993. Efficacy and timing of insecticide sprays for control of white pine weevil (Coleoptera: Curculionidae) in high-value pine plantations. Journal of Economic Entomology 86(3): 1171-1177. de Groot, P., and B.V. Helson. 1994. A review of chemical insecticides for control of Pissodes strobi (Peck). Pp. 270-284 in: The White Pine Weevil: Biology, Damage and Management. FRDA Report 226, R.I. Alfaro, G. Kiss, and RG. Fraser (eds.). Proceedings of a symposium held January 19-21, 1994 in Richmond, British Columbia. Dillon, W.R., and M. Goldstein. 1984. Multivariate Analysis: Methods and Applications. John Wiley and Sons, New York. 587 pp. Dillwith, J.A., and CE. Soos. 1994. Application of RAPD-PCR to characterize aphid populations. Forty-first Annual Meeting of the Entomological Society of America, Dallas, Texas. Dixon, W.N., and M.W. Houseweart. 1978. Location and importance of feeding by the white pine weevil, Pissodes strobi (Peck). School of Forest Resources, Life Sciences and Agriculture Experiment Station, University of Maine, Orono, CFRU Research Note 1. Dixon, W.N., and M.W. Houseweart. 1982. Life tables of the white pine weevil, Pissodes strobi, in central Maine. Environmental Entomology 11: 555-564. Droska, J.S., F.B. Knight and M.W. Houseweart. 1983. Phototactic responses of the white pine weevil to ultraviolet light. CFRU Research Note 13. Maine Agricultural Experiment Station. Miscellaneous Report 288. 6 pp. Drouin, J.A., CR. Sullivan, and S.G. Smith. 1963. Occurrence of Pissodes terminalis Hopp. (Coleoptera: Curculionidae) in Canada: Life history, behavior and cytogenetic identification. The Canadian Entomologist 95: 70-76. Edwards, O., and M. Hoy. 1994. Genetic structure of Hawaiian Diglyphus begini populations determined using RAPD-PCR. Forty-first Annual Meeting of the Entomological Society of America, Dallas, Texas. Eisenbeis, R.A., and R.B. Avery. 1972. Discriminant Analysis and Classification Procedures: Theory and Applications. Lexington Books, Lexington, Massachusetts. 254 pp. Escote-Carlson, L.J. 1991. Plant Molecular Biology Laboratory Course Manual. Biotechnology Laboratory Teaching Laboratory, University of British Columbia. 127 pp. Felenstein, J. 1974. The evolutionary advantage of recombination. Genetics 78: 737-756. Fondrk, M.K., R.E. Page Jr., and G.J. Hunt. 1993. Paternity analysis of worker honeybees using random amplified polymorphic DNA. Naturwissenschaften 80: 226-231. 196 Fraser, R.G., and D.G. Heppner. 1993. Control of white pine weevil, Pissodes strobi on Sitka spruce using implants containing systemic insecticide. The Forestry Chronicle 69(5): 600-603. Fraser, R.G., and S. Szeto. 1994. Insecticides applied with Ezject will protect young Sitka spruce from white pine weevil attack. Pp. 239-247 in: The White Pine Weevil: Biology, Damage and Management. FRDA Report No. 226, R.I. Alfaro, G. Kiss, and R.G. Fraser (eds.). Proceedings of a symposium held January 19-21, 1994 in Richmond, British Columbia. Futuyma, D.J. 1986. Evolutionary Biology. Sinauer Associates, Inc., Sunderland, Massachusetts. 600 pp. Gara, R.I., R.L. Carlson and B.F. Hrutfiord. 1971. Influence of some physical and host factors on the behavior of the Sitka spruce weevil, Pissodes sitchensis, in Southwestern Washington. Annals of the Entomological Society of America 64: 467-471. Giles, BE. 1994. Genetic biodiversity: analyzing the data. Tree 9(9): 317-319. Godwin, P. A., and T.M. ODell. 1967. Experimental hybridization of Pissodes strobi and P. approximatus (Coleoptera: Curculionidae). Annals of the Entomological Society of America 60(1): 55-58. Godwin, P. A., H.T. Valentine, and T.M. ODell. 1982. Identification of Pissodes strobi, Pissodes approximatus, and P. nemorensis (Coleoptera: Curculionidae), using discriminant analysis. Annals of the Entomological Society of America 75: 599-604. Hale, W.G., and J.P. Margham (eds). 1991. The Harper Collins Dictionary of Biology. HarperCollins Publishers, New York. 569 pp. Hamel, M., E. Bauce, and R. Lavallee. 1994. Feeding and oviposition interspecific preferences of adult white pine weevil (Coleoptera: Curculionidae) in Quebec. Environmental Entomology 23(4): 923-929. Hand, D.J. 1982. Kernel Discriminant Analysis. Research Studies Press: A Division of John Wiley & Sons Ltd, Chichester, Great Britain. 253 pp. Hanover, J.W. 1975. Physiology of tree resistance to insects. Annual Review of Entomology 20: 75-95. Harman, D.M. 1970. Determination of larval instars of the white-pine weevil by head-capsule measurement. Annals of the Entomological Society of America 63(6): 1573-1575. Harman, D.M., and G.A. Kranzler. 1969. Sound production in the white-pine weevil, Pissodes strobi, and the northern pine weevil, P. approximatus. Annals of the Entomological Society of America 62: 134-136. Harman, D.M., and H.M. Kulman. 1966. A technique for sexing live white pine weevils, Pissodes strobi. Annals of the Entomological Society of America 59:315-317. Harman, D.M., and H.M. Kulman. 1967. Flight and dispersal of the white-pine weevil. Journal of Economic Entomology 60(6): 1682-1687. 197 Harris, L. J., J.H. Borden and R.I. Alfaro. Close range orientation of Pissodes strobi as influenced by chemical and physical characteristics of its host, Sitka spruce (unpublished report). Hartl, D.L., and A.G. Clark. 1989. Principles of Population Genetics, Second Ed. Sinauer Associates, Inc., Sunderland, Massachusetts. 682 pp. Hedrick, P. 1992. Shooting the RAPDS. Nature 355: 679-680. Heppner, D.G., and P.M. Wood. 1984. Vancouver Region Sitka spruce weevil survey results (1982-1983) with recommendations for planting Sitka spruce. British Columbia Ministry of Forests, Forest Service Internal Report PM-V-5. 30 pp. Hillis, D.M., A. Larson, SK. Davis, and EA. Zimmer. 1990. Nucleic acids III: Sequencing. Pp 318-370 in: Molecular Systematics, D.M. Hillis and C. Moritz (eds.). Sinauer Associates, Inc., Sunderland, Massachusettes. Hopkins, A.D. 1911. Technical papers on miscellaneous forest insects. I. Contribtuions toward a monograph of the bark-weevils of the genus Pissodes. United States Department of Agriculture, Bureau of Entomology, Technical Series 20, Part 1: 1-68. Hopping, R. 1920. A new species of the genus Pissodes (Coleoptera). The Canadian Entomologist 52: 132-134. Hulme, M.A. 1989. Laboratory assessment of predation by Lonchaea corticis (Diptera: Lonchaeidae) on Pissodes strobi (Coleoptera: Curculionidae). Environmental Entomology 18(6): 1011-1014. Hulme, M.A. 1990. Field assessment of predation by Lonchaea corticis (Diptera: Lonchaeidae) on Pissodes strobi (Coleoptera: Curculionidae) in Picea sitchensis. Environmental Entomology 19(1): 54-58. Hulme, M. 1992. Biocontrol of spruce weevil. Pp. 17-19 in: Spruce Weevil Symposium Proceedings, Terrace, B.C., March 12, 1992. British Columbia Ministry of Forests, Prince Rupert Region, Smithers, B.C. Hulme, M.A. 1994. The potential of Allodorus crassigaster for the biological control of Pissodes strobi. Pp. 294-300 in: The White Pine Weevil: Biology, Damage and Management. FRDA Report 226, R.I. Alfaro, G. Kiss, and R.G. Fraser (eds.). Proceedings of a symposium held January 19-21, 1994 in Richmond, British Columbia. Hulme, M.A., A.F. Dawson, and J.W.E. Harris. 1986. Exploiting cold-hardiness to separate Pissodes strobi (Peck) (Coleoptera: Curculionidae) from associated insects in leaders of Picea sitchensis (Bong.) Carr. The Canadian Entomologist 118: 1115-1122. Hulme, M.A, and J.W.E. Harris. 1989. How Lonchaea coticis Taylor may impact broods of Pissodes strobi (Peck) in Picea sitchensis (Bong.) Carr. Pp. 161-166 in: Insects Affecting Reforestation: Biology and Damage. Edited by R.I. Alfaro, and S.G. Glover, Forestry Canada, Victoria. Proceedings of a meeting of the IUFRO working group on Insects Affecting Reforestation held under the auspices of the XVIII International Congress of Entomology, Vancouver, British Columbia, Canada, July 3-9, 1988. Hulme, M.A., J.W.E. Harris, and A.F. Dawson. 1987. Exploiting adult girth to separateeidae) Pissodes strobi (Peck) (Coleoptera: Curculionidae) from associated insects in leaders of Picea sitchensis (Bong) Carr. The Candadian Entomologist 119: 751-753. 198 Hunt, G.J, and R.E. Page Jr. 1995. A linkage map of the honey bee, Apis mellifera, based on RAPD markers. Genetics. In Press. Innis, M.A., D.H. Gelfand, J.J. Snisky, and T.J. White (eds). 1990. PCR Protocols. A Guide to Methods and Applications. Academic Press, New York. 482 pp. Jandel. 1994. SigmaStat™: Statistical Software. Jandel Scientific. Jefferey, C. 1977. Biological Nomenclature. Second Edition. Camelot Press Ltd., Southhampton, Great Britain. 72 pp. Kambhampati, S., W.C. Black IV, and K.S. Rai. 1992. Random amplified polymorphic DNA of mosquito species and populations (Diptera: Culicidae): Techniques, statistical analysis and applications. Journal of Medical Entomology 29(6): 939-945. Karl, S.A., and J.C. Avise. 1993. PCR-based assays of Mendelian polymorphisms from anonymous single-copy nuclear DNA: Techniques and "applications from population genetics. Molecular Biology and Evolution 10(2): 342-361. King, J.N. 1994. Delivering durable resistant Sitka spruce for plantations. Pp. 134-149 in: The White Pine Weevil: Biology, Damage and Management. FRDA Report No. 226, R.I. Alfaro, G. Kiss, and R.G. Fraser (eds.). Proceedings of a symposium held January 19-21, 1994 in Richmond, British Columbia. Kiss, G.K., and A.D. Yanchuk. 1991. Preliminary evaluation of genetic variation in interior spruce in British Columbia. Canadian Journal of Forest Research 21(2): 230-234. Kriebel, H.B. 1954. Bark thickness as a factor in resistance to white pine weevil injury. Journal of Forestry 52: 842-845. Kwok, S., and R. Higuchi. 1989. Avoiding false positives with PCR. Nature 339: 237-238. Landry, B.S., L. Dextraze, and G. Boivin. 1993. Random amplified polymorphic DNA markers for DNA fingerprinting and genetic variability assessment of minute parasitic wasp species (Hymenoptera: Mymaridae and Trichogrammatidae) used in biological control programs of phytophagous insects. Genome 36: 580-587. Langor, D.W., and F. AH. Sperling. 1994. Diagnostics of the Pissodes strobi species group in western Canada using mitochondrial DNA. Pp. 174-183 in: The White Pine Weevil: Biology, Damage and Management. FRDA Report 226, R.I. Alfaro, G. Kiss, and R.G. Fraser (eds.). Proceedings of a symposium held January 19-21, 1994 in Richmond, British Columbia. Lansman, R.A., R.O. Shade, J.F. Shapira, and J.C. Avise. 1981. The use of restriction endonucleases to measure mitochondrial DNA sequence relatedness in natural populations. III. Techniques and potential applications. Journal of Molecular Evolution 17: 214-226. Lewis, K.G. 1995. Genetic variation among populations of Pissodes strobi (white pine weevil) reared from Picea and Pinus hosts as inferred from RAPD markers. M.Sc. Thesis, University of British Columbia, Vancouver, B.C. 199 Lewis, KG., J.E. Carlson, and J.A. McLean. 1994. Using RAPD markers to investigate genetic diversity of the white pine weevil (Pissodes strobi). Pp. 184-202 in: The White Pine Weevil: Biology, Damage and Management. FRDA Report 226, R.I. Alfaro, G. Kiss, and R.G. Fraser (eds.). Proceedings of a symposium held January 19-21, 1994 in Richmond, British Columbia. Li, W.-H., and D. Graur. 1991. Fundamentals of Molecular Evolution. Sinauer Associates, Inc., Sunderland, Massachusettes. 284 pp. Logan, K.T. 1962. Growth of white pine seedlings beneath an aspen stand. Canadian Department of Forestry, Forest Research Branch, Ottawa, Technical Note 121. 13 pp. Lu, R., and H. Rank. 1995. Use of RAPD analyses to estimate populations genetic parameters in the alfalfa leafcutting bee, Meagachile rotundata. Submitted to Molecular Ecology. Lynch, M., and B.G. Milligan. 1994. Analysis of population genetic structure with RAPD markers. Molecular Ecology 3: 91-99. MacAloney, H.J. 1932. The white-pine weevil. United States Department of Agriculture, Washington, D.C., Circular No. 221. 30 pp. Manly, B.F.J. 1992. Multivariate Statistical Methods: A Primer. Chapman & Hall, London, Great Britain. 159 pp. Manna, G.K., and S.G. Smith. 1959. Chromosomal polymorphism and inter-relationships among bark weevils of the genus Pissodes German Nucleus11(2): 179-208. Manville, J.F., J. Nault, E. von Rudloff, A. Yanchuk, and G.K. Kiss. 1994. Spruce terpenes: Expression and weevil resistance. Pp. 203-217 in: The White Pine Weevil: Biology, Damage and Management. FRDA Report 226, R.I. Alfaro, G. Kiss, and R.G. Fraser (eds.). Proceedings of a symposium held January 19-21, 1994 in Richmond, British Columbia. Martin, E. (ed.). 1985. Dictionary of Biology. Sphere Books Ltd, London, Great Britain. 256 pp. Martin, G.B., J.G.K. Williams, and S.D. Tanksley. 1991. Rapid identification of markers linked to a Pseudomonas resistance gene in tomato by using random primers and near-isogenic lines. Proceedings of the National Academy of Sciences of the Unites States of America 88: 2336-2340. Marty, R, and D.G. Mott. 1964. Evaluating and scheduling white-pine weevil control in the Northeast. United States Forest Service, Northeast Forest Experimental Station, Forest Research Paper NE-19. 56 pp. Maxson, L.R., and R.D. Maxson. 1990. Proteins II: Immunological techniques. Pp. 127-155 in: Molecular Systematics, D.M. Hillis and C. Moritz (eds.). Sinauer Associates, Inc., Sunderland, Massachusettes. Maxwell, A.E. 1961. Canonical variate analysis when the variables are dichotomous. Educational and Psychological Measurement 21 (2): 259-271. 200 McLean, J.A. 1989. Effect of red alder overstory on the occurrence of Pissodes strobi (Peck) during the establishment of a Sitka spruce plot. Pp. 167-176 in: Insects Affecting Reforestation: Biology and Damage. Edited by R.I. Alfaro and S.G. Glover, Forestry Canada, Victoria. Proceedings of a meeting of the IUFRO Working Group in Insects Affecting Reforestation held under the auspices of the XVIII International Congress of Entomology, Vancouver, British Columbia, Canada, July 3-9. McLean, J.A. 1992. Overstory manipulation. Pp. 20-21 in: Spruce Weevil Symposium Proceedings, Terrace, B.C., March 12, 1992. British Columbia Ministry of Forests, Prince Rupert Region, Smifhers, B.C. McLean, J.A. 1994. Silvicultural control of the white pine weevil at the UBC Malcolm Knapp Research Forest. Pp. 248-253 in: The White Pine Weevil: Biology, Damage and Management. FRDA Report 226, R.I. Alfaro, G. Kiss, and R.G. Fraser (eds.). Proceedings of a symposium held January 19-21, 1994 in Richmond, British Columbia. McMullen, L.H. 1976a. Effect of temperature on oviposition and brood development of Pissodes strobi (Coleoptera: Curculionidae). The Canadian Entomologist 108: 1167-1172. McMullen, L.H. 1976b. Spruce weevil damage. Ecological basis and hazard rating for Vancouver Island. Canadian Forestry Service Report No. B.C.-X-141. 7 pp. McMullen, L.H., A.J. Thomson, and R.V. Quenet. 1987. Sitka spruce weevil (Pissodes strobi) population dynamics and control: A simulation model based on field relationships. Canadian Forestry Service, Pacific Forestry Centre, Victoria, British Columbia BC-X-288. 20 pp. Michelmore, R.W., I. Paran, and R.V. Kesseli. 1991. Identification of markers linked to disease resistant genes by bulked segregant analysis. A rapid method to detect markers in specific genomic regions by using segregating populations. Proceedings of the National Academy of Sciences of the United States of America 88: 9828-9832. Miller, A.V., and S.M. Craig (eds). 1980. Handbook For Pesticide Applicators And Pesticide Dispensers. Province of British Columbia, Ministry of Environment, Pesticide Control Branch, Victoria. 233 pp. Moritz, C, and D.M Hillis. 1990. Molecular Systematics: Context and Controversies. Pp. 1-10 in: Molecular Systematics, D.M. Hillis and C. Moritz (eds.). Sinauer Associates, Inc., Sunderland, Massachusettes. Murphy, R.W., J.W. Sites, Jr., D.G. Buth, and CH. Haufler. 1990. Proteins I: Isozyme electrophoresis. Pp. 45-126 in: Molecular Systematics, D.M. Hillis and C. Moritz (eds). Sinauer Associates, Inc., Sunderland, Massachusettes. Namkoong, G. 1994. Breeding strategies of resistance. Pp. 218-221 in: The White Pine Weevil: Biology, Damage and Management. FRDA Report 226, R.I. Alfaro, G. Kiss, and R.G. Fraser (eds.). Proceedings of a symposium held January 19-21, 1994 in Richmond, British Columbia. Nei, M. 1972. Genetic distances between populations. American Naturalist \06: 283-292. Nei, M. 1978. Estimation of average heterozygosity and genetic distance from a small number of individuals. Genetics 89: 583-590. 201 Nei, M. 1987. Genetic variation within species. Pp. 176-207 in: Molecular Evolutionary Genetics. Columbia University Press, New York. Overhulser, D.L., and R.I. Gara. 1975. Spring flight and adult activity of the white pine weevil, Pissodes strobi (Coleoptera: Curculionidae), on Sitka spruce in western Washington. The Canadian Entomologist 107: 251-256. Paran, I., R. Kesseli, and R. Michelmore. 1991. Identification of restriction fragment length polymorphism and random amplified polymorphic DNA markers linked to downy mildew resistant genes in lettuce, using near isogenic lines. Genome 34: 1021-1027. Peck, W.D. 1817. On the insects which destroy the young branches on the pear-tree and the leading shoot of the Weymouth-pine. Massachusetts Agricultural Journal 4: 205-211. Peckham, DJ. 1969. A serological comparison of Pissodes strobi and P. approximatus (Coleoptera: Curculionidae). The Canadian Entomologist 101: 78-90. Peirson, H.B. 1922. Control of the white pine weevil by forest management. Harvard Forest Bulletin 5, Petersham, Massachusetts. 42 pp. Perring, T.M., AD. Cooper, R.J. Rodriguez, CA. Farrar, and T.S. Bellows Jr. 1993. Identification of a whitefly species by genomic and behavioral studies. Science 259: 74-77. Phillips, T.W. 1984. Ecology and systematics of Pissodes sibling species (Coleoptera: Curculionidae). Ph.D. dissertation, State University of New York, Syracuse. Phillips, T.W., and G.N. Lanier. 1983a. Biosystematics of Pissodes Germar (Coleoptera: Curculionidae): Feeding preferences and breeding site specificity of P . strobi and P. approximatus. The Canadian Entomologist 115:1627-1636. Phillips, T.W., and G.N. Lanier. 1983b. White pine weevil, Pissodes strobi (Coleoptera: Curculionidae), attack on various conifers in New York. The Canadian Entomologist 115: 1637-1640. Phillips, T.W., and G.N. Lanier. 1985. Genetic divergence among populations of the white pine weevil, (Pissodes strobi (Coleoptera: Curculionidae). Annals of the Entomological Society of America 78: 744-750. Phillips, T.W., and G.N. Lanier. 1986. Interspecific activity of semiochemicals among sibling species of Pissodes (Coleoptera: Curculionidae). Journal of Chemical Ecology 12(7): 1587-1601. Phillips, T.W., S.A. Teale, and G.N. Lanier. 1987. Biosystematics of Pissodes Germar (Coleoptera: Curculionidae): Seasonality, morphology, and synonymy of P. approximatus Hopkins and P. nemorensis Germar. The Canadian Entomologist 119: 465-480. Phillips, T.W., J.R. West, J.L. Foltz, R.M. Silverstein, and G.N. Lanier. 1984. Aggregation pheromone of the deodar weevil, Pissodes nemorensis (Coleoptera: Curculionidae): Isolation and activity of grandisol and grandisal. Journal of Chemical Ecology 10(10): 1417-1423. 202 Pimentel, D. 1963. Introducing parasites and predators to control native pests. The Canadian Entomologist 95: 785-792. Plank, G.H., and H.D. Gerhold. 1965. Evaluating host resistance to the white-pine weevil, Pissodes strobi, (Coleoptera: Curculionidae) using feeding preference tests. Annals of the Entomological Society of America 58: 527-532. Prince, A.M., and L. Andrus. 1992. PCR: How to kill unwanted DNA. BioTechniques 12: 358-360. Puterka, G.J., W.C. Black IV, W.M. Steiner, and R.L. Burton. 1993. Genetic variation and phylogenetic relationships among worldwide collections of the Russian wheat aphid, Diuraphis noxia (Mordvilko), inferred from RAPD-PCR markers. Heredity 70: 604-618. Rafalski, J.A., S.V. Tingey, and J.G.K. Williams. 1991. RAPD markers- a new technology for genetic mapping and plant breeding. AgBiotech News and Information 3(4): 645-648. Rankin, L.J., and K. Lewis. 1994. Effectiveness of leader clipping for control of the white pine weevil, Pissodes strobi, in the Cariboo Forest Region of British Columbia. Pp. 262-269 in: The White Pine Weevil: Biology, Damage and Management. FRDA Report 226, R.I. Alfaro, G. Kiss, and R.G. Fraser (eds.). Proceedings of a symposium held January 19-21, 1994 in Richmond, British Columbia. Reiter, R.S., J.G.K. Williams, KA. Feldman, J.A. Rafalksi, S.C. Tingey, and PA. Scolnik. 1992. Global and local genome mapping in Arabidopsis thaliana by using recombinant inbred lines and random amplified polymorphic DNAs. Proceedings of the National Academy of Sciences of the United States of America 89: 1477-1481. Retnakaran, A., G.G. Grant, T.J. Ennis, P.G. Fast, B.M. Arif, D. Tyrrell, and G.G. Wilson. 1982. Development of environmentally acceptable methods for controlling insect pests of forests. Canadian Forestry Service, Department of the Environment. Information Report FPM-X-62. 22 pp. Retnakaran, A., and L. Jobin. 1994. New observation on adult behavior of the white pine weevil and implications on control with Diflubenzuron. Pp. 270-284 in: The White Pine Weevil: Biology, Damage and Management. FRDA Report No. 226, R.I. Alfaro, G. Kiss, and R.G. Fraser (eds.). Proceedings of a symposium held January 19-21, 1994 in Richmond, British Columbia. Reyment, R.A, R.E. Blackith, andNA. Campbell. 1984. Multivariate morphometries. Second Edition. Academic Press, London. 233 pp. Roberts,D.R. 1994. Somatic embryogenesis for mass propagation of weevil resistant spruce. Pp. 169-173 in: The White Pine Weevil: Biology, Damage and Management. FRDA Report 226, R.I. Alfaro, G. Kiss, and R.G. Fraser (eds). Proceedings of a symposium held January 19-21, 1994 in Richmond, British Columbia. Romesburg, H.C. 1984. Cluster Analysis for Researchers. Lifetime Learning Publications, Belmont, California. 334 pp. Sahota, T.S., J.F. Manville, and E. White. 1994. Interaction between Sitka spruce weevil and its host Picea sitchensis (Bong) Carr: A new mechanism for resistance. The Canadian Entomologist 126: 1067-1074. 203 Saiki, R.K., D.H. Gelfand, S. Stoffel, S.J. Scharf, R. Higuchi, GT. Horn, KB. Mullis, and HA. Erlich. 1988. Primer-directed enzymatic amplification of DNA with a thermostable DNA polymerase. Science 239: 487-491. Santamour, F.S., Jr., and D.F. Zinkel. 1976. Weevil induced resin crystallization related to resin acids in eastern white pine. Proceedings 23rd Northeastern Forest Tree Improvement Conference, Rutgers University, New Jersey, pp. 52-56. SAS Institute. 1989. SAS User's Guide: Statistics. SAS Institute, Cary, North Carolina. Shaw, C.R., and R. Prasad. 1970. Starch gel electrophoresis of enzymes - a compilation of recipes. Biochemical Genetics 4: 297-320. Shepherd, R.F., and IS. Otvos. 1986. Pest management of Douglas-fir tussock moth: Procedures for insect monitoring, problem evaluation and control actions. Canadian Forestry Service, Information Report BC-X-270, Pacific Forestry Centre, Victoria. 14 pp. Sieben, B.G. 1991. Spruce weevil, Pissodes strobi Peck, hazard in selected biogeoclimatic subzones in the Cariboo and Prince George Forest Regions. British Columbia Forest Service Minor Services Contract Report No. 20653. 22 pp. Sieben, B.G. 1995. Climatically based hazard rating system for spruce weevil, Pissodes strobi, in the Prince George Forest Region under present and climate change conditions. M.Sc. Thesis, Faculty of Forestry, University of British Columbia, Vancouver (in preparation). Silver, G.T. 1968. Studies on the Sitka spruce weevil, Pissodes sitchensis in British Columbia. The Canadian Entomologist 100:93-110. Smith, J.H.G, and J.A. McLean. 1993. Methods are needed to prevent devastation of Sitka spruce plantations by the Sitka spruce weevil, Pissodes strobi Peck. Pp. 81-93 in: Proceedings of the IUFRO International Sitka spruce provenance experiment Edinburgh, Scotland 1984, C.C. Ying and LA. McKnight (eds.). British Columbia Ministry of Forests, Victoria, Canada. Smith, S.G. 1956. Chromosomal polymorphism in a bark beetle. Nature 111: 386. Smith, S.G. 1962. Cytogenetic pathways in beetle speciation. The Canadian Entomologist 94: 941-955. Smith, S.G. 1970. Chromosomal polymorphism in North American Pissodes weevils: Structural isomerism. Canadian Journal of Genetics and Cytology 12: 506-540. Smith, S.G, and BA. Sugden. 1969. Host trees and breeding sites of native North American Pissodes bark weevils, with a note on synonymy. Annals of the Entomological Society of America 69(1): 146-148. Smith, S.G, and Y. Takenouchi. 1962. Unique incompatibility system in a hybrid species. Science 138: 36-37. Smith, S.G, and Y. Takenouchi. 1969. Chromosomal polymorphism in Pissodes weevils: Further on incompatibility in P. terminalis. Canadian Journal of Genetics and Cytology 11(4): 761-782. 204 Sokal, R.R., and CD. Michener. 1958. A statistical method for evaluating systematic relationships. University of Kansas Science Bulletin 28: 1409-1438. Solbrig, O.T„ and D.J. Solbrig. 1979. Introduction to Population Biology & Evolution. Addison-Wesley Publishing Company, Reading, Massachusetts. 468 pp. Sommer, S.S., AR. Groszbach, and C.D.K. Bottema. 1992. PCR amplification of specific alleles (PASA) is a general method for rapidly detecting known single-base changes. BioTechniques 12: 82-87. Stevenson, R.E. 1967. Notes on the biology of the Engelmann spruce weevil, Pissodes engelmanni (Curculionidae: Coleoptera) and its parasites and predators. The Canadian Entomologist 99: 201-213. Stiell, W.M. 1979. Releasing unweevilled white pine to ensure first-log quality of final crop. Forestry Chronicle 55: 142-143. Stiell, W.M, and A.B. Berry. 1985. Limiting white pine weevil attacks by side shade. Forestry Chronicle 61: 5-9. Strauss, S.H., J. Bousquet, V.D. Hipkins, and Y.-P. Hong. 1992. Biochemical and molecular genetic markers in biosystematic studies of forest trees. New Forests 6: 125-158. Stroh, R.C., and H.D. Gerhold. 1965. Eastern white pine characteristic related to weevil feeding. Silvae Genetica 14: 160-169. Sullivan, CR. 1959. The effect of light and temperature on the behaviour of adults of the white pine weevil, Pissodes strobi Peck. The Canadian Entomologist 91:213-232. Sullivan, CR. 1960. The effect of physical factors on the activity and development of adults and larvae of the white pine weevil, Pissodes strobi (Peck). The Canadian Entomologist 92: 732-745. Sullivan, CR. 1961. The effect of weather and the physical attributes of white pine leaders on the behaviour and survival of the white pine weevil, Pissodes strobi, in mixed stands. The Canadian Entomologist 93(9): 721 -741. Swofford, D.L. 1989. BIOSYS-1 (Release 1.7): User's Manual. Champaign, Illinois. Swofford, D.L., and R.B. Selander. 1981. BIOSYS-1: A FORTRAN program for the comprehensive analysis of electrophoretic data in population genetics and systematics. Journal of Heredity 72: 281 -283. Tabachnik, B.G, and L.S. Fidell. 1983. Using Multivariate Statistics. Harper and Row Publishers, New York. 509 pp. Taylor, S.P., R. Alfaro, C. Delong, and L.J. Rankin. 1994. The effects of overstory shading on white pine weevil damage to interior white spruce. Pp. 254-261 in: The White Pine Weevil: Biology, Damage and Management. FRDA Report 226, R.I. Alfaro, G. Kiss, and R.G. Fraser (eds.). Proceedings of a symposium held January 19-21, 1994 in Richmond, British Columbia. 205 Taylor, S.P., and R.D. Cozens. 1994. Limiting white pine weevil attacks by side and overstory shade in the Prince George Forest Region. Journal of the Entomological Society of British Columbia 91: 37-42. Tizado, E. J., and J.M. Nieto Nafria. 1994. A new species of Uroleucon (Horn. Aphididae) on Andryala spp.: A multivariate analysis. The Canadian Entomologist 126: 1251-1261. Tomlin, E.S., and J.H. Borden. 1994a. Relationship between leader morphology and resistance or susceptibility of Sitka spruce to the white pine weevil. Canadian Journal of Forest Research 24: 810-816. Tomlin, E.S., and J.H. Borden. 1994b. Development of a multicomponent resistance index for Sitka spruce resistant to the white pine weevil. Pp. 117-133 in: The White Pine Weevil: Biology, Damage and Management. FRDA Report 226, R.I. Alfaro, G. Kiss, and R.G. Fraser (eds.). Proceedings of a symposium held January 19-21, 1994 in Richmond, British Columbia. Trudel, R., R. Lavallee, E. Bauce, J. Cabana and C. Guertin. 1994. Variations in ground white pine bark concentrations in artificial diet in relation to egg laying, feeding, and mortality of Pissodes strobi (Coleoptera: Curculionidae). Journal of Economic Entomology 87(1): 96-100. Tulsieram, L.K., J.C. Glaubitz, G. Kiss, and J.E. Carlson. 1992. Single tree genetic linkage mapping in conifers using haploid DNA from megagametophytes. Bio/Technology 10: 686-690. VanderSar, T.J.D. 1978. Resistance of western white pine to feeding and oviposition by Pissodes strobi Peck in western Canada. Journal of Chemical Ecology 4(6): 641-647. VanderSar, T.J.D., and J.H. Borden. 1977a. Aspects of host selection behavior of Pissodes strobi (Coleoptera: Curculionidae) as revealed in laboratory feeding bioassays. Canadian Journal of Zoology 55: 405-414. VanderSar, T.J.D., and J.H. Borden. 1977b. Role of geotaxis and phototaxis in the feeding and oviposition behavior of overwintered Pissodes strobi. Environmental Entomology 6(5): 743-749. VanderSar, T.J.D., and J.H. Borden. 1977c. Visual orientation of Pissodes strobi Peck (Coleoptera: Curculionidae) in relation to host selection behavior. Canadian Journal of Zoology 55: 2042-2049. VanderSar, T.J.D., J.H. Borden, and J.A. McLean. 1977. Host preference of Pissodes strobi Peck (Coleoptera: Curculionidae) reared from three native hosts. Journal of Chemical Ecology 3(4): 377-389. VanDyke, E.C. 1927. New species of North American Rhynchophora (Coleoptera). The Pan-Pacific Entomologist 4: 11-17. Vanlerberghe-Masutti, F. 1994. RAPD fingerprinting in population genetics of insect host-parasitoid associations. Forty-first Annual Meeting of the Entomological Society of America, Dallas, Texas 206 Wallace, D.R., and CR. Sullivan. 1985. The white pine weevil, Pissodes strobi (Coleoptera: Curculionidae): A review emphasizing behavior and development in relation to physical factors. Proceedings of the Entomological Society of Ontario 116 (supplement): 39-62. Weir, B.S. 1990. Intraspecific differentiation. Pp. 373-410 in: Molecular Systematics, Hillis, D.M., and C. Moritz (eds.). Sinauer Associates, Inc., Sunderland, Massachusettes. Welsh, J., and M. McClelland. 1990. Fingerprinting genomes using PCR with arbitrary primers. Nucleic Acids Research 18(24): 7123-7128. Welsh, J., C. Pretzman, D. Postic, I.S. Girons, G. Baranton, and M. McClelland. 1992. Genomic fingerprinting by arbitrarily primed polymerase chain reaction resolves Borrelia burgdorferi into three distinct phyletic groups. International Journal of Systematic Bacteriology 42(3): 370-377. Werman, S.D., M.S. Springer, and R.J. Britten. 1990. Nucleic acids I: DNA-DNA hybridization. Pp. 204-209 in: Molecular Systematics, D.M. Hillis and C. Moritz (eds.). Sinauer Associates, Inc., Sunderland, Massachusettes. Wiley, E.O. 1981. Phylogenetics: The Theory and Practice of Phylogenetic Systematics. John Wiley & Sons, Inc., New York. 439 pp. Wilkinson, R.C 1979. Oleoresin crystallization in eastern white pine: relationships with chemical components of cortical oleoresin and resistance to the white pine weevil. Forestry Service Research Paper NE-438, Northeastern Forest Experiment Station, Broomhall, Philadelphia. Wilkinson, R.C. 1983. Leader and growth characteristics of eastern white pine associated with white pine weevil attack susceptibility. Canadian Journal of Forest Research 13: 78-84. Williams, J.G.K., A.R. Kubelik, K.J. Livak, J.A. Rafalski, and S.V. Tingey. 1990. DNA polymorphisms amplified by arbitrary primers are useful as genetic markers. Nucleic Acids Research 18(22): 6531-6535. Wood, P.M. 1987. Development of Sitka spruce phenotypes resistant to the spruce weevil. A summary of recent and planned research projects in B.C. British Columbia Ministry of Forests, Forest Service Internal Report PM-V-10. Wood, R.O., and L.H. McMullen. 1983. Spruce weevil in British Columbia. Canadian Forest Service Pest Leaflet, FPL 2. 4 pp. Wright, S. 1951. The genetical strucuture of populations. Annals on Eugenics 15: 323-354. Wright, S. 1978. Evolution and the Genetics of Populations. Vol. 4, Variability Within and Among Natural Populations. University of Chicago Press, Chicago. 580 pp. Wright, J.W., and W.J. Gabriel. 1959. Possibilities of breeding weevil-resistant white pine strains. Unites States Department of Agriculture, North East Forest Experimental Station, Forest Service Station Paper 115. 35 pp. 207 Ying, C.C., and T. Ebata. 1994. Provenance variation in weevil attack in Sitka spruce. Pp. 98-109 in: The White Pine Weevil: Biology, Damage and Management. FRDA Report 226, R.I. Alfaro, G. Kiss, and R.G. Fraser (eds.). Proceedings of a symposium held January 19-21,1994 in Richmond, British Columbia. Zar, J.H. 1984. Biostatistical Analysis. Second Edition. Prentice-Hall, Inc., Englewood Cliffs, New Jersey. 718 pp. 208 Appendix I Taxonomic history of the Pissodes strobi species group This taxonomic review focuses on the four Pissodes species which have been informally grouped into the P. strobi species group (Smith and Takenouchi 1969): P. strobi, P. nemorensis, P. schwarziii and P. terminalis. These species are sympatric throughout much of their range. The most complicated taxonomic profile for a member within this group has occurred with P. nemorensis Germar. Initially recognized as four distinct species by Hopkins (1911), P. approximatus, P. canadensis and P. deodarae have all now been synonymized under P. nemorensis (Table 1). The first amalgamation occurred when P. canadensis was placed under P. approximatus based on cytogenetic evidence (Smith and Sugden 1969). Both of these species exhibited chromosome numbers ranging from 30 to 34 (2N) (Smith 1956) and thus were recognized as the single species, P. approximatus - one of two chromosomally polymorphic hybrid complexes present in Pissodes (Smith and Sugden 1969). The other polymorphic hybrid complex is P. terminalis. Smith (1970) showed that the P. approximatus complex formed a numerical-morphological karyocline in NA with groups in the southeast predominantly carrying a diploid complement of 30 and those in northeast 34. Having established that viable hybrids were produced from crossing P. approximatus with P. strobi, P. nemorensis and P. yosemite (Smith 1962), Smith and Sugden (1969) surmised that P. approximatus was a hybrid swarm resulting from the cross of a 2N=34 species (P. strobi suggested) with that of a 2N=30 species (P. nemorensis suggested). The latter portion of this hypothesis was refuted, however, when Phillips et al. (1987) concluded that P. nemorensis and P. approximatus were conspecific (one species) based on overall morphological, ecological (Phillips and Lanier 1986), physiological (Nei's genetic identity = 0.984 (Phillips et al. 1984)) and behavioral similarities (Phillips et al. 1987): the amalgamation of, P. deodarae under P. approximatus having already occurred due to genetic similarities displayed in Phillips (1984) allozyme study. Prior to the consolidation of P. approximatus with P. nemorensis, numerous studies investigated the association between P. strobi and P. approximatus (i.e., were they one species?) (Godwin and ODell 1967; Harman and Kranzler 1969; Peckham 1969; Booth and Lanier 1974; and Phillips and Lanier 1983a). Although shown to breed in different locations (i.e., P. strobi in leaders, P. approximatus in boles, branches and root collars (Phillips and Lanier 1983a)), these two species attacked many of the same tree species and were virtually identical morphologically Appendix I Continued 209 (Phillips and Lanier 1983a). The species status of these two groups had direct impact on forest pest management practices, as indicated by the United States Forest Service. They feared that if, P. strobi and P. approximatus were indeed the same species, efforts to control the white pine weevil would be in vain since populations of P. approximatus bred in stumps and logs could potentially shift to healthy white pine leaders the next spring (Phillips and Lanier 1983a). Viable hybrids were produced from crosses of P. strobi with P. approximatus in both Smith's (1962) and Godwin and ODell's (1967) laboratory investigations. Differentiation between P. strobi and P. approximatus could not be demonstrated in either sound producing mechanisms (Harman and Kranzler 1969) or serological profiles (Peckham 1969) and furthermore, both species were shown to produce and respond to the same aggregation pheromones (Booth et al. 1983; Phillips et al. 1984). Godwin et al. (1982) did develop linear discriminant functions using measurements from seven morphological characters, to distinguish between P. strobi, P. approximatus and P. nemorensis, but these functions were subsequently tested and found to be of questionable value. Finally, Phillips and Lanier (1983a) argued that breeding site separation provided substantial evidence of reproductive isolation and thus concluded that P. strobi and P. approximatus were indeed separate sibling species. Modifications of intraspecific taxonomic status within P. schwarzii, to the author's knowledge, has been limited to those made by Smith and Sugden (1969). P. yosemite has been recognized as a junior subjective synonym of P. schwarzii (Smith and Sugden 1969) since both displayed/exhibited identical karyotypes (2N=28) (Manna and Smith 1959). P. terminalis weevils have also been demonstrated to have 28 chromosomes, but only in the females. Males of this species are heterozygous for an autosomal fusion, which results in all males carrying 29 chromosomes (i.e., females XC/XC, males XY/Ccc). Cross-breeding experiments (Smith and Takenouchi 1962) suggested that the hybrid species P. terminalis had originated by the crossing of female P. strobi with male P. schwarzii. The remarkable anomoly/phenomenom of dual sexual dichotomy displayed in P. terminalis is quite unique in genetics (Smith 1962) and has been the focus of numerous investigations (Drouin etal. 1963; Manna and Smith 1959; Smith 1962; Smith and Takenouchi 1962; Smith and Takenouchi 1969). The evolutionary relationships/hierarchy within the P. strobi species group is currently under hot debate (Boyce et al 1994; Langor and Sperling 1994). While allozymes, cytogenetics, Appendix I Continued 210 morphology and ecology have all suggested that P. strobi is most related to P. nemorensis,.recent phylogenetic analysis based on 71 different mitochondrial DNA (mtDNA) haplotypes indicated P. strobi as the sister lineage to all others in the group (Boyce et al. 1994). Estimated mtDNA divergence between P. strobi and P. nemorensis was shown as 2.87% and between P. nemorensis and P. terminalis as 1.83% (Boyce et al. 1989), thereby indicating P. nemorensis as more closely related to P. terminalis than to P. strobi. Boyce et al. (1994) have thus refuted Smith and Takenouchi's (1969) hypothesis concerning the origin of P. terminalis, and have suggested P. nemorensis, not P. strobi, as the likely female parent (i.e., assuming P. terminalis resulted from a hybridization event). Results from Langor and Sperling's (1994) explorations of Restriction Fragment Length Polymorphisms (RFLP) within a 1585 base pair (bp) segment of mtDNA also placed P. nemorensis closer to P. terminalis than to P. strobi. Thus the phylogeny of the P. strobi species group remains problematic and somewhat unresolved. V) ii U 3 73 CU u o h. a •a e o CCS o s-o hi cn cu u C cn "3 s 93 CU e cu W) o © a S -- > 2 < CCS -73 S — O « 73 CU O 73 C « © w e s cu - t -» 4 2 73 = ccs cu •™ cn JS ii « u S- D, 2 73 CO o a X 73 S CU a. a a e • mm* ii Vi 3 l a CCS s cu c CU em *• CU 73 S cu B •5 * © *» v J -CCS «g •se O cn <U F Q. S H n o s s is * OX) 73 cn cu ts •JS « — .2 'C 4 8 Vi ii « w CT 73 *• *s * 5 g •o w 5 *- « .2 e A 5 CCS w > cn 8 5^ cu C P e S « £ c u cu «•> > cu CU 0X1 CO BO < 8 Id w < § 2 H 60 1-* I« ^ 8 •a 3. P H 0 CD o 3 •a CD 8 211 V—' cn cd .3 t 0 II CD .3 H cd .a to ^ 8 g « 3 M 0 s '2 'S 2 s E ' B i c <u o y y o £ s P (A § 5 5 | * fe « •2 3 — W W5 c > to cn 3 0 o •a so 1 i 3 S In y S B S £ 2 s « > u 2 3 < 0< 1> c« cn 1 - X C O CO o> - a « 60 t> 3 5 5 3 § § 6 0 * 2 ^ 1 s I III S "Si ^ .s M •I .2 «« £ § -a PH S o cn 0) .2 '3 S (L> 0j> e x 3 <n 60 O o _ -a 4-1 -<-^  <n w a ^ 73 o S p 3 6 60 O o S e -s & II • o cn ta (J TO g g 3 ST o 11 i -° .2 y fi s g ( Q U O o o tn 3 .2 If o * U 1 cn M 1> J3 e f t 2 " i § 5^ 3 & S t-i S5 P -Sr cy 3 cs 3 o ex 3 o o I ca c2 o i-H T3 cd T3 cu T3 IS 2 00 3 1> in N •B '2 O o 60 X> s c O W D 8 « 1) o e x o O 60 <u eg W cvo - H C N c i w m •S 2 •2 §* £ P -CD CD O P 0 t> 1 ^ 2 o e x o E cd C «u -5 S ® .5 w « £ 3 cw b >> cd =5 o 2 O * * 4> 2 60 cS < u PH 1 1 1 • T5 s e o U T J e a a < 00 3 <- to a £ •e ° s o > U . 2 <l> 1 -S o 1 60 G a * § TJ f B cn •a OS I I E S <u a -§ 212 S rt J5> "3 i O a •B » a, eg a. s TJ 2 "5.13 w -rj y 1^  W 3 g —1 <J >, ' I * & 2 c & o S >-H rt <L> CD TJ rt fe> a o •a 2 5 5 s p a i H ° 3 a-, o 3^ o § e *3 5 rt « « 2 O w w £ t i w ID a a , D § &f 1 § W Q 5 e o S -a c .2 JS tS "Si » 'C a S * 3 •= * S & H s 9 « & •at -© E CM to I I <3 to e 8 § § > o 3 O w 4> D.'g ON 3 < § IS .8 •a •5 - 1 g a £ K J< a a) ON O ON O TJ <D § g, •3 to1 & a o ^ 8.1 to o U es o 13 6 § 5 CD 3 to es 0 8 6 1 S -S3 1 ° 8, ,2 CD a ~ & 8 <D ON * i3 4> ° 3 3 -3 •a e3 I 1 if TS O " & •3 ,3 . . T ) .22 £ ^ • ON tO s s .s g s s ^ I-N O c— ON ON -5 & 3 .S * to O i l l N - cj o ?J •3" « i2 •£ & S S "g 5 I if | f l <4-N o <D 3 O •3 < S2 3 CD •3 z Q S -5 CD 3 in . 3 ."3 M TJ O 6 <D E to £ 3 3 CD rt — c ID ^ 3 o j> > es 5? . 3 eS O fJ TJ rt <D - a 3 (2 -a o g g rt CD X> O tw CD <J >-> <D 33 ° «5 g § to •* a i s 8 iD 3 j3 -a C/3 6 0 ON CO - g 9 Q 1J <3. 3 Pd u rt s z s>-> 11 V9 8 « -a o o (f l O M '3 c CD Z q 214 Appendix in Flow chart of laboratory work done in the search for male specific genetic markers in Pissodes strobi Rationale For Each Step To achieve max. yield, high quality DNA extractions which produced clear, consistent results when used in RAPD assays Procedure Followed • Bulk DNA extraction technique perfected • Quantification of DNA by absorbance at 260 nm To determine most cost effective conditions which produced clear, highly resolved specific amplification products (minimizing spurious ) amplification products) RAPD assay conditions optimized for weevil DNA Amplification products resolved on 1% Synergel/ 2% agarose gels To identify primers which would target male specific loci when used in RAPD assays t - 70 primers screened using bulked DNA extracted from 100 weevils (each sex) collected from Ss Attempt to increase the number of loci amplified per primer (relative to bulked DNA from 100 weevils) • 104 primers (59 unique to those above) screened using bulked DNA extracted from 20 weevils collected from Ss To examine banding profiles produced by 8 primers targeting putative male specific markers using three different bulked DNA extracts - 8 primers tentatively ID'd which resulted in amplification products specific to the Y chromosome (primers = 219, 291, 365, 374, 376, 383, 386 and 387): Used in 3 separate RAPD assays with male and female: a) bulked DNA from 20 weevils b) bulked DNA from 100 weevils c) bulked DNA generated by pooling equal amounts from 12 individuals Appendix III Continued Rationale For Each Step Procedure Followed To establish the presence of putative male specific loci in individual male weevils and complete absence of these genetic markers in individual female weevils To show that, due to homology, putative male specific markers used as hybridization probes bound only to male DNA and not to female DNA on dot-blots Primers 219, 374 and 376 used in RAPD assays using DNA extracted from 12 individual males and females (Not bulked DNA) • 4 most distinct putative male specific markers (amplified from primers 219, 374 and 376) and used as hybridization probes of dot-blots containing DNA extracted from 46 male weevils and 45 female weevils (individuals from 4 different populations used: 2 from Ss, 1 from Sw and 1 from Se) Appendix IV 216 Preparation of putative male specific DNA fragments for use as hybridization probes: Although steps B-E illustrate the process for only primer 219, the entire procedure was also done with primers 374 and 376 A. Primary amplification done: o j> o J1 ? ? o 7 <? ? o J1 Bulked DNA Each primer used twice with t o t o f d Y bulked male and female DNA W W W W W W Primer 219 Primer 374 Primer 376 B. Products resolved by electro-phoresis: 2 agarose slices of each putative male specific DNA fragment obtained E x a m p l e o f l o a d i n g s h o w i n g r e a c t i o n s f r o m o n e o f the 4 p r i m e r s u s e d B a n d to be u s e d for p r o b e C. DNA Fragments retrieved by centrjfugal force (high speed spin- one minute) 100 bp ladder Filter cup Electrophoresis buffer and DNA fragment D. Retrieved Solution diluted with 600 u,L of TE E. Secondary amplifications done on retrieved DNA fragments: 20 and 25 cycles investigated; total reaction volume 25 uJL F. Five u.L aliquot of reaction resolved and quantified on gel 5 ^ . 5jlL 20 cycles I 25 cycles 2lJ ' '—= 1 3>4 3?V' 3% L J-15 cycles -100 bp ladder 100 bp ladder' 50 rjg DNA used for labeling reaction except for large fragment from primer 376 Appendix V 2 1 Additional preparation of putative Y chromosome specific marker- larger of two DNA fragments retrieved from reaction containing primer 376- for use as a hybridization probe G. 25 Cycle reaction containing the 765 bp fragment retrieved from reaction using primer 376 undergoes additional 10 cycle amplification V Additional 10 cycles in Thermocycler Larger Y-Specific Fragment from primer 376 H. 5 u,L of final product resolved on gel: Additional DNA fragments visualized _ Slaodardi g Q no* of -Iiyteresf z 100 bp ladder 100 bp ladder I. Steps A-F (Appendix IV - not illustrated herein) repeated in further'attempt to isolate sufficient quantity of pure 765 bp fragment (primer 376) J. Step I unsuccessful, therefore steps A-C repeated on replicates (as recapped to the right) vwwwvvvvvvv Primary amplification with primer 376 done on seven replicates Entire products resolved on gel 6 slices containing 765 bp fragment cut DNA retrieved by centrifiigal force K. Liquid reduced to 23 u.L in speed vacuum: entire quantity used as source of template DNA in labeling reaction 218 Appendix VI Sample source (sample site and leader number within each location from which adult emerged) of weevils used in the dot-blot procedure (bolded numbers indicate only 50 rjg extracted DNA used for binding to membrane- all other numbers indicate 100 t|g DNA used) Row DNA Sample in on Each Well (12 Blot wells per row) Leader Number Weevils Obtained From Sample Source A B C D E F G H 12 males: 12 males: 12 males: 10 males: 23 4 5 6 11 14 18 19 25 29 30 24 6 10 11 12 15 18 26 29 31 40 3 5 6 11 16 17 30 34 38 45 51 54 3 5 8 15 18 23 25 27 33 37 well H-dHjO; well 12 -control 12 females 12 females 11 females well 12-dH20 10 females: 2 3 4 5 6 11 14 18 19 25 29 30 24 6 10 11 12 15 18 26 29 31 40 1 3 5 15 18 25 26 33 36 37 37 5 6 11 16 17 30 34 38 45 54 W2_MCBRIDE-4&5 S VAN 1 GOLDRI VER E2GOLDEN-119 SMIIIRESFOR W2JVICBRIDE-4&5 SVAN1 GOLDRI VER E2_GOLDEN-119 SMIII RESFOR well ll-dH20: well 12 +control U_ N t D t D t D C n i O O O NT n W t l l N N C O O O O c O L D N t n o i r i N O r*i • - w t y f n Q ' - ' - o (_) o • O • • o ' o N j c o t r i N r i r i c D O O i n m c v j o c o o i n c o c o f n r o ^ r m o o r n 0 0 0 0 0 0 - - - -f S <u o u o 03 -o <U c •o 51 •o c cu D. a < c © T3 " « o e o c o •a < U P H E o 3 a •4-4 3 o C O co cn >» OJ (0 >< a c •i— o <c t_ _j ro c X Co OJ c C o o C M a •r- l-H E *-» C _ ) O «J o (_> _j CP L. nj L, LL a O Ol U u OJ c i—I u i _ . o a. - J r o c o i o c o o c n i n c n OJ O) t \ l O PI ' t O) o o o o o o o o . . o O — i O ' m c u r - o o o ' - N -co in o C V J to n T - r - ^ - o o o o n in O O (o in to in to C O ' - O ' - N W O W co (\i o m a m to ' - W O ' - O O ' - W c o o o o c o c o t o c o c o O ' — cu ro ro OJ cu r o o c u c o * - c o i n t n ' - O W ' - O O O W * - . o o o O C D i n t - n r - r f K O i D c o n c n m o n co T - w to w co 0 * - * - * - 0 0 0 * -roo0^fOr* . i* - . r* . C M i ^ r o ^ - r o ^ t i n c o ' - - - f t l t - o o O ' -' - n c o ^ r o j c o c o c n ro^rcotoiocnojtN* c n c o o ) ' - ^ - o ) i n ^ ' O O O O O O O O o o o o t o c o m o j >N . c n c o o in r - ( o tn co cu in n O f l l M N P > O t O W O O O ' - O ' - O O co'ttcom«*ojo*-ro <- CO N W Ifl N O) r n o i N o n o ' - o ) O O O ' - O O O O U - I H T C D Q U J U . — — — o o o o c u c u o j o j r o r o r o r o H H H H H H H H u u u u u u u u O O O O O O O O n N ( D t o ^ o ^ ' - O N i D c o c o c n c o i n » - « - N f n c o T - c o ( D C \ j o . i O ' T T f n o i n r s O t - f s w o o f ' i N c o c O ' - m n i n e o N O ^ ( D O N ^ c n n c n N i B W c y t n c n w o o c o w o f n c o c o t - c n f o ^ i n c y i ^ t - r o r n ^ c o ^ o i m c o c o » - c u r n ^ ^ m ^ o ) r ^ r n o N W N O ' J i v o t n i n c o o ' n t n N O w ^ ' t o c o o w r n i n f ' . o j 0 ^ ^ o j c \ j c \ j n n n r n r j r r ^ T j i i i i n i n i f l i r ) c o c D t Q C O C D c o O O O O O O O O O O O O O O O O O O O O O O O O O n CO o CO CO r-» CO f-x CD *- r- CO o CO o to CO in cn CO CO o CO CO in OJ in in CO O OJ ro o CO tn o m CO CM m f- m fx. CD f^ . cn CD cn OJ cn to O m cn in o ro r-. m *~ m ro CM o CO CO tn «T n ro CM OJ o o> CD CO CO r*- CO CO in ro cn m CVI Oi OJ OJ OJ OJ OJ OJ OJ OJ cu OJ O o o O O O o O o O O O O O O O o o o O o O o o o O o o O O O O O O O o o O O O o o o o o O O o o o r n ^ ^ O T C M i n r ^ r n i n c o ^ ^ c n ^ u i C M O T c o i n c D i n ^ - i N . c D i n ' - r n o ' - o i n i n o t D O o i o w c n N W T - i n o i i n c y c o N C U C 0 a i « * t O C U C 0 C 0 r * - O J » — C 0 » — ^CDCOCDCOintOCOCO*— CO m o J C D ' - O ' - O ' - O O O O O O O O O O O O O O O O * ~ o o o o o o o o o o o o o o o o o o o o o o o o o o >-*- co in T- CM o CO CD CO ro CO fv. o in ro •ft CO CM CO in co ro tO O CD CO in m r-». OJ CM CM CM OJ O O ro O CO ro ro tn CO C D 0 ( 0 0 ) 0 CO o CO C D o ro o CD o CD tn in CO CM in O ro in ro to CM (D CO N cn o O in in o ro CO in CD CD tn r— 0 0 CM o r-» m cn to o O) r*. to in in ro ro OJ OJ CM CM r- O o O CD cn V CVI CM OJ O O o »— cu rn * in to s co cn o oj ro in ^ w r n ^ i n j o s c o c n T - t - ^ r - ^ ^ T - ^ T - r - w w w w f t j w 2 Z Z 2 2 Z 2 2 Z 2 Z Z 2 2 2 Z 2 2 Z 2 Z Z 2 Z Z orccorcccrccDrccaoraQrtrccccQrc C L C L O _ 0 _ O . Q . C L C U C L C L Q . C U D _ D _ 0 _ 0 - D _ 0 ^ u_ ' - o r o i n t o N ' - o CJ < o < o ' o o • s t o o m i n o o i - -t a c v J c o c o S - ^ t o o i o o o — O ' - O — T f ON i — I 1-CU O •«-> u o I D ' — O J C O - ^ O ' — r--O ' - i ' m i n o o o ) O O O O O O ' - ' -> o o o < * - i o f o n i n r n o N M n T - o - ' - O O O O l ^ • e o i n o O ' t ' - T - m c o s o o i - U K D T - N c o c O ' - o m in N CM a>CMO)^co^ to^cMr^T -» -cMcocoOTtncocQ^r« '» .co incMto i n c o m o o e o N c o N c n c O ' - h . W N Q O o i n a D N O O ' - c n o N O W ' - % t ( D N C O ^ O I f > O l t ' C O N C O O ) C J ) < D N C D ' * ' - N ^ O ) a ) n a ) ' - ^ N o n c o c o T - r n i r ) N O ) ' - r n i n N O ) ' - n ' j c o N 0 0 0 6 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 8? 73 w <U C 73 CU cH 73 cu •(-I •tj .s > a, ••3 «• e CU a, *» <1 o e s 73 13 •-< o ' s o CU a o 73 < U (U S o 3 3 O t/J «< O O ^ O C O C O ^ L O G) N c o o i o i o n m r j ) »— C U C M - - — T - O ' — O ' — O J n (_) * O ' * - ' O ' O c 0 X 0 _ l l_ CO X in in 0 ^ s N co 0 01 O J C O O O I N ^ O P ) 2- 0. CO T— C M « — o c n m o c o L D 0 >•. CM T - r - O r - T - O O W c l _ (—1 0 D_ to X CJ O • * - • O O O ' c •i— 0 i-> < i_ _ l (0 +J ro 0 c Li- CO O tO O 0) *~ CM'O l_-0 en T - O O t D C O ' - O N L. CD c c T— n o * - - t f a > m c v j c o 0 0 0 0 CM ' ^ O ' - ' - O O O ' - u c a CD E +-» (_> . r- 1 0 • 0 0 0 CO l_ O <a O - C CD O _ l *-> 4 -CO «f-t_ 4 -to i_ tu o a j i n o c n m o i ' - 0 0 a 0 cn O ^ - W N K C O ' c r O ) •t~ 0 t — o m c \ i m n - c j ( D N 0) 0 CM O T - r - t M t - O O - - ai c • H CJ *- • 0 t 0 • 0 ' i_ O A3 a. _ J > a> . c cu a r - T - S O l f l O S t O O ) CD cn o j C M o i m m c u m ^ •*— > O i ' - r - . i n * — c o f * - c o LU c CM 0 0 0 0 0 0 0 0 CO • H DJ O • 0 0 0 0 0 < • O LU _ l (_) O w t o m c o c n i o K cn ' - c o c o N N c n m r v j i— t n f t l O N O I D C D CM O O O O ' - ' - O O CJ o cn m ro in co *d-inror-.cMco*-cM T-mh-CMfoinmm ^ - 0 0 0 0 0 0 0 n i D C D N W C D C O n O O f l f O i - l D N l O o i o i ' - i n O ' - n ' -* - o c v j i - c M » - T - m O tu u. CD CD CD CD CT) O O 0 CM CM <M CM CM ro ro n H H H H H h-i 1—1 t-H U U U U U CJ CJ CJ O O O O O 0 O 0 —1 —1 1 1 1 _1 _ j _ J ^ • ^ ^ i n o ^ r ^ o o o o ) 0 ) ^ M ^ - c M i o i n » - 0 " * 0 ) ^ T C M C M i r ) c n m t D ^ w c o ^ f f l c o ' f f n o O ' J w o c o f ' i c o o c n c n c n s . n i n w N W o c o ^ o i - c n w ( o i n i n c \ i o ) o m c o r , i c n o N O N O w c o m w o o ) N ( D ^ t f t n w r ' - o c n c n c o c O N K ( 0 ( D L n rotn^mmmCMCMCMCMCMCMCMCMCMCMT-T-T-i-T-T-T-*-,-0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1— O O t*- 0 CO CD r-» 0 CD 1— O O m GO 00 ai 1- CD T- m CO 0 m to 1— 0 CM cn in n CO 10 cn O O O in CM m O CM CM CO m CO CO *t CM CO IO r*. CD CD 0 r-- CO CO ro O 0 0 m t*- CO 0 O t*» •<T m CO in T— T— CM CM m T- in 1— cu CM CD O 0 r~ O r- O 0 0 0 0 O O O O O O 0 O 0 0 0 <- O O O 0 O O O O 0 O 0 O O O O O O O 0 O 0 0 0 CO CM r- O O ro CM CD CM in CO CO 10 O 0 «— r»- CD r - CM r- CM CM CO to O O r«- CM 0 O m CD m 0 T- CO m ro CD co CM CD CO in m 0 in CO -tf m r-. CM CO CO 0 CM m -«* CD CD CM CO cn to r». •^ f m m CM CD CM in 01 in T— tn 0 in in m Ol r-- CM 0 CO CO CD O CD CO 1-- CO in m CM CM CM *" T~ O O O 0 01 cn *r co CM O 0 O ' - w r T c t L O i o N t D c n O ' - f A i r n f i o ^cMm^tncor«..cocn*-*--*-*-*-*-*-*-*-*-CMCMCMCMCMCU 01 "O Vi <u 3 H o <s «H •a cu I . a. Vi CU 3 U OS > <J PH in <s *J u C » - m ( D C \ I C N J ( \ l C T C \ j T - ^ C \ I O 1 C D M C 0 r ^ O T O ^ ^ c y w c u w n n n n ^ ^ ^ ^ ^ i n i n i n i f l i f i i n t o c o t o c o i o i O N K o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o C cn Ol CO •cr OJ y— CM «xf fx. CO in cn CO cn rx CO cn ro CO in *— «t * - in CO OJ o cn X o ro tn o o fx. CM oi CM CO CO fx. fx. CO CM «* fx. CO CD o to CO ro in co t— o T— CO *r rx. ro in cn fx. cn CO O OJ in O CO CM O fx. fx. o cn ro CM CO in OJ t_ in o CO o Ol fx. fx. CO in rj- m CM o Ol Ol CO CO fx. fx. co in ID ro cn ro +-> 1_ fx. tn rr m OJ CM OJ CM OJ OJ OJ OJ OJ OJ OJ 01 CD o o o o O o O O O o o O O o o o o o o o o o o o o o o o o o o a CO o o o o O o O o O o o O O O o o o o o o o o o o o o o o o o O >> c L_ o 0_ to c +-< < CO O c 1_ 0) L_ O o o o Ol fx. O CD O fx. tn CO O OJ fx. CO CO O CO CM in in fx. o f*- , — CM CD in Ol c O o m cn in co fx. CO CM cn CO CO CD *r Ol •xt OJ *r OJ m fx. TJ- co fx. O r*. to o t_> c fx. O fx. OJ CO fx. CO CO CO fx. to in CO CM ro o o r»- CM CM Ol O fx. Ol CO a o m CO in O »— CM in fx. <xt o in •xr in «* cn y— in ro CM co in o tn 0 J T— T— E O l_ rx. OJ o O o o o i— O o o O o o o o o o  o o o o o o o o O J= o o +•* >•- *~ o T - o o O o o o o o o o O o o o o O o o o o o o o o o o ,— !Jz pa O a CO o O c* 3 t_ (0 0- > 0) CM in in CD CO Ol cn tn in ro CO CO o o ro co r - ro CO in ro fx. CM ro c GO m Ol fx. cn CO cn o fx. o CO O o in CO CM O CO m »- rx. o in o Ol in CD i— OJ -^ cn CO 00 *— CO o CM CO Ol cn in fx. 1— CM O fx. in in in rx. in CM CM OJ tf in co Ol CD CO in CO OJ CM CO ro in o Ol *t Ol in Ol in O in o fx, ID co rn m Ol CO in ro r -> »— i— O o> CO CO fx. fx. in in ro ro ro ro OJ OJ * - — O O O Ol Ol cn Ol Ol LU C in m m c\j c\j - • O O O O O O ' - w c i w i o t D N o a m O ' - N P i ^ i n i D s o o r o o »- w tn v i o to N co oi » - *- r~ i- T- »- r- ojc\JC\JC\lc\]C\JC\jcMC\JC\jm Z Z Z Z Z Z Z Z Z Z Z Z Z Z Z Z Z Z Z Z Z Z Z Z Z Z Z Z Z Z CJ JO C\ v o W ) e Vi s. e o < E o 3 a •<-» 3 O cn < cn 222 O O o o ^ O ' - o o i o W ' - o t f O J ' - O N ^ L O t o c o n o n w T - O N O O W O O L D O l O O ^ ' - t t O l M W O P l ' - f t J O O i t c n O O ) o o o o o o o o o o o o o o o o o o o o o o o o o i n o ^ - f A J O t o t M c o ' - m o o w o o w o i o o t o o o o o o o o o o t ' - r - o ) ( \ j o i D o o o ( \ i o O ' - ' " f n o , " ' - o ) ^ o o s o • ^ • T - T - O T O J O I O O O O O CD o LU cn 00 Ol GO rs. in m CM CO CO Ol in o cn in o in O T-o «* y- CO -<* O) 00 CT) 00 rs. in ro CM CO CO l_ CvJ r— CO in u CM CM «— O cn T— o r— «— CO CT) CM o ro in CO m T— 00 o O cn co TT i_ *r CM CM T— o Ol O T— ca CM O) 2 CO rs. O rs. o o CM CO cn CM in CO in CM o o o CT) m rs. o oo CO rs. •*r O rs. o O CM CO CO r-» m • *— ro CM CO CO o ^~ o r— T— rs. CO in O CO N . CM ts. CO in y- in rs. 3 »— m CM CO CO o *— o T— CT in *r +J o CJ r-- *— O CO cn in CM CO CT) CO T— i— CT) O T— CO O) o ' - ' - t O CT rs. *— O CO O l in T— CM CO O) W in CO o fs. cn CM cn cn CM in «<? Ol CM Ol in CM CO CO CM o cn CO GO CM rs. cn CM Ol CT) CM in cn CM o o CO ro CT) in CO o in CO CO CO m rs. GO CO m CM in CO CO rs. o CO Ol O) o CO ro Ol in CO O in CO CO CO Cu C *~ *~ a: o c CD in O *~ o CT) CM in n CM *~ in O) m CM CM *~ CO 00 CO O CM O c Ol in O *~ rs. o cn CM in ro Z3 © GO m 0) CO CO CO m CM CO in T-CO o O CM in r - ro in CM in o co o 0) CO 00 CO •*r m fs. CM GO in TJ 2 2 ro CO GO CT) Ol CM »- *- m »— *- CO CO 2 cn CO *- 00 Ol Ol o cn o in in rs. (Si o i c o o i c o r s - i n m c M c o c o Ol LO O CO LO O LO fs. O *- o •<* *- to *t CO o i G o o i c o r s . i n r o c M c o c o c 01 CO rs. m > CM CO • M - C M C M ' - O C n ' - O ' - ' - CO oi CM o ro in co ro 00 O O Ol CO ^ cn • q - C M C M i - O O l ' - O ' - ' -•r— L . CO CM Ol CM l s V O r s O ' - ^ O C \ l ( D ro CM in GO in CM o O O oi ro rs. o co N < f O N O ' - ' J O W C D CO in O in CJ ro * - r o c M t o c o o * - o * - * - rs. CO LO o CO rs. CM rs. rs. to in in rs. t-t • - f O C M C O C O O ' - O ' - ' -in -<* Ol in r s . r - o c o o i i n ' - c M c o o i (D T~ *t Ol O *~ co m O ' - ' - ' J O n r s . * - o c o a i i n * - C M c o o i n CT CT) in •* CO C M 0 1 0 1 C M * - i n ^ t 0 1 C M » - Ol r- in CM CO CO CM *r O ro to co CM fs. C M O > O I C M T - W ^ - O I C M » -CO CO ro CO a c o r o a i i n c o o i n c o o o c o m rs. co co ro CM in co CO fs. o CO O) Ol o c o r o o i i n c o o t n c o o o c o t_ CO in o i i o o ^ - r s - o o i C M i n r o CM «- in ai ro CM CM CO CO CO O CM O © o i i n O ' - r s . O c n c M t n r o 4- r- D. c o CM CM in c o c o ^ r c o ^ - r o r s - C M c o i n co o o CM in fs. rs. ro in CM in o co o >. c o c o ^ c o ^ r o r s . c M o o i n © O O o ro co »- co oi oi CM » - i— m r- co co r— n co »— co cn T* oi Sum CM o CM CM ro CM CM CM CM o • - i_ i. 0) o 3 TJ l_ O O L. CO 2 LU 0 ' - c \ J m * a - i n c o f - . o o c n o * - c y r o ^ t i o • - ( v j f i ^ L O a j M a o i ' - i - ' - ' - ' - ' - T - ' - ' - i - w t M w w w w Z Z Z Z Z Z Z Z 2 Z Z 2 Z Z Z Z Z Z Z 2 Z Z 2 Z 2 o r c c c c a : Q : a : c c r £ o r c c a : c c f r Q i c r a : Q : r r a ; r r a : Q : Q : a : a : Q_ Q_ 0_ Q_ A- 0_\a_ Q i a Q . Q . a . L a . Q . Q . Q . Q . C L Q . Q . Q . C L Q . ( L T - c M r o ^ j - m c D f s . o o o ) ' -croccrcrcrttcccEcro: Q _ Q . Q _ 0 _ Q - 0 _ 0 _ 0 _ 0 _ C L c o ^ o r o i n o t o o i o t M ' - c v j f x . r x . f x , c \ j ' - O N v r m [ D i D n o c , i ( \ j « - O N T - T t c o w w o n ' - w o o v c n o o . o o o o o o o o o o o o o o o **- O O W N n L n N C D C O ( D C O i n c O O ) ^ M - ^ O ^ C D t D O ) ' - C O C O C D o c n o o i n N i i i i r i c o o J c n f n N i r i n n Q N L n a D n i o i i i t D N N O J i_ m t o i n n o o j c y i n N O « D O ) c o o ^ N C 3 ) ^ ' t ' - r n o i , < i ' ' - i o c o o r o n t D N O ^ m ^ w ^ c D i n f n ^ c D a j O N W O o i i n N n c o ^ c o c n c o o o n m r n ^ ^ ^ ^ i n i n w i f i t D t o m t D N N K N c o c o c o LU *-< y-0y-y-r~y~y-T-y-y-r-y-T-T-T-y-r-r~y-y-y-r-r~y-r~r-01 T J UJ o o o o o o o o o o o o o o o o o o o o o o o o o o C M O O ' - ' — r o o r - T — c n T f o o ^ o o i n o ^ w o t o w i o ^ r n c O ' - o f o i n o t D o i o w ' - c y N N N o ^ O ' - o o c o W ' - O ' J W T - O N ' j i n c D i D m o n c x j ' - O N o o o w o o m o i o o , * ' - ^ c o w ( \ i o ' Y i ' - w o O T c n o o i o o o o o o o o o o o o o o o o o o o o o o o o o o OJ Ol tn o m tn o in fx- O y-o tf * - CO tf t_ ao Ol CM O m in CO m *— CO o o oi to tf <a n CM tn co tn CM o tf o o oi cn rx. o CO h- CO tn o to fx. CM tr fx. to in * - in fx. cr t* CO T— T— tf cn o »— CO Ol o »- tf O CO Ol tn CM t f CO to CM tf o m to co CM fx. m fx. GO CO cn CM in CO CO fx- O CO Ol Ol O CJ c CM T— in oi cn CM •xT CM T— to CO CO O CM O L ca ZJ CD T— tn O O CM in r - . fx. cn in CM in o co O TJ 2 CM *- *~ cn t ~ CO -3- CO o • • II O I-X o i w i D ' - n i o o w o p i o ^ o O ' - ' - ' - O ' - ' - n w o o w o if) oi in O cn in o in fx. o o tf to tf in 00 Ol CM O cn in CO en CO o O Ol CO tf en CM in oo tn CM o tf o o Ol n f x o CO hH fx. to in o to fx. CM t j fx. fx. CO in «- in fx. 1—< tf CO T— *— tf Ol o »— CO cn o tf O t-A oi <- tn CM tt CO CO CM tf o m to CO CM fx. m fx. CO CO cn CM tn CO to fx. o CO Ol Ol o CD CM * - in oi m CM tf CM *~ CO CO CO O CM o >» ' - t o o o w i n s s m i n w i n o t D O i— CM * - m * — »- CO tf CD r x . o m t j - t o c o t f o i t t o o t f t o » - o t f c n o i t 3 - i n p i f O c n i n i n c D O a J v t c o o o j f n N ^ ' C O c o i o c o c o ' - m a i o t a ' t o c o t f o i m t o ^ - i n m c M t f o o j t f t f » - r x , * - c n t f t - r x , o i o o i * - o i C M t 3 - c D t M r n r - u i i J ^ ^ o i ' - C M i n c o i o o j c D t O ' - c n t D t o c M r x . ( D r - ^ ^ o ^ c n r f T — c M f x . r x . o i r f c O ' — O J c n m o ^ f t f t o o o » - ' - o i m c M r x . o i m o i c o c M C M c o r x - c o r o o r ^ c o c d f ^ ^ r - c n c M r n f ^ c o c M i n c n fx. in co co O cn CM in r m tf J tf oo r tf rx. - Ol CO • m r -tn CM in tf CO tf m r— m o CM in in in oi oi to fx. tf oi CO Ol CO T - CM CM oi * - tn ^ r - O o in o c o o o o o o o o o o o o o o o o o o o o o o o o o o » - C M c n t f i n t o f x . c o o i O ' - t M m t f u i r - T - r - i - t - ^ r - r - r-C\JW(\JC\|lM(\J Z Z Z Z Z Z Z Z Z Z Z 2 2 2 Z o 0 ^ t M c n t f i n t o r x . c o o i o * - C M r n t 3 - i o C C ^ C M m t f i n t O t ^ C O C T J ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ C M C M C M t M t M t M U J Z Z Z Z Z Z Z Z Z Z Z Z Z Z Z 2 2 2 2 Z Z Z Z 2 Z z o r o r c c o r o z o r a r o r c c c r • ^ c u o . C L D _ C L a . o - C L O - C L a . r j . C L C ^ 224 O 73 CU *E w e e T3 C 5 "3 a. o a. i . TJ cn o TJ LU cn < - . - L O * - C O ' - C O C O » - ' -c n c o ^ t r n ^ r o m o c o c n c o m L n r ^ i n c D C J — t ^ » - t o r s - t ^ c n o c n c o c o i n c o r ^ c o a s i ^ o i c o • - o o o o o o o o o o o o o o o o o o o o o o o CU CO s '3 2 ^ CU 1 - 1 CA •< .a a a •s » a a * £ CU « 3- > cu T3 »- a a © <— 0) co nj — — c Co Q < •- •-> > O) o s Q. CO CO 0) c — o 0) CO ~ - > <-> (B a> to CO <tt n •-o u > o o > in •«-> o +• • • ta ) O E i— i— r - CO O O O O O O O O O O O O o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o T - c v j f O ^ f L O c o r ^ c o c n O T - c u 13 a << • M a « a u u cn CU CA a CU o 3 a +^  3 o CA) < 03 225 A O T— r— U o o o n in o o o o < o o o t_ 0_ o o o tU TJ C o ro OJ O) HI 10 o CM fx. flj t_ o tf L. <0 •<- Ol T— 00 to a c to tf i n i n > c r o — CO o OJ < cn c cu CO CVI CO CO t-O (-T3 cu 3 _C "•C e o U a H e cu a a «< a 0} cn * a. t>. co v 10 TJ XI J3 0) T3 10 « 4-» cn 0} Q) n • cvi o) co r~ CD ao o o o o o o O o o o o o o o o o C D m CO r - o O J fx. co rx. tf o 00 O Ol CO CO tr tf ro co r-x Ol CO O o *— O J ro ro ro o o o o O o o o o o o o o o o o o o o o o o o o o o o o O) in O J oi o CO O ro O J tn o C O •xT C M fx. in ro * - to C D O * - r-». in O J O J C M ro CD O) fx. O J O J C O C M Ol O J oo ro OJ C M Ol tf in in to CO fx. r-» tf tf tj- tt tf tf tr o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o O O O O O O O O O O O O O O O O O O O O O O O O O O O O O o o o o o o o o o o o o o o o o o o o o o o o o p o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o o ' o o o o o o o o o o o o o o o o o o o o o o o o CO fx. CO CO Ol fx. Ol CO * -in O J co ro ro tf o m O J » - O CO OJ fx. C M C M Ol Ol fx. CO C M rx. ro C M OJ tj- O ro CM i n * -O O O o i n O J to O J t r CM tf O J Ol co ro ro i - CM Ol 00 CO tf o o o o o o in co tf to rx. oi * - in oi r~ ro tf t j to oi ro tn o * - ro Ol CM fx. OJ CO ro ro C M C M O o O o O O O O O O T - r o t r r x . c o o o o i t f r o o i t o t n r o i n r x . e o a i c M t f c o t o c o o r o c o o t f o o i o i c M i n t M i - o i c o c o m i n T - t - T - 0 0 0 0 0 O O O O O O O O O O O O O O O O O CM CD CO i n to co oi CO tt CO Ol i n o m * -tf tf ro ro O o O o o o o o o o o o co t f co rx. * - i n i n i n oi co fx. o o oi tf ro co ro *- * - tr oo i n C M o oo co tf C M C M C M C M * - T- * -o o o o o o o o o o o o o o o o o o o o o ro o 00 o O CO ro r-O o o o o o o o o o o o o O O O O O O O Q Q Q O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O ^ t n t f | x % T - o i C M t f » - o e M r x . t f r o 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 * - 0 , _ * - C M « - i - O J r o r o O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O o j r ^ r x . o J o ^ ^ c n t o i n t f i n o j o i c o c M c i i t r o o t o t r c o t D C M r o r ^ o i t f c n m c D c o t t t r t o L o o r o o ^ o c T > c M r o t 3 ) o i r o c o r x - O t r OTttrx.tocMCMOJOjmcotMOJ^tjT-cooDro t f o i o i t M o o i o o f x - t o i n i n t r ^ t t t f r o r o r o m m r r i r o CO C M *- i -t r t D O J t o c o o i i n o i t o i n i n c o o i r o - x r i n o i C D i n ' - co i n ^ ' O ^ ^ o ' ^ ^ ^ ^ ^ n o ^ m S - ^ S S r l t r a i c D ^ c o c u c o o o r o r - ^ r o ^ c o o i n c o c o t x . c M O c o i n r n o 0 ^ c o c o r x . c o c o t r t f t T r o ^ c M r x i n t f r o m r o c M C M C M C M C M ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ , . . . . . . ^ c M r o ^ i n t o r x - M O ^ O ^ t M c o ^ i n t o ^ c o t o S S s ^ s S w O T O T m i n t M O i i n r o i n m r o t f o C M t ^ 3 ^ 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 226 Vi cu "3 c cu a a. < lm CCS lm © Vi CU W e CCJ -w Vi •3 cu s cu WD e CU CU CU u Vi T3 O x: CU = OH a B CU l a cu cu .o s CU cu o c o • M "cu h. s-o u CU B CU -S a o U a> a cf O CD O cfi CU to O o a o 3 fc o CJ CD CU .2 & ° | 1 c/5 j CU Q . O u <: CD CJ C cd -t-> CU cu CU ccJ ts Q C N r-O N oo < ©' o £ ON V O o V O O —< 00 ON 00 o o o CN «n CN O cN m O N O N oo o o o V O 00 oo ON ^ oo oo •< o o Iz; o r-- o oo m o r- r- o © o o >n vo vo o oo o o o © V O T f cN m J O N O N • < o o i — (N O O N r~ cN 00 00 00 o o o 00 O N i— i O N t— C O 00 OO 00 o o o CN ~* ~ T f O N O N < o o £ CN O N <-i CN V O O N O N 00 © o © O m o Ti-<-< O N O N O N o © o oo oo O N T f ^ 00 00 < © ' © ' £ O N 00 — i r~ vo vo r- r- vo © d © V O C O T f — l T f oo oo d d © CO o > CU l _ PH -1—» CO g -b cu s *° £ o tu « 'O CU go S KJ CU 'c9 -2 CCj CO > .-_ ccJ JU -4—1 CCS H H £ o m cu _Z c •o CU go ^cS.2 > "c3 CO P H O £ c3 o H <: o V O CO <u "e3 < o V O Q ^ CO g T 3 <U H * 0 CCJ I—( ^ o w g "O <u go CO | | J ® "e3 CO £ c^-S CO <u 6 CU o V O co c; T3 <U I - H £ o m cu T3 CU go ^c2-2 9 is M Q *J—» CO g cd i—i ^ o [ T i CU -a cu go " c O . 2 ® *c3 CO £ c ^ ' S +-» o H o V O CO CU 13 o V O ON ON < o o Z ; 00 vo CN CN O N O N O N 00 d d d m >n cN ro ro O O N O N O N ©" d d S i ? CO g T3 <U b CO h H ^ o ~o -a n j CU CU go co us 3 > "c3 CO £ c« -S CO cu s CU CQ o V O 227 Appendix XI Dendrograms produced via the Distance Wagner Procedure using Prevosti's genetic distance (A: total data set 60B; B: females only 60B; C: males only 60B). Trees were rooted at the midpoint of the longest path. A. Total Data Set 60B Total length of tree = .801 Cophenetic correlation coefficient = .912 Distance from root .00 .02 .05 .07 .09 .11 .14 .00 .02 .05 El GOLDEN-2 » E2_GOLDEN-119 W3 PRTNCEGEORGE W2 MCBRJLDE-4&5 Wl MCBRIDE-26 P2 GOGAMA2 Pl_GOGAMAl ******* SMII_KITIMATn """"""" SMI KITIMATI S V A N I G O L D R T V E R ** S V A N 2 _ N G O L D R I V E R ********* SVAN3 TOFEvo .07 .09 ^ + + + .11 .14 Appendix XI Continued 228 B. Females only 60B Total length of tree - .887 Cophenetic correlation coefficient = .931 .00 + — .02 .05 .07 .00 .02 .05 .07 Distance from root .09 .12 .14 • — + — + — + — + »** ******* El GOLDEN-2 E2 GOLDEN-119 Wl_MCBRIDE-26 W2_MCBRIDE-4&5 ' W3_PRTNCEGEORGE ************** P2 GOGAMA2 PI G O G A M A 1 S M I I KITIMATII SMI_KITIMATI ™ ™ * * * S V A N I G O L D R T V E R S V A N 2 _ N G O L D R T V E R i > i > < i i > * * > i > i > < i > t A A i > * * > * > * » * S V A N 3 X O F I N O .09 .12 .14 229 Appendix X I Continued C . Males only 6 0 B Total length of tree = .961 Cophenetic correlation coefficient = .898 .00 .02 + + .00 .02 .05 .05 Distance from root .07 .10 .12 .15 -+ El G O L D E N - 2 E 2 _ G O L D E N - 1 1 9 ** W2 M C B R I D E - 4 & 5 W3 P R T N C E G E O R G E Wl MCBRIDE-26 P2 G O G A M A 2 PI G O G A M A 1 SMII_KITEVIATn *********** SMI KITIMATI • SVAN I G O L D R I V E R SVAN2_NGOLDRTVER ******* SVAN3 T O F I N O — + — + — + — + .07 .10 .12 .15 230 o CU < © VO CU c n CS -4-1 « •a o fe cu o PQ c n . , D . cu a x X -a c CU a. Q. < "O cu CQ c n r s CQ U CQ ir> 5 CN • -e CQ fc. CU •*-< CQ CU fc. OX) fe _© B CU > CU c n cn cu S CU S cr cu G O 3 a, o P H "<3 < 2 ecj O P i J E 8 z w ,3 s8 z w Q s 8 til Q w Q 5 ICQ 5 i s < & > 2 ,3 co Z w > Z Q CO o O T f vo m in 00 CN in IT) in in CN t> O N 1—1 00 CO vo vo CO CO V O CO V O CN r- CO V O co V O © O N © O N © O N o O N © © © o ' d © © ©' © d ©' © © © T f vo VO CO ro vo O © ro in T f ro vo © © oo CN 00 T f in © © C N oo vo ro >-H 00 © © <N 00 T f in i—i 00 © © ' © © O N — i CN r-d ©' , - C o V O T f CS C-» r- C N © ' © 00 CN co vo CO N O © © r- co I - H 00 © O N o" © T f V O vo co ro vo © © O. CT V O ro ro r--00 ^ m T f © © © ~ © ~ H O ^ © © - H © ^ m m T f m m T f © © ' T f vo ro vo © O N © © 00 CN vo ro © O N © © o ^ o —' r- l O N m T f © O N © ' © ' © ~ © —i o ~ © — ' © I-* o- cr 00 0O CN vo ro © O N vo f~ ro CN CN ro o © © V O T f T f vo CO V O ro vo © O N © O N d © ' © ' © r- ro 00 © O N © ' © ' © r~ ro CO oo l—H 00 o O N o O N d © ' © © vo T f ro V O © O N o" d o 00 CN CO 00 r—< 00 © O N © O N © © © © ro 1—( 00 o O N © d © i—i O N m T f © O N o © © ' O N 00 ro CN r- ro V O CN T f m © ' © ' © © vo T f O © O N © 00 CN <—' 00 ro V O © ' © © ' © ' 00 CN 00 CN vo ro 00 •—« oo T f in d d o" d O H cr o . cr 00 «t-l V O ro ro ro ro ro 00 CN V O CO © O N d d r-~ co V O CO CN © d 00 CN r- CN T f m © © ' V O T f 00 <—i o Ov d © © —i © —i © ~ H © <-< © -^c m in C O vo O O N © ' © ' © — I O H C T t-~ CO f H oo © O N © © O N i - H CN o ^ © —' © •-' © * H © © —I © - H T f V O <—i 00 CO V O © © ' 00 CN O O N T f m © © ' © o ^ O N in T f ©' d O H CT O N CN 231 o o H GO CD 'o CD cx co CU CD CO O H 1 cn CD £ CD Q 00 cn CU 'o cu o-cn CD CU cn CU S CU Q t-- r—1 vo 0 0 0 0 vo ON cn ON i n r—1 0 CN O O CN O 0 O O O O O O VO <n cn O 00 CN ON VO IT) cn r-- vo CN CN cn cn CN CN cn O d © O O d O Ov O N CN CN vo Os cn i n 0 CN ON CN O O CN 0 CN d d d d d d d •a 60 oo C+-< vo VO CN vo cn O N o\ t-- CN cn cn cn cn cn cn cn CN CN cn 3 o o h-l ed o H 00 cn a. 3 o O ccj 0 H 1 cn CD s CD Q on cn O -3 O 6 cn CD s CD Q CN , — 1 CN CN CN 0 0 cn ON cn «—' CN <—1 CN d d d d d C N C N O cn d d VO cn cn O CN CN ON VO cn f- CN CN T f cn cn CN CN cn d d O O O O d cn ON cn 0 0 0 < — 1 O cn iri 0 vo CN O O CN 0 CN O d d d d d d d -0 <+-i 60 60 u-t ••p vo VO CN VO cn ON ON r- r-- CN cn cn 1—1 1—1 cn cn cn cn cn CN CN 232 Appendix XDH Nei's Genetic Identities (1972) Part A: RAPD MARKERS- TOTAL DATA SET 60A Region/Group3 Region/Group SsVan.h. SsMainiand Se-SwCo„,piex jPine SSvanls 0.963 (3 populations) (0.954-0.981) SSMaWand 0.928 0.972 (2 populations) (0.907-0.943) Se-Swcompiex 0.926 0.924 0.968 (5 populations) (0.895-0.941) (0.914-0.939) (0.950-0.986) jPine 0.931 0.922 0.955 0.975 (2 populations) (0.908-0.951) (0.910-0.930) (0.934-0.976) a: Ranges indicated in bracket Part B: RAPD MARKERS- TOTAL DATA SET 60A Group SSvan.Is. Group3 Sw-Se-jP SSvan.Is. (3 populations) 0.963 (0.954-0.981) ^ "^ Mainland (2 populations) 0.928 (0.907-0.943) 0.972 Sw-Se-jPine (7 populations) 0.928 (0.895-0.951) 0.923 (0.910-0.939) 0.962 (0.934-0.986) a: Ranges indicated in brackets Part C: ALLOZYMES- PHILLD?S AND LANEER (1985) Region3 Region Pacific Coast Rocky Mountains Northeast Pacific Coast 0.949 (3 populations) (0.922-0.963) Rocky Mountains 0.945 0.974 (5 populations) (0.899-0.989) (0.951-0.983) Northeast 0.953 0.973 0.986 (9 populations) (0.918-0.988) (0.938-0.995) (0.965-0.998) a: Ranges indicated in brackets 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0075158/manifest

Comment

Related Items