Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Genetic variation in western red cedar (Thuja plicata Donn) seedlings Cherry, Marilyn L. 1995

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

Item Metadata


831-ubc_1995-982675.pdf [ 4.33MB ]
JSON: 831-1.0088835.json
JSON-LD: 831-1.0088835-ld.json
RDF/XML (Pretty): 831-1.0088835-rdf.xml
RDF/JSON: 831-1.0088835-rdf.json
Turtle: 831-1.0088835-turtle.txt
N-Triples: 831-1.0088835-rdf-ntriples.txt
Original Record: 831-1.0088835-source.json
Full Text

Full Text

GENETIC VARIATION IN WESTERN RED CEDAR (Thujaplicata Donn) SEEDLINGSbyMARILYN L. CHERRYB.Sc.F., University ofNew Brunswick, 1981A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE STUDIESDepartment of Forest SciencesWe accept this thesis as conformingto the required standard.THE UNIVERSITY OF BRITISH COLUMBIAApril 1995© Marilyn L. Cherry, 1995In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives, It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature)Department of F(ThS7 SQjEr’JcI5The University of British ColumbiaVancouver, CanadaDate 1Z’)pfi ai,DE-6 (2/88)ABSTRACTTo determine whether the apparent lack of genetic variation in western redcedar (Thujaplicata Donn), as previously inferred by isozyme and terpene studies,would hold true for quantitative seedling traits, a provenance study was initiated toinvestigate patterns of variation in seedling growth and survival characteristics, coldtemperature acclimation, and response to inbreeding.Seedlings from ten coastal and ten interior provenances, half with familystructure (five families / provenance), were grown for three years at one coastal(Vancouver) and one interior (Salmon Arm) location. Twenty-three potted cloneswere both self-pollinated and polycrossed at Cowichan Lake; resulting progeny weremonitored for growth and frost hardiness.Genetic variation could be detected from the first year, and increased annually.The narrow-sense individual heritability, assuming some inbreeding, of final heightsof trees growing in Vancouver was 0.38. Height, root collar diameter, acclimation,and deacclimation exhibited mainly within-population variation, while variation indry weight measurements, foliar nutrient content, survival at Salmon Arm, andmaximum cold hardiness was evident mainly between populations. Coastal / interiordifferences were noted in first-year heights, branch number, height, survival, andcrown dieback at Salmon Arm following a severe winter in which trees sufferedmajor desiccation damage, and in acclimation and deacclimation. In general, adaptivetraits appeared to show more between-population differences, while traits under lessselective pressure showed mainly within-population variation.11Provenances displaying the greatest variation at the family level were thosefrom Vancouver Island. Between-population variability appeared to be highest in theB.C. interior, and lowest in northern B.C. populations.Elevation influenced all traits displaying provenance variation. Locationeffects occurred, and some genotype by environmental interactions were noted.Plasticity was evident in timing of growth initiation and cessation, timing ofacclimation and deacclimation, and in depth ofmaximum hardiness reached per year.Early traits showed little evidence of inbreeding depression, but there seemedto be a trend towards gradual expression of inbreeding depression over time, at leastin traits under selective pressure.This research showed that western red cedar is much more complex thanpreviously believed, and substantial genetic variation exists in several traits of thisspecies.111TABLE OF CONTENTSPageABSTRACT iiLIST OF TABLES ViiLIsT OF FIGuis ixACKNOWLEDGEMENTS xiDEDICATION xii1. INTRODUCTION 11.1. Objectives 11.2. Literature Review 51.3. Literature Cited 112. SEEDLING GROWTH 132.1. Introduction 132.2. Methods 162.2.1. Seed Sources 162.2.2. Seed Testing 192.2.3. Greenhouse Design 192.2.4. Germination and Nursery Culture 202.2.5. Outplanting 212.2.6. Seedling Data Collection 222.2.7. Foliar Nutrient Analysis 242.2.8. Data Analysis 252.3. Results 312.3.1. Seed Traits 312.3.2. Height and Root Collar Diameter 342.3.3. Dry Weights 422.3.4. Foliar Nutrient Analysis 422.3.5. Seedling Survival 432.3.6. Analysis by Seed Collection, GeographicalRegion, and by Cluster 472.4. Discussion 512.5. Literature Cited 60ivPage3. RESISTANCE TO ENVIRONMENTAL STRESSES 633.1. Introduction 633.2. Methods 743.2.1. Frost Hardiness Profile over a Broad Range of Temperatures 743.2.2. Effect of Crown Position on Frost Hardiness 753.2.3. Effect ofRate of Freezing on Measures of Cell Injury 753.2.4. Frost Testing over Two Winters 763.2.4.1. Greenhouse Design and Nursery Culture 763.2.4.2. Frost Hardiness Tests 783.2.4.3. Data Analysis 813.2.5. Variable Chlorophyll Fluorescence and Frost Hardiness 853.2.6. Calorimetry 873.2.7. Foliar Nutrient Analysis and Frost Hardiness Testing 883.3. Results 903.3.1. Temperature Profiles 903.3.2. Crown Position 923.3.3. Freezing Rate 923.3.4. Midsummer Resistance to Cold Temperatures 943.3.5. Coastal Hardiness Cycles 953.3.6. Acclimation 1033.3.7. MaximaiHardiness 1133.3.8. Deacclimation 1213.3.9. Variable Chlorophyll Fluorescence 1223.4. Discussion 1283.5. Literature Cited 1364. SELF-FERTILIZATION VS POLYCROSSING 1404.1. Introduction 1404.2. Methods 1424.2.1. Seed Sources 1424.2.2. Seedling Measurements 1454.2.3. DataAnalysis 1464.3. Results 1484.3.1. Cone and Seed Traits 1484.3.2. Seedling Growth: Height, Root Collar Diameter, and Dry Weight 1484.3.3. Frost Hardiness 1544.4. Discussion 1594.5. Literature Cited 162VPage5. CoNcLusioNs 1645.1. Literature Cited 168APPENDICES 169Appendix 1 General expected mean square equations 170Appendix 2 Mean squares: seed traits 181Appendix 3 Mean squares: height and root collar diameter 184Appendix 4 Mean squares: dry weights 190Appendix 5 Mean squares: grouping by geographic region 193Appendix 6 Mean squares: foliar nutrient analysis 194Appendix 7 Mean squares: survival traits 195viLIST OF TABLESPageTable 2.1. Geographic variables of provenances used in this study 18Table 2.2. Summary of measured growth traits 32Table 2.3. Germinant abnormalities noted during the first growing season 35Table 2.4. Percentage of the total variance attributable to provenanceand to family within provenance for measured growth traits 40Table 2.5. Results of individual provenance analyses of measurementstaken at UBC after the 1992 growing season onprovenances having family structure 41Table 2.6. Provenance rankings of final height at UBC (in cm) and% live foliage and % survival at Skimikin 45Table 2.7. Final UBC height (cm), RCD (mm), and dry weight (g)means ± standard deviation, percentage of the totalvariation attributable to provenance, and standard error(s.e.e.) of the provenance variance component by region 49Table 2.8. Comparison of geographical regions with groupings ofprovenances based on cluster analysis of certain traits 50Table 3.1. Analysis of variance results of frost hardiness testingindex of injury over two winters, plus one summertest, for all provenances 105Table 3.2. Analysis ofvariance results of frost hardiness testingindex of injury over two winters, plus one summer test,for provenances having family structure 106Table 3.3. Analysis of variance significance levels ofLT50 by testover two seasons 107Table 3.4. Analysis of variance results of frost testing index of injuryon three dates of seedlings growing at Salmon Arm 109vi’PageTable 3.5. Analysis of variance results of frost LT50 of trees grownat Skimikin compared to trees grown at UBC 112Table 3.6. Temperatures at which exotherms were observed duringcalorimetry experimentation on Feb. 15, 1992 in twotrees I provenance, plus provenance mean compared to LT50of these provenances as determined in normal frost testing 117Table 3.7. Summary of analyses from the various tests used to assessfrost hardiness, using the date closest to maximumhardiness where seasonal monitoring was done 120Table 3.8. Comparison ofprovenances based on ranking by LT50estimated from electrical conductivity testing and by variablechlorophyll fluorescence results from two test dates 126Table 4.1. Maternal parents used in the selfing / outcrossingtrials, classified by tester used in the polycross 143Table 4.2. Pollen contributions of the two testers used 144Table 4.3. Treatment (self-pollinated vs polycrossed) means± standard errors and ranges of family means ofstudied first-year growth traits 149Table 4.4. Summary of significance levels found throughanalysis of variance 150Table 4.5. Family variance components ± standard error, and%CYF/02Ttl for first year height and root collardiameters of selfed and polycrossed seedlings 152Table 4.6. Significance levels of treatment (S vs P) found whenseparate ANOVA’s of dry weight parameters wereperformed per family 155Table 4.7. ANOVA results of frost testing selfed and outcrossed seedlingsgrowing at Cowichan Lake during 1991 /92 (six families) andat Jordan River during the winter of 1992 / 93 (eight families) 156viiiLIST OF FIGURESPageFigure 2.1. Provenances and test sites within the species range 17Figure 2.2. Mean provenance final heights of trees growingat UBC contrasted with provenance elevation 36Figure 2.3. Mean provenance final heights and range of family /provenance means of trees growing at UBC 38Figure 2.4. Mean provenance survival after the winter of1991 / 92 of trees growing at Salmon Armcontrasted with provenance elevation 46Figure 3.1. Example of plotted index of injury vs testtemperature used to estimate LT50’s persample per test date 82Figure 3.2. Mean provenance index of injury profilecurves resulting from frost testing between-1O°Cand-50°ConDec. 19,1991 91Figure 3.3. Mean provenance LT501sofseedlings tested atabout twice the standard freezing rate comparedto seedlings tested at the standard rate of freezing 93Figure 3.4. Mean zonal LT501softrees growing at UBC overtwo winters of testing 96Figure 3.5. Mean provenance LT50’s oftrees growing at UBCover two winters of testing 97Figure 3.6. Mean family LT50’s of trees growing at UBCduring 1990/91 98Figure 3.7. Mean family LT501softrees growing at UBCduring 1991/92 99Figure 3.8. Daily maximum and minimum temperaturesatUBC during 1990/91 100ixPageFigure 3.9. Daily maximum and minimum temperaturesat UBC during 1991 / 92 101Figure 3.10. Apportionment ofa2Provenance anda2Familyof index of injury for each often test datesper year over two years of frost testing 102Figure 3.11. Family variance components of frost testindex of injury over two years for eachprovenance having family structure 104Figure 3.12. Mean LT501softrees growing at Salmon Armcompared to trees at UBC during 1991 / 92 110Figure 3.13. Daily maximum and minimum temperaturesat Salmon Arm during 1991 /92 111Figure 3.14. Foliage temperature, as monitored bycalorimetry, plotted against ambient testtemperature for one tree sample (a); thesame sample minus the control channel (b) 118Figure 3.15. FvIFm and Fv/F0 per test temperature in January 123Figure 3.16. Fv/Fm and Fv/F0 per test temperature in April 124Figure 4.1. Mean shoot, root, and total dry weights per familyfor selfed vs polycrossed one-year old progeny 153Figure 4.2. Mean frost test LT50 for selfed vs polycrossedprogeny during 1991 /92 157Figure 4.3. Mean frost test LT50 for selfed vs polycrossedprogeny during 1992/93 158xACKNOWLEDGEMENTSMany people have helped me throughout the preparation of this dissertationand deserve recognition here; I sincerely apologize if I have inadvertently overlookedsomeone.The National Science and Engineering Research Council and the ScienceCouncil of B.C. both provided funding for this work. Seed collections were providedby John Russell, Mike Carlson, and Barry Jacquish of the B.C. Ministry of ForestsResearch Branch, and by Gerry Rehfeldt of the USDA Forest Service in Moscow,Idaho.Zika Srejic and Lee Charleson provided some materials and cheerfullyassisted with seedlings growing at UBC’s South Campus Nursery. Much thanks isdue to John Russell and his people at the MOF Cowichan Lake Research Station forall the work done out of that site. Mark Martinez did a great job in helping totransport seedlings to Salmon Arm and in helping to prepare that site fortransplanting. Mike Carison kindly provided a planting crew for the Salmon Armsite, and Keith Cox must be thanked for all his generous assistance with the treesgrowing at the MOF Skimikin Seed Orchard at Salmon Arm. Yousry El-Kassabyprovided facilities at the Pacific Forest Products site in Saanich, and offered muchhelpful advice (and funding) in his role as industry collaborator.Edith Camm made available her lab’s chlorophyll fluorometer for this project.Al Balisky volunteered his lab’s datalogger equipment. Frank, Katie, Trevor, Signy,Tyson, and Joy were greatly appreciated for taking on an oftentimes tedious jobwithout complaint.My supervisory committee members are appreciated for their interest in andassistance with this project. John McLean deserves thanks for his continued supportand encouragement. Much thanks is due to Vera for all her support, and mostespecially for the many hours of dogsitting.My biggest thanks goes out to my supervisor, Gene Namkoong, for takingmyself and my project on in midstream, and for being such a kind and inspiringmentor.xiDEDICATION— Dedicated to Doug —Many a night I saw the Pleiads, rising thro’ the mellow shade,Glitter like a swarm of fire-flies tangled in a silver braid.Here about the beach I wander’d, nourishing a youth sublimeWith the fairy tales of science, and the long result of Time- Alfred Lord Tennyson, Locksley Hall, 1842The great tragedy of Science - the slaying of a beautiful hypothesisby an ugly fact.- T.H. Huxley, Biogenesis andAbiogenesis, 1870xli1. INTRODUCTION1.1. OBJECTIVESWestern red cedar (Thujaplicata Donn) is not a major commercial species inBritish Columbia (at 7.67 million m3 volume harvested, comprising 9.68 % of theProvince’s total log production in 19931), but is nevertheless considered to be animportant value-added timber species on the coast, as the wood is valuable andexhibits some unique properties. Demand for western red cedar wood products hasrisen in recent years (Minore, 1983), resulting in increased harvesting levels andhence a greater demand for planting stock. Current B.C. nursery sowing requestlevels for this species call for 10.712 million seedlings provincewide in 1994, up from2.99 1 million seedlings which were requested in 1983 (Scott Lohnes, B.C. Ministryof Forests, pers. comm., 1994).The extent of genetic variability in western red cedar is unknown, althoughthis species has widely been presumed until now to be genetically depauperate,largely based on the results of isoerizyme and leaf extractive studies. However,isozyme variability may not reflect amounts of variation in metric traits (Muona,1988) and thus inferences about the latter cannot be made based on the former. It isnot known whether quantitative traits of interest have the potential for geneticimprovement in this species. An understanding of the genetics ofwestern red cedar isI From unpublished statistics of the B.C. forest industry, as compiled by the Council ofForest Industries, Vancouver, 1994.1therefore desirable to provide knowledge which may later be of use in forestregeneration, forest maintenance, and in the understanding of biodiversity at thespecies level.The intent of the present research was to study the extent and pattern ofgenetic variability in certain growth and adaptive traits of western red cedar seedlingsto provide knowledge which will further the understanding of the genetics, growth,and physiology of this species, aid in seed zone delineation, offer information whichmay be useful in the development of breeding programs for this species, and aid indeveloping seedlings which may be better suited to their plantation environment.The objectives of this research were the following:- to estimate the extent and allocation of genetic variation found in growth,morphological, and survival traits ofwestern red cedar between coastal andinterior range populations and between and within provenances and familiesusing quantitative genetic analyses- to investigate the effect of test location on growth and cold hardiness traits- to study the effects of self-fertilization on growth and cold hardiness of seedlings- to aid in the delineation of seed transfer guidelines and to make preliminaryrecommendations regarding selection and breeding strategies for the westernred cedar breeding programThe generalized null hypothesis was that no detectable genetic variation ingrowth, morphological, and survival traits exists between coastal and interiorprovenances or between or within populations. The null hypothesis for frost testingwas that genetic differences in the responses to cold temperatures do not occur andthat the range of variation between and within populations does not increase ordiminish from the time when trees begin hardening until total dehardening hasoccurred. The null hypothesis for self-fertilization studies was that no differences2could be detected between selfed and outcrossed progeny.The first objective was to partition any variability found in the characteristicsstudied into genetic and environmental components. Estimates could then be derivedfor how broad or narrow seed zones should be and whether it is worthwhile todevelop intensive breeding programs for this species. Knowledge of seedling growthand physiology gained by this study could be applicable in the regeneration of thisspecies.The second objective was to determine whether differences could be detectedbetween seedlings growing in a relatively unstressful, favourable coastal site andthose grown in a relatively harsh interior environment, and to detect the presence orabsence of any genotype by environment interaction. Again, this knowledge wouldaid in achieving the last objective.Ifwestern red cedar were bereft of genetic heterogeneity, inbreeding might notbe detrimental. The intent of the third objective was to investigate such effects ingrowth and adaptive traits of the progeny of self-pollinated and outcrossed maternalparents.The fourth objective, to provide knowledge which would assist in thedelineation of seed transfer zones and to make preliminary recommendations aboutselection and breeding of this species, would be based on the results found throughthe achievement of the first three objectives.The genetics of western red cedar seedlings were examined by studyingpopulations from the coastal and interior ranges (henceforth referred to as zones) ofthis species. Provenances (geographic origins of the seed sources) having available3nonbulked seed were selected from each zone, halfwith family (open-pollinatedmaternal half-sib) structure. Seedling samples were grown at one of two sites: acoastal site representative of a mild, favourable environment and an interior siterepresenting a relatively harsh and stressful climate.To investigate whether variability in seedling growth traits was evident,seedling heights were measured for three years. After the third growing season hadconcluded, root collar diameters were measured on all trees. A subsample of treeswere selected for dry weight measurements, while another subsample was tested forfoliar nutrient content. Adaptive traits were represented by seedling survival at twosites and by frost hardiness of trees growing at two locations. By determining thelevels of any variation, it may be possible to discover whether or not differentprocesses are acting on different types of characteristics.To compare the effects of inbreeding with outcrossing, maternal parents wereboth self-pollinated and polycrossed; the resulting progeny were grown and thenmeasured for height, root collar diameter, dry weight, and cold hardiness. Shouldevidence of inbreeding depression be found, this might imply that inbreeding may notbe the primary mating strategy in natural stands, and some genetic variation mustoccur in measured traits.41.2. LITERATURE REVIEWWestern red cedar, a member of the family Cupressaceae, is one of twomembers of the genus Thuja native to North America, and the only Thuja speciesnaturally found in the western part of the continent. Western red cedar, actually anarborvitae and not a true cedar, is sometimes referred to as western redcedar.Western red cedar has a fairly widespread natural range. The coastal portionof the range extends along the Pacific coast from southeastern Alaska, at a latitude of56°30’, to northern California at a latitude of 40°10’ (Burns and Honkala, 1990). Aswell, this species has an inland range occurring along the western slope of the RockyMountains from west of Prince George, B.C. (at 54°30’ latitude), to northern Montanaand Idaho (45°50’ latitude) (Minore, 1983; Burns and Honkala, 1990). The coastaland interior populations are separated except for a small contiguous region comprisedof scattered patches extending along the B.C. - Washington border.Interior western red cedar can be found at elevations from 320 m to 2,130 m(Burns and Honkala, 1990). Coastal western red cedar grows from sea level to about2,300 m in Oregon. In coastal B.C., the upper elevational limit is lower, at about1,200 m. In Alaska, the upper altitudinal limit is still less, at 910 m (Minore, 1983;Burns and Honkala, 1990), and at its northernmost limit, the presence ofwestern redcedar stops abruptly at 300 m elevation (Pojar and MacKinnon, 1994).Although western red cedar may form pure stands, it is usually associated withother tree species, and may be present in any stage of forest succession. On the coast,coniferous associates include yellow cedar (Chamaecyparis nootkatensis [D. Don]Spach), Port-Orford cedar (Chamaecyparis lawsoniana [A. Mum] Parl.), coastal5Douglas-fir (Pseudotsuga menziesii var. menziesii [Mirb.] Franco), Abies Mill.species, Sitka spruce (Picea sitchensis [Bong.] Carr.), western hemlock (Tsugaheterophylla [Raf.] Sarg.), mountain hemlock (Tsuga mertensiana [Bong.] Carr.),lodgepole pine (Pinus contorta Dougi.), western white pine (Pinus monticola Dougi.),and Pacific yew (Taxus brevfolia Nutt.) (Bums and Honkala, 1990). In the interior,western red cedar grows with interior Pinus and Picea species, interior Douglas-fir(Pseudotsuga menziesii var. glauca [Beissn.] Franco), western hemlock, western larch(Larix occidentalis Nutt.), and Pacific yew (Bums and Honkala, 1990).In B.C., western red cedar is predominantly found in the Coastal Douglas-fir,Coastal Western Hemlock, and Interior Cedar Hemlock biogeoclimatic zones (Krajinaet al., 1982; Meidinger and Pojar, 1991). The ecological amplitude ofwestern redcedar appears broad with regard to soil moisture and richness. This tree grows inpoor to rich soil nutrient regimes and in dry to very wet available soil moistureregimes, and is very resistant to prolonged spring flooding (Krajina et al., 1982;Rushforth, 1987).This species has a high shade tolerance, although less than that ofwesternhemlock, Pacific yew, and Pacific silver fir (Abies amabilis [Dougl.] Forbes). On drysites, western red cedar is shade-requiring (Krajina et aL, 1982). The frost toleranceof this species is not as high as that of some of its associated coniferous species, andwhere sufficient precipitation is present, its range seems to be limited by lowtemperature extremes (Bums and Honkala, 1990). Winter desiccation damage canalso be severe (Miller, 1978). Adams and Mahoney (1991) observed thattranspirational stress was more detrimental to this species than light competition,although growth reductions were influenced by below-ground competition.6Western red cedar, as with all members of the Cupressaceae, hasindeterminate shoot growth, without winter bud formation or preformed shoots. Thusmeristems do not benefit from the protection of bud scales in winter. The foliage haslittle cutin and wax and hence is poorly protected from excessive transpiration (Burnsand Honkala, 1990). However, shoot growth is opportunistic and elongation occursas long as climatic conditions are favourable. Shoots have a longer period of growththan that of any associated conifers (Bums and Honkala, 1990). This species hasbeen known to undergo shoot elongation into December in coastal B.C., but shootgrowth in the spring takes longer to commence than does that of its determinateassociates.Western red cedar contains extractives, notably the thujaplicins, in itsheartwood, which act as natural fungitoxins and make western red cedar one of themost resistant species to pathogen attacks, contributing to the longevity of thisspecies. However, over the centuries, biodegradation of these extractives by fungimay gradually occur (Bums and Honkala, 1990). Nevertheless, this species can livefor about 1,000 years.Naturally occurring asexual propagation is thought to be common, at least inIdaho stands (Minore, 1983). Polyploidy has been noted in members of theCupressaceae (Minore, 1983). Haploid and triploid members of Thujaplicata havebeen found, as have 4n Juniperus. Wright (1976) speculated that ploidy may conferan adaptational strategy for the invasion ofnew habitats, as polyploids often exhibitfaster than normal growth and so may have an advantage in occupying a site.The recent history ofwestern red cedar has been difficult to reconstruct, inpart because pollen among members of the family Cupressaceae is indistinguishable7and also difficult to separate from Taxus pollen (Hebda and Mathewes, 1984) and inpart because of incomplete and sporadic pollen records. Critchfield (1984), inexamining the impact of the Pleistocene on the genetic structure of conifers in NorthAmerica, could find no explanation in fossil records for the presumed reducedvariability ofwestern red cedar as reported in earlier studies.Due to severe climatic conditions during the last ice age, even in nonglaciatedareas, it is unlikely that any populations ofwestern red cedar were able to survive onnunataks which have been hypothesized for parts of the Queen Charlotte Islands,Alaska, and Vancouver Island. It is not known whether this species took refuge inone or in several disjunct refugia south of the ice sheets’ borders.The last glaciation was believed to have reached as far south as 47° to 48°latitude in western North America (Booth, 1987). Near the coast, the Puget lobe ofthe Cordilleran ice sheet reached to just south of the Puget lowland. In Idaho, the icesheet reached as far south as about the southern shore of Lake Pend Oreille. A recordof pollen dating back to 13,800 years before present (BP) at Davis Lake, situated justsouth of the Puget lobe and hence beyond the limit of the last glaciation, shows thepresence of Cupressaceae pollen back to 13,800 years BP (Baker, 1983). Sporadicpollen records of the Cupressaceae are evident from Davis Lake to northernCalifornia, but go no further south than that, at about 39° latitude (Baker, 1983).The ice sheets receded in an inland direction from the west coast ofNorthAmerica beginning about 14,000 years BP (Hughes, 1987; Heusser, 1989), and beganreceding in Idaho sometime between 14,000 and 11,000 years ago (Waitt andThorson, 1983). Thus species reinvasion could have occurred at that time accordingto rates ofmigration had conditions been favourable. However, it appears that8climatic and edaphic conditions were not favourable for western red cedar habitation(Heusser, 1983). It is believed that warming began about 12,500 years BP (Barnoskyet aL, 1987), and from about 10,000 to about 6,000 years BP, the climate was verywarm and dry, with summer droughts (Wright, 1983). Western red cedar is believedto have spread into B.C. from southwestern Washington from about 10,000 years agoonwards, but was only present in low amounts until about 5,500 years ago along thecoast and on the Queen Charlotte Islands, when it is believed that weather patternsbecame cooler and moister and hence generally more favourable for this species(Bamosky eta!., 1987). An increase in the abundance ofwestern red cedar hasoccurred over the last 3,000 years (Mathewes, 1989), with no immediate explanation.Western red cedar’s extensive range is similar to that ofwestern hemlock. Therefugial ranges of both species were believed to be similar (Baker, 1983) and bothwere restricted in reoccupation of previous territories until the warm, dry climaticconditions began to change (Heusser, 1983). Where western red cedar is common, itspollen forms a large percentage of the pollen cloud (Baker, 1983), and thus should becapable of rapid population expansion (Yeh, 1988).Yeh (1988) suggested that gene flow among neighbouring populations is quiteextensive, with no effective barriers; expectations would be that divergence betweenpopulations would lessen with extensive gene flow. The isozymes studied by Yeh fitHardy-Weinberg expectations; from those results, and from the fact that this speciesis widespread, he surmised that severe inbreeding does not occur. Assuming ageneration time of 20 years for western red cedar, he calculated that about 500generations have occurred since a bottleneck about 10,000 years ago. However, Yeh’sestimated generation time of 20 years for western red cedar is extremely questionable.9Using an approximate generation time of about 200 years (longer than that on thecoast and possibly shorter than that in the interior), which is a more realistic figurewhen the ecology of the species is considered, and a time of about 6,000 years, whichis when western red cedar became more prevalent in formerly glaciated areas ofB.C.,only about 30 generations would have occurred to the present time, with fewer thanthis on the coast, and more than this in the interior.In summary, western red cedar has various features which distinguish it frommany of its associated conifer species. Its indeterminate nature, longevity, initiallyhigh resistance to pathogens, ability to withstand flooded soil conditions and low lightlevels, comparative inability to withstand very low temperatures and conditions ofhigh desiccation and sunscald, and its presumed paucity of genetic variation allcontribute to the uniqueness of this species. Western red cedar is inferred to havesurvived in one to several refugia south of the extent of the Pleistocene ice sheets, andbegan reoccupying its previously held ranges from about 10,000 years ago, with themajor influx into previously held territory in B.C. around 5,500 to 6,000 years ago.These ecological, silvical, and evolutionary history concepts must be understood andused as a foundation upon which knowledge of the genetics and physiology of thisspecies should be superimposed.The genetic architecture ofwestern red cedar, as revealed in seedling growth,survival, and responses to environmental stress (the latter two being consideredadaptive traits) were investigated. Since inbreeding depression is common in mostconifers, but lethals may have been purged in this species considering its evolutionaryhistory, the responses to inbreeding were also examined.101.3. LITERATURE CITEDAdams, D.L. and R.L. Mahoney. 1991. Effects of shade and competing vegetationon growth ofwestern redcedar regeneration. West. J. App!. For. 6(1): 2 1-22.Baker, R.G. 1983. Holocene vegetational history of the western United States. Pp109-127 In Wright, H.E. Jr., ed. Late-quatemary environments of the UnitedStates 2: 277 pp.Barnosky, C.W., P.M. Anderson, and P.J. Bartlein. 1987. The northwestern U.S.during deglaciation; vegetational history and paleoclimatic implications. Pp289-32 1 In Ruddiman, W.F. and H.E. Wright Jr., eds. North America andadjacent oceans during the last deglaciation. The geology ofNorth AmericaK-3, 501 pp.Booth, D.B. 1987. Timing and processes of deglaciation along the southern marginof the Cordilleran ice sheet. Pp 7 1-90 In Ruddiman, W.F. and H.E. WrightJr., eds. North America and adjacent oceans during the last deglaciation. Thegeology ofNorth America K-3, 501 pp.Bums, R.M. and B.M. Honkala. 1990. Silvics ofNorth America Vol. 1, Conifers.USDA FS Agric. Handbook 654, Wash. D.C., 675 pp.Critchfield, W.B. 1984. Impact of the Pleistocene on the genetic structure ofNorthAmerican conifers. Pp 70-118 In Lanner, R.M., ed. Symposium on historicaland genetic components of geographic variation patterns. Proc., 8th N.A. For.Biol. Workshop, Utah, 196 pp.Hebda, R.J. and R.W. Mathewes. 1984. Holocene history of cedar and native Indiancultures of the North American Pacific coast. Science 225: 711-713.Heusser, C.J. 1989. North Pacific coastal refugia - the Queen Charlotte Islands inperspective. Pp 91-106 In Scudder, G.G.E. and N. Gessler, eds. The outershores. Proc., 1St Inter. Sci. Symp., Queen Charlotte Islands Museum Press,327 pp.Heusser, C.J. 1983. Vegetational history of the northwestern United States includingAlaska. Pp 239-258 In Wright, H.E. Jr., ed. Late-quatemary environments ofthe United States 1: 407 pp.Hughes, T. 1987. Ice dynamics and deglaciation models when ice sheets collapsed.Pp 183-220 In Ruddiman, W.F. and H.E. Wright Jr., eds. North America andadjacent oceans during the last deglaciation. The geology ofNorth AmericaK-3, 501 pp.11Krajina, V.J., K. Klinka, and J. Worrall. 1982. Distribution and ecologicalcharacteristics of trees and shrubs in British Columbia. UBC Press,Vancouver, 288 pp.Mathewes, R. 1989. Paleobotany of the Queen Charlotte Islands. Pp 75-90 InScudder, G.G.E. and N. Gessler, eds. The outer shores. Proc., 1St Inter. Sci.Symp., Queen Charlotte Islands Museum Press, 327 pp.Meidinger, D. and J. Pojar. 1991. Ecosystems of British Columbia. BC Ministry ofForests Special Report #6, Crown Publ. Inc., 330 pp.Miller, P.R. 1978. Abiotic diseases. Pp 5-41 In Bega, R.V., ed. Diseases of Pacificcoast conifers. USDA FS Agric. Handbook No. 521, 206 pp.Minore, D. 1983. Western redcedar - a literature review. USDA FS Pacific NW For.and Range Exper. St. Gen. Tech. Rep. PNW-150, 70 pp.Muona, 0. 1988. Population genetics in forest tree improvement. Pp 282-298 InBrown, A.H.D., M.T. Clegg, A.L. Kaliler, and B.S. Weir, eds. Plantpopulation genetics, breeding, and genetic resources. Sinauer Assoc. Inc.Pubi., Mass., 449 pp.Pojar, J. and A. MacKinnon. 1994. Plants of coastal British Columbia. Lone PinePubl., B.C., 527 pp.Rushforth, K.D. 1987. Conifers. Christopher Helm Publ. Ltd., 232 pp.Waitt, R.B. Jr. and R.M. Thorson. 1983. The Cordilleran ice sheet in Washington,Idaho, and Montana. Pp 53-70 In Wright, H.E. Jr., ed. Late-quaternaryenvironments of the United States 1: 407 pp.Wright, H.E. Jr. 1983. Introduction. Pp xi-xvii In Wright, H.E. Jr., ed. Latequaternary environments of the United States 2: 277 pp.Wright, J.W. 1976. Introduction to forest genetics. Academic Press, 463 pp.Yeh, F.C. 1988. Isozyme variation of Thujaplicata (Cupressaceae) in BritishColumbia. Biochem. System. and Ecol. 16(4): 373-377.122. SEEDLING GROWTH2.1. INTRODUCTIONGenetic variation in shoot elongation has been studied in a few Cupressaceaespecies. Zobel (1983) studied the twig elongation patterns of Port-Orford cedar(Chamaecyparis lawsoniana [A. Murr.] Pan.) and found little genetic variability inphenology. Harry (1987) tested shoot elongation and growth plasticity of incense-cedar (Calocedrus decurrens [Torr.] Florin) seedlings to age three and obtainedsimilar results in that species. A further, recent study of this species (Rogers et a!.,1994) noted among-region variation in growth attributes at age 12 that had not beenapparent in younger seedlings. Studies of genetic variation in yellow cedar(Chamaecyparis nootkatensis [D. Don] Spach) showed significant family variation inseedling height (Cherry and Lester, 1992; Russell, 1993).There is a paucity of infonnation on the genetics ofwestern red cedar,although many horticultural varieties, involving crown form and foliage colouration,are known (Hillier, 1981; den Ouden and Boom, 1982; Rushforth, 1987). Poiheim(1970; 1972; 1977a; 1977b) studied chlorophyll and shoot form somatic mutationsand survival in haploids, diploids, and triploids of cultivated varieties. Nault (1986)found much tree-to-tree variation in thujaplicin content of old growth and secondgrowth cedar, and suggested that variation in this trait may be influenced genetically.A small number of disconnected studies of various characteristics, most ofwhich included only a few populations, provide mixed conclusions about the level ofgenetic variation in this species. Three early studies found variation between genetic13entries. Larsen (1953) found differences between two western red cedar clones inresistance to both the cedar leaf blight (Keithia thujina Durand) and to frost damage;these two traits were positively correlated in the clones tested. Søegaard (1966) testedresponses to cedar leaf blight in western red cedar and Japanese thuja (Thujastandishii [Gordon] Carriere). He concluded that resistance to the blight wasconferred mainly by a single gene, which is dominant in the T. standishii tested andrecessive in the T plicata used in the test, and suggested that resistance to frost maybe inherited by a similar mechanism. Ilmurzynski et a!. (1968) noted that growth ofan Alaskan provenance was inferior to one provenance from Oregon and one fromIdaho when grown in Polish nurseries and plantations.Other studies have shown evidence of little genetic variation among measuredtraits. Von Rudloff and Lapp (1979) obtained no significant difference in leaf oilterpene composition between coastal and interior provenances, and concluded thatthis species has one of the lowest degrees of variability in North American conifersinvestigated to date, along with red pine (Pinus resinosa Ait.). Von Rudloff et al.(1988) reanalyzed their leaf oil terpene data nearly a decade later using discriminantanalysis, and this time found minor differences among coastal provenances.Copes (1981) investigated isozyme variability of nine enzyme systems inseven provenances ofwestern red cedar, while Yeh (1988) studied variation in 15enzymes from eight provenances of this species; both found little geneticpolymorphism among populations. Yeh (1988) found that 14 of 19 loci weremonomorphic.Bower and Dunsworth (1987) obtained no significant differences among threelocal low elevation Vancouver Island western red cedar provenances in survival or14height growth of seedlings growing at three Vancouver Island plantations.A very recent study on inland western red cedar has just been published(Rehfeldt, 1994) involving 41 interior bulked provenances from seven major riverdrainages in Idaho, two drainages in Montana, and two in southern interior B.C., plusone Idaho provenance with family structure, all growing at three Idaho sites.Although Rehfeldt found both between- and within-population differences in three- orfour-year height at two out of three sites, in presence ofwinter injury at one site, andin mortality at one site, none of the traits studied varied significantly over all testsites. Weak elevational dines were detected in four-year height at one site and inpresence of winter injury at the same site, with r2 = 0.10 and 0.16 respectively. Awider sampling ofprovenances and families, spanning more of the species’ range, isneeded to define the structure of genetic variation in western red cedar and to obtain abroader picture of possible geographical influences.Early estimates of genetic variation in branch angle and in the allocation ofphotosynthate between the stem, branches, and root are desirable, and are known togive reasonable predictions of these traits in the mature tree. No reports have beenpublished on the genetics of shoot morphology in western red cedar, despite the manyknown horticultural varieties of this species having varying shoot forms.The purpose of this chapter was to investigate the genetic patterns andrelationships that may be found in western red cedar seedling growth (as measured byseedling height, root collar diameter, and dry weight allocation), morphology (branchangle), foliar nutrient content, and survival characteristics between and within zonesand provenances encompassing a wide range.152.2. METHODS2.2.1. Seed SourcesThe coastal range ofwestern red cedar was represented by B.C. Ministry ofForests (Cowichan Lake Research Station (CLRS), Vancouver Island) conecollections from Prince Rupert, the Queen Charlotte Islands, Vancouver Island, andthe Lower Mainland (Figure 2.1; Table 2.1). Unfortunately, seed could not beobtained from the Washington or Oregon portions of the coastal range of this speciesor from the coastal / interior transitional populations at the time that the study wasinitiated. The interior western red cedar range was represented by seed collected fromSalmon Arm, B.C. (B.C. Ministry of Forests, Kalamalka Research Station, Vernon,B.C., collections) and from northern Idaho and Montana (USDA Forest Service,Intermountain Research Station, Moscow, Idaho, collections). Within a provenance,cones were collected from individual trees which were at least 100 m apart, but withina one km range, to minimize relatedness between seed sources.Initially, ten provenances from the coastal range ofwestern red cedar and tenprovenances from the interior range were included in the study of seed and seedlingtraits. Due to a prohibitive number of seedlings involved, not all provenances wererepresented by family structure; provenances where families were bulked wereincluded so that the coastal and interior ranges could be more fully represented.Within each range, five provenances had family structure (five families perprovenance), with the remaining five provenances per zone having no familystructure, for a total of 60 genetic entries. Provenance abbreviations, as usedthroughout the thesis text, figures, and tables, are listed at the bottom of Table 2.1.16Figure 2.1. Provenances (.) and test sites (A) within the species range (after Lowery,1984).17Table 2.1. Geographic variables ofprovenances used in this study (F indicates familystructure); for analysis, latitude and longitude minutes were converted tohundredths of a degree, but their whole minute values are given here.Mean # Mean Mean dailyElev. growing annual JanuaryProvenance rn) Latitude Longitude prec. (cm) temp. (°C)CoastalMasset’ 20 54°00’ 132°00’ 160 125 2.5Oliver Lake 65 54°17’ 130°15’ 200 150 2.5Quinsam 250 49°58’ 125°32’ 140 150 2.5Tofino 50 49°01’ 125°35’ 230 300 5.0Mt. Benson 700 49°08’ 124°02’ 180 100 2.5Mill Bay 100 48°37’ 123°33’ 220 125 2.5Squamish 100 49°47’ 123°08’ 220 200 0.0Cheakamus 690 50°05’ 123°03’ 120 250 -5.0Whonnock 205 49°13’ 122°26’ 200 175 2.5Hope 600 49°28’ 121°17’ 200 125 -2.5InteriorMt. Mara Low 550 50°40’ 118°45’ 100 50 -5.0Mt. Mara Mid 1,100 50°43’ 1 18°47’ 80 65 -7.5Mt. Mara High 1,950 50°45’ 1 18°49’ 60 75 -10.0Benton Flat 686 48°21’ 1 16°50’ 120 65 -4.0Kaniksu 1,219 48°03’ 116°11’ 100 90 -4.5Lob 945 47°09’ l14°56’ 100 60 -6.5St. Joe 1,097 47°03’ 116°37’ 140 65 -1.5Pierce 1,509 46°29’ 1 15040t 100 100 -3.5Kooskia 411 46°08’ 115°40’ 130 90 -1.51 The following three-letter provenance abbreviations have been used throughout the thesis:Mas Mib (F) mmL (F) LolOil Squ mmM (F) StjQui (F) Che (F) mmH (F) PieTof (F) Who Bfl (F) KooMtb Hop (F) Kan182.2.2. Seed TestingIn order to determine whether seed traits influenced subsequent growth, andwhether variation could be detected between genetic entries even prior to the seedlingstage, certain seed characteristics were investigated. Seed weights were taken on 100cleaned nondewinged seeds per seed source.A random sample of 100 seeds per genetic entry was X-rayed for percentageof seed containing a complete embryo (% seed fill). A Softex Supersoft X-rayApparatus model TV-25-1 machine, set to run at 15 kVp and 3.0 mA for 7.0 secondsper exposure, was used for seed X-raying.A subsample of one hundred seeds per seed source was tested for percentagegermination. Seed from each source was placed into a plastic petri dish on amoistened Whatman filter paper and placed under broad-spectrum fluorescent growlights under a 12-hour photoperiod at room temperature. Petri dishes were kept moistas required to allow germination. The germinants were counted as they appeareduntil no further germination occurred. A tally of abnormal germinants was recordedwhen such abnormalities were noted.2.2.3. Greenhouse DesignThe initial greenhouse design was a randomized complete block designcommon garden study with ten replications (reps). Each genetic entry wasrepresented by one randomly assigned nine-seedling row per rep.However, very little germination occurred in provenance Silver Star from19Vernon, B.C. In one Silver Star family, most of the germinants had their cotyledons,rather than their radicles, emerge first from the seedcoat; these germinantssubsequently died. Silver Star could thus not be included. Only two of the Mt. MaraHigh elevation (near Salmon Arm, B.C.) families produced germinants. Thus theinterior zone was represented by only 22 genetic entries (instead of 30) for a total of52 genetic entries over both zones (rather than 60). Some families or bulkedprovenances did not have enough germinants to sow all ten replications; therefore thedesign contained missing cells.2.2.4. Germination and Nursery CultureAs seed stratification is not necessary for western red cedar, seeds were storedin a cooler at about 3°C until they were utilized. All seeds were pregerminated priorto sowing. Starting on May 7, 1990, seeds were put put into plastic petri dishes onmoistened Whatman filter papers and placed under a light table at the University ofBritish Columbia (UBC), having full-spectrum fluorescent lights and a 12-hourphotoperiod, at room temperature. The filter papers and lid of the petri dish were keptmoist by periodically squirting them with a water bottle as needed.When the emerging radicle of a seed was as long as the seedcoat, thegerminant was dibbled into a filled styroblock cavity using tweezers and narrowcavity spoons. Blocks used were Styro Vent 91 blocks, having 7 x 13 = 91 cavitiesper block and a volume of 133 cc per cavity. The soil mix used was the standard mixfor forest nurseries in B.C.: 2 bales peat: 1 bag vermiculite: 1.125 kg dolomite lime:225 g trace elements per m3 of soil. Each block could accomodate seven genetic20entries, which were planted in lengthwise rows, with one empty cavity after every twotrees per row to help minimize block edge effects.The first of the germinants were sown on May 16, 1990 and every daythereafter until all replications were filled or no more genninants were available for aparticular genetic entry. Due to unusually wet weather during the spring, some seedsrotted and hence some cavities had to be resown once or twice if extra seed wasavailable. The last germinant was planted on June 18, 1990.Styroblocks were set up in a raised outdoor compound at UBC’s SouthCampus Nursery in Vancouver, B.C. (elevation: 40 m; latitude: 49°1 5.5’ N; longitude:123° 13.8’ W; mean annual frost-free period: 244 days; mean annual precipitation:128.9 cm (Environment Canada, pers. comm., 1994); biogeoclimatic subzone:CWHdm (Gordon Kayahara, UBC, pers. comm., 1994)). Circular overheadsprinklers were used for irrigation and fertilization purposes. Trees were grown undera normal container nursery fertilization regime throughout the 1990 growing season.The seedlings were overwintered on the ground in the outdoor compound,partially covered with sawdust which was also banked around all blocks. Snowfallduring the winter of 1990/91 was sufficient enough to blanket the seedlings and thusfurther insulate them.2.2.5. OutplantingDuring early to mid May 1991, seedlings were transplanted from thestyroblocks. All odd-numbered replications (1, 3, 5, 7, and 9) were relocated to a21transplant bed at UBC’s South Campus Nursery. Even-numbered replications (2,4, 6,8, and 10) were shipped to the B.C. Ministry of Forests Skimikin Seed Orchard(Skimikin) at Salmon Arm, within the interior range ofwestern red cedar, andtransplanted into a bed within the seed orchard (elevation: 540 m; latitude: 50°47’ N;.longitude: 11 9°25’ W; mean annual frost-free period: 121 days; mean annualprecipitation: 49.3 cm (Envir. Canada, pers. comm., 1994); biogeoclimatic variant:IDFmw2 (Keith Cox, Ministry of Forests Skimikin Seed Orchard, pers. comm.,1994)). The latter group of seedlings were trucked to Salmon Arm while still in thestyroblocks to minimize handling and risk of injury.Both fields had been plowed and harrowed prior to transplanting. Trees wereplanted 1/3 m apart in straight rows using a flat planting shovel. At both sites, thebest eight out of nine seedlings per genetic entry per replication were outplanted ineight-tree row plots, maintaining the randomized complete block design with fivereplications per location.No further fertilization treatments were given to the seedlings. However,seedlings were occasionally watered at both sites during severely dry weather toprevent severe drought stress. Occasional weeding between rows and between treesin a row was also carried out.2.2.6. Seedling Data CollectionSeedling height measurements on all live trees were taken, to the nearest mm,on August 8 and 9, 1990 and at the end of the 1990 growing season after heightelongation had stopped. Seedling height measurements to the nearest 0.5 cm were22taken after the 1991 and 1992 growing seasons at the UBC and Skimikin locations.Root collar diameter (RCD) was measured to the nearest mm on all trees at UBC andSkimikin after the end of the 1992 growing season.Seedling dry weight measurements were taken on a subsample of trees at UBCafter the end of the 1992 growing season. Initially, dry weight measurements werealso planned for a subsample of trees growing at Salmon Ann. Due to a severe winterduring 1991 / 92 at the latter site, virtually all seedlings underwent severe desiccationdamage, and suffered either high mortality or foliage dieback and hence major shootdeformity. Thus it was decided that the dry weights of such stressed trees would notbe representative of trees growing at an interior site, and so these measurements werenot talcen.At UBC, trees were selected for destructive dry weight sampling after finalheight and root collar diameter measurements had been taken. For provenanceswithout family structure, all eight seedlings per replication were selected forsampling, for a maximum of 40 trees over the five replications.For provenances having family structure, four families out of the five familiesper replication were selected in each replication. The four families chosen rotatedover each replication, so that each of the original five families was tested in four ofthe five replications, and skipped in one of the replications. In each replication, twotrees per family were selected; wherever possible, trees #3 and #6 from the 8-tree rowpiots were chosen. If tree #6 was the last tree in a row, tree #5 was taken instead.Trees with double leaders or with insect damage were judged not to be trulyrepresentative and a neighbouring intact tree was sampled in its place. Thus eachprovenance with family structure was represented by two trees in each of four23families over five replications, for a maximum of 40 trees over all replications.Tree samples were labelled with tags and dug up with the root system intactusing a wedge-shaped planting shovel. Seedlings were then placed in plastic garbagebags and transported to the lab at UBC.For each seedling, the number of lateral branches coming off the main stemwas counted. In all trees from Replication #3, average branch angle per tree wasrecorded. Each tree was then cut up using garden pruners and separated into roots,main stem, and lateral branches with all of their foliage. Each of the three sectionsper tree was placed into a labelled paper bag. The bags were then placed into a BlueM Stabil-Therm drying oven and samples were dried for 24 hours at 65°C.Samples were weighed immediately after being removed from the oven.Contents were removed from the paper bags and poured into a tared plastic tub on abalance. Samples were weighed to the nearest 0.01 g.Seedling survival at both UBC and Salmon Arm was recorded for all years.2.2.7. Foliar Nutrient AnalysisTo estimate provenance level variation in foliar nutrient content of certainmacro and micronutrients, and to see whether any relationships between nutrientuptake and growth could be observed, a subsample was taken. On September 30,1992, foliage samples from nine trees in each of six provenances (Oliver Lake, MillBay, Tofmo, Mt. Mara Low elevation, Mt Mara Mid elevation, and Benton Flat) werecollected from 2-year old seedlings at UBC’s South Campus Nursery transplant bed.24All provenances except for Oliver Lake and Benton Flat were represented by threetrees from each of three families; the latter two provenances were represented by ninetrees (no family structure). Samples were placed in labelled paper bags and put into aBlue M Stabil-Therm drying oven, where they were dried at 60°C for 48 hours. Theywere then ground up using a Braun coffee grinder. Each sample was then sievedthrough a 1 mm (#18 mesh) screen and then placed into labelled envelopes of doublethickiiess.Samples were shipped to the MacMillan Bloedel laboratory at Nanaimo, B.C.where foliar nutrient analysis was carried out. Samples were analyzed for % N, % P,% K, % Ca, % Mg, ppm Mn, ppm Fe, ppm Cu, and ppm Zn.2.2.8. Data AnalysisThe overall means and standard deviations were calculated for all parameters.Means were also determined by zone, provenance, and family where applicable.Seed parameters were not replicated, so only minimal analyses could be done.Correlations were estimated on seed traits (percentage germination, percentageseedfill, and seed weight). Regressions were calculated to see whether height androot collar diameter measurements were related to seed parameters measured. Inparticular, seed weight can be assumed to be maternal in nature; relationshipsbetween seed weight and growth would indicate maternal influences on early growth.Germination and seedfihl are indicators of seed viability, and should these traits haverelationships with growth, it may be inferred that more viable seed sources alsoproduce germinants which have superior growth capabilities, at least initially.25Regressions were carried out on the provenance least-squares means of finalheight measurements taken at UBC and certain geographic parameters associated witheach provenance to study clinal or spatial patterns in variation which may occur.Due to the badly unbalanced nature of the height and root collar diameter(RCD) data, a subset of provenances having family structure was used in analyses inwhich variability was investigated only at the zone and provenance levels, and familystructure was ignored. For trees at UBC, the same individuals as chosen for dryweights were selected, as described above (Section 2.2.6). A similar selection methodas described for dry weight sampling was used to choose a subsample of individualsfrom provenances having family structure at Salmon Arm. Thus all provenances,both those with and without family structure, were represented by a maximum of 40seedlings over all replications per site in provenance-level analyses. Separateanalyses were also carried out with only those provenances having family structure, inwhich all seedlings from these provenances were included, and variability at thefamily level was investigated.Data were analyzed by analyses of variance (ANOVA) (SAS® PROC GLM),using Type IV Sums of Squares as the data still contained imbalances, and inparticular contained some missing cells. The GLM (general linear model) procedure,using the method of least squares, allowed for the following: testing of random andmixed effect models, control over hypothesis tests, covariate analyses, specificcontrasts, and analysis ofunbalanced data sets; other procedures did not have thesecombined capabilities.Zones were considered to be fixed effects, as all possible zones (coast,interior) for this species were included, with all other variables in the model being26considered random. Where necessary, Satterthwait&s Pseudo-F tests (Neter et a!.,1990) were calculated. Seedling age (in days) was used as a covariate in the modelsfor height and root collar diameter, as sowing took such a long period to complete andthus seedlings differed slightly in age. Replication by zone and replication byprovenance per zone were lumped into the replication by family per provenance perzone variable in most analyses, as these terms were not found to be significant andwere not of primary importance to the analyses.An analysis which included the variable location was carried out on heightstaken at the end of 1991 at both sites; the data for 1991 were also analyzed separatelyper location. At the end of 1990, all seedlings were still at UBC, so location was notpertinent. The 1992 growing seasons ofUBC and Salmon Arm were not comparableas the Salmon Arm site had suffered such severe desiccation damage during thewinter of 1991 / 92. Hence height data were analyzed separately by site for 1992. Ananalysis including the variable location was also carried out for root collar diameter.Variance components were estimated (SAS® PROC VARCOMP) for heightand RCD measurements on provenances having family structure. SAS® PROCUNIVARTATE and SAS® PROC DISCRIM were also performed; the former wasused to test for normality of the data while the latter tested for homoskedasticity, orequal variances of the dependent variable Lack of fit tests (residual plots) were alsoperformed to test for linearity of the models.From the dry weight data collected, shoot weights (foliage + stem weight),total dry weights (shoot + root weight), Shoot/root ratios, stem ‘/total wt. ratios, andshoot Wt./total wt. ratios were calculated. Correlations were estimated between alldry weight parameters, number of lateral branches, and height and root collar27diameter of trees sampled for dry weight measurements. Analyses of variance at theprovenance and family levels were carried out on all parameters, and variancecomponents were determined at the family level for these traits. Univariate anddiscriminant analyses and residual plots were also carried out to test for normality andheteroskedasticity of the data. Separate correlations and analyses of variance wereestimated on samples from Replication 3 in which branch angle had been measuredand was included in the analyses.Correlations were estimated between all macro and micronutrients analyzed.Analyses of variance at the provenance and family levels were also estimated.Correlations were estimated on the amount of foliage per tree escapingdesiccation damage during the winter of 1991 / 92 at Salmon Arm and tree heights atthe end of 1991 and 1992 and RCD after the 1992 growing season. Correlations werealso carried out on mean height and survival after 1991 and 1992 and on mean heightafter 1991 and survival after 1992 by site; a regression was also estimated todetermine whether survival in 1992 was dependent on mean height at the end of 1991at Salmon Arm. Genetic covariances between 1992 height and survival per locationwere calculated.Regressions were estimated between the mean percentage of trees planted in1990 still alive after 1992 and the mean percentage of live foliage after 1992 perprovenance at Skimikin and certain geographic parameters associated with each seedsource, again to investigate clinal or spatial trends in the variation of these traits.Analyses of variance were performed on foliage escaping desiccation per treeafter the 1991 / 92 winter, number of trees alive in 1991 and in 1992, percentage of28trees living in 1991 that were still alive at the end of 1992, and the percentage of treesinitially planted in 1990 still surviving after the 1992 growing season at each location.The latter trait was also tested in an analysis which included both locations. Variancecomponents were estimated for provenances with family structure for the amount offoliage surviving desiccation, number of trees living in 1991 and in 1992, andpercentage of trees alive after 1991 still living after 1992.Narrow-sense individual heritabilities were calculated for traits which gavesome evidence of family variation. The purpose of calculating heritabilities was notto reflect an accurate estimation for the species; rather, it was to give an indication asto whether earlier claims of little to no variation present in this species were supportedby this research. Thus it was considered of little importance to look at the effects ofzone, provenance, and location in these determinations, and so these effects were notincluded in the model.It is suspected that some unknown degree of relatedness, in particular someself-pollination, may be a factor in natural stands ofwestern red cedar. To beconservative, it was arbitrarily decided that thea2Family estimated /8 of the additivevariance rather than 1/4 of the additive variance as is normally used in half-sibanalyses. The ratio of /8 was thus midway between half-sib and full-sib estimations.The general form ofheritability calculations was:h2= 2.7*&F(PZa2E +oR(L)*F(P Z) +a2F(P Z)where: F family; P = provenance; Z = zone; R = replication; L = location;E error29Standard eors of the Family variance components were calculated.For seed data, 1992 height and RCD data, and dry weight data, provenanceswere split into five regions: the north coast of B.C., Vancouver Island, the lowermainland ofB.C., the B.C. interior, and Idaho / Montana populations. For eachregion, overall means and standard deviations were calculated. These statistics werealso determined for the seed data by seed collection source. Analyses of variancewere conducted in which region was a variable in the model. The proportion ofa2Provenance out of theo2Total was estimated from variance component estimates.Hierarchical cluster analysis for provenance groups was carried out on UBCfinal height, Skimikin survival, and Skimikin live foliage after winter desiccationdamage for comparison with the five geographic regions, using SAS® PROCCLUSTER with SAS® PROC TREE used to draw the output.302.3. RESULTS2.3.1. Seed TraitsExpected mean square equations that were used to determine the appropriateerror terms to be used in F tests for each factor are given in Appendix 1. Meansquares and significance levels (*** = P <0.001; ** = 0.001 <P < 0.01; * = 0.01 <P<0.05) as estimated by analysis of variance (ANOVA) are given in Appendix 2through Appendix 7. Where analysis of covariance was performed (denoted as*yjjj term in the linear model), the covariate term was adjusted to its mean(often shown as f3*(X - X) in the literature). Table 2.2 summarizes the overallmeans plus or minus the standard deviation, variables found to be significant atprovenance and family levels by ANOVA, narrow-sense individual heritabilities, andthe family variance component plus or minus standard error for measured growthtraits.ANOVA showed that zonal differences could be detected in percentage seedfill and seed weight, and provenance differences were evident in percentagegermination for those with family structure (Appendix 2.1). All seed parameters werecorrelated with each other to some degree. Germination varied widely between seedprovenances. Cones were collected in different years, and by different people. Also,evidence of insect damage to a few seed lots was found. It is possible that a fewweeks of stratification may have aided in causing better uniformity in speed ofgermination. Although it is believed that differences between provenances in seedviability are real, further analysis is needed for rigid conclusions regarding theviability of seed sources to be drawn.31Table 2.2. Summary ofmeasured growth traits; see text for a full description ofvariables and the units they were measured in.Overall ANOVA resultsVariable Mean ± st. dev. Prov. level Fam. level h2 a2E ± s.e.% Fill 68.8 20.53 ZWeight, g 0.0013 0.00023 Z%Germ. 55.5 23.43 PHeight‘90 Plugs 12.5 2.93 Z P Z P F 0.28 0.681 0.201‘91 UBC 37.4 8.06 P P F 0.33 6.732 2.255‘92 UBC 72.6 14.74 P P F 0.38 24.795 8.250‘91 Skim 29.0 7.45 P P F 0.12 3.048 1.167‘92 Skim 29.2 9.90 P Z F 0.12 3.856 2.486‘91,2Loc. 33.3 8.84 P P F 0.20 3.738 1.360RCD‘92 UBC 9.93 1.98 P F 0.16 0.181 0.078‘92 Skim 9.56 2.06 P F 0.23 0.330 0.163‘92,2Loc. - - P 0 0 0Dry WeightBr. angle 42.6 8.60- - -#laterals 18.7 3.31 Z 0.06 0.221 0.426Stemwt 8.4 3.58 P P 0.17 0.653 0.566Foliarwt 15.5 7.87 Z P P 0.25 4.209 2.584Shootwt 23.9 10.98 Z P P 0.22 7.523 5.136Rootwt 7.5 3.09 Z P 0.12 0.435 0.513Totalwt 31.4 13.46 Z P P 0.20 10.845 7.934Shootfp,oot 3.21 0.90 P P 0.29 0.065 0.037Shoot,-j-0 0.75 0.055 P P 0.20 0.0002 0.0001Stem/j’otal 0.27 0.049 P P F 0.34 0.0002 0.0001Table 2.2 continued...Nutrients%N 1.25 0.131 P F%P 0.26 0.022 P P%K 0.97 0.160 P P F% Ca 0.95 0.123 P P% Mg 0.21 0.032 PppmMn 114.46 23.345 Pppm Fe 48.02 20.043 Fppm Cu 7.32 2.629 Pppm Zn 15.89 7.500 P0.21 0.024 P PK/IN 0.78 0.149 P PK/Ca 1.05 0.278 P P FSurvivalUBC:%90/92’ 91.99 11.09 P ZSkim:%90/921 41.89 26.97 P%91/922 42.7 27.47 P 0 0 0#Live’923 2.9 1.91 P 0 0 0Fol. des.4 16.52 28.55 P Z P F 0.05 11.340 8.170‘92, 2 Loc.:%90/921 -1 Percentage of trees planted in 1990 still alive at the end of 19922 Percentage of living trees at the end of 1991 still alive at the end of 1992Number of trees alive at the end of 1992“ Percentage of foliage per crown surviving desiccation damage of winter 91 / 9233However, differences between seed sources were detected in the amount andtype of abnormal germinants. Table 2.3 lists the abnormalities which were notedduring the first growing season. Cotyledon morphology differences betweenprovenances could be seen. For example, germinants from Mill Bay had very short,thin cotyledons while germinants from Hope uniformly exhibited long cotyledons.These observations were obvious, and further analysis was deemed unnecessary.2.3.2. Height and Root Collar DiameterHeight and RCD measurements were not found to be dependent on seedparameters, except in a few cases where the coefficient ofdetermination values werevery low (all r2 were <0.12). These weak relationships were found for RCD withgermination and seed weight, and for heights measured in August 1990 and at the endof 1990 with seed weight. Thus maternal influences and growth advantages ofgenetic entries having greater seed viability did not materialize, and seed level effectscould be ignored.Stepwise regressions of final height of seedlings grown at UBC on elevation,latitude, longitude, estimated average annual days in the growing season, annualprecipitation, and daily temperature in January indicated that elevation was the mostinfluential factor of those tested on height, alone accounting for 30.9 % of thevariation in height between provenances (Figure 2.2). An elevational dinedetermined through simple linear regression found that 1 cm in height gain occurredfor every 167 m decrease in elevation. However, the best model fitted, according tothe stepwise procedure and to Mallow’s C, statistic, using a significance level of34Table 2.3. Germinant abnormalities noted during the first growing season.Abnormality Provenance Family # countedalbino Squamish - 2Oliver Lake - 2Masset - 1Mill Bay 1 1Mt. Mara Mid elev. 4 1chlorotic Mill Bay 1 5Squamish - 1double leader Cheakamus 3 1MilIBay 2 1Mt. Mara Mid elev. 4 1two radicles Whonnock - 1adventitious lateral Quinsam 5 1cotyledons emerged 1St Silver Star Low elev. 2 mostsingle fused cotyledon Whonnock - 1three cotyledons Kooskia 6Squamish 4Masset - 3Mill Bay 1 2Mill Bay 4 2Tofino 1 2Tofino 5 2I’ Oliver Lake - 2Mill Bay 2 1Quinsam 3 1Quinsam 5 1Cheakamus 1 1Cheakamus 4 1Mt. Mara Low elev. 4 1Mt. Mara Mid elev. 5 1Hope 5 1Benton Flat 2 1Lob - 1Pierce - 13585 80—....—‘75—a.70’IU65-.60-I‘05001,0001,5002,000Elevation(m)Figure2.2.MeanprovenancefinalheightsoftreesgrowingatUBCcontrastedwithprovenanceelevation.a. = 0.05 for entry into and staying in the model, was:HtUBC = 156.545 - 0.010*Elev. - 0.652*Longitude; R2 = 0.466; P = 0.0066For all height and RCD measurements, provenance differences were evidentfor provenances without family structure (Appendix 3.1), and provenance (except infinal height at Skimikin) and family differences occurred in provenances with familystructure (Appendix 3.2; Figure 2.3). Zonal differences were only observed in firstyear heights and in final heights at Skimikin.Location differences were obvious when 1991 heights at UBC and SalmonArm were analyzed together, but were not evident in root collar diameter (Appendix3.3; Appendix 3.4). The seedlings growing on the coast were able to take advantageof a much longer growing season. After 1991, trees growing at UBC were about23 % taller than those at Skimikin; after 1992, trees at UBC were more than double inheight compared to trees at Skimikin. Replication and rep by family interactionswere significant, indicating probable microsite differences. Genotype by environmentinteractions were noted at the family level for 1991 height (for location* family,P = 0.0208) and at the zonal level for RCD (P = 0.0072).The trees at Salmon Arm which were still alive to be measured by the end of1992 were virtually no taller than they had been one year earlier, due in part to foliagedieback over the previous winter and in part to the fact that the trees had beenseverely stressed and grew in height little or not at all in the season after damage hadoccurred. Root collar diameters measured at Salmon Arm after 1992 were about thesame as those measured at UBC during the same year. It is believed that the treesincreased in caliper in 1991 even after seedling height growth had ceased. Caliper37CoastalInterior100-90-80-—t+t70-0060-50—MibTofHopMasSqummLmmHKanStjKooQuiChe011MtbWhommMBflLolPieProvenancesFigure2.3.Meanprovenancefinalheights(horizontalbars)andrangeoffamilyIprovenancemeans(verticalbars)oftreesgrowingatUBC.growth could have occurred in 1992, at least in the base of the stem, if the lowercambium was not damaged.Table 2.4 shows the apportionment of total variance attributed to provenanceand to family at the family level of analysis. After one growing season, the plugseedlings showed a higher provenance variance component for heights than familycomponent, but after three seasons, the transplanted seedlings at both sites displayed agreater proportion of variance between families within provenances than betweenprovenances. Root collar diameters measured after 1992 at UBC showed a greaterpercentage of variance attributable to family than to provenance, although theopposite was true at Salmon Arm.Analyses were carried out on each provenance separately to compare theamount ofwithin-provenance variation in height and root collar diameter ofUBCtrees between the provenances (Appendix 3.5; Appendix 3.6). The three VancouverIsland provenances contained family variation in height (Table 2.5), while familyvariation in RCD was evident at one Vancouver Island provenance (Quinsam), atCheakamus, and at Mt. Mara Mid elevation.Narrow-sense individual heritabilities of seedling height calculated on treesgrowing at UBC increased yearly from 0.28 after one year to 0.38, with a standarderror of 0.13, after three years. Although these numbers cannot be treated as absoluteindications of the amount of heritable variation in western red cedar, they do indicatethat heritability in this species is not zero. The heritability of height at Skimikin wasmuch lower, at 0.12 for both 1991 and 1992. Heritabilities of RCD at both locationsafter 1992 were lower than those ofUBC heights, but large enough to indicate realvariation.39Table 2.4. Percentage of the total variance attributable to provenance and to familywithin provenance for measured growth traits.Trait %a2P(Z) %2F(PZ)Height‘90 plug 17.77 > 8.23‘91 UBC 11.95 > 10.09‘92 UBC 9.06 < 10.66‘91 Skimikin 8.57 > 5.23‘92 Skimikin 2.55 < 3.88‘91,2 Locations 6.50 > 3.84RCD‘92 UBC 3.08 4.57‘92 Skimikin 10.37 > 7.34Dry Weight#laterals 1.83 < 2.08Stem wt 9.54 > 4.56Foliar wt 11.72 > 6.67Shootwt 9.19 > 5.94Rootwt 5.37 > 4.07Totalwt 8.67 > 5.87Shoot,’j00 8.80 > 6.89Shoot,q’01 8.11 > 5.40Stem/Totaj 18.18 > 9.09SurvivalFoliar desiccation 14.90 > 1.4440Table 2.5. Results of individual provenance analyses’ of measurements taken atUBC after the 1992 growing season on provenances having familystructure (values appear in brackets where Family was not significant).Region Prov. Fam. signif. , y2 %a2F/a2[HeightVan. Is. Quinsam F 22.812 21.712 12.8Tofino F 24.275 19.841 14.1Mill Bay F 93 .024 63.772 36.2Mainland Cheakamus (11.718) (13.582) (5.8)Hope (1.388) (8.061) (0.7)B.C. Inter. Mt. Mara Low 0 0 0Mt. Mara Mid (8.742) (10.609) (4.3)Mt. Mara High 0 0 0Idaho Benton Flat (6.910) (9.552) (4.5)Root Collar DiameterVan. Is. Quinsam F 0.645 0.498 16.2Tofino 0 0 0Mill Bay (0.182) (0.219) (5.1)Mainland Cheakamus F 0.403 0.355 9.2Hope 0 0 0B.C. Inter. Mt. Mara Low 0 0 0Mt. Mara Mid F 0.229 0.2 13 7.1Mt. Mara High 0 0 0Idaho Benton Flat (0.086) (0.144) (2.9)Provenance = + Rep + Familyf + R*Frf +412.3.3. Dry WeightsBranch angle was highly variable, even within a tree (Appendix 4.3). Nodifferences were noted in branch angle at any level through ANOVA. This may onlybe true for juvenile trees though. The number of lateral branches per stem variedbetween the coastal and interior zones, but no differences were evident betweenprovenances or families.Correlations between dry weight parameters of all test seedlings were allhighly significant, with the only negative relationship between root dry weight andshoot/root dry weight ratio. For provenances without family structure, most dryweight parameters differed at either the provenance or zonal level or at both levelswhen analyzed by ANOVA (Appendix 4.1; Appendix 4.2). For provenances havingfamily structure, all dry weight traits differed at the provenance level. Familydifferences occurred only in the stem/total dry weight ratio. Heritabilities rangedfrom 0.12 to 0.34 on these variables. Standard errors were calculated for theheritability of the stem/total (h2 ± s.e. = 0.34 ± 0.11) and shoot/root (h2 ± s.e.= 0.29 ± 0.11) ratios, indicating that heritabilities were significantly greater than zero.2.3.4. Foliar Nutrient AnalysisANOVA of the foliar nutrient analysis of all provenances showed provenancedifferences for all macro and micronutrients analyzed except for Fe (Appendix 6.1).For provenances having family structure, family differences were apparent in N, K,Fe, and the K/Ca ratio, while provenance differences were noted in P, K, Ca, and theratios of fN, K/N, and K/Ca. Nine pairs of nutrients were correlated, with the42significant correlations ranging from r = -0.53 to r = Seedling SurvivalAlthough little mortality to seedlings occurred over the duration of the studyat the UBC site, significant differences in survival at the zonal level were detected inthe ANOVA of seedlings growing at UBC on provenances having family structure,and provenance differences were noted in provenances without family structure(Appendix 7.1; Appendix 7.2).At the Salmon Arm site, differences existed between provenances havingfamily structure in the number of living trees at the end of 1992, percentage ofgerminants sown in 1990 still alive after 1992, percentage of trees living at the end of1991 (before the winter where major damage occurred) which were still alive by theend of 1992, and in the percentage of foliar damage occurring to a tree’s crown overthe winter of 1991 I 92. For foliar desiccation damage, zonal and family differencescould also be detected (Appendix 7.3; Appendix 7.4). Although a higher amount ofthe variation was found at the provenance level than at the family level (Table 2.4), aheritability of 0.05 was estimated for foliar damage.When an analysis of the percentage of germinants sown in 1990 still aliveafter 1992 was carried out in which both locations were included (Appendix 7.5;Appendix 7.6), genotype by environmental interactions were noted at bothe the zonal(P = 0.0462) and provenance (P <0.0001) levels.Foliage escaping desiccation damage per tree in 1992 at Salmon Arm was43negatively correlated to tree height at the end of 1991 (r -0.19; P <0.0001) andpositively correlated to height (r 0.58; P <0.0001) and RCD (r = 0.47; P <0.0001)after the 1992 growing season, after damage had occurred. The percentage of theoriginal genninants still surviving after 1992 was negatively correlated to the meanheight in 1991 at Skimikin (r = -0.12; P <0.0389) but positively correlated at UBC(r 0.26; P <0.0001). Genetic correlations of seedling height and survival at UBCand Skimikin turned out to be meaningless due to negative variance components.Table 2.6 compares provenance rankings for the following traits: height at UBC,survival at Skimikin, and % live foliage at Skimikin.Stepwise regressions ofmean percentage survival of the original number oftrees planted still alive after 1992 and ofmean percentage of live foliage perprovenance after the winter of 1991 I 92 at Skimikin with elevation, latitude,longitude, estimated average annual days in the growing season, annual precipitation,and daily temperature in January indicated that elevation was the most important ofthe independent variables on both traits. For survival, elevation alone accounted for51.6 % of the variation (Figure 2.4); for live foliage, elevation was associated with38.9 % of the variation. Elevational dines estimated from simple linear regressionshowed that 1 % increase in survival at Skimikin occurred for every 43 m increase inelevation, while 1 % increase in live foliage occurred for every 58 m increase inelevation. The best models which could be fitted, based on stepwise regression andMallows C, statistic, using a significance level of a. 0.05 for entry into and stayingin the model, were:Skimsurv = -153.556 + 0.028*Elev. + 3.567*Latit.; R2 0.682; P <0.0001Livefol = -211.126 + 0.023*Elev. + 4.3l0*Latit.; R2 = 0.726; P <0.000144Table 2.6. Provenance rankings of final height at UBC (in cm) and % live foliageand % survival at Skimikin; ranking of height is from largest to smallest,while ranking for live foliage and survival is from smallest to largestmean value.Height Live Survivaln) foliage (%)Mib (80.6) Mib (2.3) Mib (16.9)Squ (79.5) Squ (3.2) Mas (19.8)Who (78.6) Tof (4.5) Squ (21.8)Hop (78.4) Who (5.5) Koo (22.5)mmL (78.1) Stj (5.6) Tof (24.1)Che (74.0) Koo (8.5) Stj (28.4)Koo (72.2) Kan (10.4) Who (32.5)Mtb (71.9) Qui (10.4) Qui (36.6)Qui (71.8) Che (11.3) Pie (38.3)Tof (70.5) Pie (14.1) Che (39.9)Lol (69.3) Mas (16.2) Oil (45.0)Bfl (69.2) Lol (17.2) Mtb (48.2)Mas (66.6) Hop (17.3) Kan (48.3)Pie (66.3) mmL (19.6) mmL (48.4)Kan (66.3) Mtb (21.5) Bfl (49.8)mmM (65.6) 011 (23.9) Lol (52.1)Stj (64.9) Bfl (25.5) Hop (57.6)011 (63.2) mmM (34.0) mmlvi (58.3)mmH (61.2) mmH (68.3) mmH (90.6)45100-90-80-70-0 ©60-.n50-Cl)-.40-•30—.20-10—‘05001,0001,5002,000Elevation(in)Figure2.4.Meanprovenancesurvivalafterthewinterof1991/92oftreesgrowingatSalmonAnncontrastedwithprovenanceelevation.2.3.6. Analysis by Seed Collection, Geographical Region, and by ClusterWhen the seed data were separated into the three seed collections that made upthis study, differences between collections in germination, percentage seed fill, and inseed weight were found in the ANOVA when all lots were looked at (Appendix 2.2).However, when provenances not having family structure were removed and the datareanalyzed, no significant differences were evident between collections in any of theseed parameters.The same data were regrouped into broad geographical regions, whichconsisted of splitting the coastal seed collection into three regions. Thus the fiveregions were: northern B.C. coast, Vancouver Island, B.C. mainland, B.C. interior,and Idaho / Montana. ANOVA results showed that, as found for collections, regionsdiffered in all seed parameters when all provenances were analyzed but did not differin any seed trait when analyzing only those provenances with family structure(Appendix 2.3).Dry weight parameters plus heights and RCD taken at UBC after the 1992growing season were grouped into the same five regions. ANOVA was performed onall provenances by region (Appendix 5.1). Regions varied significantly in stem dryweight, foliar weight, shoot (foliage + stem) weight, root weight, total dry weight,stem/total dry weight ratio, and in RCD at UBC after 1992. The shoot/root andShOOt/total dry weight ratios and height after 1992 at UBC did not vary significantlyby region, although a north / south trend in the Shoot/root ratio was noted. However,as the representation of provenances by region was unbalanced, and not abundant inone case, these results should be viewed with caution.47Table 2.7 lists the parameter means plus or minus the standard deviation,percentage of the total variance (a2T) attributed to provenance (a2P) , and thestandard error of estimate (s.e.c.) on the provenance variance component per region.The regions were represented respectively by 2, 4, 4, 3, and 6 provenances.The percentage of variation attributed to provenances and s.e.e. of the a2Pappeared to vary greatly from region to region. The highest percentage ofa2P ands.e.c. of the a2P for all parameters occurred in the B.C. interior region. Noprovenance variance or s.e.c. of the a2Pwas found in any parameter except the shootto root dry weight ratio in the northern B.C. coastal region. However, this region wasrepresented by only two provenances, neither ofwhich had family structure, so theseresults probably do not reflect the true amount of provenance variance in this region.Although these results cannot be taken as absolute, they do indicate that realdifferences do occur between regions for western red cedar.Hierarchical cluster analysis produced somewhat different clusters than theabove geographically based regions. Table 2.8 presents the clustering of provenancesby different variables and Duncan analysis ofmean differences in UBC final height.When UBC fmal heights were clustered into four groups, all provenances within ageographical region except for the Mt. Mara provenances were clustered within twoconsecutive clusters. The same did not hold true for the other traits.48Table 2.7. Final UBC height (cm), RCD (mm), and dry weight (g) means ± standarddeviation, percentage of the total variation attributable to provenance, andstandard error (s.e.e.) of the provenance variance component by region.Region1 2 3 4 5Trait N coast Van. Is. Mainland BC inter. Idaho± s.d.Ht’92 64.54±12.85 74.19±14.67 76.70±13.89 70.43±16.18 68.54±12.26RCD ‘92 9.44±1.88 10.34±1.90 10.42±2.01 9.55±2.00 9.11±1.70Stem wt 6.87±2.95 9.08±3.70 9.83±3.71 8.22±3.74 7.15±2.81Foliarwt 19.16±8.48 17.74±8.17 18.10±7.76 14.93±7.70 10.93±5.02Shootwt 26.03±11.25 26.82±11.26 27.93±11.09 23.14±11.15 18.07±7.57Totalwt 33.70±13.68 34.87±13.71 36.81±13.54 30.47±13.51 24.17±9.38Shootj’p00 3.41±0.84 3.41±1.00 3.22±0.83 3.15±0.87 3.01±0.86a2P&2THt ‘92 0 8.30 1.37 20.03 1.54RCD’92 0 1.95 2.04 5.64 3.01Stem wt 0 4.46 1.63 8.68 2.74Foliarwt 0 12.36 6.97 21.88 2.59Shootwt 0 6.08 4.87 22.37 2.91Total wt 0 4.14 4.24 22.05 3.42Shoot,’Root 2.46 5.12 5.09 21.92 6.41s.e.e. ofa2Ht’92 0 14.435 5.003 47.774 3.612RCD ‘92 0 0.099 0.104 0.278 0.093Stem wt 0 0.811 0.629 2.627 0.246Foliarwt 0 7.868 4.019 12.157 0.832Shootwt 0 9.331 6.908 26.171 1.934Totalwt 0 12.124 9.390 37.473 3.211Shoot/R0o 0.106 0.056 0.044 0.172 0.049#trees.Ht/RCD 59 569 431 391 320#trees,drvwt 55 151 155 98 19549Table 2.8. Comparison of geographical regions with groupings of provenances basedon cluster analysis of certain traits; UBC height is ranked from tallest toshortest provenance mean, while survival and living foliage at Skimikinare ranked from smallest to largest percent, due to expected negativecorrelations with height; regions are ranked by average group height.Geographical UBC final Duncan, Skimikin Skimikin UBC ht., Skim.region height UBC ht. survival% live foliage% surv.. live fol.B.C. Mainland Mib a Mib Mib MibSquamish Squ ab Mas Squ SquWhonnock Who a b Squ Tof WhoHope Hop a b Koo WhoCheakamus mmL a b Tof Stjb KooVancouver Island b TofMill Bay Che c b Stj Koo MasMt. Benson Koo c d Who Qui StjQuinsam Mtb c d e Qui KanTofmo Qui c d e Pie CheTof cdef Che CheB.C. Interior Lol c d e fg QuiMt. Mara Low Bfl c d e fg Pie PieMt. Mara Mid d e f g 011 Mas KanMt. Mara High d e fg Mtb LolMas hdefg Kan HopIdaho. Montana Pie h d e fg mmL HopKooskia Kan h d e f g Bfl mmLLob mmM h e f g Lol mmL MtbBenton Flat Stj h f g Mtb LolPierce h g 011 BflKaniksu h g Hop Bfl mmMSt. Joe 011 h g mmM OilmmH hNorthern B.C. mmMMasset mmH rmnHOliver LakemmH502.4. DiscussioNIt is apparent that significant levels of genetic variation exist in Thuja plicataseedling growth attributes. Although only a sampling of populations could be testedand only two test locations were used, and hence heritability estimates were onlyrepresentative ofmeasured seedling traits at these specific sites, it is safe to infer thatsignificant variation in this species exists. Heritability estimations were conservativeto guard against the possibility of inbreeding, although indications are that adultwestern red cedar populations may not be inbred to a great degree. In the unlikelyevent that the estimates were still not conservative enough, it can still be said with ahigh amount of confidence that the heritabilities are not zero. These heritabilitiescertainly fall within the normal range for growth traits of other coniferous trees,roughly 0.2 to 0.3 (J.S. Brouard, pers. comm., 1994).Differences between populations from the coastal versus the interior rangesexisted in initial plug seedling height (during the early seedling establishment stage),in height and amount of crown dieback at the interior location after a severe winter,and in some dry weight parameters. Regional differences appeared in many dryweight parameters and in RCD measured at the coastal site. While survival and dryweights showed mainly between-population differences, the trend in height at themild coastal site appeared to be towards more within- than between-populationdifferences. Broken down by provenance, the Vancouver Island populations showedfamily (within-population) differences in height; root collar diameter showed familydifferences at Cheakamus, Quinsam, and Mt. Mara Mid elevation.As gene flow seems to be extensive in this species, and outcrossing isassumed to be the normal reproductive strategy, one would expect to find greater51within- than between-population differences. This seems to be the case for height androot collar diameter.For the traits where between-population differences are greater than thosewithin populations, broad environmental selection pressures may be the mostimportant factors shaping the genetic structure of this species since the last glacialretreat, in contrast to mutation and random drift. As greater between-populationdifferences were exhibited in survival traits, this would lend credence to the abovetheory. The separate analyses by provenance also seem to support this theory. Ifdisruptive (diversifying) selection is occurring, then variation would be increasing inthose characteristics that were affected.Within-population differences were noted on Vancouver Island and near thesouthern mainland coast where environments are most favourable to western redcedar growth, and less limited by environmental (i.e. temperature, moisture)extremes. Gene flow may be higher in these locales.As the apical meristem is not protected by bud scales and foliage has littlecutin and wax for protection from excessive transpiration, it would be expected thatwestern red cedar would be quite susceptible to frost damage, transpirational stress,and foliar sunburn, and would be expected to be opportunistic as far as shootexpansion. Thus the most probable selective pressures are those imposing conditionsof transpirational stress, very low winter temperatures, and foliar sunscald. If climaticconditions are not limiting, western red cedar could be expected to undergo rapidexpansion due to extensive gene flow.The severe winter damage to trees growing at Salmon Arm during the winter52of 1991 /92 may have been so devastating in part because the seedlings were growingunder full sunlight without the benefit of protection from an overstory. Thus theamount of damage and mortality, and major growth reduction the following growingseason, may not truly represent conditions of a naturally regenerated or planted foreststand in the interior (the former becoming established below a protective overstory,and the latter probably having some degree of brush which would partially shadeseedlings, at least during part of the year).An increase in elevation appears to be the most important geographicalelement of those studied in limiting western red cedar growth. This concurs with thefindings ofRehfeldt (1994), although the current study observed much strongerelevational relationships than those found by Rehfeldt. The discrepancy would mostlikely be due to the far greater diversity and much wider range of locales sampled inthe current research. The Mt. Mara elevational dine near Salmon Arm in the interiorof B.C. gives evidence of the degree to which an interior population source will varyby elevation.Besides the elevational dine at Mt. Mara, a latitudinal dine occurred in Idaho,unfortunately confounded by the difference in elevation between the two populations.Pierce is 21’ north of Kooskia (and 1,098 m higher in elevation). Pierce was theshorter of the two at the UBC site but was taller at Skimikin after 1992, and survivalat Skimikin was slightly better for Pierce than for Kooskia.The populations from the U.S. are the only ones of the current study to havepossibly escaped glaciation. The locations where the Benton Flat and Kaniksupopulations come from in northern Idaho were almost certainly glaciated. St. Joe inwestern Idaho and Lob in western Montana were near the inferred glacial boundary53and so it is unknown whether these areas would have been covered by ice or not.Pierce and Kooskia, however, were probably not glaciated, although they may haveundergone severe flooding during deglaciation. It is impossible to tell at this timewhether sources ofwestern red cedar trees reoccupying formerly glaciated areas inIdaho and trees coming from nonglaciated regions had the same ancestral origin. It isalso unknown if coastal and interior populations migrated from the same refugium.One baffling question to be addressed is the apparent discrepancy between theamount of variation evident in quantitative traits as observed in this study and the lackof variation seen in former studies involving isozymes, leaf extractives, or a verysmall, or nonrangewide, sampling ofpopulations.Studies have often found metric traits of silvicultural importance to beuncorrelated with allozyme variation. Geographic dines may differ in location ordirection among loci or traits; thus factors other than migration must be influential insuch cases (Namkoong and Kang, 1990). Many studies have suggested thatquantitative traits are much more differentiated between populations than areisozymes (e.g. see Muona, 1988).Only about 0.1 % of all nucleotide substitutions in the total genome, or about20 % of the 0.5 % of the genome which codes for all proteins in a eukaryoticorganism, can be detected by electrophoresis (Powell, 1975, as cited in El-Kassaby,1991). No study ever examines all of the proteins in any species; thus only a smallportion of the total variation present is ever investigated with this method. Enzymessurveyed are usually those found in high concentrations in tissue, and excluded arethe products of regulatory loci and loci which code for ribosomal proteins and transferRNA (Falkenhagen, 1985). Poor laboratory technique will limit even further the54detection of isozyme variation. Between DNA and the protein are the steps oftranscription and translation; isozymes are thus phenotypes, and the genotypes canonly be inferred from them, as they are not directly observed (Hattemer, 1991).Lewontin (1984) also showed that the power of statistical tests discriminatingbetween quantitative traits of populations and species and those discriminatingbetween gene frequencies at individual loci are vastly different, and thus directcomparisons between the two should not be made. He stated that the probability offinding differences between populations and species in quantitative traits is muchhigher than that of gene frequences.The variability of quantitative traits is governed more by selection pressuresthan is the variability of isozymes (Muona, 1988). Isozyme variation gives littleinsight into adaptive patterns of variation in quantitative characters and hence shouldnot be used to make inferences about them, or to determine seed transfer and breedingzone delineations.Assuming a mutation rate of about 10-6 per gamete per generation for a singlelocus, on the order of 100 to 1,000 generations are required after reclaiming territoryafter a genetic bottleneck for the heritable variance in quantitative traits of a species tobe restored to previous levels, while 10 to 106 generations would be necessary torestore neutral alleles such as those studied by isozyme analysis (Lande, 1988).Western red cedar is believed to be still undergoing this process of restoration offormer variability, so some level of variability must have been retained through theice age.None of the abovementioned ideas could be conclusively proven to be the55governing factor in why the results of this current study would indicate that muchmore variation is present in western red cedar than was inferred by previous isozymestudies of this species, but perhaps certain inferences might be suggested. It is alsonot known why many other associated coniferous species exhibit greater levels ofisozyme variation than western red cedar, although perhaps other species ofindeterminate nature may show similar trends. Less variation in neutral alleles mayhave been present in western red cedar even prior to glacial retreat. Alternatively, thisspecies may have been reduced to lower levels of neutral alleles during glaciation,from which it is still recovering, as relatively few generations have occurred since thelast ice age. Western red cedar may be slower to adapt to environmental changes thanother species exhibiting higher heritabilities in quantitative traits, or else may be moreplastic in certain traits, with less specialized genetic variation in these traits.All but one of the populations tested in this study were part of largecontiguous ranges (either coastal or interior), so random drift probably does not affectthese provenances. However, isolated patches ofwestern red cedar do occur betweenthe the coast and interior ranges and along the eastern edge of the range; random driftmay affect such populations to some degree in the distribution of variation within thespecies. The provenance from Lob in Montana is situated on a noncontiguous patch,but is close to the Idaho range and so some degree of gene flow from the main rangeprobably occurs.A paper on Picea omorika (Panc.) Purk., a species also previously believed tobe genetically depauperate with a high degree of self-fertility, reported much geneticvariation in both enzyme and quantitative loci, and high levels of inbreedingdepression in 24-year old trees of this species; the authors theorized that the onset of56selection against inbreds occurs later than for other species (Kuittinen et al. ,1991).They concluded that neither post-glaciation mutation rates nor random drift couldhave accounted for the levels of variation that were observed in their study.Some evidence of genotype by environment interaction, at the family level forheight and the zonal level for RCD, and at the zone and provenance levels forseedling survival, was found in measured traits between two very different sites. Atthe coastal location, with a high number of frost-free days and an average amount ofannual precipitation in comparison with provenances used in the study, the three bestperformers in height growth were the three closest provenances, also with longgrowing seasons, having average or above average annual precipitation levels, andcoming from similar biogeoclimatic units. From this it can be stated that provenancesfrom the southern coast probably perform best in relatively unstressful environmentssuch as that found at the UBC site.The interior test site was characterized by less annual precipitation than any ofthe provenances tested except for the nearest low elevation provenance, and a lownumber of frost-free days in comparison to most provenances tested. After the 1991growing season, the provenance with the best average height was Mt. Mara Lowelevation, the closest provenance at low elevation. The local mid and high elevationprovenances, Mt. Mara Mid and Mt. Mara High, were ranked 16th and 19threspectively out of 19 provenances. Provenances ranked 2nd and 3rd at this time wereSquamish (coastal) and Hope (eastern edge of the coastal range).After the 1992 growing season, after experiencing severe desiccation theprevious winter, Mt. Mara Low elevation was still ranked the tallest. However, Lobfrom Montana and Pierce from Idaho (all interior wet belt provenances) were now57ranked and 3rd The local mid and high elevation Mt. Mara provenances werenow ranked 4th and 8th in height. Survival was greatest in Mt. Mara High, followedby Mt. Mara Mid with Mt. Mara Low ranked 6th after the severe winter damage atSkimikin. Generally provenances where the mean January temperature wascomparatively low survived the best.Specialist species are those with between-population differences, which areadapted to current specific environments (Rehfeldt, 1984). The genotype is expressedphenotypically, so selection acts on the genotype. Specialists may not have the abilityto adapt quickly to catastrophes, due in part to coadapted gene complexes which mayexist. Generalist species are those exhibiting plasticity (the ability of an organism toalter its phenotype in response to changes in environmental conditions) orhomeostasis (canalization), which can function under a broad range of conditions,with many genotypes potentially being expressed as similar phenotypes. As selectionwould act against the phenotype, the genotype should be conserved over time.Generalists can sometimes undergo rapid gene flow and are more resilient toenvironmental fluctuations, particularly in time, and hence strong clinal variationshould not occur (Rehfeldt, 1984).Western red cedar seems to lie somewhere between a specialist and ageneralist. Timing of shoot initiation and cessation is extremely plastic; in that sense,western red cedar is an opportunist, taking full advantage of favourable growthconditions. Seasonal height growth in general followed the pattern of a generalist, aswithin-population differences seemed to be greater than those between populations.Survival under stressful conditions and dry weight traits were morespecialized in nature. Population differences were apparent in these traits, whereas no58significant within-population differences were detected. Foliar desiccation at SalmonArm following the severe winter showed both between- and within-populationdifferences, but between-population differences were much greater than within-population differences.Western red cedar showed similar plastic shoot elongation patterns to those ofother indeterminate conifer species studied (Zobel, 1983; Harry, 1987; Cherry andLester, 1992; Russell, 1993). Variation levels appeared to be not unlike those of otherassociated conifer species. For example, apportionment of variation between andwithin populations for first year container seedling heights of this study was verysimilar to that ofwestern hemlock container seedlings of about the same age (Kuserand Ching, 1981), the latter species having a very similar species range and presumedrecent refugial history to that ofwestern red cedar. Response ofwestern red cedar toharsh environments seemed to be severe when compared to seedlings of other speciesundergoing similar conditions.592.5. LITERATURE CITEDBower, R.C. and B.G. Dunsworth. 1987. Provenance test ofwestern red cedar onVancouver Island. Pp 13 1-135 In Smith, N.J., ed. Western redcedar - does ithave a future? Conference Proceedings. UBC Faculty of Forestry, Vancouver,177 pp.Cherry, M.L. and D.T. Lester. 1992. Genetic variation in Chamaecyparisnootkatensis from coastal British Columbia. West. J. Appl. For. 7(1): 25-29.Copes, D.L. 1981. Isoenzyme uniformity in western red cedar seedlings fromOregon and Washington. Can. J. For. Res. 11(2): 45 1-453.El-Kassaby, Y. 1991. Genetic variation within and among conifer populations:review and evaluation ofmethods. Pp 61-76 In Fineschi, S., M.E. Malvolti, F.Cannata, and H.H. Hattemer, eds. Biochemical markers in the populationgenetics of forest trees. SPB Academic Publ., the Hague, 251 pp.Falkenhagen, E.R. 1985. Isozyme studies in provenance research of forest trees.Theor. and Appl. Genetics 69: 335-347.Harry, D.E. 1987. Shoot elongation and growth plasticity in incense-cedar. Can. J.For. Res. 17: 484-489.Hattemer, H.H. 1991. Genetic analysis and population genetics. Pp 5-22 InFineschi, S., M.E. Malvolti, F. Cannata, and H.H. Hattemer, eds. Biochemicalmarkers in the population genetics of forest trees. SPB Academic Publ., theHague, 251 pp.Hillier, H.G. 1981. Hillier’s manual of trees and shrubs, 5th ed. David and CharlesPubl. Ltd., 575 pp.Ilmurzynski, E. et al. 1968. [Investigations on the growth and development ofcertain North American tree species in Polish nurseries and plantations.] Pr.Inst. Badaw. Lesn. 364, 84 pp [In Polish]. In FA 30: 2121.Kuittinen, H., 0. Muona, K. Karkkainen, and Z. Borzan. 1991. Serbian spruce, anarrow endemic, contains much genetic variation. Can. J. For. Res. 21: 363-367.Kuser, J.E. and K.K. Ching. 1981. Provenance variation in seed weight, cotyledonnumber, and growth rate ofwestern hemlock seedlings. Can. J. For. Res. 11:662-670.60Lande, R. 1988. Genetics and demography in biological conservation. Science 241:1455-1460.Larsen, C. S. 1953. Studies of the diseases in clones of forest trees. Hereditas 39:179-192.Lewontin, R.C. 1984. Detecting population differences in quantitative characters asopposed to gene frequencies. Amer. Nat. 123(1): 115-124.Lowery, D.P. 1984. Western redcedar an American wood. USDA For. Serv. FS261, 6 pp.Muona, 0. 1988. Population genetics in forest tree improvement. Pp 282-298 InBrown, A.H.D., M.T. Clegg, A.L. Kahler, and B.S. Weir, eds. Plantpopulation genetics, breeding, and genetic resources. Sinauer Assoc. Inc.Publ., Mass., 449 pp.Namkoong, G. and H. Kang. 1990. Quantitative genetics of forest trees. PlantBreeding Rev. 8: 139-188.Nault, J.R. 1986. Radial distribution of thujaplicins and thujic acid in old growth andsecond growth western redcedar (Thujaplicata Dorm). MSc Thesis, UBC,Vancouver, 61 pp.Neter, J., W. Wasserman, and M.H. Kutner. 1990. Applied linear statistical models,3rded. Irwin Inc., 1,181 pp.den Ouden, P. and B.K. Boom. 1982. Manual of cultivated conifers, 3rd ed.Martinus Nijhoff, 520 pp.Poiheim, F. 1977a. [Investigations on bud variation in Cupressaceae. 8. Are ever-sporting periclinal chimeras possible in haploid Thujaplicata “Gracilis”?]Flora 166(2): l77186 [In German]. In Plant Br. Abs. 47: 10,856.Polheim, F. 1977b. [Selection of a new growth type in the haploid Thujaplicata“Gracilis”.] Arch. fur Zuchtungsforschung 7(4): 311-313 [In Danish]. InPlant Br. Abs. 48: 5,981.Polheim, F. 1972. [Survival rate and shoot variation due to mutation following Xirradiation of haploid and diploid plants of Thujaplicata.] BiologischeRundschau 10(3): 200-201 [In German]. In Plant Br. Abs. 43: 706.Poiheim, F. 1970. [Triploidy in Thujaplicata excelsa Timm.] Biol. Rundschau 8:402-403 [In German]. In Plant Br. Abs. 41: 6,258.Powell, J.R. 1975. Protein variation in natural populations of animals. EvolutionaryBiology 8:79-119.61Rehfeldt, G.E. 1994. Genetic structure ofwestern red cedar populations in theInterior West. Can. J. For. Res. 24: 670-680.Rehfeldt, G.E. 1984. Microevolution of conifers in the northern Rocky Mountains: aview from conmion gardens. Pp 132-146 In Lanner, R.M., ed. Proc., 8th N.A.For. Biol. Workshop, Logan, Utah, 196 pp.Rogers, D.L., D.E. Hany, and W.J. Libby. 1994. Genetic variation in incense-cedar(Calocedrus decurrens): I. Provenance differences in a twelve-year-oldcommon-garden study. West. J. Appi. For. 9(4): 113-117.Rushforth, K.D. 1987. Conifers. Christopher Helm Pubi. Ltd., 232 pp.Russell, J.H. 1993. Genetic architecture, genecology and phenotypic plasticity inseed and seedling traits of yellow-cedar (Chamaecyparis noorkatensis [D.Don] Spach). Ph. D. Thesis, Univ. B.C., 168 pp.Søegaard, B. 1966. Variation and inheritance of resistance to attack by Didymascellathujina in western red cedar and related species. Pp 83-87 In Gerhold, H.D.,E.J. Schreiner, R.E. McDermott, and J.A. Winieski eds. Breeding pest-resistant trees. Pergamon Press, 505 pp.von Rudloff, E. and M.S. Lapp. 1979. Population variation in the leaf oil terpenecomposition ofwestern red cedar, Thujaplicata. Can. J. Bot. 57(5): 476-479.von Rudloff, E., M.S. Lapp, and F.C. Yeh. 1988. Chemosystematic study of Thujaplicata: multivariate analysis of leaf oil terpene composition. Biochem.System. and Ecol. 16(2): 119-125.Yeh, F.C. 1988. Isozyme variation of Thujaplicata (Cupressaceae) in BritishColumbia. Biochem. System. and Ecol. 16(4): 373-377.Zobel, D.B. 1983. Twig elongation patterns of Chamaecyparis lawsoniana. Bot.Gaz. 144(1): 92-103.623. RESISTANCE TO ENVIRONMENTAL STRESSES3.1. INTRODUCTIONThe ability of trees to withstand cold temperatures and associated stressessuch as desiccation during the winter without being damaged (to the extent thatsubsequent growth is adversely affected) or killed is one measure of adaptability.Winter conditions would be expected to impose selective pressures upon populations,and damage and mortality at any one locale might be expected to be related to thegeographic origin of the tree seed. However, the physiology of frost hardiness isquite complex, and direct responses of trees to their environs may not be obvious.The terms stress resistance, frost hardiness, and dormancy are oftenincorrectly used interchangeably. Resistance to stress, often termed hardening off oracclimation, is acquired by northern temperate forest tree species in a gradual processat the end of every growing season. In the spring gradual dehardening, ordeacclimation, of the tree occurs. Hardy trees are more resistant to many forms ofstress, not just that of cold temperature; stress resistance refers to all portions of theorganism (Lavender, 1985). Frost hardiness is the ability of a tree to withstandsubfreezing temperatures without damage, and is often expressed as the minimumtemperature at which 50 % of a group of seedlings are killed or injured, the “lethaP’temperature 50 (LT50) (Glerum, 1985). Dormancy is simply the cessation of shootgrowth in tissues; this refers only to the shoot apical meristem, as diameter growthmay still occur. Roots, however, do not become dormant; feeder roots may growwhen temperatures are not too severe (Perry, 1971).63The onset of dormancy occurs in mid-July in B.C., and may be induced bymoisture stress (Lavender, 1985), although dormancy can be broken at this point ifenvironmental conditions are favourable. Shoot elongation ceases and terminal budsform in determinate species. Eventually, during late summer, the tree becomesunresponsive to favourable environmental conditions, and the shoot and then thecambium will no longer begin growing in response to external stimuli.The shortening photoperiod after the summer solstice is commonly the firststimulus acting on most temperate tree species to induce hardiness (Weiser, 1970),usually around early September in B.C. Trees at this stage become more responsiveto temperatures at and just below freezing; such temperatures initiate a second stageof acclimation, in which large increases in degree of hardiness take place (Glerum,1985). A third acclimation stage may be reached in very hardy woody species(Weiser, 1970). This stage is induced by prolonged exposure to very low (i.e. -30°Cto -50°C) temperatures, and this stage is quickly lost.In the northern hemisphere, temperate forest trees are generally most hardyduring January. After a certain chilling requirement has been met, they begin to losedeep dormancy. The environmental cues which act to initiate deacclimation are notknown, although Silim and Lavender (1991) obtained strong correlations in whitespruce (Picea glauca [Moench] Voss) seedlings between exposure to temperatures ofabout 5°C and both the release of bud dormancy and development of dehardeningpotential. Dehardening lags behind release of dormancy (Lavender, 1985).Dehardening can occur very rapidly once it has been initiated (Glerum, 1985),although trees have been known to deepen in hardiness if adverse environmentalconditions are imposed at this time (R. Guy, UBC, pers. comm., 1994).64Under freezing conditions, intracellular and / or extracellular freezing occursin plants. Intracellular freezing, which occurs when freezing is rapid, is essentiallyalways fatal, probably due to physical injury to the cell, as in cell membrane ruptureby ice crystals (Weiser, 1970).Extracellular freezing results when freezing is slow, and is sometimes, but notalways, damaging (D. Lavender, UBC, pers. comm., 1987). The mechanisms ofinjury are not fully understood for this type of freezing, although hypotheses abound.During gradual freezing, water is drawn out from the inside of cells, and freezes in theintercellular spaces (Levitt, 1980). Tree cells are known to increase their cellmembrane permeability during acclimation (Weiser, 1970), thus allowing water toexit the cell more easily during the period when cold temperatures are expected.Cell dehydration occurs progressively as water becomes frozen externally.Plasmolysis, in which the protoplast withdraws away from the cell wall, occurs ascellular water moves outward. During deplasmolysis, water reentering the cell uponthawing may cause cell lysis, which is the most probable cause of death and injury innonhardened tissue (Steponkus, 1984). Hardened plant cells may form exocytoticextrusions which can be reincorporated without lysing during rehydration(Thomashaw, 1990), but rapid temperature changes could cause hardy cells to lyse.Risk of frost injury varies with stem position and tissue type. Basal stemshave been found to be less hardy than upper stems of forest tree species, with thedifference between stem position lessening with increasing hardiness (Sakai andOkada, 1971). In a test comparing various portions of one western red cedar tree,foliage samples were more resistant than twigs, which in turn were slightly hardierthan buds [sic] (Sakai and Okada, 1971). Douglas-fir stem tissues were more65sensitive to subfreezing temperatures than buds during the periods of hardening anddehardening, but were less sensitive than buds during the time of peak hardiness(Ritchie, 1991). Sakai and Okada (1971) found that cambial cells and surroundingphloem cells were the least frost sensitive; cambial cell resistance increased with agein Japanese larch (Larix leptolepis [Sieb. and Zucc.] Gordon) seedlings.Frost damage may be compounded by desiccation due to an imbalancebetween water loss and absorption. Wind may increase transpiration by sweepingaway the thin surface boundary layer ofwater vapour from the foliage; this effect iscompounded in bright conditions when stomata are open and thus offer littleresistance to transpiration (Kramer and Kozlowski, 1979; Salisbury and Ross, 1978;Sakai, 1970). When the tree roots or the stem below the snowline are very cold orfrozen, water cannot be absorbed at a fast enough rate to replenish that which is lostdue to transpiration; thus shoots become desiccated and may be killed (Kozlowski,1971; Miller, 1978).Metabolic changes have been noted in trees during hardening, especiallyduring the second and third stages of acclimation. Growth inhibitors, anthocyanins,fats, and phenols are accumulated, while growth hormones (indole acetic acid, IAA,and the gibberellins, GA’s) are at low levels in midwinter (D. Lavender, UBC, pers.comm., 1987; Perry, 1971). Changes have been noted in sugar, amino acid, nucleicacid, lipid, and protein levels (Glerum, 1985; Weiser, 1970), and in configurations ofthe latter two (Blum, 1988); some of these changes may confer cryoprotectantproperties (Ritchie, 1991).It is generally believed that the nutrient status of a plant affects its ability toharden, but how this occurs is ambiguous. Many contradictory results have been66published, probably because nutrients affect growth rate and hence only indirectlyaffect hardiness (Glerum, 1985). Of importance is undoubtedly the amount ofnutrients available (adequate levels being better than deficient or excessive levels),and their relative proportions (Glerum, 1985; Pellett and Carter, 1981). Plantgenotypes have been found to differ in the uptake and utilization ofmineral nutrients(Fisher and Mexal, 1984). Schaedle (1991) discussed heritable traits which could beselected for in relation to nutrient uptake and utilization.The stimuli inducing dormancy in trees apparently turn some genes on andothers off (Perry, 1971), affecting the synthesis of certain enzymes. Thomashaw(1990) noted the expression of certain genes (COR, or cold-regulated, genes) in plantsafter exposure to cold temperatures, whose functions are as yet unknown but whoseexpression seems to parallel freezing tolerance. He speculated that levels of abscisicacid (ABA) increase in response to cold, which in turn could regulate the expressionof COR genes.Trees may be genetically adapted to withstand cold damage by having anearlier or greater response to the stimuli influencing cold hardiness in the autumn, byhaving a delayed response to favourable growing conditions in the spring where thereis a threat of late spring frosts, by having a greater degree ofhardiness which can beattained, by having a greater phenotypic plasticity to environmental conditions, or bydifferences in the synthesis of enzymes (such as by the COR genes mentioned above).Generally, more northerly provenances, or those from higher elevations ormore exposed or windswept sites, would be expected to have the greatest frosthardiness and slowest growth rates. Such was found to be the case for interiorDouglas-fir (Rehfeldt, 1978) and lodgepole pine (Rehfeldt, 1980).67The range in frost hardiness among provenances seems to vary with the stageof acclimation. The greatest differentiation between populations in hardiness leveloccurred during the autumn, before trees were able to withstand very coldtemperatures, for lodgepole pine (Rehfeldt, 1980) and for Sitka spruce (Cannell andSheppard, 1982). In the latter study, more northerly provenances became hardysooner, but began to deharden later in the spring, than more southerly provenances.Knowledge about frost hardiness patterns ofwestern red cedar has onlyrecently begun to emerge. It is known that this species is not very frost resistant incomparison with other coniferous species that grow in the same range (Krajina eta!.,1982; Minore, 1983). In the interior, western red cedar becomes established mainlyin areas where snowcover occurs before the ground becomes solidly frozen (Krajinaet aL, 1982). The primordia ofwestern red cedar, as with all Cupressaceae species,are surrounded by only a few scalelike tissues, without the benefit of the resincovering found on buds of determinate species, and hence have little protection fromwater loss and thus from desiccation damage (Sakai, 1983). Sunscald may occurwhen a period of subfreezing temperatures is followed by sunny days, where theleader and youngest branches above the snowline may die, especially lateral brancheson the southern side of the crown (Miller, 1978).It has been repeatedly demonstrated that short photoperiods and moisturestress have little effect in inducing cold hardiness in western red cedar (Vaartaja,1959; Silim, 1991; Krasowski and Owens, 1991; and Folk et a!., 1994). The mainfactor found to date which seems to induce hardiness in this species is cold, inparticular subfreezing, temperatures.Weger et aL (1993) observed that synthesis of the carotenoid pigment68rhodoxanthin, and to a lesser degree lutein, was correlated to hardiness in western redcedar, and was largely responsible for the colour change observed in foliage duringthe winter. They proposed that the pigmentation reduces the light intensity reachingthe photosynthetic apparatus and hence helps to prevent winter photodamage. Theirresults coincide with those of Columbo and Raitanen (1991) for Thuja occidentalis,who found that seedlings whose foliage had turned brownish were hardier than treeswhose foliage had remained green.Genetic testing of cold hardiness in this species has been very limited. Aninland Idaho western red cedar provenance was found to be more frost hardy than twocoastal provenances from Corvallis, Oregon and Seattle, Washington (Sakai andWeiser, 1973), presumably in response to differences in climatic conditions.Methods of assessing cold hardiness in plants are numerous. The most directmethod is to freeze whole intact plants and then visually assess the amount of damageto tissues that shows up over a period of a few weeks after thawing. Drawbacks ofthis method include the subjectivity of ocular assessment of discolouration and tissueflaccidity, the necessity of destructively sampling whole seedlings, and the lengthyperiod it takes for damage to be expressed.The electrical conductivity method, to be explained in detail in Section3.2.4.2, is based on the fact that when injury such as frost damage occurs, the cellmembrane loses its selective penneability, and thus electrolytes in the cytoplasmleach out of injured tissues when submersed in water in proportion to the severity ofthe injury. The electrical conductivity of a sample can easily be measured. Thismethod utilizes tissue segments, not the whole plant.69Very good correlations between the whole tree browning method andelectrical conductivity testing (estimated LT50’s: - 18.5°C ± 0.5°C with electricalconductivity; -18.9°C ± 1.2°C with cambial assessment) were found for western redcedar, with visual assessment of the cambium being more closely related than foliarbrowning (Silim, 1991). Even if not as accurate compared to testing whole plants,this test should be the more precise of the two, and it is precision which is desired fordetecting genetic differences.Other methods sometimes used to assess cold hardiness are variablechlorophyll fluorescence, calorimetry, and nuclear magnetic resonance (NMR), thefirst two methods which will be described below. NMR and calorimetry bothmeasure cellular ice formation. NMR is a type of spectroscopy which measuresspectral differences between frozen and liquid water. However, it is expensive, and ithas been reported (Burke et al., 1976) that attempts to correlate the quantity ofboundunfreezable water with cold hardiness have been unsuccessful.When foliage is exposed to light, normally some of the light energy which isabsorbed by the chlorophyll pigments is used in photochemical water splittingreactions which lead to carbon assimilation, with some of the excess energy beingdissipated as heat, and some reemitted as fluorescence (Vidaver et aL, 1991; Taiz andZeiger, 1991). The variable chlorophyll fluorescence test measures the fluorescenceemitted in response to light exposure, but is only indirectly linked with frosthardiness.The initial baseline fluorescence emission, F0, represents the amount offluorescence prior to any light excitation, and so indicates those processes which areindependent of any photochemical events. F0 has been found to increase with plant70hardiness in one seedlot ofwestern red cedar (Weger et al., 1993). When a pulse ofhigh-intensity saturating light is given, the fluorescence level rises to a maximum(Fm). The difference between Fm and F0 is the variable fluorescence, F, which occursonly in photosynthetically active tissue (Vidaver et aL, 1991). The ratio Of Fv/Fm, orsometimes lv/F43, is used as a probe for photoinhibition, indicating stressfulconditions, with a lower ratio indicating a greater amount of photoinhibition.Plants under duress such as low temperatures are sensitive to photoinhibition,even at moderate light levels (Oquist and Huner, 1991). Current theory holds thatphotoinhibition ofphotosynthesis has two components (R. Guy, UBC, pers. comm.,1994): the first has photoprotective effects, while the second, photooxidation, isdetrimental to the plant. The first photoinhibitory response is a reduction inphotosynthetic capacity occurring when light levels exceed photosyntheticrequirements (Ogren, 1991), preventing electron buildup. However, thisphotoprotective mechanism does not work well under cold conditions.If carbon assimilation is halted due to low temperatures, but an excess ofunutilized electrons are present, photooxidation damage occurs (Oquist and Huner,1991). Resulting toxic oxygen species may cause injury to the chloroplastmembranes, photosynthetic pigments, lipid membranes, and electron transport chainproteins (Blum, 1988; Ritchie, 1991; Powles, 1984). Photodamage may beirreversible and even lethal.Calorimetry, or thermal analysis, measures the latent heat of fusion occurringwhen water freezes within a plant tissue by using a single thermocouple attached tothe plant as the test chamber is progressively cooled (Salisbury and Ross, 1978).Temperature exotherms are produced at each freezing event. One or two large71exotherms are produced when normally noninjurious extracellular water freezes,somewhere around -2°C to -10°C (Weiser, 1970).Deep supercooling occurs when water molecules have no nucleus aroundwhich to crystallize, and hence ice formation cannot occur. Hardy plants are believedto be able to deep supercool to about -3 8°C in the absence of heterogeneousnucleating agents in the tissue environment (George et a!., 1974). This temperaturecorresponds with the point to which pure water can be supercooled if ice nucleationcan be prevented, after which it spontaneously nucleates (Burke et a!., 1976;Salisbury and Ross, 1978). Ice is most likely then formed within living cells, killingthem. Low temperature exotherms (LTE’s), associated with the point where injury ordeath occurs, have been documented by calorimetry in some cold-hardy species downto -40°C, which coincides with the range limit ofmany of the species tested (Georgeeta!., 1974; Ritchie, 1991). Deep supercooling may be an avoidance mechanism forsurviving freezing injury in some tissues (Salisbury and Ross, 1978).The purpose of this chapter was to explore the genetic relationships in coldhardiness ofwestern red cedar, in particular genetic variation in and correlations oftraits related to cold hardiness. Variation was studied at three levels: between zones,provenances, and families within provenances.Specific objectives of cold hardiness research were:- to estimate the extent and allocation of genetic variation found in cold hardinesstraits ofwestern red cedar between coastal and interior zones and between andwithin provenances and families using quantitative genetic analyses- to investigate the influence of environmental factors on levels of cold hardinessand on rates of acclimation and deacclimation- to compare different methods of assessing cold hardiness72The seasonal stages of hardening, attaining maximum hardiness, anddehardening were investigated over two winters and at two locations, one relativelymild and one relatively harsh. Hardiness attributes (index of injury, LT50 frosthardiness curves, base, variable, and maximal fluorescence levels, low temperatureexotherms, and influence of foliar nutrient status on hardiness) in this species werestudied, using a number of available techniques: electrical conductivity testing,variable chlorophyll fluorescence, and calorimetry, and by comparing nutrientanalyses with the LT50, as found by electrical conductivity, within the same tree.733.2. METHODS3.2.1. Frost Hardiness Profile over a Broad Range of TemperaturesTo test for differences between provenances in the electrical conductivity testindex of injury (‘i) of acclimating western red cedar seedlings over a broad range oftemperatures, a test was carried out on December 19, 1991. The frost hardiness of sixprovenances (the same as those used in pretesting; see Section at -5°Cintervals from -10°C to -50°C was monitored. Four seedlings per provenance weretested at each of the nine test temperatures.Samples were frozen normally (see Section over the range oftemperatures, and electrical conductivity tests carried out. The It values perprovenance were plotted against each temperature to produce profile curves. Thesegraphs were scrutinized for pattern and degree of change with temperature decrease inthe six provenances. An analysis of variance was carried out on the It values.Nonlinear regression, using the logistic function due to the sigmoid nature of the frostcurve profiles, was carried out following the methods of Sit and Poulin-Costello(1994), using the following functional form:Y a1 + ebcXThe solutions found for a (index of injury at -50°C), b (intercept), and c(slope), where Y I and X temperature, for each tree per provenance were thenanalyzed by analysis of variance and multivariate analysis of variance.743.2.2. Effect of Crown Position on Frost HardinessTo determine whether within-tree differences in degree of frost hardiness wereapparent, a study was set up to compare branch samples from different positionswithin the seedling crown. The same six provenances as used in pretesting wereselected for this study (Section Four seedlings were tested per provenance;seedling samples were kept separate from each other.Branches were taken from the upper third and lower third of each tree.Samples were placed into labelled plastic bags and processed as per regular frost testsamples, with the exception of keeping separate jars for branches from the uppercrown and from the lower crown.Seedling samples were frost tested on December 2, 1991 to three testtemperatures: -12°C, -2O°C,and---28°C; temperatures were selected based on theknowledge of how cold hardy these provenances were at that time according to theregular frost tests. A date near to when maximum cold hardiness could be expectedwas desired, but because of freezer scheduling constraints, this test was performed alittle prior to the desired time frame.An electrical conductivity test was done as per usual (Section, andanalysis of variance was used to determine whether differences in hardiness occurredbetween the upper and lower crown.3.2.3. Effect of Rate of Freezing on Measures of Cell InjuryA frost test was carried out in which samples were frozen at a rate of at least75twice as fast as normal frost tests to see whether a difference in degree of damage wasnoted. A higher degree of damage at faster rates of freezing would indicate damagesuch as cell lysis or intracellular freezing.Six provenances were chosen for this study, the same provenances as used inpretesting (Section Four trees per provenance were tested at threetemperatures: -15°C, -22°C, and -29°C. Temperatures were based on the degree ofhardiness found in the normal frost tests at that time. The test was performed onFebruary 24, 1992.Freezing rates were limited by the ability of the freezing unit to cool down.Between +2°C and -15°C, the freezer cooled at a rate of-i 1.86°C per hour. Between-15°C and -22°C, the cooling rate was -12.73°C per hour. Between -22°C and -29°C,the cooling rate was only -9.77°C per hour. Thus the overall cooling rate averagedout to about -12°C per hour.An electrical conductivity test was conducted on the samples. EstimatedLT501swere compared to those of the normal frost tests on adjacent dates.3.2.4. Frost Testing over Two Winters3.2.4.1. Greenhouse Design and Nursery CultureOf the ten provenances per zone used in the western red cedar growth studies,four provenances per zone were chosen for studying seedling frost hardiness over twowinters. Half of the provenances per zone maintained family structure (three familiesper provenance) while the other half of the provenances per zone had no family76structure. Provenances and families used for frost testing were as follows:Coast InteriorTofino 1, 3, 5 Mt. Mara Low elev. 2, 3,4Mill Bay 1, 2,4 Mt. Mara Mid elev. 2,4, 5Oliver Lake Benton Flat (families bulked)Squamish KooskiaEntries were chosen to give a broad representation of each zone; however,provenance selection was limited to those that had adequate seed germination toprovide enough seedlings for the study.Seedling germinants were taken from the petri dishes full of germinants usedin the growth studies. Germinants were dibbled into Styro Vent 91 blocks, filled withnormal nursery soil mix as used in the growth studies, using tweezers and a cavityspoon. Two blocks per genetic entry were sown for frost test purposes. Blocks sownfor frost hardiness testing were grown along with and under the same cultural regimeas the blocks grown for growth studies.For frost hardiness pretesting prior to every frost test, six genetic entries werechosen, three from the coast and three from the interior: Oliver Lake, Squamish, MillBay (families bulked), Mt. Mara Low elev. 3, Mt. Mara Mid elev. 2, and Kooskia.Ten styroblocks were split into thirds and sown for pretesting samples. Eachprovenance was sown into 1/3 of each of five blocks. Seedlings were sown and grownas described above for normal frost test seedlings. The first winter of frost testing,1990 I 91, used samples from these containerized plug seedlings.77Seedlings to be used for pretests and frost tests during the winter of 1991 / 92were outplanted into a transplant bed adjacent to the transplant bed used for growthstudies at UBC’s South Campus Nursery. A subset of trees, for frost hardiness testingon three test dates, was transplanted at Skimikin Seed Orchard; the same eightprovenances, half having family structure, were used as per the samples at UBC. Atboth locations, seedlings were planted in whole unreplicated blocks at 1/3 m spacing. Frost Hardiness TestsOne week prior to running a frost test, a pretest run was carried out on sixprovenances to give a rough estimate of the weekly rate of hardening and to estimatea range within which the LT501softhe different provenances fell. These pretestresults were used in estimating the range of temperatures to be used in the larger frosttests.Pretesting was done at four test temperatures, with an equal interval betweenall temperatures. Samples from four trees per provenance were used. The main frosttests used four trees per genetic entry and three test temperatures; the remainingmethodology was the same for both tests. However, only three samples per geneticentry from Skimikin were used, due to limitations in material available at that site.For each sample, lateral branches from the upper crown were cut off of thetrees at the stem, bagged, and brought back to the lab for immediate processing. Inthe lab, all samples were stored in a cooler(--S+3°C) for the short period (a matter ofminutes to an hour) until they were processed.78During 1990 / 91, the first winter in which frost testing was done, the materialwithin each genetic entry was mixed together and the bulked foliar material dividedinto four replicates. During 1991 / 92, the second winter of cold hardinessmonitoring, the tree samples were kept separate so that individual analyses for eachtree could be estimated.Frost testing was performed using the electrical conductivity test which wasmodified from the methods of Glerum (1985) and Silim (1991). Upon removal fromthe cooler, branches were rinsed with distilled, deionized water. The tips of allfoliage to be used were snipped off and discarded. The remaining foliage was cut intopieces about 5 mm long. Each piece had two cut ends; this would facilitate an easierflow of electrolytes during testing.Approximately five foliage pieces were placed into labelled 20 ml glassscintillation vials which had been previously washed and rinsed with distilled,deionized water. A small amount of silver iodide and a small squirt of distilled,deionized water were added to the jars prior to adding sample pieces to act as an icenucleator to aid the onset of freezing. Four replicated jars were used per genetic entryper test temperature. Vials to be frozen to different test temperatures were givenwashed, colour-coded plastic lids with plastic inner liners. Samples for each testtemperature were placed into matching colour-coded test tube racks which wouldallow for unrestricted air circulation between vials.Frost testing was carried out in a programmable semi-customized FormaScientific Biofreezer unit. Control trees which were not frozen were kept in adarkened cooler at +3°C while the tests were being run. Test trees were put into thefreezer and kept at +2°C for one hour to allow the freezer to stabilize prior to79beginning the test.The temperature was then decreased at a rate of -5°C per hour until the firstdesired test temperature was reached. The freezer was held for one hour at thattemperature, and then the samples which were to be tested only to that temperaturewere removed with all overhead lights turned off. The removed vials within theirracks were placed into a dark plastic garbage bag to minimize photoinhibitory effects.The vials were placed in the darkened cooler and allowed to thaw gradually overnightat +3°C.The remaining vials in the freezer were cooled at -5°C an hour until the nexttest temperature was reached, held at that temperature for one hour, and then the nextgroup of samples was removed. This process was repeated until all test temperatureshad been reached and all vials were removed from the freezer and placed into thecooler.The following morning all vials, including those of the unfrozen controls,were removed from the cooler. Fifteen ml of distilled, deionized water was added toeach jar. The vials were then left at room temperature for 24 hours.The next morning all jars, including controls, were then shaken, and theelectrical conductivity measured with a Cole-Parmer 148 1-60 conductivity meter.Samples were then completely killed by placing the vials, still in their racks, into a60°C Fisher Scientific waterbath for 45 minutes. Jars were then left at roomtemperature for a further 24 hours; a second conductivity measurement was thentaken on each jar.The amount of injury due to freezing to a certain temperature was calculated80for each sample using the following formula (Columbo et at., 1984):It RCfrozen - RCcontroi1 - (RC00ii ioo)where:It = index of injury (0- 100 %)RCfrozen = Relative Conductivity of frozen sample= Electrical Conductivity of frozen sample * 100Electrical Conductivity of frozen killedRCconfrol Relative Conductivity of control sample= Electrical Conductivity of control sample * 100Electrical Conductivity of control killedThe It values were plotted against test temperatures. From these graphs, thetemperature at which 50 % of injury would occur (LT50, lethal temperature 50) wasestimated. Figure 3.1 illustrates one such graph used in the interpolation ofLT50’s.Trees at UBC were frost tested ten times over each of two winters. Trees atSkimikin were frost tested three times over the winter of 1991 I 92. For each of thelatter tests, snipped branches were bagged, placed into a box containing frozenicepacks, and shipped to the lab at UBC via Greyhound bus for testing. Data AnalysisAnalysis of variance (SAS® PROC GLM) was carried out on the results of all8100V100 0V-2-3-4-5-6-7-8-9Temp,C-10Figure3.1.ExampleofplottedindexofinjuryvstesttemperatureusedtoestimateLT501spersamplepertestdate.tests separately for each date. All provenances were tested by the following modelper date:= + Temperatures + Zone + T*Z + Prov(Z)P(Z) + T*P(Z)tp(z) +Provenances having family structure were also analyzed by the following model:Zfl = + T, + Z + T*Z + P(Z)P(Z) + T*P(Z),(z) + Family(P Z)PZ)+ T*F(P Z)tgpz) +Temperature and zone were treated as fixed effects, while provenance and familywere treated as random variables, with the appropriate F tests being constructedaccordingly (Appendix 1.9 and Appendix 1.10). Variance components weredetermined (SAS® PROC VARCOMP) for the purposes of heritability estimationwhere family was a significant effect.Curvilinear regressions were estimated on the curves produced by graphingestimated LT50 values over each season. Due to their shape, LT50 curves wereanalyzed by weighting models by the number ofweeks from the beginning of thetesting season and by weeks2,which describe a general parabolic curve (seconddegree polynomial). The log ofLT50 was used as the dependent variable. During thefirst winter of frost testing, samples were bulked together for each genetic entry, soonly simple analyses could be attempted due to nonreplications. The models testedfor 1990 / 91 were:Log LT5o= + Week + Week2+ +Log LT5o= + Week + Week2+ P, + E(p)nThe former model was used on all provenances, with the latter being used on83provenances having family structure. Provenance was not nested within zone in thelatter model, as only two provenances per zone having family structure remained.The models tested for the 1991 / 92 season were:Log LT5o + Week + Week2+ Zz + P(Z)P(Z) +5(zp)nLog LT5,= + Week + Week2+ P, + F(P)) +with the former model being used with all provenances, and the latter model beingused with provenances having family structure.Analysis of variance was also done on the estimated LT501sper date. For testsdone in 1990 / 91, as seedling samples per genetic entry were bulked to give fourmixed replications, family level analysis was not possible. Zone was left out of themodel because only four provenances, the ones with family structure, had replicatesfor the purposes ofLT50 analysis. During 1991 I 92, the seedlings were kept separateand each of four seedlings per genetic entry were treated as replicates, so analysiscould be done at the family level as well as at the provenance level, and provenancesnot having family structure could be included in the latter analysis.To investigate possible relationships between the geographical distribution ofprovenances and frost hardiness, hierarchical cluster analysis was carried out on theLT50 values for January 21, 1991 and January 27, 1992, the dates per year closest tothe maximum hardiness reached. Correlations were also estimated comparing theLT501sfrom these dates with environmental attributes of the provenances. Stepwisemultiple linear regressions were also done to determine whether the maximum LT50per year was dependent on certain environmental conditions related to eachprovenance.84Frost tests of seedlings from Salmon Arm were analyzed by ANOVA per testdate as per the trees from UBC that were tested. An analysis of variance was done foreach of the dates when trees from Skimikin were tested, in which the LT50 of treesfrom Skimikin was compared to that of trees from UBC at the same time period.Thus the effect location (treated as a fixed effect) and its interactions were added tothe standard model. The expected mean squares were determined as shown inAppendix 1.7 and Appendix 1.8 by substituting Location for Temperature.3.2.5. Variable Chlorophyll Fluorescence and Frost HardinessTo determine whether damage to the photosynthetic apparatus (i.e. to thechioroplast) due to cold temperatures paralleled the results obtained using theelectrical conductivity method, six genetic entries (the same as those used forpretesting, Section were tested by variable chlorophyll fluorescence at twotest dates. Four trees per genetic entry per temperature per test date were used. Treeswere tested at three temperatures on January 13, 1992 and April 1, 1992; testtemperatures were chosen based on how hardy the normal frost tests showed theseprovenances to be. In January, seedlings were tested to -15°C, -25°C, and -35°Cwhile in April, seedlings were tested to -2°C, -6°C, and -10°C.Seedlings which had been grown for one year in a styroblock Vent 91 andthen one year in a transplant bed at the UBC forest nursery were snipped off at theroot collar on a cloudy afternoon and brought directly into the lab. The cut portion ofthe stem was wrapped in wet cotton wool. Trees were placed into large size plasticfreezer bags.85Frost testing was carried out at night using the Biofreezer. Control seedlingswere dark adapted (left at +3°C in total darkness for over two hours) prior tomeasuring. Sample trees were frozen to the desired temperatures under darkconditions. After being maintained for one hour at the appropriate temperature, treesof each treatment were removed from the freezer in the dark. Samples were placed ona lab bench and given 20 minutes of bright light. To accomplish this, an overheadprojector was turned on its side on the bench so that the light was pointing towardsthe seedlings which were fanned out, leaders towards the lamp, around the circle ofthe beam. A light meter was used to check that seedlings were exposed to similarirradiance. A rectangular glass tank was filled with water and placed directly in frontof the projector bulb to reduce exposure to infrared light. After the light treatment,trees were given 20 minutes of total darkness.The variable and maximum fluorescence were then measured in the dimmedlab using a pulse modulated Heinz Walz PAM 101 chlorophyll fluorometer. The 1cm diameter fiberoptic probe was placed on the tree foliage, trying to cover as muchleaf surface area as possible. The optical tip initially shone red light onto the foliage;this gave a reading for the minimal chlorophyll fluorescence level, F0. A Xenon highintensity lamp was then activated for one second of illuminating saturating flash ofabout 3400 j.tmol m2 s’; this gave Fmimaj (Fm) for the tree sample.Fviab1e (F) was calculated as the difference between Fm and F0 (F = Fm - F0).The fluorescence ratios Of F/Fm and Fv/F0 were calculated for all samples. Analysisof variance was carried out on these ratio values, as well as Ofl F0, Fm, and Findividually (Appendix 1.7).863.2.6. CalorimetryThermal analysis, or calorimetry, was carried out to see whether thetemperatures at which extracellular freezing and internal supercooling occurred couldbe detected, and if so, whether there were differences between provenances in thetemperature at which these exotherms occurred. Two seedlings from each of fourprovenances (Oliver Lake, Mill Bay, Kooskia, and Mt. Mara Mid elev.) wereselected. Testing was done in mid-February, when trees were starting to deacclimate.Seedling branch samples were collected from UBC’s transplant bed onFebruary 15, 1992 and brought into the lab. A datalogger was set up with onethermocouple probe attached to each of nine separate channels. The probe of each ofeight channels was embedded into the foliage of a sample so that the foliage touchedboth sides of the thermocouple. The probe was taped into place, ensuring that theportion with the probe tip was not obstructed.Samples were placed into the freezer. One channel was not hooked up to anyseedlings, but measured ambient air temperature (the control). The cooler wasprogrammed to run for one hour at +2°C and then to steadily decrease in temperatureat a rate of -5°C per hour until -50°C was reached.Once the temperature of the freezer reached 0°C, the datalogger was started;temperatures were sampled on all nine channels every five seconds for ten hours.Temperatures were recorded throughout the test using a connected laptop computer.The freezer was only able to reach -49°C, as its compressors could not maintain adecrease in temperature of -5°C / hour as the temperature approached -50°C. Datawere later transferred to a Sun computing system. For each sample, graphs of foliage87temperature over time and sample temperature minus control temperature weredeveloped. Analysis of variance was carried out to test for provenance differences inthe temperatures at which the first and second exotherms occurred, the size in degreesC of the two exotherms, and the temperature to which each sample rose to after eachexothermic event.3.2.7. Foliar Nutrient Analysis and Frost Hardiness TestingA study was done to determine whether there was any correlation betweenfrost hardiness of a seedling and foliar nutrient content of certain macronutrients(% N, % P, % K, % Ca, and % Mg).Five seedlings from each of eight provenances were tested for both frosthardiness and for foliar nutrient analysis, using the same seedlings for both tests inorder to observe any relationships from the individual tree level and upwards.Provenances tested were: Oliver Lake, Squamish, Mill Bay, Tofino, Mt. Mara Lowelev., Mt. Mara Mid elev., Benton Flat, and Kooskia.Foliage samples were collected from sample trees on January 25, 1993, placedinto individual labelled bags, and brought into the lab. Portions of the branch sampleswere processed as per usual for frost testing (Section Remaining portionswere processed for foliar nutrient analysis, as described previously (Section 2.2.7).Frost testing was done at four temperatures (-28°C, -3 6°C, -44°C, and -52°C),based on how hardy the trees were deemed to be at the time. Foliar nutrient analysiswas carried out at the MacMillan Bloedel lab in Nanaimo, B.C.88As provenance Tofino was found to be far less hardy than anticipated, its LT50could not be estimated from this test. Therefore Tofino was retested the followingweek on February 2, 1993, using test temperatures of -20°C, -28°C, and -36°C.Correlations were conducted between the macronutrient concentrations andthe estimated LT50 ofthe same tree. A regression and an analysis of variance werealso performed, with LT50 as the dependent variable (Y value) in both cases.893.3. RESULTS3.3.1. Temperature ProfilesAnalysis of six provenances frost tested on December 19, 1991 at -5°Cintervals between -10°C and -50°C showed differences between provenances, but notbetween zones, in index of injury (‘i). A nonsignificant interaction term ofprovenance by temperature, where temperature was used as a covariate, indicated thatthe slopes of provenance curves did not differ when provenance It was plotted againsttemperature.Nonlinear regressions of the frost profile curves were estimated based on thelogistic function (Sit and Poulin-Costello, 1994), as plotted curves were sigmoid inshape (Figure 3.2). For this analysis, a few obvious outlier points on a small numberof the individually plotted tree curves were manually smoothed (i.e. a few of the Ivalues at -10°C were higher than those at -15°C). Values obtained per tree perprovenance were then used in an analysis of variance. Provenances did not differsignificantly in index of injury at -50°C, curve slopes, or intercepts, although the leastsquares mean a (I5) ofMt. Mara Mid elevation differed significantly from those ofMt. Mara Low elevation and Oliver Lake according to pairwise t-tests. The inflectionpoints of the curves (X b/c) did differ between provenances, with mean provenanceinflection points ranging from -26.5°C to -32.6°C.Examination of the curves showed all provenances approaching the upperhorizontal asymptote between -35°C and -40°C, regardless of what the index of injurywas. For the majority of all provenance profiles, between the test temperatures ofabout -20°C and -40°C, the curves were linear. Thus it was hoped that by careful90100Mib90Oil80Squ70mmL60mrnMEL—50Koo40 30 20 10 0-10-50Temp,CFigure3.2.Meanprovenanceindexofinjuryprofilecurvesresultingfromfrosttestingbetween-10°Cand-50°ConDec.19,1991.-15-20-25-30-35-40-45selection of frost test temperatures to include all provenance LT50’s but which werenot much broader in range than necessary, only the linear portion of the curves wouldbe sampled. Thus pretest results obtained the week prior to every frost test werecarefully heeded as a check of the current status of seedling cold temperaturesusceptibility.3.3.2. Crown PositionAnalysis of variance on six provenances tested for differences in hardinessfrom different crown positions within a single tree indicated that the upper third of acrown was significantly less hardy than the lower third of a crown, although the meandifference in index of injury was only 2.8 %. A paired t-test comparing the upper andlower crown confirmed that crown position was significant in its effect on frosthardiness.Subsequent foliage sampling for frost tests was always done from the samearea within a crown, namely the upper third of the crown but not the uppermostbranches.3.3.3. Freezing RateFigure 3.3 shows the estimated LT50 values of the six provenances tested atabout twice the freezing rate as normal tests, compared to estimated LT50 values asdetermined from freezing tests the week prior to (Feb.17, 1992) and the weekfollowing (Mar. 2, 1992) the doubled freezing rate test (Feb. 24, 1992). An estimate9225—(j3In20—15 10—ElNormal•DoubledFigure3.3.MeanprovenanceLT501sofseedlingstestedatabouttwicethestandardfreezingratecomparedtoseedlingstestedatthestandardrateoffreezing.Mib011SqummLmmMKooof the hardiness of trees frozen at the normal -5°C / hour was obtained byextrapolating LT50 values from lines drawn between the Feb. 17 and Mar. 2 testresults.A t-test to test for differences in LT50’s between the two freezing rates showedthat differences were significant (P <0.0098). Samples frozen at twice the normalrate had LT50sabout 3.2°C wanner than trees frozen at the normal rate; i.e. damageto 50 % of the cells occurred 3.2°C higher for samples frozen at the faster rate.Winter frost testing over two seasons was constantly done at a temperaturedecrease of -5 °C / hour so that freezing rate did not adversely affect test resultinterpretations.3.3.4. Midsummer Resistance to Cold TemperaturesA frost hardiness test carried out in midsummer when the trees were activelygrowing and were deemed to be in their least cold hardy state found an overall meanLT50 of -4.1°C, with family / provenance means ranging from -2.8°C in a Mill Bayfamily to -6.4°C in a Mt. Mara Low elevation family. The coastal average LT50 was-4.2°C, while the interior average LT50 was -4.1°C. Analysis of variance ofmidsummer I values found no variation at the zonal, provenance, or family levels, orin any of the interaction terms; however, as would be expected, test temperature wassignificant.943.3.5. Coastal Hardiness CyclesFrost hardiness curves from the main frost tests conducted over two winterson seedlings growing at UBC, depicting estimated LT50 over the whole winterseasons, are shown for zonal means for both winters, provenance means over bothwinters, family means for 1990 / 91, and family means for 1991 / 92 in Figure 3.4,Figure 3.5, Figure 3.6, and Figure 3.7 respectively. Curves indicated that differencesin hardiness between genetic entries are maximal when trees are most hardy, andnonexistent when trees are least hardy during the summer.Daily maximum and minimum temperatures at UBC for the winters of 1990 /91 and 1991 /92 are given in Figure 3.8 and Figure 3.9. Notably, the first below-freezing temperatures of the season occurred earlier in the year in the fall of 1991 (atthe end of October) than in 1990 (late December).In the curvilinear regression analyses ofLT50 curves over each test season,weighting terms were significant in all models tested (P <0.0001). Zonal differenceswere apparent during 1990 / 91, but not during 1991 / 92. Provenance differences inLT50 curves were seen during both seasons (P <0.0001). Family differences were notevident during the second season of testing.Variance components were estimated for provenances having family structureon the dates where the family term was found by analysis of variance to besignificant. Unless significant (in which case they remained in the model),interactions were dropped from the model for the purposes of variance componentdetermination to increase the power of the error term. Figure 3.10 compares thepercentage of the provenance and family variance components between the two test95LT, C-55 I I I I I I I I I I I I I I I I I P I I I I I10123 11/6 11/20 12/4 12/16 1/21 2/11 3/4 3/18 4/8Date— Coastal “ InteriorLT, C—55. p , i i i i i i10)21 11/4 11125 12/16 1/6 1/27 2/17 3/9 3430 4/13DateCoastal InteriorFigure 3.4. Mean zonal LT501softrees growing at UBC over two winters of testing(top: 1990 / 91; bottom: 1991 / 92).96LT, Ci—55 i10/23 11/6 11120 12/4 12/16 1121 2/11 3/4 3/18 4/8DateCoastal ““ Interior jLT, C—55. , . , i i i ,10121 11/4 11125 12/16 116 1127 2/17 3/9 3/30 4/13DateCoastal InteriorFigure 3.5. Mean provenance LT501softrees growing at UBC over two winters oftesting (top: 1990 / 91; bottom: 1991 / 92).97LT, C—55 i i , i i i i i i i10t23 11/6 11t20 12/4 12/16 1/21 2111 3/4 3/18 4/8DateTofino MillBayLT, C—55 ,10t23 11/6 11120 12/4 12/16 1/21 2111 3/4 3/18 4/8DateMMLow “ MMMIdFigure 3.6. Mean family LT50’s of trees growing at UBC during 1990 / 91 (top:coast; bottom: interior).98LT, C—55 i i i10121 11/4 11t25 12/16 116 1127 2/17 3/9 3/30 4/13DateToflno “ MiliBayLT, C—55 . , , ,10121 11/4 11/25 12/16 116 1/27 2/17 3/9 3130 4113DateMM Low MM MidFigure 3.7. Mean family LT501softrees growing at UBC during, 1991 / 92 (top:coast; bottom: interior).99Temp,C30 25 20 15 10 5 0 -5 -10-15Date9/110/111/112/11/12/13/14/1Figure3.8.DailymaximumandminimumtemperaturesatUBCduring1990/91.30 25 20 15 10Temp,C11ILtiAIh1’5.0•-5.-10--15III 1I,“VITV,9/110/111/112/11/12/13/14/1DateFigure3.9.DailymaximumandminimumtemperaturesatUBCduring1991/92.Figure 3.10. Apportionment ofa2Provenance anda2Family of index of injury foreach of ten test dates per year over two years of frost testing (top:1990 I 91; bottom: 1991 / 92).100%80%• P(Z)40%60%• F(P Z)20%c’ 00C Z -100%IE.P(Z)20%40%I I0%‘0 00Q z z0102years, and Figure 3.11 compares the % a2F for each of the provenances over two testseasons. Variance components, and heritability estimates, were highly variable fromtest to test; they were presumed to be imprecise due to the low number ofprovenances having family structure and the low number of families per provenance,so heritabilities are not included.3.3.6. AcclimationPrior to the onset of hardening, no variability occurred in the level ofhardiness maintained by a genetic entry; differences did not appear until around lateOctober. While testing did not commence until after acclimation had already begun,acclimation appeared to commence at the same time at the coast for all genetic entriesduring both test years. Once the hardening process had begun, interior provenancesacclimated at a faster rate than coastal provenances (Figure 3.5); this was particularlynoticable during the second test year.ANOVA results of the frost index of injury for all provenances (Table 3.1)and for only those provenances with family structure (Table 3.2) showed that zonaldifferences were apparent on some dates during acclimation. Provenance differenceswere evident when all provenances, irrespective of family structure, were analyzed.However, when only those provenances with family structure were investigated, mostof the variation appeared to be within provenances during acclimation. Whiletemperature was significant on all test dates, temperature interactions occurred in afew tests during the time when trees were actively hardening.ANOVA results of the LT50 per test date are shown in Table 3.3. Except for103Mill Bay------MML----MMM----Tofin10080Nov. 6 Nov. 20 Dec. 16 Jan. 21 Feb. 11 Mar. 4 Mar. 18Mill Bay MML — - - — MMM — — — Tofrno100Nov. 4 Nov. 25 Dec. 16 Jan. 27 Feb. 17 Mar. 9 Mar. 30Figure 3.11. Family variance components of frost test index of injury over two yearsfor each provenance having family structure (top: 1990 / 91; bottom:1991 /92).104Table 3.1. Analysis of variance1 results of frost hardiness testing index of injuryover two winters, plus one summer test, for all provenances, where: * * *P <0.001; ** = 0.001 <P <0.01; * = 0.01 <P <0.05; ns 0.05 <P.12I Z £(Z) I I!Z T*P(Z)Oct. 23/90 ns * *** ns *Nov. 6/90 ns ns nsNov. 20/90 ** *** *** ns nsDec. 4/90 ns ns nsDec. 16/90 ns ‘‘ ‘‘ ns nsJan.21/91 ns ns nsFeb. 11/91 * *** *** ns nsMar. 4/91 ** *** *** ** nsMar. 18/91 ns *** *Apr. 8/91 * ns * * * * 115Oct. 21/9 1 * * *** ns nsNov. 4/91 ns ns nsNov. 25/91 * *** *** ns nsDec. 16/91 * *** *** ns nsJan. 6/92 ns * * * * * * ns nsJan. 27/92 ns * * * * * * ns nsFeb. 17/92 * ns ns nsMar. 9/92 ns ns nsMar. 30/92 ns ns nsApr. 13/92 ns ns ns nsJuly 13/92 ns ns ns ns1 The full linear model tested was as follows:It. = + T + Z + T*Z+ P(Z)P(Z) + T*P(Z)tp(z) +5(tzp)nwhere: It = index of injury, T = temperature, Z = zone, and P = provenance. The interactionterms on most test dates were not significant, so those nonsignificant terms weresubsequently eliminated from such models.105Table 3.2. Analysis of variance’ results of frost hardiness testing index of injuryover two winters, plus one summer test, for provenances having familystructure, where: ***P<000I; **f)•()() <P<0.01; * =0.01 <P<0.05; ns = 0.05 <P.i2t I PJZ F(P Z) I I2 T*P(Z) T*F(P Z)Oct. 23/90 * ns ns ns ns nsNov. 6/90 ns * * ns ns nsNov. 20/90 * ns ns ns *Dec. 4/90 ns ** ns ns nsDec. 16/90 ns ns ns ns nsJan. 21/91 ns ns ns ns nsFeb. 11/91 * ns *** ns * nsMar. 4/91 * us *** * ns nsMar. 18/91 us us us ns usApr. 8/91 ns us ns ns ns usOct. 21/91 ns ns ns * * * ns ns nsNov. 4/91 ns us ns ns nsNov. 25/91 * us ** ns ** usDec. 16/91 * us ns ns usJan. 6/92 ns * * * ns * * * ns ns nsJan. 27/92 ns * *** ns ns nsFeb. 17/92 * ns ** ns ns nsMar. 9/92 * ns ns ns *Mar. 30/92 ns ns ns ns *Apr. 13/92 ns ns ns * * * ns ns *July 13/92 ns ns ns ns ns nsThe full linear model tested was as follows:Jt. = + T + Z + T*Z+ P(Z)p(Z) + T*P(Z)tp(z) + F(P Z)fo,Z) + T*F(P Z)Z) +8(tzpf)nwhere: It = index of injury, T = temperature, Z = zone, P = provenance, and F = family. Theinteraction tenus on most test dates were not significant, so those nonsignificantterms were subsequently eliminated from such models.106Table 3.3. Analysis of variance significance levels ofLT50 by test over two seasons(test dates per year are given in Table 3.2) where: = P < 0.001;**JJ <P<0.01; * =0.01 <P<0.05; ns=0.05<P.1990/911 1991/92 Model 12 1991/92Model23It i? Z PiZ P P01 ns ns ns * fl52 ** ns ** * *3 ** * *** ** *4 * * ** ***5 ns ns ns6 ** ns ns7 ** ns ns ns ns8 ns ** ns *9 ns ns * ns ns10 ns ns ns ns ns1 1990 / 91 Model: LT5o = j. + P, +2 1991 / 92 Model 1: LT5o11= .t + Z + P(Z)P(Z) + E(pz)n1991 / 92 Model 2: LT5o = p. + P, + F(P)f,) +where: Z = zone; P = provenance; and F = family107the first test per year, provenance differences occurred in LT50 estimates. Zone andfamily differences also were found during acclimation in the second year except onthe first test in late October.Frost testing of trees growing at Skimikin showed similar ANOVA results inindex of injury to those found with trees growing at UBC during the same test year(Table 3.4). In November, frost index of injury differed at the provenance and familylevels; while zones did not differ, a zone by temperature interaction occurred. TheLT50 differed between provenances, but not between zones or families.A comparison ofLT50 values from Salmon Arm and trees at UBC during thesame year (Figure 3.12) indicated that trees at Skimikin became hardier much quickerthan trees growing at UBC. By comparing the daily temperatures at Skimikin (Figure3.13) with those at UBC (Figure 3.9), it is evident that below-freezing temperaturesoccurred much earlier at Salmon Arm (late September) than at Vancouver (end ofOctober), although warm daytime temperatures were occurring at both locationsduring the fall. By the first testing of trees from Skimikin, Salmon Arm had justexperienced temperatures down to about -15°C.Analysis of variance of the LT50 values obtained from tests of Skimikin treesvs tests ofUBC trees at the same time are given in Table 3.5, and confirm theobservations made regarding Figure 3.12. During November, location differenceswere obvious. Provenance differences were apparent when all provenances wereanalyzed, but when only those provenances having family structure were investigated,variation appeared to reside mainly within provenances. Location interactions did notoccur during the period of acclimation.108Table 3.4. Analysis of variance results of frost testing index of injury on three datesof seedlings growing at Salmon Arm, where: *** = P <0.001;**0001<P<001; *=0.0l<P<0.05; ns=0.05<P.Variable Nov. 8 / 91 Jan. 23 I 92 Mar. 25 / 92All Provenances’Zone ns ns nsProv(Z) *** ***Temprature *** *** ***T*Z ns ns nsT*P(Z) ns ns nsProvenances withFamily Structure2Zone ns ns nsProv(Z) * ** **Family(P Z) ** * nsTemperature *** *** ***T*Z * 115 fl5T*P(Z) ns ns flST*F(PZ) ns ns ns1 Linear model: = + T + Z + T*Z + P(Z)p(Z) + T*P(Z)q,(z) + E(tpz)n2 Linear model: It = + T + Z + T*Z + P(Z)p(Z) + T*P(Z),(z) + F(P Z)f,Z)+ T*F(P Z)tz) + E(trpz)n109-5—-10--15—-20--25--30-000 I0-35--40-IIIIIIIIIII02468101214161820222426WeekFigure3.12.MeanLT50softreesgrowingatSalmonArmcomparedtotreesatUBCduring1991/92.0 -5 -10-15Temp,CDate-209/110/111/112/11/12/13/14/1Figure3.13.DailymaximumandminimumtemperaturesatSalmonAmduring1991/92.Table 3.5. Analysis of variance results of frost LT50 of trees grown at Skimikincompared to trees grown at UBC, where: = P <0.001;**=3 <P<0.01; * =0.01 <P<0.05; ns=0.05<P.Test Date: UBC / SkimikinVariable Nov. 4/8, 91 Jan. 27/23, 92 Mar. 30/25. 92All Provenances1Zone ns ns nsProvenance(Z) *** *** ***Location *** * ***L*Z ns ns nsL*P(Z) ns * ***Provenances withFamily Structure2Zone ns ns nsProvenance(Z) ns *4* *Family(P Z) * ns nsLocation *** ns **L*Z ns ns nsL4P(Z) us * usL*F(PZ) us ns us1 Linear model: LT5ol = + L1 + Z + L*Zi + P(Z)p(Z) + L*P(Z)Ip(z) + (Ipz)n2 Linear model: LT5olfp = ii + L1 + Z + L*Zi + P(Z)p() + L*P(Z)lp(z) + F(P Z)Z)+ L*F(P Z)lZ) + E(1f)1123.3.7. Maximal HardinessDuring midwinter (mid-December to mid-February), when the trees were at ornear their maximum levels of frost hardiness, analysis of all provenances showed thatvariation in I between provenances occurred on almost all test dates (Table 3.1).When provenances having family structure were analyzed (Table 3.2), it appeared thatmost of the variation associated with provenances occurred at the family level. Onthe one midwinter date in the first year when family was not significant, provenancesand the temperature by family interaction terms were both significant, and on the onemidwinter date in the second season where no family variation occurred, strongprovenance differences were observed. Zonal differences did not occur at the timewhen trees were most hardy.Analysis ofLT50 values indicated provenance differences during bothmidwinters except in mid-December of the first test season (Table 3.3). Figure 3.5shows slightly less divergence in the LT50 curves on that date; the rate of acclimationmay have been slowing considerably at that time. Zonal differences did not occur atmaximal hardiness. Family variation was also not evident during midwinter inestimated LT50.Lower maximum LT50’swere obtained during the second year of frost testingat UBC. Provenance rankings over the two seasons were similar, but the degree ofhardiness was deeper for 1991 /92 in all cases. The date when maximum hardinessoccurred each season was the same for all provenances.Midwinter It (Table 3.4) and LT50 at Skimikin varied between and withinprovenances, but not between zones. In comparing midwinter tests at UBC with113Skimikin (Table 3.5), location effects were significant when all provenances wereexamined, but not when only provenances with family structure were analyzed.Provenance differences were apparent in both analyses, and zone did not vary ineither case. A significant location*provenance interaction indicated that provenancerankings changed somewhat between locations, causing genotype*environmentinteractions. Testing in late January showed that trees from Salmon Arm no longerhad a lower LT50 than trees from UBC; in fact the latter had a slightly lower LT50.Hierarchical cluster analysis was carried out on the provenance mean LT501sfound on Jan. 21, 1991 and Jan. 27, 1992 at UBC, the tests closest to the times ofmaximum hardiness. Arbitrarily grouping these two traits into three clusters, the leasthardy cluster consisted ofKooskia, Squamish, Mill Bay, Tofino, and Mt. Mara Lowelev., with Oliver Lake and Benton Flat forming the middle cluster and Mt. Mara Midelev. forming the most hardy cluster. When survival at Skimikin following winterdesiccation, foliar damage at Skimikin, and the final height of trees growing at UBCwere added to the midwinterLT50’s, clustering into three groups produced the sameresults as above with the exception ofMt. Mara Low elev., which moved to themiddle cluster.The mean UBC midwinter LT50’swere correlated to fmal height at UBC (in1991, r= -0.700; in 1992, r= -0.598), mean survival (r= 0.787 for 1991; r= 0.668 for1992) and foliar desiccation (r = 0.829 and 0.750 for 1991 and 1992 respectively) atSkimikin, elevation (r 0.800 and 0.711 for the two years), average number ofgrowing degree-days (r -0.680 and -0.523 for 1991 and 1992), and mean Januarydaily temperature (r = -0.651 and -0.564), while the mean UBC midwinter LT50 for1992 was also correlated to latitude (r = 0.429) and mean annual precipitation114(r = -0.3 58). Stepwise regressions, using a significance level of c = 0.05 for entryinto and staying in the model, found the following models to be the best:LT50Jan91 = 13.823 - 0.008*Elev. - 0.318*Longit.; R2 = 0.729; P <0.0002LT50.Jan92 = 109.359 - 0.028*Elev. - l.061*Longit. - 0.048*Growing days;= 0.759; P <0.0001Hierarchical cluster analysis of the estimated January LT50 of Skimikin treesinto three clusters placed Tofino alone in the least hardy group, Benton Flat and Mt.Mara Mid elev. in the most hardy group, with the middle group being comprised ofthe other five provenances. When survival at Skimikin following winter desiccation,foliar damage at Skimikin, and the fmal height of trees growing at UBC were addedto the midwinter Skimikin LT50, clustering was identical to that for the two midwinterLT501sfor UBC, that is Mt. Mara Mid elev. in the hardiest group, Benton Flat, Mt.Mara Low elev., and Oliver Lake in the middle group, and Kooskia, Squamish, MillBay, and Tofmo in the least hardy group.The mean midwinter LT50 of trees at Skimikin was correlated most of thesame variables as the trees at UBC during 1992 (height: r = -0.343; survival: r =0.583; foliar desiccation: r = 0.644; elevation: r = 0.685; growing days: r = -0.575;mean January temperature: r = -0.622; longitude: r = -0.415; and mean annualprecipitation: r = -0.509). A stepwise multiple regression, using a significance levelof a = 0.05 for entry into and staying in the model, found the following model to bethe best:LT50JanSkim = -30.3 79 - 0.010*Elev.; r2 = 0.469; P <0.0001115As with growth traits of this study, elevation was the most importantindependent variable affecting hardiness in all cases. Elevational dines estimated bysimple linear regression for the January LT50’s ofUBC trees during 1991 and 1992and Salmon Arm trees during 1992 showed that 1°C of hardiness would be obtainedfor every 175 m, 79 m, and 98 m increase in elevation respectively.Calorimetry results are presented in Table 3.6. An example of a graph offoliage temperature over time produced from one tree sample, plus the differencebetween this same sample and the ambient temperature during the period when theexotherms were observed, is given in Figure 3.14.Two exotherms were detected for each tree sample; the first exothermoccurred between -3.6°C and -11.3°C, while the second exotherm occurred between-7.4°C and -13.5°C. No exotherms were seen at temperatures approximating the LT50of these provenances at that particular time (which ranged from about -23.0°C to-27.7°C) as estimated from electrical conductivity frost testing. The meantemperatures per provenance at which the first and second exotherms occurred werein general ranked similarly to the LT50 rankings per provenance, with the hardierprovenances having slightly lower temperatures at which exothermic events tookplace and slightly larger spikes at each exothermic occurrence.No provenance differences were detected in ANOVA of the temperatures atwhich either of two exotherms occurred, the size (°C) of the two exotherms, and thetemperature to which each sample rose to after each exothermic event, although a testinvolving a larger sample size might have been able to detect stronger trends.The first and largest exotherm per sample seen in the current study would116Table 3.6. Temperatures at which exotherms were observed during calorimetryexperimentation on Feb. 15, 1992 in two trees / provenance, plusprovenance mean (size of exotherm in °C is given in brackets) comparedto LT50 of these provenances as determined in normal frost testing.Provenance Tree LT50 1t Exotherm 2nd ExothermOliver Lake A -27.7 -8.89 (1.77) -10.41 (4.75)B -11.30 (5.86) -13.46 (4.30)-10.09 -11.93Mill Bay A -23.1 -4.10 (1.14) -8.07 (1.78)B -5.19 (1.92) -8.04 (3.32)X -4.64 -8.05Kooskia A -23.0 -3.61 (1.52) -9.49 (4.69)B -7.13 (0.31) -8.66 (4.51)5 -5.37 -9.07Mt. MaraMid A -27.3 -5.41 (1.88) -12.08 (4.78)B -6.00 (1.94) -7.42 (3.02)-5.70 -9.7511700EFU01)F-Figure 3.14. Foliage temperature, as monitored by calorimetry, plotted againstambient test temperature for one tree sample (a); the same sampleminus the control channel (b).-10a2 4 6 8 10Time (hours)-30-4006543.21-0—1 Ib0 1 2Time (hours)3 4118correspond to when readily available water is frozen extracellularly. The secondexotherm was probably due to a particular chamber or compartment of the cellfreezing, and not due to intracellular water migrating out of the cell and freezingextracellularly, as the latter occurs gradually, not suddenly (R. Guy, UBC, pers.comm., 1994). Alternatively, two exotherms may have represented the two foliagepieces sandwiching each probe undergoing extracellular freezing at different times.The lack of a low temperature exotherm in this study could be attributable toinsufficient detection by the apparatus. Alternatively, Ritchie (1991) has suggestedthat some boreal and high elevation species do not supercool, although he suggestedthat species which can reach low levels of hardiness without supercooling are able toharden because of an ability to tolerate extreme cytoplasm dehydration, which wouldintuitively seem unlikely for western red cedar. However, these results do echo thoseof George et al. (1974) who found no low temperature exotherm for Thujaoccidentalis.A summarization comparing results of the various tests used to investigatemaximal frost hardiness is given in Table 3.7. For this summary, where seasonaltesting was carried out, the tests done nearest to the time ofmaximum hardiness wereused.A joint study of foliar nutrient analysis and frost hardiness of the same sampletrees found that the LT50 of seedlings was not correlated to any of the macronutrientsanalyzed (N, P, K, Ca, and Mg) or to the ratios of ‘/N, K/N, or K/Ca. However,when separate correlations were done by zone, trees from the interior zone showedcorrelations between the LT50 and N, K, and K/Ca. When correlations were carriedout by provenance, the LT50 was correlated to N, Ca, and ‘/N at Tofino, to P and /N119Table 3.7. Summary of analyses from the various tests used to assess frost hardiness,using the date closest to maximum hardiness where seasonal monitoringwas done, where - indicates that means were not estimable because valuesoccurred over more than one temperature, or values were not estimable atthe family level; see text for definitions of abbreviations.Overall standard All Prov. withVariable Mean ± di Prov. Fam. structure 2p %a2FUBC It 91 - - P F 25.65 26.73UBCI92 - - P P F 56.58 3.10Skim It 92 - - P P F 39.94 10.24UBC LT50 91 -28.19 2.89 P - 73.29 -UBC LT50 92 -38.84 7.04 P P 82.20 0.87Skim LT50 92 -34.62 5.89 P P 52.92 0Jan. Fv/F1 0.49 - P - - -JanYv/F0 1.00 - P - - -Calorimetry2 - - all ns - - -I Fluorescence ratio means are based on values obtained at -35°C, the test temperatureclosest to the estimated LT50 at the time.2 Calorimetry refers to the temperatures at which the two exotherms per sample occurred,the size in degrees C of the exotherms, and the difference between the temperaturesat which exothermic reactions occuffed and the size of the exotherm produced.120at Mill Bay, and to Ca at Benton Flat. A stepwise elimination regression ofLT50 onthe macronutrients and ratios found none of the variables significant to the model atthe 0.05 level; N was significant at P = 0.11 with r2 = 0.067.Analysis of variance of the frost hardiness I nutrient status test found that theLT50 varied between zones but not between provenances per zone. Allmacronutrients and tested ratios varied between provenances but not between zonesexcept for Ca, which did not vary significantly at either level. These results aresimilar to those obtained in the nutrient study covered in Chapter DeaccilmationAll provenances at UBC appeared to begin dehardening around the same date,and are presumed to have completed deacclimation at the same time. Deacclimationon the coast seemed to occur at the same rate for all genetic entries.Patterns of variation during deacclimation seemed similar to those duringacclimation. During deacclimation, differences between zones in It were observed.While variation occurred at the provenance level when all provenances wereinvestigated (Table 3.1), most of this variation appeared within populations whenonly provenances with family structure were analyzed (Table 3.2). For LT50 analysis,zones did not vary; provenances varied on a few dates, and families were significantonly in early March of the second test year (Table 3.3).The nonsignificance of families and provenances near the end of the hardeningcycle can easily be explained, as it is evident that significant differences do not occur121while trees are nonhardy throughout the summer, and variation in LT50 is low. Thenonsignificance of provenances on Feb. 17, 1992 when all provenances were lookedat is harder to explain, but frost curves show slightly less divergence on that date,maybe because the rate ofdehardening was accelerating; zones varied on that date.At Salmon Arm, provenance differences in It and in LT50 were evident in lateMarch (Table 3.4), but zone and family differences were not. At this time, trees fromUBC were dehardening much more rapidly than trees at Skimikin (Figure 3.12).When comparing Vancouver and Salmon Arm in late March, locations variedsignificantly, as did provenances. A location*provenance (genotype*environment)interaction was significant in March where all provenances were examined.3.3.9. Variable Chlorophyll FluorescenceVariable chlorophyll fluorescence analysis results were obtained at testtemperatures similar to expected LT50 values on two dates, once close to maximumseedling hardiness and once when seedlings were almost completely dehardened.Figure 3.15 shows the ratios of FvIFm and Fv/F0 respectively plotted against testtemperatures on January 13, 1992, while Figure 3.16 shows the same ratios asobtained on April 1, 1992.The expected trend for the fluorescence graphs would be decreasing ratioswith lower temperatures, and for the nonfrozen controls to have the highest ratios,indicating that as temperatures decrease, the photoprotective effects ofphotoinhibition are less apparent, and more damage is possible to the plant. This was1220.8 -_______Mibpoil0.7-Squ0.6 - -::. ...... mmLmmME •-k. .- ...Koo0.4-0.3 -0.2- I I I I I I5 0 -5 -10 -15 -20 -25 -30 -35Temp,C3.Mib\oil2.5-Squ2 - E. mmLmmMo4Koo0.5 -0- I I I I5 0 -5 -10 -15 -20 -25 -30 -35Temp, CFigure 3.15. FvIFm (top) and Fv/F0 (bottom) per test temperature in January.1230.8 -________MbpOilTetnp, C3.Mibpoil25. Squ...mmL2.0. ...... nunM..........+..,.............. Koo1.5-...:-•.......0.5I I I2 0 -2 -4 -6 -8 -10Temp, CFigure 3.16. FvJFm (top) and Fv/F0 (bottom) per test temperature in April.124the case for April 1. On the January 13 graphs, both ratio values at - 15°C were lowerthan those at -25°C. The unexpected dips in these curves appear to have been due tosome factor affecting only those samples tested to -15°C, but could alternatively havebeen due to some factor occurring with the -25°C samples. A low light was brieflyturned on by accident just prior to the measuring of fluorescence of the -25°Csamples; however, this short light exposure was too brief to have had any effect.Analysis of variance ofmeasured variables showed no zonal differences inany trait on either test date. Provenance differences were seen in the FV/Fm and FvJF0ratios on both test dates. While F0 in January and F and Fm in April varied byprovenance, these values are meaningless, as these absolute values will differ with theamount of tissue surface area analyzed per seedling, which could not be constant fromtree to tree (R. Guy, UBC, pers. comm., 1994).General provenance rankings based on Fv/F and F/}?0 at the temperatureestimated to be closest to the expectedLT501s(January 13: -3 5°C; April 1: -6°C) werecompared to mean provenance LT50 rankings of these same provenances as found byelectrical conductivity frost tests taken near the dates of the variable chlorophyllfluorescence tests (Table 3.8). Provenance rankings of FvfFm and FvJF0were notsimilar to LT50 rankings.When the average LT50 per provenance was compared to the fluorescenceresults obtained for each provenance, the ratio of Fv/Fm at the test temperature closestto that at which the LT50 occurred ranged from about 0.47 to 0.53 in January andabout 0.57 to 0.67 in April; the Fv/F0 ratio ranged from about 0.91 to 1.13 in Januaryand about 1.45 to 2.09 in April. It is evident that none of the ranges overlapped. Itseems that the relationship between fluorescence and electrical conductivity125Table 3.8. Comparison of provenances based on ranking by LT50 estimated fromelectrical conductivity testing and by variable chlorophyll fluorescenceresults from two test dates; ratios for January 13 are from samples testedto -35°C, while ratios for April 1 are from samples tested to -6°C.Variable chlorophyll fluorescencei Ii0c F1Jan. 13 /92 Mib -30.0 Mib 0.53 Mib 1.13Squ -32.0 mmL 0.50 mmM 1.02Koo -33.0 mmM 0.50 mmL 1.01mmL -33.0 011 0.49 011 1.00011 -37.0 Squ 0.48 Squ 0.95mmM -43.0 Koo 0.47 Koo 0.91April 1 /92 Mib -5.1 Squ 0.67 Squ 2.09Squ -5.2 mmL 0.65 mmL 1.88Koo -5.4 011 0.62 011 1.69mmL -6.0 Mib 0.61 Mib 1.67mmM -6.0 mmM 0.61 mmM 1.60011 -6.7 Koo 0.57 Koo 1.45126measurements of cold hardiness is thus not clearcut and hence the predictability offrost damage from the fluorescence test is not immediately obvious.1273.4. DiscussioNThe interpretation of frost hardiness data was hampered by the low number ofprovenances and families per provenance which were able to be tested. For instance,comparison of provenance and family variance components of hardiness traits ondifferent dates did not give a very cohesive story. However, it is reasonable toconclude that genetic variation does exist in the cold hardiness ofwestern red cedar,both between and within provenances. It appears that more of the variation foundwhen seedlings are acclimating or deacclimating occurs within rather than betweenprovenances, and greater differences between entities occur when the trees are at theirmaximum stage of hardiness. Zonal differences and significant temperatureinteractions only occurred at the times when acclimation or deacclimation was rapid.In observing the hardiness curves of the same genetic entries grown at thesame mild coastal location over two winters, no differences in date when acclimationcommenced or when deacclimation had terminated were apparent between zones,provenances, or families; nor could differences in the date of estimated maximumhardiness be found between genetic entries. Once hardening had begun, provenancesfrom the interior range acclimated at a faster rate. However, both coastal and interiorprovenances growing at UBC all dehardened at the same rate. It appears thatacclimation began about a week earlier in 1991 than in 1990, although deacclimationappeared to have been completed at about the same time for both test seasons.Timing of the onset of hardiness responses thus seemed to be plastic in nature,following the pattern of a generalist, and hence it can be inferred that timing ofacclimation, and probably deacclimation, is not under direct genetically variable(although plasticity itselfmay vary genetically).128With an everchanging state of hardiness occurring between autumn andspring, it is not surprising to find one test out often carried out per year where nowithin-population differences could be found. The two dates when this occurred forindex of injury, in early winter of the first test year and in midwinter of the secondyear of testing, coincided with a period just prior to the time when maximumhardiness was reached.By comparing cold hardiness of trees grown at UBC vs trees grown at SalmonArm during 1991 / 92, a comparison of the same genetic entries of the same ageduring the same year was being made between seedlings grown at a mild coastal siteand those grown at a relatively harsh interior location. Trees at Skimikin beganhardening sooner in the fall and dehardened later in the spring, although the degree ofmaximum hardiness reached in those trees was not deeper than trees from UBC.Seedlings at both UBC and Skimikin which were tested during the winter of1991 I 92 became hardier than the same genetic entries tested the previous year atUBC. While seedling age may have been a factor, it seems more likely that theestimated maximumLT50tsobtained during 1991 /92 from the two locationsrepresent the maximum hardiness capability ofwestern red cedar, as no other study todate reports hardiness levels lower than the levels found in this study for western redcedar. The maximum degree of hardiness thus appears to be a plastic response in thisspecies.The graphical representations of daily maximum and minimum temperaturesat the two locations where seedlings were growing may offer some explanation intoseedling responses. During the first test season, the minimum temperature had notdipped below +11°C until the very end of of September 1990, when the temperature129fell to +7.3°C. The temperature did not go below +5°C until November 1, andfreezing temperatures were not encountered until December 19. Interior provenanceswere acclimating fairly rapidly from late October onwards, while coastal provenancesdid not acclimate at a faster rate until about November 20, about three weeks aftertemperatures below +5°C were experienced. Acclimation rates slowed in allprovenances from about mid-December.By contrast, in the fall of 1991, temperatures fell below +10°C on September14, reached down to +7°C on September 21, and went below +5°C on October 18.There was an increase in the number of hours when the temperature was below +5°Cduring early November compared to the same time period in the previous year. Thefirst below-freezing temperature occurred on October 28. Interior provenancesacclimated during the fall of 1991 at a steady rate until January 1992. Coastalprovenances increased in rate of acclimation around November 25, about the sametime as in the previous year, but did not slow until January. All provenances werehardier than they had been a year earlier by about December 16.At Salmon Arm, temperatures below +10°C occurred intermittentlythroughout July and August of 1991, below +7°C on August 26, and below +5°C bySeptember 3. The first frost occurred on September 22. By the first week inNovember, the average LT50 of seedlings growing at Skimikin, about -28.8°C, was onaverage -15.4°C lower than that of trees at UBC. Trees at Salmon Arm onlydecreased in hardiness a further -5.8°C for the winter. Thus under extreme weatherconditions, seedlings of all provenances acclimated much sooner and reachedmaximum hardiness much earlier, although seedlings were maintained at maximalhardiness at least until the end of January, when maximal hardiness was believed to130take place in UBC trees.The occurrence and timing of low temperature thus appear to be the factorsmost strongly responsible for triggering hardening in this species. If temperaturesbelow +5°C are critical for inducing acclimation, then the results of this study wouldindicate that changes in the rate of acclimation can be observed roughly three to fourweeks after such temperatures are encountered, at least in trees growing in mildlocations. It is suspected that a more sudden response may be found in harsherenvironments. These conclusions seem to be supported by those of Silim (1991), whoreported a lag phase ofmetabolic adjustment ofwestern red cedar in response to a lowtemperature stimulus of about three weeks.Below-freezing temperatures undoubtedly affect the depth of hardinessreached in any one year. However, as much lower winter temperatures wereexperienced during midwinter 1990/91 (to -13.4°C on December 29) than in 1991 /92 (to -2.9 on January 7) at UBC, but trees did not become as hardy in the first testyear as in the second, timing of low temperatures seems to be the critical element.Shoots of at least some seedlings were still elongating by December 1, 1990,prior to any frost. As stated earlier, western red cedar is an opportunist and will growas long as conditions are favourable. Thus acclimation commences before shootgrowth has ceased, although deep hardening probably cannot occur until growth hasstopped.At UBC, minimum temperatures began rising by the first of February in 1991(however, late frosts happened in early March). In 1992, the last frost was on January19. During both years, deacclimation was underway by mid-February, and was131complete by about mid-April. By contrast, temperatures substantially below freezingwere occurring at least until the end of April 1992 at Salmon Arm. By March 25,seedlings at Skimikin were still hardy to about -20.8°C.It is obvious that a period of above-freezing temperatures is necessary for theonset of deacclimation to take place. These results concur with those of Silim (1991)and Krasowski and Owens (1991) for one western red cedar seed source each. Silim(1991) also noted a lag of only five days after exposure to warm temperatures beforedehardening commenced in this species. There appears to be genetic uniformity inthe response to the environmental cues affecting dehardening.While the response ofwestern red cedar to the external stimuli responsible forinitiating acclimation and deacclimation and the winter maximum hardiness levelreached appear to fit Rehfeldt’s (1984) definition of a generalist exhibiting phenotypicplasticity, specialist behaviour was also observed in the form ofwithin-populationvariation in hardiness per test date. Ifwestern red cedar is an outcrosser with goodgene flow, within-population differences would be expected to be higher thanbetween-population differences.However, selective pressures are obviously present in the form ofwinterenvironmental conditions. These selective pressures could be low temperatureextremes, factors contributing to winter desiccation, and possibly some unknownfactors affecting the relationship between a trees genotype and its ability to survivethe winter with minimum damage. It seems that the severity ofwinter conditions isthe most influential factor shaping the patterns of frost hardiness variation in westernred cedar.132Between-provenance differences were evident in January 1992 when seedlingsgrown at UBC were at their hardiest and were also evident on all dates on trees grownat Skimikin. Thus when conditions are most severe, between-population differencesseem to overshadow within-population differences, and adaptive variation is at theprovenance level. While these results could be compared to survival at Skimikin aftera severe winter, another indicator of adaptive variation in which provenancedifferences were apparent, it must be kept in mind that frost analysis was carried outon a small number of provenances having family structure. Thus the determination ofwhether variability in degree ofmaximal hardiness obtained in any one year is mainlyallocated between or within populations cannot be definitively made at this time.Scrutiny of individual provenance performance showed that Kooskia in Idaho,the only provenance tested which can safely be assumed to have remained unglaciatedduring the last ice age, was one of the least hardy of the interior provenances, alongwith Mt. Mara Low elevation. Mt. Mara Mid elevation was consistently the mosthardy of all provenances from about November to March. From the coastalprovenances, the high latitude Oliver Lake near Prince Rupert, B.C. was the hardiestthroughout the two winters. These observations were corroborated by cluster analysisof the mean midwinter provenance LT501s.Although it seems that elevational dines do exist in hardiness traits, too fewprovenances were sampled to get an accurate picture of the steepness of such dines.It is quite possible that latitudinal dines also exist, especially in the interior wheretemperature is not moderated by the ocean.The exploratory investigations into using alternative methods for assessingcold hardiness were not promising. While variable chlorophyll fluorescence may give133indications of relative frost hardiness on any one date, this method appeared to have apoor predictive ability compared to the electrical conductivity method. No obviousrelationships between the ratios FvIFm and Fv/F0 and the LT50 as found by electricalconductivity were observed.Calorimetry was also unpromising. The two exotherms that occurred persample tree were not related to hardiness levels or to differences between geneticentries. Reasons for the absence of a low temperature exotherm indicating the pointof serious injury are unknown. Although an exotherm may have occurred but was toosmall for detection, it is probable that western red cedar does not supercool its cellularwater, so no water is available at very low temperatures for freezing and henceproducing an exotherm.Although provenance variability occurred in the nutrient levels of trees testedfor frost hardiness, as was noted in the growth study, no relationships betweennutrient status and frost hardiness were observed. None of the macronutrients were atcritically low or excessively high levels. If nutrients were at limiting or toxic levels,it is possible that hardiness may be affected.The positive correlations between frost hardiness parameters and seedlingsurvival and the inverse relationship between frost hardiness and seedling height havepractical implications for selection within this species, as all of these traits would bedesirable. Silvicultural techniques such as planting seedlings under an overstorymight be considered as tools to assist in survival and minimize the effects of coldtemperatures on trees selected for other traits such as growth. Alternately,correlation-breakers could be sought, selection indices developed, or tandem selectionbe practiced.134In summation, the degree of adversity ofwinter environmental conditionsseems to be the most important factor shaping the variability of cold hardiness inwestern red cedar. During the times when unfavourable environmental conditionswere most extreme, between-population differences were higher than within-population differences, although the opposite was true during the most rapid periodsof acclimation and deacclimation. Maximum differences between populationsoccurred when the trees were at their annual maximum hardiness levels.Low temperatures experienced in the autumn appeared to initiate the inductionof the acclimation process, with the timing of these cold temperatures seeming todetermine the depth of tree hardening reached over a winter. The severity ofwinterconditions at the test locations was a factor in the timing ofhardening anddehardening (which were plastic in nature), and in the absolute hardiness levelsreached (down to a lower limit). Of the geographic aspects associated with eachprovenance that were studied, elevation was the most influential.1353.5. LITERATURE CITEDBlum, A. 1988. Plant breeding for stress environments. CRC Press Inc., Florida,223 pp.Burke, M.J., L.V. Gusta, HA. Quamme, C.J. Weiser, and P.H. Li. 1976. Freezingand injury in plants. Ann. Rev. Plant Physiol. 27: 507-528.Burr, K.E., R.W. Tinus, S.J. Waliner, and R.M. King. 1990. Comparison of threecold hardiness tests for conifer seedlings. Tree Physiol. 6: 351-369.Cannell, M.G.R. and L.J. Sheppard. 1982. Seasonal changes in the frost hardiness ofprovenances ofPicea sitchensis in Scotland. For. 55: 137-153.Columbo, S.J., C. Glerum, and D.P. Webb. 1984. Frost hardiness testing: anoperational manual for use with extended greenhouse culture. Ont. Mm. Nat.Res., 14 pp.Columbo, S.J. and E.M. Raitanen. 1991. Frost hardening in white cedar containerseedlings exposed to intermittent short days and cold temperatures. For.Chron. 67(5): 542-544.Fisher, J.T. and J.G. Mexal. 1984. Nutrition management: a physiological basis foryield improvement. Pp 27 1-299 In Duryea, M.L. and G.N. Brown, eds.Seedling physiology and reforestation success. Martinus NijhoffPubl., 325pp.Folk, R.S., S.C. Grossnickle, J.E. Major, and J.T. Arnott. 1994. Influence of nurseryculture on western red cedar II. Freezing tolerance of fall-planted seedlingsand morphological development of fall- and spring-planted seedlings. NewForests 8: 231-247.George, M.F., M.J. Burke, H.M. Pellett, and A.G. Johnson. 1974. Low temperatureexotherms and woody plant distribution. Hort Sci. 9(6): 5 19-522.Glerum, C. 1985. Frost hardiness of coniferous seedlings: principles andapplications. Pp 107-123 In Duryea, M.L., ed. Evaluating seedling quality:principles, procedures, and predictive abilities ofmaj or tests workshopproceedings. Oreg. St. Univ., Corvallis, 143 pp.Kozlowski, T.T. 1971. Growth and development of trees Vol. I. Academic Press,443 pp.136Krajina, V.J., K. Klinka, and J. Worrall. 1982. Distribution and ecologicalcharacteristics of trees and shrubs in British Columbia. UBC Press,Vancouver, 288 pp.Kramer, P.J. and T.T. Kozlowski. 1979. Physiology of woody plants. AcademicPress, 811 pp.Krasowski, M.J., and J.N. Owens. 1991. Growth and morphology ofwestern redcedar seedlings as affected by photoperiod and moisture stress. Can. J. For.Res. 21: 340-352.Lavender, D.P. 1985. Bud dormancy. Pp 7-15 In Duryea, M.L., ed. Evaluatingseedling quality: principles, procedures, and predictive abilities ofmajor tests.Oregon State Univ., 143 pp.Levitt, J. 1980. Responses ofplants to environmental stresses Vol. I. Chilling,freezing, and high temperature stresses. Academic Press, 497 pp.Miller, P.R. 1978. Abiotic diseases. Pp 5-41 In Bega, R.V., ed. Diseases of Pacificcoast conifers. USDA FS Agric. Handbook No. 521, 206 pp.Minore, D. 1983. Western redcedar - a literature review. USDA FS PNW For. andRange Exper. St. Gen. Tech. Rep. PNW-150, 70 pp.Ogren, E. 1991. Prediction ofphotoinhibition ofphotosynthesis from measurementsof fluorescence quenching components. Planta 184: 538-544.Oquist, G. and N.P.A. Huner. 1991. Effects of cold acclimation on the susceptibilityofphotosynthesis to photoinhibition in Scots pine and in winter and springcereals: a fluorescence analysis. Functional Ecology 5: 91-100.Pellett, H.M. and J.V. Carter. 1981. Effect of nutritional factors on cold hardiness ofplants. Hort. Rev. 3: 144-171.Perry, T.O. 1971. Dormancy of trees in winter. Science 171: 29-36.Powles, S.B. 1984. Photoinhibition ofphotosynthesis induced by visible light. Ann.Rev. Plant Physiol. 35: 15-44.Rehfeldt, G.E. 1984. Microevolution of conifers in the northern Rocky Mountains: aview from common gardens. Pp 132-146 In Lanner, R.M., ed. Proc., 8th N.A.For. Biol. Workshop, Logan, Utah, 196 pp.Rehfeldt, G.E. 1980. Cold acclimation in populations ofPinus contorta from thenorthern Rocky Mountains. Bot. Gaz. 141(4): 458-463.Rehfeldt, G.E. 1978. Genetic differentiation of Douglas-fir populations from thenorthern Rocky Mountains. Ecol. 59(6): 1264-1270.137Ritchie, G.A. 1991. Measuring cold hardiness. Pp 557-582 In Lassoie, J.P. and T.M.Hinckley, eds. Techniques and approaches in forest tree physiology. CRCPress Inc., Florida, 599 pp.Sakai, A. 1983. Comparative study on freezing resistance of conifers with specialreference to cold adaptation and its evolutive aspects. Can. J. Bot. 61: 2323-2332.Sakai, A. 1970. Mechanism of desiccation damage of conifers wintering in soil-frozen areas. Ecology 5 1(4): 657-664.Sakai, A. and S. Okada. 1971. Freezing resistance of conifers. Silv. Gen. 20: 91-97.Sakai, A. and C.J. Weiser. 1973. Freezing resistance of trees in North America withreference to tree regions. Ecol. 54(1): 118-126.Salisbury, F.B. and C.W. Ross. 1978. Plant physiology, 2nd ed. Wadsworth Pubi.Co., 436 pp.Schaedle, M. 1991. Nutrient uptake. Pp 25-59 In van den Driessche, R., ed.Mineral nutrition of conifer seedlings. CRC Press, 274 pp.Silim, S.N. 1991. Regulation of cold hardiness in seedlings ofwestern red cedar,yellow cedar and white spruce. Ph. D. Thesis, Univ. B.C., 142 pp.Silim, S.N. and D.P. Lavender. 1991. Relationship between cold hardiness, stressresistance and bud dormancy in white spruce (Picea glauca [Moench] Voss)seedlings. Pp 9-14 In Proc. of the 1991 Forest Nursery Assoc. ofB.C.meeting, 153 pp.Sit, V. and M. Poulin-Costello. 1994. Catalog of curves for curve fitting. B.C.Ministry of Forests Research Branch Biometrics Handbook No. 4, 110 pp.Steponkus, P.L. 1984. Role of the plasma membrane in freezing injury and coldacclimation. Ann. Rev. Plant Physiol. 35: 543-584.Taiz, L. and E. Zeiger. 1991. Plant Physiology. Benjamin! Cummings Pubi. Co.Ltd., 565 pp.Thomashaw, M.F. 1990. Molecular genetics of cold acclimation in higher plants. Pp99-13 1 In Scandalios, J.G., ed. Advances in genetics Vol 28. Genomicresponses to environmental stress. Acad. Press Inc., USA, 308 pp.Vaartaja, 0. 1959. Evidence of photoperiodic ecotypes in trees. Ecol. Monog. 29(2):91-111.138Vidaver, W.E., G.R. Lister, R.C. Brooke, and W.D. Binder. 1991. A manual for theuse ofvariable chlorophyll fluorescence in the assessment of theecophysiology of conifer seedlings. B.C. Ministry of Forests FRDA Report163, 60 pp.Weger, H.G., S.N. Silim, and R.D. Guy. 1993. Photosynthetic acclimation to lowtemperature by western red cedar seedlings. Plant, Cell and Envir. 16: 1-7.Weiser, C.J. 1970. Cold resistance and injury in woody plants. Science 169: 1269-1277.1394. SELF-FERTILIZATION VS POLYCROSSING4.1. INTRODUCTIONMost temperate zone coniferous forest tree species studied to date have beenfound to have high levels of genetic diversity. Although partial self-fertilization doesoccur in most species (Muona, 1990), conifers are predominantly wind-pollinatedoutcrossers, with the average proportion of outcrossed progeny for a species usuallyclose to or greater than 0.90, and average individual population estimates usuallygreater than 0.80 (Hanirick et al., 1979; Muona, 1990; Adams and Birkes, 1991). Forsuch species, inbreeding depression is common, and may be manifested in low filledseed production, low germination rates, poor growth, higher mortality rates, anddifferential physiological response to stress (Franklin, 1970; Park and Fowler, 1982;Woods and Heaman, 1989; Yazdani and Lindgren, 1991; Blake and Yeatman, 1989),with the most commonly reported effects being the first two.Members of the genus Thuja that have been studied have been found to havelower outcrossing rates than those ofmany other conifers. Mean multilocusoutcrossing rates of 0.75 for Thuja orientalis (Xie et al., 1991) and 0.635 for Thujaoccidentalis (Perry and Knowles, 1990) were observed over seven and four locirespectively. Perry et al. (1990) found a mean heterozygosity of 0.094 over for Toccidentalis, three times that found by Yeh (1988) for one Vancouver Island andseven southern interior B.C. populations of Thujaplicata. El-Kassaby et al. (1994)obtained an outcrossing rate for two loci of 0.32 in a western red cedar seed orchardpopulation consisting of 28 trees, and suggested that western red cedar has a high140level of natural inbreeding, with possibly as high as a 50 % selfing rate. Asmentioned earlier, other studies have found little variation in isozymes (Copes, 1981)and leaf oil terpenes (von Rudloff and Lapp, 1979; von Rudloff et al., 1988) ofweestern red cedar.Owens et a!. (1990) found that selfed seed from two clones ofwestern redcedar had only a slightly higher embryo abortion rate than outcrossed seed, andselfmg had no apparent effect on amount of filled seed produced. A further smallnonreplicated study (Colangeli, unpubi. manuscript) also found similarities in numberof full-sized seeds, percentage filled seed, and in germination rates between selfedand outcrossed seed.It is unknown whether the similarities between selfed and outcrossed seed willhold true in western red cedar for a larger sample population. It is also not knownwhether any effects of inbreeding depression resulting from self-pollination will startto become apparent in a nursery environment and later in a field plantationenvironment. The purpose of the current preliminary study was to investigate theseissues in a small number of families.Self-pollinated and polycrossed maternal half-sib seedlings formed the samplepopulation. Cone and seed attributes were monitored to determine whether they wererelated to subsequent growth. Seedling growth was represented by height, root collardiameter (RCD), and dry weight measurements. Seedling frost hardiness wasinvestigated, and served as an example of an adaptive trait.1414.2. METHODS4.2.1. Seed SourcesDuring the spring of 1990, 23 potted six-year old western red cedar graftedclones growing at the B.C. Ministry of Forests CLRS (elevation: 200 m; latitude:48°49’ N; longitude: 124°10’ W; mean annual frost-free period: 173 days; meanannual precipitation: 214.7 cm (Envir. Canada, pers. comm., 1994); biogeoclimaticvariant: CWHxm2) were selected as maternal parents. Sixteen of these were from therelatively dry CDFmm, CWHxm, and CWFTdm subzones (Table 4.1), with theremaining seven from the very wet CWHvh and CWHvm subzones (see Meidingerand Pojar, 1991 for details on the biogeoclimatic ecosystem classification ofB.C.).Each female was selfed by maintaining male and female strobili inside onepollination bag and outcrossed by injecting, with a hypodermic needle, a 10-clonepolycross pollen mixture ofunrelated genotypes into a second pollination bag onanother branch in which all male strobili had been removed. A separate tester mix(Tester 1), where contributing pollen parents were from the same three subzones asthe female parents, was used with females from the dry subzones than that used withthe trees from the wet subzones (Tester 2). The latter tester was made up ofpollenfrom the CWHvh, CWHvm, and CWHwh subzones (Table 4.2).Seed was collected and sown into plug styroblock 415 containers on April 11,1991 in a randomized complete block design. Seedlings were grown in a heatedfiberglass greenhouse under a normal growing regime for this species. A subset of100 seeds per female parent (herein referred to as family) and treatment (self vspolycross) were X-rayed; unfilled seeds were then separated out and the remaining142Table 4.1. Maternal parents used in the selfing / outcrossing trials, classified bytester used in the polycross.BGC Subzone / Seed / Dry FrostClone Variant Height Weight HardinessTester 1 (dry) 181 CDFmm x398 CDFmm x x x400 CDFmm x435 CDFmm x355 CWHxm1 x432 CWHxm1 x x x438 CWHxm1 x439 CWHxm1 x x x421 CWHxm2 x x445 CWHxm2 x518 CWHxm2 x519 CWHxm2 x198 CWHdm x x x206 CWHdrn x408 CWHdm x411 CWHdm xTester 2 (wet) 307 CWHvh1 x312 CWHvh1 x330 CWHvm1 x341 CWHvm1 x344 CWHvm1 x x x367 CWHvm1 x486 CWHvml x143Table 4.2. Pollen contributions of the two testers used.Clone BGC Subzone / VariantTester 1 (dry) 441 CWHxm2 20.2200 CWHdm 15.1182 CDFmm 15.1431 CDFmm 15.1437 CDFmm 7.6514 CWHxm2 7.6180 CWHxm1 7.6520 CWHxm1 6.0436 CDFmm 3.8506 CWHxm2 1.9Total CDFmm 41.6CWHxm2 29.7CWHxm1 13.6CWHdm 15.1Tester 2 (wet) 271 CWHwh1 20.0166 CWHvml 13.3293 CWHvm1 13.3413 CWHvm1 13.3174 CWHvml 6.7337 CWHvm1 6.7453 CWHvm1 6.7487 CWHvm2 6.7466 CWHvh1 6.7493 CWHvh1 6.7Total CWHvm1 60.0CWHvm2 6.7CWHvh1 13.3CWHwhl 20.0144filled seeds weighed.4.2.2. Seedling MeasurementsSeedling heights were taken repeatedly throughout the growing season on fourseedlings per family / treatment combination in each of six blocks for the purposes ofexamining growth curves. Measurements were taken every two weeks, starting fiveweeks after sowing, until October 31, 1991, at which time growth had slowedconsiderably but not completely. Analysis was only carried out on the final heightmeasurements taken near the end of the growing season, as it was deemedunnecessary to analyze heights on each test date, but growth curves were generatedfrom the biweekly data. Root collar diameters were also taken at the end of themeasurement period.The choice of families to be used in tests involving subsets of families werelimited by inconsistencies in the amount of available seed per family; more of thefamilies from Tester 1 had surplus germinants for such studies. Dry weightmeasurements were taken in January 1992 on selfed and outcrossed offspring of asubset of five families (ofwhich all but one were from Tester 1). Twenty-fourseedlings were measured from each family / treatment combination. Samples werecut at the root collar, bagged, and dried in a convection oven at 100°C for 24 hoursand then immediately weighed upon removal from the oven.During the winter of 1991 / 92, cold hardiness testing was carried out on threeseparate dates on six families, all but one ofwhich were from Tester 1. Four separate,nonbuilced trees per family / treatment combination were tested at three temperatures145per test date. The electrical conductivity method (Glerum, 1985; also see Section3.2.4.2 in Chapter 3) was used to assess relative cold hardiness. Freezing was doneusing a programmable freezing unit. Test dates were November 13, 1991, February5, 1992, and March 17, 1992.A second winter of frost testing involving eight selfed and polycrossedfamilies was carried out during 1992 I 93. Seedlings had been transplanted to a siteclose to Jordan River on Vancouver Island, very close to the ocean at about 300 melevation. Frost testing was done six times over the winter. Test dates weresomewhat limited by an inability to access the site throughout the winter due tosnowfall on unplowed secondary logging roads, making them impassable.4.2.3. Data AnalysisCorrelations were estimated between cone and seed data, between dry weightparameters, and between mean seed, final height, and dry weight data to determine towhat extent certain traits were influenced by others, e.g. whether maternal effects ofseed affected early growth, and whether biomass allocation shifted with changes intotal weight. Due to nonrepliction of samples, only separate ANOVA’s on cone andseed data could be done, where either treatment was used as the error term to testfamily effects, or where family was used as the error term to test treatment effects.Analyses of variance were performed on tree height measurements and frosttest data collected on all test dates and on dry weight measurements, using SAS®PROC GLM. In all tests, treatment effects were assumed to be fixed; family effectswere treated as random where all families were tested (cone, seed, and growth) and146were treated as fixed where a subset of the families were tested (dry weight and frost),as members of these subsets were chosen for a particular reason and were notassumed to represent the total population. Tester was not included in the models fordry weight and frost injury, as only one family from each of these small data sets wasfrom Tester 2. Tester and its interactions were not significant for height and RCD, sothese terms were subsequently removed from the model. Replication (rep)interactions were lumped into the error term of the height and root collar diametermodels, as they too were not significant. General forms of the linear models were:Sf = Jt + Mm + 1’(M)m) + E(fin)tSf = j. + Mm + T(M)t(m) +Ht = JI+Rr+Tt+Ff+T*Fff+s(flDw = p+T+Ff+T*Fff+flIt = + C + T + C*TCt + Ff+ C*FCf + T*Fff+ C*T*Fcff+5(ctf)nwhere: S = cone and seed traits; Ht = seedling first year height (cm) and rootcollar diameter (mm); Dw = dry weight parameters; It = frost index ofinjury; R = replication; M = male tester; F = family; T = treatment; andC = temperature (°C)Expected mean square equations for height and root collar diameter are givenin Appendix 1.11. For frost testing, all factors were considered to be fixed, so themean square error term was the approriate denominator for testing all factors.1474.3. RESULTS4.3.1. Cone and Seed TraitsTable 4.3 lists the overall means and standard errors ofmeasured seed,growth, and dry weight traits by treatment, and gives family ranges of traits.Maternal parents from the CWHvh and CWHvm subzones (Tester 2) appeared tohave higher average values for all cone and seed traits, although this could not betested due to nonreplication of samples. Families seemed to have large ranges for allof these traits except for percentage of female strobili (herein referred to as ‘flowers”)maturing into cones, as most flowers did develop into a cone. Differences betweentreatments did not seem great for any of these parameters.All cone and seed parameters were greater in Tester 2 than in Tester 1;however, these differences were found to be statistically significant only inpercentage seedfihl and in number of female flowers counted. Significant familydifferences were apparent in all cone and seed traits except percentage of femaleflowers maturing into cones (Table 4.4). No treatment differences occurred in anycone or seed trait. Seed weight and percentage filled seed were not correlated witheach other; nor were they found to be correlated to the percentage of female flowerswhich developed into cones, or to the mean seedling height at the end of one year.4.3.2. Seedling Growth: Height. Root Collar Diameter. and Dry WeightSignificant differences between families in final height and root collardiameter (P <0.001) occurred (Table 4.4). However, there was no evidence of height148Table 4.3. Treatment (self-pollinated vs polycrossed) means ± standard errors andranges of family means of studied first-year growth traits.Irjt # families Poly (± s.e.) Self (± s.e.) Family Range#Female flowers 23 139.9 (14.01) 146.0 (15.01) 49 ... 272% flowers/conesi 23 76.6 (5.59) 57.9 (6.53) 31.5 ... 100Seeds/cone 23 10.8 (0.94) 10.0 (0.94) 4.9 ... 20.6Seed fill % 23 33.5 (4.44) 33.7 (4.04) 7.0 ... 73.9Seed weight (g)2 23 0.15 (0.007) 0.15 (0.009) 0.10 ... 0.20Height (cm) 23 28.8 (0.24) 28.5 (0.24) 23.2 ... 33.2RCD (mm) 23 3.65 (0.03) 3.60 (0.03) 3.19 ... 4.25Shoot dry wt (g) 5 2.69 (0.063) 2.49 (0.059) 2.49 ... 2.72Rootdrywt(g) 5 1.33 (0.036) 1.19 (0.033) 1.15... 1.33Total dry wt (g) 5 4.02 (0.093) 3.68 (0.088) 3.68 ... 4.05ShOOt,’R00 5 2.11 (0.046) 2.17 (0.043) 2.14 ... 2.281 % female flowers developing into fulisized cones was assumed to be an indication of coneabortion rate, and seed fill % an indication of seed viability; # female flowers andseed weight are maternal in nature, and are included here as an indication of initialtest conditions2 Seed weight is based on 100 seeds149Table 4.4. Summary of significance levels found through analyses of variance,where ***P<000l; **J1 <P<0.0l; <P<0.05; ns0.05 <P.Trait # Families Family Treatment (S vs P) Fam*TreatFemale flower 23 - -Female flower 23 - ns -% flowers/cones 23 ns - -% flowers/cones 23 - ns -Seeds/cone 23 *** - -Seeds/cone 23 - ns -Seed fill 23 * - -Seed fill 23 - ns -Seed weight 23 * - -Seed weight 23 - ns -Height, 1 St year 23 * * * ns * * *RCD, 1St year 23 *** ns **Shoot dry weight 5 ns *Root dry weight 5 ns ** *Total dry weight 5 ns ** nsShoot/Root ratio 5 ns ns150differences between selfed and cross-pollinated half-sib offspring from the samefamily. Growth curves of overall mean height per treatment were almost identical,although mean heights per treatment per family diverged slightly. Interactionsinvolving family*treatment were significant in ANOVA’s performed on allmeasurement dates, as selfs were larger than outcrosses in some families.Family variance components for polycrosses were about 2/3 those of the selfs(Table 4.5). A Variance Ratio F test (Zar, 1984) showed that the ratio of the familyvariance components of selfs to polycrosses was significantly less than the expected4:1 ratio for both height and root collar diameter. Assuming no dominance orinbreeding in the parental generation or in the polycrosses, it would be expected thatthe family variance components of the selfed progeny should be four times greaterthan those of the half-sib outcrosses (Namkoong, 1966; Wilcox, 1983). The same testshowed that the family variance components for height and root collar diameter didnot significantly differ between the two levels of inbreeding.The heights of trees sampled for dry weight parameters were again found byANOVA to significantly vary between families but not between self! polycrosstreatments. Root collar diameters of trees sampled for dry weights did not differbetween either families or treatments. Treatment effects were not found in theshoot/root ratio, but were apparent in shoot, root, and total shoot + root dry weights(Figure 4.1). No family differences were noted in any dry weight parameter.Family*treatment interactions were significant in root dry weight and in theShoot/root ratio; in one out of five families for shoot, root and total dry weights (twoout of five families for the Shoot/root ratio), the average weight was greater in selfsthan outcrosses, causing the interactions.151Table 4.5. Family variance components ± standard error, and % (T2Fami1y /a2Totalfor first year height and root collar diameters of selfed and polycrossedseedlings.Trait Treatment cy2Fi± s.e.) cTF / a2THeight Polycross 4.478(1.619) 14.56Self 6.590 (2.211) 20.86Root collar diameter Polycross 0.037 (0.0 17) 6.25Self 0.053 (0.021) 9.20152Figure 4.1. Mean shoot, root, and total dry weights per family for selfed vspolycrossed one-year old progeny. dry wt (g)L398198 344 432 4391. 1Root dry wt (g)398 432198 344 439Total dry wt (g) iii198 344 398 432 439153All dry weight parameters were correlated to each other, and to the root collardiameter of trees sampled for dry weight, except for shoot dry weight with theshoot/root ratio. Seedling height of trees sampled for dry weight was correlated to theroot collar diameter and to all dry weight measurements except for root dry weight.Correlations involving the shoot/root ratio were negative except with height. Table4.6 shows results from analyzing each family individually for dry weight traits.4.3.3. Frost HardinessA significant difference in the frost testing index of injury was found betweenfamilies on all three test dates (Table 4.7). Treatment differences were not significanton November 13, 1991, but were significant on February 5, 1992 and March 17,1992. It appeared that selfed trees were less hardy than the polycrosses from at leastthe time ofmaximum hardiness until trees had fully dehardened in the spring (Figure4.2). Some families appeared to harden at different rates. Family by treatmentinteractions were significant in all tests; on the last two dates, selfs were more hardythan polycrosses in one family out of the six tested.A second winter of frost testing involving eight selfed and polycrossedfamilies showed even greater differences between the two treatments at maximumhardiness in the second season of testing, with polycrosses being significantly hardierthan the selfs during the periods ofmaximum hardiness and dehardening (Figure 4.3).154Table 4.6. Significance levels of treatment (S vs P) found when separate ANOVA’sof dry weight parameters were performed per family, where: *** = p0.001; ** = 0.001 <P <0.01; * = 0.01 <P <0.05; ns = 0.05 <P.Family Shoot dw Root dw ShOOt/Root Total dw198 * ns ns *344 ns ns ns ns398 ns ns ** ns432 ns ns * ns439 ** ** *155Table 4.7. ANOVA results of frost testing selfed and outcrossed seedlings growingat Cowichan Lake during 1991 / 92 (six families), and at Jordan Riverduring the winter of 1992 / 93 (eight families), where: = P < 0.001;**0001<P<001; *=0.01<P<0.05; ns=0.05<P.Em Treat I Temp (C) E I F*T*CNov. 13/91 ns * *** ** ns nsFeb. 5/92 *** * ** 115 * nSMar. 17/92 *** * *** * nsSept. 25/92 *** * *** *** ** ** ***Oct. 2/92 ns ns ns ** nsOct. 30/92 *** ns ns ns *Nov. 28/92 *** *** *** *** * ns nsMar. 3/93 *** *** *** *** *** *** nsApr. 7/93 * * *** * ** ns156-5 -10--15-ç—-20-(‘I0-25--30-IIIIIIIIIIIIIIINov13/91Feb5/92Mar17/920Figure4.2.MeanfrosttestLT50forselfedvspolycrossedprogenyduring1991/92.-5 -10--15—-20--25--30-a060 .00IIIIIIIIIIIIIIIIIIIISept25Oct16Oct30Nov28Mar3Apr7Figure4.3.MeanfrosttestLT50forselfedvspolycrossedprogenyduring1992/93.4.4. DiscussioNFamily differences occurred in most of the traits studied: seed traits, seedlingheight, root collar diameter, and frost hardiness index of injury. Although provenancestructure was not adhered to in the selfing study, these results concur (excepting seedtraits) with the results ofChapter 2 and 3, involving a completely different set offamilies, in which height, RCD, and frost hardiness index of injury on most test datesvaried at the family level. In both sets of families, dry weights were not found todiffer at the family level. In traits where all selfed vs outcrossed families weremeasured, it appears that families from the wetter subzones (Tester 2) were superior;these environments are presumed to be the better sites for this species.Greater family variance in polycrosses than expected was found whencompared to selfs. The polycross family variance component should be an estimateof 1/4 of the additive variance, while the selfed family variance component should beequal to all of the additive variance (plus 1/4 of the dominance variance if dominanceoccurs). The discrepancy may partly be due to the inequality of contributions of themales in the two testers: in Tester 1, four of the ten males contributed about 65 % ofthe pollen, while in Tester 2, four out often males contributed about 60 % of thepollen. Thus the relationship between the cross-pollinated progeny of a female wouldbe somewhat greater than that of half-sibs, possibly approaching that of full-sibs.The significant difference between treatments in dry weight parameterscompared to the nonsignificance of treatments in height and RCD might imply thatwood specific gravity, bark thickness, or amount of foliage or branches differ betweentreatments, though these hypotheses would need to be tested.159For seed and early seedling growth traits, self-pollinated western red cedaroffspring seemed to exhibit few signs of inbreeding depression when compared tocross-pollinated half-sib offspring of the same family, lending credence to the resultsof Owens et al. (1990). However, differences in frost hardiness which occurred afterthe first and second growing seasons, with polycrosses being hardier to lowertemperatures, could be an indication that self-pollinated progeny may prove to havelower survival rates over time. It would also not be unanticipated if differences inseedling growth began to appear, with the outcrosses expected to begin outperformingthe selfs; evidence for this was shown by the higher dry weight measurements ofoutcrossed seedlings compared to those of selfs.It appears that western red cedar is a species which is self-fertile, but inherentdetrimental effects due to inbreeding begin to be exhibited sometime later in the lifecycle. This would parallel the findings of Perry and Knowles (1990), who found alower than normal effect of inbreeding depression on embryo survival (—70 %) ofopen-pollinated Thuja occidentalis seeds, while the proportion of heterozygotes in theparent stands did not deviate from Hardy-Weinberg expectations, leading the authorsto speculate that much of the genetic load in this species may be expressed aftergermination and during development to a mature tree.Other studies (e.g. Libby et a!., 1981; Kuittinen et a!., 1991; also see citationsin Perry and Knowles, 1990 and Muona, 1990) have also suggested that themanifestation of inbreeding depression may be delayed until later in the life cycle ofvarious coniferous species. One study (Sorensen and Miles, 1982) observed nodetrimental effects on survival in three tree species while seedlings were growingunder very favourable conditions in a nursery, but upon outplanting, severe160environmental conditions eliminated inbred seedlings prior to maturing. Thesetheories need to be corroborated for western red cedar through more intensive studiesofmature and young stands of this species, and also by obtaining results of thecurrent study over future growing seasons.Until the severity of latent inbreeding depression, if present as suspected, canbe determined, along with the time frames when such depression may be manifested,it is suggested that the ability ofwestern red cedar to produce viable self-pollinatedseed not be interpreted to mean that inbreeding will not have any negativeconsequences in offspring. As with other species, care for now should be taken inseed orchards and breeding populations to avoid close inbreeding.1614.5. LITERATURE CITEDAdams, W.T. and D.S. Birkes. 1991. Estimating mating patterns in forest treepopulations. Pp 157-172 In Fineschi, S., M.E. Malvolti, F. Cannata, and H.H.Hattemer, eds. Biochemical markers in the population genetics of forest trees.SPB Academic Publ., the Hague, 251 pp.Blake, T.J. and C.W. Yeatman. 1989. Water relations, gas exchange, and earlygrowth rates of outcrossed and selfed Pinus banksiana families. Can. J. Bot.67: 1618-1623.Copes, D.L. 1981. Isoenzyme uniformity in western red cedar seedlings fromOregon and Washington. Can. J. For. Res. 11: 451-453.El-Kassaby, Y.A., J. Russell, and K. Ritland. 1994. Mixed mating in anexperimental population ofwestern red cedar, Thujaplicata. J. Hered. 85(3):227-23 1.Franklin, E.C. 1970. Survey ofmutant forms and inbreeding depression in species ofthe family Pinaceae. USDA For. Serv. Res. Paper SE-61, 21 pp.Glerum, C. 1985. Frost hardiness of coniferous seedlings: principles andapplications. Pp 10-123 In Duryea, M.L., ed. Evaluating seedling quality:principles, procedures, and predictive abilities ofmajor tests workshopproceedings. Oregon State Univ., Corvallis, 143 pp.Hamrick, J.L., Y.B. Linhart, and J.B. Mitton. 1979. Relationships betwen life historycharacteristics and electrically-detectable genetic variation in plants. Ann.Rev. Ecol. Syst. 10: 173-200.Kuittinen, H., 0. Muona, K. Karkkainen, and Z. Borzan. 1991. Serbian spruce, anarrow endemic, contains much genetic variation. Can. J. For. Res. 21: 363-367.Libby, W.J., B.G. McCutchan, and C.I. Millar. 1981. Inbreeding depression in seWsof redwood. Silv. Gen. 30(1): 15-25.Meidinger, D. and J. Pojar. 1991. Ecosystems ofBritish Columbia. BC Ministry ofForests Special Report #6, Crown Publ. Inc., 330 pp.Muona, 0. 1990. Population genetics in forest tree improvement. Pp 282-298 InBrown, A.H.D., M.T. Clegg, A.L. Kahier, and B.S. Weir, eds. Plantpopulation genetics, breeding, and genetic resources. Sinauer Assoc. Inc.,Mass., 449 pp.Namkoong, G. 1966. Inbreeding effects on estimation of genetic variance. For. Sci.12: 8-13.162Owens, J.N., A.M. Colangeli, and S.J. Morris. 1989. The effect of self-, cross-, andno pollination on ovule, embryo, seed, and cone development in western redcedar (Thujaplicata). Can. J. For. Res. 20: 66-75.Park, Y-S. and D.P. Fowler. 1982. Effects of inbreeding and genetic variances in anatural population of tamarack (Larix laricina [Du Roil K. Koch) in easternCanada. Silv. Gen. 31(1): 21-26.Perry, D.J. and P. Knowles. 1990. Evidence of high self-fertilization in naturalpopulations of eastern white cedar (Thuja occidentalis). Can. J. Bot. 68: 663-668.Perry, D.J., P. Knowles, and F.C. Yeh. 1990. Allozyme variation of Thujaoccidentalis L. in northwestern Ontario. Biochem. Syst. Ecol. 18(2-3): 111-115.Sorensen, F.C. and R.S. Miles. 1982. Inbreeding depression in height, heightgrowth, and survival of Douglas-fir, ponderosa pine, and noble fir to 10 yearsof age. For. Sci. 28(2): 283-292.von Rudloff, E. and M.S. Lapp. 1979. Population variation in the leaf oil terpenecomposition ofwestern red cedar, Thujaplicata. Can. J. Bot. 57: 476-479.von Rudloff, E., M.S. Lapp, and F.C. Yeh. 1988. Chemostatic study of Thujaplicata: multivariate analysis of leaf oil terpene composition. Biochem. Syst.Ecol. 16(2): 119-125.Wilcox, M.D. 1983. Inbreeding depression and genetic variances estimated fromself- and cross-plooinated families ofPinus radiata. Silv. Gen. 32(3-4): 89-96.Woods, J.H. and J.C. Heaman. 1989. Effect of different inbreeding levels on filledseed production in Douglas-fir. Can. J. For. Res. 19: 54-59.Xie, C.Y., B.P. Dancik, and F.C. Yeh. 1991. The mating system in naturalpopulations of Thuja orientalis. Can. J. For. Res. 21: 333-339.Yazdani, R. and D. Lindgren. 1991. The impact of self-pollination on production ofsound selfed seeds. Pp 143-147 In Fineschi, S., M.E. Malvolti, F. Cannata,and H.H. Hattemer, eds. Biochemical markers in the population genetics offorest trees. SPB Academic Pubi., the Hague, 251 pp.Yeh, F.C. 1988. Isozyme variation of Thujaplicata (Cupressaceae) in BritishColumbia. Biochem. Syst. Ecol. 16(4): 373-377.Zar, J.H. 1984. Biostatistical Analysis, 2nd ed. Prentice-Hall, Inc., 718 pp.1635. CoNcLusioNsIt is evident that enough genetic variation in measured western red cedarseedling traits has been identified to negate the claim that little to no variability existsin this species. Heritabilities, while not absolute, are certainly within the realm ofestimates made for other coniferous forest species. Typical narrow-sense individualheritabilities for growth traits of other coniferous species range from 0.1 to 0.4 (e.g.see Zobel and Talbert, 1984) and are most commonly between 0.2 to 0.3 (J.S.Brouard, pers. comm., 1994).This study has disproven earlier inferences that western red cedar is almost orcompletely genetically depauperate in all traits (von Rudloff and Lapp, 1979; vonRudloffet a!., 1988; Copes, 1981; Yeh, 1988; Bower and Dunsworth, 1987). Muchmore heterogeneity has been found in growth and survival characteristics than thatreported by Rehfeldt (1994), and elevational dines are much stronger for traitsmeasured in this study than those found by Rehfeldt.Seedling growth, as measured by height and root collar diameter, and seedlingacclimation and deacclimation with respect to low temperatures predominantlyexhibited within-population variation; also, about a third of sampled foliar nutrientsexhibited variability at the family level. Variation between populations was evidentin seedling dry weights, foliar nutrient levels, tree and foliage survival at an interiorsite, and in cold hardiness attributes when seedlings were experiencing the harshestconditions, the latter three traits which could be considered adaptive in nature andunder the influence of selective pressure. Differences between the coastal and interior164zones were observed in first year container plug heights, certain dry weight traits,cold acclimation and deacclimation, and in the final heights and amount ofdesiccation damage to foliage at the interior site after a severe winter.From these studies, provenance differences appear to be most stronglyinfluenced by elevation; later investigations may also find latitude to be of someimportance as well, at least for some traits. Location differences were significant.Genotype by environment interactions were found at the zonal level in root collardiameter and seedling survival, at the provenance level for seedling survival andmidwinter and spring hardiness, and at the family level for height after the 1991growing season.The provenances with the largest amounts of family variation were those ofVancouver Island. Although a small and unbalanced sample was used, when lookingat all provenances found in a broad geographical region, it appears that the mostbetween-population variation could be found in the B.C. interior, while the least wasfound in coastal northern B.C.It may be speculated that more generations after the last glaciation haveoccurred in the southern interior than the northern coast, affecting inherent levels ofvariation through long term genetic mechanisms. However, a very generalizedemerging pattern seems to be that where selective pressures are the most severe,between-population differences predominate, and within-population differencesemerge where selective pressures are less critical.As so few generations ofwestern red cedar have occurred since the lastglaciation, refugial populations probably retained residual levels of variation, because165it is unlikely that enough time has passed for populations to build up to the currentlevels of variability found within and between populations if the species was leftnearly depauperate during the last ice age. It can be hypothesized that this species isprobably in a state of constant change.Early juvenile traits showed slight evidence of inbreeding depression due toselfmg, albeit there are indications that inbreeding depression may be progressivelyexpressed later in the life cycle ofwestern red cedar. If delayed inbreedingdepression is manifested, this would indicate the presence of dominance variance insuch traits. However, this study has concentrated solely on the effects of additivegene action and no estimate of nonadditive genetic variation can be made. Shouldlatent inbreeding depression be exhibited later in the life of this species, as issuspected, then an excess of homozygotes should not be found in the adultpopulation, and the proportion of heterozygotes in mature stands should not deviatefrom Hardy-Weinberg expectations, assuming of course that all assumptions ofHardy-Weinberg equilibrium have been met.The populations investigated in this study displayed phenotypic plasticity withrespect to timing of growth, timing of acclimation and deacclimation, and maximumdegree of hardiness attained per year (down to a limit, which appears to be geneticallycontrolled). This opportunistic plasticity in the responses to timing of climatic eventsis to some degree a result of the indeterminate nature of this species, as energy is notredirected into forming an annual bud and bud flushing, and growth can continue aslong as conditions are favourable.The provenances which can be identified from this study as being generallyfavourable across all traits for which they were measured include Mt. Mara Low166elevation, Hope, and Mt. Benson. Benton Flat was only average for height, but wassuperior in other traits. No outstanding families based on height and frost hardinessresults were obvious due to the inverse relationship of the two traits, although one ofthe families from Mt. Mara Low elev. (#4) did look promising.It is strongly suggested that different breeding programs be initiated forcoastal vs interior populations. Trees destined for coastal sites of lower elevationsand adequate moisture need be less concerned with hardiness levels, and gains inheight and volume can be made. Interior populations on the other hand should bealso selected on their ability to withstand relatively harsh conditions, with average tobetter than average heights desired in these trees.Recommendations for seed transfer are preliminary at best, as only two siteswere used in this study. However, elevation and possibly latitude have beenidentified as being important in setting transfer limits. As only a sampling ofpopulations were tested, the steepness of dines or spatial patterns ofvariation cannotbe predicted with certainty from this data.As substantial variation has been found to exist in quantitative seedling traitsofwestern red cedar, the next step would be to better define responses at more thanone location per zone. It is hoped that an even more widespread sampling ofprovenances and families be included in such testing.1675.1. LITERATURE CITEDBower, R.C. and B.G. Dunsworth. 1987. Provenance test of western red cedar onVancouver Island. Pp 131-135 In Smith, N.J., ed. Western redcedar - does ithave a future? Conference Proceedings. UBC Faculty of Forestry, Vancouver,177 pp.Copes, D.L. 1981. Isoenzyme uniformity in western red cedar seedlings fromOregon and Washington. Can. J. For. Res. 11(2): 45 1-453.Rehfeldt, G.E. 1994. Genetic structure ofwestern red cedar populations in theInterior West. Can. J. For. Res. 24: 670-680.von Rudloff, E. and M.S. Lapp. 1979. Population variation in the leaf oil terpenecomposition ofwestern red cedar, Thujaplicata. Can. J. Bot. 57(5): 476-479.von Rudloff, E., M.S. Lapp, and F.C. Yeh. 1988. Chemosystematic study of Thujaplicata: multivariate analysis of leaf oil terpene composition. Biochem.System. and Ecol. 16(2): 119-125.Yeh, F.C. 1988. Isozyme variation of Thujaplicata (Cupressaceae) in BritishColumbia. Biochem. System. and Ecol. 16(4): 373-377.Zobel, B. and J. Talbert. 1984. Applied forest tree improvement. John Wiley andSons, Inc., 505 pp.168APPENDICESAPPENDIX 1. GENERAL EXPECTED MEAN SQUARE EQUATIONSAppendix 1.1. Expected mean squares equations for analyses of variance ofprovenances without family structure for the following model:Yyp = J.L + R + Zz + P(Z)p(Z) + R*P(Z)rp(z) +- Zone is a fixed effect; all other factors are considered to be random effects.Rep(Location) interactions were lumped together.Rep = a2E+na2RP(Z) + zpna2RZone = o2E+naRP(Z) +moP(Z) + rpna2ZProv(Z) = a2E+ncy2RP(Z) + rno2P(Z)R*P(Z) = E+naRP(Z)Error = a2EThe above expected mean squares equations were used to determine theappropriate error terms to be used in F tests and pseudo-F tests for each factor.Constructed F tests were as follows (H: hypothesis effects; E: error term):H=Z E=P(Z)H=R R*Z P(Z) E=R*P(Z)170Appendix 1.2. Expected mean squares equations for analyses of variance ofprovenances with family structure for the following model:Yrzpfn = + Rr + Zz + P(Z)p(z) + F(P Z)çpz) + R*F(P Z)rpz) +- Zone is a fixed effect; all other factors are considered to be random effects.- Rep(Location) interactions were lumped together.Rep = a2E+no2RF(PZ) + zpfno2RZone = a2E+ncyRF(PZ) + rnaF(PZ) + rfncy2P(Z) + rpfna2ZProv(Z) = a2E+na2RF(PZ) + rno2F(PZ) + rfnaP(Z)Fam(P Z) = a2E +naRF(PZ) + rnaF(PZ)R*F(P Z) = a2E+nO2RF(PZ)Error = a2EThe above expected mean squares equations were used to determine theappropriate error terms to be used in F tests and pseudo-F tests for each factor.Constructed F tests were as follows (H: hypothesis effects; E: error term):H=Z E=P(Z)H=P(Z) E=F(PZ)H=F(PZ) R E=R*F(PZ)171Appendix 1.3. Expected mean squares equations for analyses of variance ofprovenances without family structure for the following model:= p. + L1 + R(L)1(l) + Z + L*Zi + P(Z)p(z) + L*P(Z)Ip(z) + R*P(L Z)rp(Ez)+- Zone is a fixed effect; all other factors are considered to be random effects.- Rep(Location) interactions were lumped together.Location = a2E +no2RP(LZ) + rno2LP(Z) + zpna2R(L) + rzpna2LRep(L) = a2E+naRP(LZ) + zpnaR(L)Zone = o2E+na2RP(LZ) +ma2LP(Z) + lrna2P(Z) + rpna2LZ+ lrpno2ZL*Z = a2E+naRP(LZ) + rnaLP(Z) + rpna2LZProv(Z) = a2E+nc2RP(LZ) + rncy2LP(Z) + lniaP(Z)L*P(Z) = a2E+naRP(LZ) +mcLP(Z)R*P(L Z) = a2E+na2RP(LZ)Error = a2EThe above expected mean squares equations were used to determine theappropriate error terms to be used in F tests and pseudo-F tests for each factor.Constructed F tests were as follows (H: hypothesis effects; E: error term):H=L*Z P(Z) E=L*P(Z)H=R(L) L*P(Z) E=R*P(LZ)H = L E = R(L) + L*P(Z) - R*P(L Z)H=Z E=L*Z+P(Z)L*P(Z)172Appendix 1.4. Expected mean squares equations for analyses of variance ofprovenances with family structure for the following model:+ Li + + Z + L*Zi + P(Z)p(z) + L*p(Z)ip(z) + F(P Z)z)+ L*F(P Z)Iz) + R*F(L P Z)rIpz) + E(frpf)n- Zone is a fixed effect; all other factors are considered to be random effects.- Rep(Location) interactions were lumped together.Location = a2E +na2RF(LPZ) +ma2LF(PZ) + rfrlcT2LP(Z) + zpflia2R(L)+ rzpfnaLRep(L) a2E+na2RF(LPZ) + zpfna2R(L)Zone a2E+noRF(LPZ) +rnaLF(PZ) + lrna2F(PZ) + rtha2LP(Z)+ lrfhaP(Z) + rpfna2LZ+ lrpfna2ZL*Z a2E+ncy2RF(LPZ) + rna2LF(PZ) + rfiia2LP(Z) + rpffia2LZProv(Z) = o2E+noRF(LPZ) + rnaLF(PZ) + lrncF(PZ) + rthaLP(Z)+ frfncyP(Z)L*P(Z) = a2E+nc2RF(LPZ) + rna2LF(PZ) + rflia2LP(Z)F(P Z) = a2E+ncyRF(LPZ) + rncyLF(PZ) + lrncyF(PZ)L*F(P Z) = a2E+na2RF(LPZ) +rna2LF(PZ)R*F(L P Z) = a2E +naRF(LPZ)Error = o2EThe above expected mean squares equations were used to determine theappropriate error terms to be used in F tests and pseudo-F tests for each factor.Constructed F tests were as follows (H: hypothesis effects; E: error term):H=L*Z E=L*P(Z)H=L*P(Z) F(PZ) E=L*F(PZ)H=R(L) L*F(PZ) E=R*F(LPZ)H=L E=R(L)+L*P(Z)R*F(LPZ)H=Z E=L*Z+P(Z)L*P(Z)H=P(Z) E=L*P(Z)+F(PZ)L*F(PZ)173Appendix 1.5. Expected mean squares equations for analyses of variance ofprovenances without family structure for the following model:Yrzp = 1.1 + Rr + Zz + P(Z)p(z) +5rp(z)- Zone is a fixed effect; all other factors are considered to be random effects.- Rep(Location) interactions were lumped together.Rep = a2E+ zpa2RZone = o2E+ra2P(Z) + rpa2ZProv(Z) = a2E+ra2P(Z)Error = a2EThe above expected mean squares equations were used to determine theappropriate error terms to be used in F tests and pseudo-F tests for each factor.Constructed F tests were as follows (H: hypothesis effects; E: error term):H=Z E=P(Z)174Appendix 1.6. Expected mean squares equations for analyses of variance ofprovenances with family structure for the following model:Yrzpf + Rr + Zz + P(Z)p(z) + F(P Z)f(pz) + Epz)- Zone is a fixed effect; all other factors are considered to be random effects.- Rep(Location) interactions were lumped together.Rep = a2E+fa2RP(Z) + zpfa2RZone = a2E+raF(PZ) + rfa2p(Z) + rpfa2ZProv(Z) = a2E+ra2F(PZ) + rfa2P(Z)Fam(P Z) = a2E+raF(PZ)Error = a2EThe above expected mean squares equations were used to determine theappropriate error terms to be used in F tests and pseudo-F tests for each factor.Constructed F tests were as follows (H: hypothesis effects; E: error term):E = P(Z)H=P(Z) E=F(PZ)175Appendix 1.7. Expected mean squares equations for analyses of variance ofprovenances without family structure for the following model:Yirzpn = + L + R(L)r(I) + Zz + L*Zi + P(Z)p(Z) + L*P(Z)Jp(z) +- Zone is a fixed effect; all other factors are considered to be random effects.- Rep(Location) interactions were lumped together.Location = a2E+mo2LP(Z) + zpna2R(L) + rzpno2LRep(L) o2E+zpnaR(L)Zone = a2E+ma2LP(Z) + lmc2P(Z) + rpnc2LZ+ frpna2ZL*Z = a2E+rnaLP(Z) + rpna2LZProv(Z) = E+maLP(Z)+lrnaP )L*P(Z) =a2E+rnaLP(Z)Error = a2EThe above expected mean squares equations were used to determine theappropriate error terms to be used in F tests and pseudo-F tests for each factor.Constructed F tests were as follows (H: hypothesis effects; E: error term):H=L*Z P(Z) E=L*P(Z)H=L E=R(L)+L*P(Z)ErrorH=Z E=L*Z+P(Z)L*P(Z)176Appendix 1.8. Expected mean squares equations for analyses of variance ofprovenances with family structure for the following model:= + L1 + R(L)) + Z + LZ1+P(Z)1)+ L*P(Z)Ip(z) + F(P Z)ipz)+ L*F(P Z)Ipz) + E(Irzpf)n- Zone is a fixed effect; all other factors are considered to be random effects.- Rep(Location) interactions were lumped together.Location a2E+ma2LF(PZ) + rflia2LP(Z) + zpfna2R(L) + rzpfna2LRep(L) = a2E+ zpfnaR(L)Zone = cr2E+rna2LF(PZ) + lrno2F(PZ) + rfncy2LP(Z) + lrfna2P(Z)+ rpfnaLZ+ lrpfna2ZL*Z = a2E+ma2LF(PZ) + rfna2LP(Z) + rpfna2LZProv(Z) = a2E+rnaLF(PZ) + lmaF(PZ) + rfnaLP(Z) + lrfna2P(Z)L*P(Z) = a2E+rna2LF(PZ) + rfha2LP(Z)F(P Z) = a2E+maLF(PZ) + lrnaF(PZ)L*F(P Z) = a2E +rna2LF(PZ)Error = a2EThe above expected mean squares equations were used to determine theappropriate error terms to be used in F tests and pseudo-F tests for each factor.Constructed F tests were as follows (H: hypothesis effects; E: error term):H=L*Z E=L*P(Z)H=L*P(Z) F(PZ) E=L*F(PZ)H=L ER(L)+L*P(Z)ErrorH=Z E=L*Z+P(Z)L*P(Z)H=P(Z) E=L*P(Z)+F(PZ)L*F(PZ)177Appendix 1.9. Expected mean squares equations for frost test index of injuryanalyses of variance of all provenances at UBC for two years and atSkimikin, plus variable chlorophyll fluorescence tests, for thefollowing model:ytzpn + T + Zz + T*Ztz + P(Z)p(z) + T*p(z)tp(z) +- Zone and temperature are fixed effects; all other factors are considered to berandom effects.Temp = a2E +na2TP(Z) + zpna2TZone = a2E+tna2P(Z) + tpna2ZT*Z = a2E+naTP(Z) + pna2TZProv(Z) = a2E+tna2P(Z)T*P(Z) = E+naTP(Z)Error = a2EThe above expected mean squares equations were used to determine theappropriate error terms to be used in F tests and pseudo-F tests for each factor.Constructed F tests were as follows (H: hypothesis effects; E: error term):H=Z E=P(Z)H=T T*Z E=T*P(Z)178Appendix 1.10. Expected mean squares equations for frost test index of injuryanalyses of variance of provenances having family structure at UBCfor two years and at Skimikin for the following model:= ji+ T +4+ T*ZtZ + P(Z)p(Z) + T*P(Z)t1,+ F(P Z)pZ) + T*FP Z)Z)+ E(pf)n- Zone and temperature are fixed effects; all other factors are considered to berandom effects.Temp a2E +na2TF(PZ) + fha2TP(Z) + zpflio2TZone = a2E+tnaF(PZ) + tfnaP(Z) + tpfiia2ZT*Z = a2E+no2TF(PZ) + fna2TP(Z) + pfna2TZProv(Z) cy2E +tnaF(PZ) + tfnaP(Z)T*P(Z) a2E+na2TF(PZ) + fha2TP(Z)Fam(P Z) = a2E+tnaF(PZ)T*F(P Z) = a2E +na2TF(PZ)Error = a2EThe above expected mean squares equations were used to determine theappropriate error terms to be used in F tests and pseudo-F tests for each factor.Constructed F tests were as follows (H: hypothesis effects; E: error term):H=Z E=P(Z)H=P(Z) E=F(PZ)H=T T*Z E=T*P(Z)H=T*P(Z) E=T*F(PZ)179Appendix 1.11. Expected mean squares equations for analyses of variance of selfedvs polycrossed seedling height and root collar diameter for thefollowing model:= i+Rf+Tt+Ff+T*Ftf+e(fl- Treatment (selfed vs polycrossed) is a fixed effect; all other factors are consideredto be random effects.- Replication interactions were lumped together.Rep a2E+ tfna2RTreat o2E +ma2TF+ rfncy2TFamily a2E + rtna2FT*F cyE+mcyTFError a2EThe above expected mean squares equations were used to determine theappropriate error terms to be used in F tests and pseudo-F tests for each factor.Constructed F tests were as follows (H: hypothesis effects; E: error term):H=T E=T*F180APPENDIX2.MEANSQUARES:SEEDTRAITsGOAppendix2.1.Meansquaresofseedtraitsforallprovenancesandforprovenanceshavingfamilystructure,wherezoneisafixedeffect.1=+Zones+2Y1=+Zoner+Provenance(Z)P(Z)+8(pz)n;whereprovenancewasusedastheerrortermtotestzoneeffectsAllprovenances’ProvenanceswithfamilyStructure2TraitZoneErrorZoneProv(Z)ErrorSeedfill%2,456.13**380.652,658.72709.71347.06Seedweight48x108**4x10820x1087x1085x108Germination%482.80550.3627.841,919.12***352.08Appendix2.2.Meansquaresbyseedcollectionofseedtraitsforallprovenancesandforprovenanceshavingfamilystructure,wherecollectionisafixedeffect.Allprovenances’ProvenanceswithfamilyStructure2IritCollectionErrorCollectionProv(Coll)ErrorSeedfill%1,440.23*379.761,335.04825.52*347.06Seedweight30x108**4x10819x1086x10-85x108Germination%3,680.38***421.222,245.571,335.99**352.08=+Collection+2=+Collection+Provenance(C)()+E(pc)n;whereprovenancewasusedastheerrortermtotestcollectioneffects.Appendix2.3.Meansquaresbyregionofseedtraitsforall provenancesandforprovenanceshavingfamilystructure,whereregionisafixedeffect.Allprovenances1ProvenanceswithfamilyStructure2TraitRegionErrorRegionProv(Reg)ErrorSeedfill%1,214.50**353.841,657.57530.10347.06Seedweight19x108**4x10817x1085x1085x10-8Germination%2,458.47***386.532,451.891,030.28*352.081Ygn=+Regiong+E(g)n2Ypgn=+Regiong+Provenance(Reg)p(g)+(pg)nwhereprovenancewasusedastheerrortermtotestregioneffects.APPENDIX3.MEANSQUARES:HEIGHTANDROOTCOLLARDIAMETERAppendix3.1.Heightandrootcollardiametermeansquaresofprovenanceswithoutfamilystructure.TraitAgRpZoneProv(Z)R*P(Z)ErrorHeight’90, plugs1,014.O1***14.18295.42*52.14***7.68***4.66Height’91,UBC1,456.44***668.94***54.75366.18***57•44**38.37Height’92,UBC4,281.91***4,649.16***1,402.441,142.19***139.51112.26Height’91,Skim1,304.53***855.89***2.71175.06***48.9740.70Height’92,Skim717.39**430.87***773.98184.62***59.3784.51RCD,’92UBC77.03***104.84***45.4811.76***3.18*2.28RCD’92,Skim39.85***17.61***10.366.72**2.863.60+[3(Age)+Replicationr+Zone+Provenance(Z)(Z)+R*P(Z)(z)+6(rpz)n-ExpectedmeansquareequationsaregiveninAppendix1.1.Appendix3.2.Heightandrootcollardiametermeansquaresofprovenanceshavingfamilystructure.00TraitAgRpZoneProv(Z)Fam(PZ)R*F(PZ)ErrorHeight’90,plugs2,420.91***34.26***1,054.80*205.09***46.03***8.21***4.23Height’91,UBC2,677.61***751.63***920.181,001.18**294.58***85.06***39.85Height’92,UBC3,829.07***10,049.92***2,586.463,013.20*1,011.70***295.75***119.85Height’91,Skim3,447.28***1,175.40***15.12392.84*162.75***52.87**40.39Height’92,Skim762.44**540.40***3,014.03**243.15131.55*86.4683.62RCD,’92UBC162.99***231.50***64.4815.5310.38***4.40***2.45RCD‘92,Skim50.20***9.5842.8116.447.78*4•54***3.05‘rzn=+(Age)+Replicationr+Zoner+Provenance(Z)()+Family(PZ)f(pz)+R*F(PZ)rgpz)+E(rfpz)n-ExpectedmeansquareequationsaregiveninAppendix1.2.Appendix3.3.Meansquaresofseedlingheightattheendofthe1991growingseasonandrootcollardiameterattheendofthe1992growingseasonattwolocationsonprovenanceswithoutfamilystructure.00ONAgLocationRep(L)ZoneLZProv(Z)L*P(Z)R*P(LZ)ErrorHeight‘912,759.49***27,123.66***762.64***34.3216.97465.39***66.6053.17**39•49RCD‘92116.87***319461.48***1.2736.23**14.58****Zi+Provenance(Z)()+L*P(Z)lp(z)+R*P(LZ)(l)+8(Irzp)n-ExpectedmeansquareequationsaregiveninAppendix1.3.Appendix3.4.Meansquaresofseedlingheightattheendofthe1991growingseasonandrootcollardiameterattheendofthe1992growingseasonattwolocationsonprovenanceshavingfamilystructure.AgLocationRep(L)ZoneLZProv(Z)L*P(Z)Fam(PZ)L*F(PZ)R*F(LPZ)ErrorHeight‘916,092.47***49,115.89***965.72***564.92269.551,191.54*154.77338.53***111.17*68.94***40.11RCD‘92213.17***93.60120.86***9.82160.30**19.8811.448.926.48447***2.61Y1=+3(Age)+Location1+Replication(L)r(l)+Zone+L*Zj+Provenance(Z)P(Z)+L*P(Z)Ip(z)+Family(PZ)f(pZ)+L*F(PZ)Itpz)+R*F(LPZ)rf’(Ipz)+-ExpectedmeansquareequationsaregiveninAppendix1.4.0000Appendix3.5.Meansquaresofseedlingheightattheendofthe1992growingseasonatUBConprovenancesanalyzedindividually.ProvenanceRpFamilyREErrorQuinsam823.961,166.88*351.54***105.23Tofinoi,849.72***1,232.82**219.75**94.22MiiiBay859.64*4,437.61***270.46**128.26Cheakamus1,217.09*822.42304.44**145.19Hopei,809.86**390.65327.28**140.14Mt.MaraLow1,875.95391.35690.71***135.35Mt.MaraMid1,955.40***575.24219.85133.78Mt.MaraHigh121.7993.9667.62124.47BentonFlat1,341.66**357.43198.62*99.90rl’n=+Replicationr+Familyf+R*Frf+E(lj.zp)n;whereR*FwasusedastheerrortermtotestRepandFamily00Appendix3.6.Meansquaresofseedlingrootcollardiameterattheendofthe1992growingseasonatUBConprovenancesanalyzedindividually.ProvenanceRpFamilyREErrorQuinsam27.96**30.23**4.22*2.41Tofino37.10***2.064.25*2.31MillBay27.25**12.755.05*2.54Cheakamus39.19***21.0l**4.642.90Hope34.60***1.584.303.06Mt.MaraLow3994**3.04754***2.63Mt.MaraMid24.07***9.11*2.702.33Mt.MaraHigh1.960.341.353.58BentonFlat25.44**6.393.70*2.10=+Replicationr+Familyf+R*F+E(l);whereR*FwasusedastheerrortermtotestRepandFamilyAPPENDIX4.MEANSQUARES:DRYWEIGHTSAppendix4.1.Dryweighttraitmeansquaresofprovenanceswithoutfamilystructure.TraitZoneProv(Z)R*P(Z)ErrorStemwt109.72***86.3447•53***8.127.80Foliarwt502.83***2,674.84**134.58**34.0036.49Shootwt1,079.18***3,722.35**210.44**62.8473.22Rootwt17.92229.61**14.428.225.63Totalwt1,332.13***5,800.96**308.18**102.93113.08Shoot,rpoot939***3.642.12*0.97**0.50Shoot,To0.04***0.010.008*0.004***0.002Stem/Total0.0020.150.03***0.004***0.001#Lateralbranches129.92***105.22*15.3913.07*8.30=p.+Replicationr+Zone+Provenance(Z)(Z)+R*P(Z)rp(z)+E(rpz)n-ExpectedmeansquareequationsaregiveninAppendix1.1.Appendix4.2.Dryweighttraitmeansquaresofprovenanceshavingfamilystructure.TraitZoneProv(Z)Fam(PZ)R*F(PZ)ErrorStemwt126.60***143.3467.24***14.4612.42***7.42Foliarwt476.23***1,454.00352.96***66.6445.46*35.08Shootwt1,079.96***2,510.40585.70***129.68102.01*69.78Rootwt8.05151.0636.61**11.6012.28***7.06Totalwt1,168.61***3,893.06828.88**194.04162.83**110.88Shoot,rRoot10.02***5.364.26**0.990.67**0.42Shoot/Total0.04***0.010.02***0.0030.003***0.002Stem/Total0.008**0.020.02***0.003*0.002*0.001#Lateralbranches72.13***25.9221.1712.881l.12**6.81+Replicationr+Zone+Provenance(Z)()+Family(PZ)f(p)+R*F(PZ),)+E(fpz)n-ExpectedmeansquareequationsaregiveninAppendix1.2.Appendix4.3.Branchanglemeansquaresofprovenanceswithoutfamilystructureandthosehavingfamilystructure,wherezoneisafixedeffect.Provenanceswithoutfamilystructure’Provenanceswithfamilystructure2ZoneProv(Z)ErrorZoneProv(Z)Fam(PZ)Error45.3683.6943.1835.3144.74138.2580.511Y1=+Zoner+Provenance(Z)(Z)+8(,PZ)fl;whereprovenancewasusedastheerrortermtotestzoneeffects2Yfpfl+Zone+Provenance(Z)()+Family(PZ)f(pZ)+E(fpz)nwhereprovenancewasusedastheerrortermtotestzoneeffects,andfamilywasusedastheerrortermtotestprovenanceeffectsAPPENDIX5.MEANSQUARES:GROUPINGBYGEOGRAPHICREGIONAppendix5.1.Meansquaresof UBCseedlingtraitsof allprovenancesgroupedbyregion.TraitProv(Reg)R*P(Reg)ErrorHeight’926,319.58***4,560.332,398.91***235.48**157.07RCD’92132.35***67.14*18.87***4.16**2.87Stemdrywt237.46***201.O1**42.43***11.65*8.83Foliardrywt1,044.64***1,316.82**251.55***43.8840.24Shootdrywt2,264.03***2,240.93**418.20***90.7680.87Rootdrywt14.39161.93**29.68**10.77*7.65Totaldrywt2,490.64***3,583.18**616.32***145.38126.54Shoot/Root23.93***3.333.12***0.81**0.54Shoot/Total0.11***0.0090.01***0.003***0.002Stem/Total0.009**0.08**0.002***0.001Yrpgn=+Replicationr+Regiong+Provenance(Reg)p(g)+R*P(Reg)p(g)+5(rpg)n-ExpectedmeansquareequationsaregiveninAppendix1.1,substitutingRegionforZone.APPENDIX6.MEANSQUARES:F0MARNUTRIENTANALYSISAppendix6.1.Meansquaresoffoliarnutrientcontentforallprovenancesandforprovenanceshavingfamilystructure,wherezoneisafixedeffect.Allprovenances’Provenanceswithfamilystructure2NutrientZoneProv(Z)ErrorZoneProv(Z)Fam(PZ)Error%N0.0030.041*0.0150.0210.0680.037**0.011%P0.0010.003***0.00020.0030.002**0.00020.0003%K0.00070.130***0.0170.0080.248*0.047**0.014%Ca0.0080.058**0.0120.0040.088*0.0200.009%Mg0.0080.016***0.0320.00020.0010.00060.0004%Mn11.572,201.41**418.09289.00439.61704.75361.03ppmFe262.24653.91383.6146.69524.25914.64*318.92ppmCu9.7116.60*6.051.4824.856.615.49ppmZn208.07151.26*45.1787.1126.8931.0623.86/N0.00050.003***0.00040.00040.005*0.00070.0004Kpj0.0070.160***0.0110.0350.297***0.0110.013K/Ca0.0120.412***0.0510.0030.796*0.119*0.045IY1=+Zoner+Provenance(Z)(Z)+E();whereprovenancewasusedastheerrortermtotestzoneeffects2Yfpzn=+Zone+Provenance(Z)(Z)+Family(PZ)p)+6(fpz)nwhereprovenancewasusedastheerrortermtotestzoneeffects,andfamilywasusedastheerrortermtotestprovenanceeffectsAPPENDIX7.MEANSQUARES:SURVIVALTRAITSU’Appendix7.1.Survivaltraitmeansquaresofprovenanceswithoutfamilystructure.Y=p.+Replicationr+Zoner+Provenance(Z)P(Z)+Eyp(z)IritZoneProv(Z)Error%Planted‘90/Alive‘92,UBC9.79403.14343.40*146.58%Planted‘90/Alive‘92,Skimikin1,807.39*250.43822.32623.43%Alive‘91/Alive‘92,Skimikin2,275.57*174.10857.95703.79#Live‘92,Skimikin977**0.024.672.49-ExpectedmeansquareequationsaregiveninAppendix1.5.Appendix7.2.Survivaltraitmeansquaresofprovenanceshavingfamilystructure.TraitRpZoneProv(Z)Fam(PZ)Error%Planted‘90/Alive‘92,UBC13.90837.97**68.92136.57105.76%Planted‘90/Alive‘92,Skimikin3,196.16***23,098.785,l81.87***299.00411.75%Alive‘91/Alive‘92,Skimikin3,305.92***24,007.435,286.28***330.75409.10#Alive’92,Skimikin15.56***75.0527.60***1.742.38=+Replicationr+Zone+Provenance(Z)()+Family(PZ)f(pz)+f(pz)-ExpectedmeansquareequationsaregiveninAppendix1.6.Appendix7.3.Skimikinfoliardesiccationdamagemeansquaresofprovenanceswithoutfamilystructure.RpZoneProv(Z)R*P(Z)Error2,955.21**590.062,117.89**731.84519.99.i+Replicationr+Zoner+Provenance(Z)(Z)+R*P(Z)rp(z)+8(ipz)n-ExpectedmeansquareequationsaregiveninAppendix1.1.Appendix7.4.Skimikinfoliardesiccationdamagemeansquaresofprovenanceshavingfamilystructure.ZoneProv(Z)Fam(PZ)R*F(PZ)Error6,333.64***150,352.17*14,988.88***1,286.46**739.27621.79Y=+Replicationr+Zone+Provenance(Z)(Z)+Family(PZ)tpz)+R*F(PZ)rpz)+pz)-ExpectedmeansquareequationsaregiveninAppendix1.2.Appendix7.5.Meansquaresofpercentageofseedlingsplantedin1990stillaliveafter1992attwolocationsonprovenanceswithoutfamilystructure.LocationRep(L)ZoneLZProv(Z)L*P(Z)Error73,757.78***908.59*645.5410.12677.32482.41388.36=+Locations+Replication(L)1.(l)+Zoner+L*Zj+Provenance(Z)(Z)+L*P(Z)Ip(z)+E(Ip)n-ExpectedmeansquareequationsaregiveninAppendix1.7.Appendix7.6.Meansquaresofpercentageofseedlingsplantedin1990stillaliveafter1992attwolocationsonprovenanceshavingfamilystructure.LocationRep(L)ZoneLZProv(Z)L*P(Z)Fam(PZ)L*F(PZ)Error242,183.18***1,605.03***7,650.4516,449.34*2,553.072,811.43***272.16163.15258.28=+Location1+Replication(L)f(l)+Zone+L*Zi+Provenance(Z)(Z)+L*P(Z)lp(z)+Family(PZ))+L*F(PZ)Jf,Z)+E(Irpf)n-ExpectedmeansquareequationsaregiveninAppendix1.8.


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items