"Forestry, Faculty of"@en . "Forest and Conservation Sciences, Department of"@en . "DSpace"@en . "UBCV"@en . "Holowachuk, Diane L."@en . "2008-09-30T16:35:59Z"@en . "1993"@en . "Master of Science - MSc"@en . "University of British Columbia"@en . "Stable carbon isotopic composition (8\u00B9\u00B3C), an index of water-use efficiency (WUE), was used to test genetic and environmental responses of Pinus contorta Dougl. populations to moisture gradients. To gain insight into patterns of natural selection for high or low WUE, trends in 8\u00B9\u00B3C were measured in genotypes from different habitats. Correlations between 8\u00B9\u00B3C and productivity were tested. Sixteen-year-old P. contorta provenance trials at three climatically different sites in British Columbia were sampled. To determine whether early selection for WUE was possible in P. contort a, greenhouse grown seedling shoots from the same seedlots as the saplings were analyzed. Correlations among seedling 8\u00B9\u00B3C, growth variables and yield were tested as well. \r\nA sampling technique to accurately reflect field 8\u00B9\u00B3C was determined by assessing 8\u00B9\u00B3C variation in foliage and wood for five open-grown P. contorta saplings at one site (Juliet Creek, latitude 121\u00B0 00' N, longitude 49\u00B0 43' W, elevation 1010-1067 m). Variation in 813C within and among trees was smaller in wood than needles. Paired north and south aspects (wood) appeared to accurately track 813C year to year variation. Therefore, field 8\u00B9\u00B3C was determined for whole wood stem cores spanning ten years growth from the northland south sides of trees. The cores were taken at stump level to avoid missing years. Depending on the site, ten or eleven populations representing a wide range of habitats (from 49\u00B0 26' to 59\u00B0 59' N latitude and 114\u00B0 25' to 132\u00B0 58' W longitude) were sampled. \r\nThere were genetic differences in 8\u00B9\u00B3C among populations. Differences were related to provenance (habitat of origin), temperature and precipitation. A particular population from the wet, maritime Pacific coast (P. contorta var. contorta) stood out from the others, which were all from the continental interior (P. contorta var. latifolia). There were no genetic differences in plasticity among populations. Indicated relative WUE increased progressively from the wettest to the driest trial site. The magnitude and direction of this increase was similar in all populations. \r\nThe relationship between 8\u00B9\u00B3C and biomass increment in controlled environments as well as in nature, was not clear. High relative WUE was related to low or high yield, depending on the population, its growing conditions, and its physiological and morphological attributes. However, correlating mean population yield at each trial site with mean 8\u00B9\u00B3Cshowed a positive correlation between high productivity and high WUE. To understand differences within and among populations, it was apparent that the physiological and morphological bases for high productivity must be measured.\r\nOf the seedling provenances, only the coastal one had a different relative WUE. Favorable growth conditions may have nullified expression of genetic differences. In P.contorta, it appeared that early selection for WUE may be possible. Mean isotopic compositions of saplings and greenhouse grown first year seedling shoots from the same seedlots were positively correlated."@en . "https://circle.library.ubc.ca/rest/handle/2429/2413?expand=metadata"@en . "4953812 bytes"@en . "application/pdf"@en . "POPULATION DIFFERENCES IN WATER-USE EFFICIENCY FOR PINUS CONTORTADOUGL. AS INDICATED BY STABLE CARBON ISOTOPIC COMPOSITIONbyDIANE LYNDA HOLOWACHUKB.Sc.N. The University of Alberta, 1973A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF SCIENCEinTHE FACULTY OF GRADUATE STUDIES(FACULTY OF FORESTRY, DEPARTMENT OF FOREST SCIENCES)We accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIASeptember 1993\u00C2\u00A9 Diane Lynda Holowachuk, 1993In 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 N./t-zzt, rThe University of British ColumbiaVancouver, CanadaDate^27, /DE-6 (2/88)AbstractStable carbon isotopic composition (813C), an index of water-use efficiency (WUE),was used to test genetic and environmental responses of Pinus contorta Dougl. populationsto moisture gradients. To gain insight into patterns of natural selection for high or low WUE,trends in 813C were measured in genotypes from different habitats. Correlations between513C and productivity were tested. Sixteen-year-old P. contorta provenance trials at threeclimatically different sites in British Columbia were sampled. To determine whether earlyselection for WUE was possible in P. contorta, greenhouse grown seedling shoots from thesame seedlots as the saplings were analyzed. Correlations among seedling 813C, growthvariables and yield were tested as well.A sampling technique to accurately reflect field 813C was determined by assessing813C variation in foliage and wood for five open-grown P. contorta saplings at one site(Juliet Creek, latitude 121\u00C2\u00B0 00' N, longitude 49\u00C2\u00B0 43' W, elevation 1010-1067 m). Variationin 813C within and among trees was smaller in wood than needles. Paired north and southaspects (wood) appeared to accurately track 813C year to year variation. Therefore, field513C was determined for whole wood stem cores spanning ten years growth from the northand south sides of trees. The cores were taken at stump level to avoid missing years.Depending on the site, ten or eleven populations representing a wide range of habitats (from49\u00C2\u00B0 26' to 59\u00C2\u00B0 59' N latitude and 114\u00C2\u00B0 25' to 132\u00C2\u00B0 58' W longitude) were sampled.There were genetic differences in 813C among populations. Differences were relatedto provenance (habitat of origin), temperature and precipitation. A particular populationfrom the wet, maritime Pacific coast (P. contorta var. contorta) stood out from the others,which were all from the continental interior (P. contorta var. latifolia). There were nogenetic differences in plasticity among populations. Indicated relative WUE increasedprogressively from the wettest to the driest trial site. The magnitude and direction of thisincrease was similar in all populations.The relationship between 813C and biomass increment in controlled environments aswell as in nature, was not clear. High relative WUE was related to low or high yield,depending on the population, its growing conditions, and its physiological and morphologicalattributes. However, correlating mean population yield at each trial site with mean 513Cshowed a positive correlation between high productivity and high WUE. To understanddifferences within and among populations, it was apparent that the physiological andmorphological bases for high productivity must be measured.Of the seedling provenances, only the coastal one had a different relative WUE.Favorable growth conditions may have nullified expression of genetic differences. In P.contorta, it appeared that early selection for WUE may be possible. Mean isotopiccompositions of saplings and greenhouse grown first year seedling shoots from the sameseedlots were positively correlated.Table of ContentsAbstractTable of Contents^ ivList of Tables viList of Figures^ viiList of Abbreviations viiiAcknowledgements1.0 Introduction^ 12.0 Literature Review 42.1 Definition of Water-Use Efficiency^ 42.2 Physiology of Pinus contorta 72.3 Intraspecific Variation in WUE for Forest Tree Species^ 92.4 Intraspecific Variation and Ecotypic Differentiation 112.5 The Relationship of WUE to the Ecology of Pinus contorta ^ .112.6 Stable Carbon Isotope Discrimination and WUE^ 12Introduction^ 12History 12Units of Measurement^ 13Theory^ 142.7 Carbon Isotope Discrimination and Plant Growth Characteristics ^ 162.8 Environmental Effects on Carbon Isotope Discrimination^ 17Light^ 17Water 18Nutrients^ 192.9 Carbon Isotope Analysis in Woody Plants^ 192.10 Variation in $513C Within and Among Trees 22iv3.0 Materials and Methods^ 243.1 Determination of Sampling Methods for Field Trials^243.2 Determination of Genetic Variation in WUE Among Sapling Provenances. . ^263.3 Seedling Experiment^ 333.4 Data Analysis 35Determination of Sampling Methods for Field Trials^ 35Determination of Genetic Variation in WUE Among Sapling Provenances. ^ 35Seedling Experiment^ 364.0 Results^ 374.1 Determination of Sampling Methods for Field Trials^ 374.2 Determination of Genetic Variation in WUE Among Sapling Provenances. . ^ 444.3 Seedling Experiment^ 695.0 Discussion^ 805.1 Determination of Sampling Methods for Field Trials^ 805.2 Determination of Genetic Variation in WUE Among Sapling Provenances. ^ 845.3 Seedling Experiment^ 926.0 Conclusions^ 957.0 Recommendations for Further Research^ 96References^ 98Appendices 106List of Tables1^Test site characteristics for 70 Mile House, Holden Lake and Salmon Lake ^ 272^Provenance characteristics and available climate data^ 313^ANOVA for five Pinus contorta saplings at Juliet Creek (discs at stump level) . . . ^ 384^ANOVA for 813C comparisons of wood and needles at stump level in five Pinuscontorta saplings at Juliet Creek^ 435^ANOVA for sample comparisons in one Pinus contorta sapling at Juliet Creek. . . ^ 486^Summary statistics for 813C in Pinus contorta sapling provenances planted at70 Mile House, Holden Lake and Salmon Lake^ 507^ANOVA for provenance trials at 70 Mile House, Holden Lake and Salmon Lake. . ^ 528^Test results for differences in 813C among nine provenances over all sites ^ 559^ANOVA for provenance trials at 70 Mile House and Holden Lake^ 5710 Test results for differences in 813C among ten provenances over two sites(70 Mile House, Holden Lake)^ 6011 ANOVA for provenance trials at Holden Lake and Salmon Lake^ 6212 Test results for differences in 813C among ten provenances over two sites(Holden Lake, Salmon Lake)^ 6513 Summary statistics for growth parameters of Pinus contorta seedling provenances. ^ 7014 One-way ANOVAs on growth parameters for Pinus contorta seedling data^ 7115 Means differences among seedling shoot weights (g) ^ 7216 Means differences among seedling root weights (g) 7417 Means differences among seedling shoot/root ratios (natural logarithms)^7518 Means differences among seedling 813C values^ 76viList of Figures1^Map showing locations of provenances and test sites^ 302^Schematised test block layout^ 323^Regression of alpha cellulose 513C on whole wood 513C^ 344^Stump wood 513C values for years 1983, 1984, 1987 and 1989 in five Pinus contortasaplings^ 395^Aspect 813C values (stump wood) for years 1983, 1984, 1987 and 1989^ 406^The 513C values in stump wood as portrayed by averaging north and southaspects in Pinus contorta saplings^ 417^The 813C values in stump wood as portrayed by averaging east and westaspects in Pinus contorta saplings^ 428^Aspect x material interactions (1989) in Pinus contorta saplings^459^Needle 513C trends with aspect (1989) in Pinus contorta saplings 4610 Wood 813C trends with aspect (1989) in Pinus contorta saplings ^4711 Needle 813C trends with aspect and position in one Pinus contorta sapling^4912 Changes in provenance 813C least squares means among three sites ^ 5413 Changes in provenance 513C least squares means between two sites \u00E2\u0080\u0094 70 MileHouse and Holden Lake^ 5914 Changes in provenance 813C least squares means between two sites \u00E2\u0080\u0094 HoldenLake and Salmon Lake^ 6415 Pinus contorta greenhouse seedling 513C means plotted against saplingleast squares means averaged over three sites ^ 7816 Pinus contorta greenhouse seedling 513C means plotted against saplingleast squares means at Holden Lake^ 7917 Illustration of geographical trends for 813C in Pinus contorta^85viiList of AbbreviationsNE^photosynthetic carbon assimilation rate/transpiration rate (i.e., instantaneouswater-use efficiency)Ng^^photosynthetic carbon assimilation rate/stomatal conductance (i.e., intrinsicwater-use efficiency)C3^plant species for which the first product of photosynthesis is a 3-carbonmoleculeDe^diffusivity of CO2De^diffusivity of water vaporei^water vapor pressure inside the leafea^ambient water vapor pressure outside the leafg^stomatal conductance to CO2cstomatal conductance to water vaporgwkPa^kilopascalLSD^least significant differenceLSM^least squares meanmap^mean annual precipitationmat^mean annual temperaturempdm^mean precipitation of the driest monthmsp^mean summer precipitationmtwm^mean temperature of the warmest monthMPa^megapascalPAR^photosynthetically active radiationPDB^fossil Pee Dee belemnitePi/Pa^intercellular partial pressure of CO2/ambient partial pressure of CO2PPFD^photosynthetic photon flux densityR^molar abundance ratio '3C/'2Cra^boundary layer resistance to diffusionviiirm^mesophyll resistance to diffusionrs^stomatal resistance to diffusionRuBisCO^ribulose-1,5-bisphosphate carboxylase/oxygenaseWUE^water-use efficiencyA^isotope discriminationAc^leaf-to-air concentration difference for CO2Ae^leaf-to-air concentration difference for water vapor513c^stable carbon isotope abundance variable43c^portion of carbon fixed that is lost to respirationOw^non-productive water loss due to incomplete stomatal closure, cuticularopenings and evaporation from the soilv^water vapor pressure difference between the inside and outside of the leafoho^per mil (parts per thousand)ixAcknowledgementsThis thesis is dedicated to my parents, Pauline and Emmett Holowachuk, who havesupported me generously in my educational endeavours.The funding for this study was provided by an NSERC Operating Grant to Dr. RobertGuy and by Challenge 90 and Challenge 91 Grants.The cooperation and assistance of Dr. C. Ying and Ms. Leslie McKnight, researchstaff for the British Columbia Ministry of Forests, were crucial to the establishment of myproject. Frank Kohlberger's conscientious efforts in the field and lab promoted accurateresults. Dr. N. Livingston (University of Victoria) generously provided technical help. Dr.A. Kozak gave statistical guidance. My supervisor, Dr. R. Guy and committee members,Drs. C. Ying, P. Burton and D. Lester offered constructive advice.11.0 IntroductionThe term, water-use efficiency (WUE), relates plant production to waterconsumption. An indicator of plant performance, WUE is used to study genetic differencesin mechanisms that maximize performance in arid environments.The concept of water use economy is not new. Drought limits vegetation growth,distribution and yield more than any other environmental factor (Kramer 1983). Plantbreeders, concerned with improved harvests under arid conditions, focus on breeding fordrought resistance rather than increased WUE, i.e., gaining higher yields from given amountsof water (Tesar 1984). In agriculture, WUE is desired because it is perhaps the foremostyield component (Martin and Thorstenson 1988). Although short-term yield usually declineswith increased WUE, total yield per given land tract will increase if water supply limitsgrowth, i.e., growth is maintained for a longer time (Fisher and Turner 1978). In cases wherewater is non-limiting, however, high WUE may restrict yield (Meinzer et al. 1990, Condon etal. 1990).Compared with agricultural crops, we know little about genetic variation in WUE orits physiological basis in woody plants. We do not know the circumstances under whichefficient water use confers competitive advantage or promotes acceptable growth rates intrees. We have yet to examine the effect of drought and the magnitude of genetic, andgenotype x environment interactions as sources of variation in WUE.Water-use efficiency varies intra-specifically in Populus L. (Blake et al. 1984), Pinusponderosa Laws. (Monson and Grant 1989) and Pseudotsuga menziesii (Mirb.) Franco (Srnitand van den Driessche 1992). Populus studies (Dickmann et al. 1992) show that therelationship between dry-weight increment and high WUE is not clear. Monson and Grant(1989) and Eickmeier et al. (1975) have used WUE in Pinus ponderosa and Tsugacanadensis (L.) Carr, respectively, as measures of differential adaptation to moisture-limitedhabitats. DeLucia et al. (1988, 1989) show that high WUE in Pinus ponderosa and Pinus2jeffreyi Grey. & Balf. may be a competitive disadvantage in water-limited environments. Indesert soils, Pinus seedlings cannot effectively compete with the shrub species. The shrubs'relatively high water-use rates deplete the soil water, thus restricting the pines to areas withgreater water availability.In agriculture, stable carbon isotope analysis effectively screens crops for variation inWUE (Farquhar et al. 1989b). Recently the application of this technique to shrubs and treeshas burgeoned (Read and Farquhar 1991; Fu et al. 1992a, b; Jackson et al. 1992; Leavitt andDanzer 1992; Livingston 1992; Newberry 1992; Zhang et al. 1993). The isotopiccomposition of a plant is typically expressed as the abundance ratio 13C/12C relative to astandard and is called 513C. The 813C of a plant changes due to discrimination against 13CO2over 12CO2 during photosynthetic carboxylation and gaseous diffusion. Isotopic compositionor 513C of plant tissue is positively related to WUE. The reasons for this are well understood.Stomatal closure causes increased instantaneous WUE since the photosynthetic rate isreduced proportionally less than the transpiration rate. At the same time, stomatal closuredecreases the partial pressure of CO2 in the intercellular air spaces (pi). Discriminationagainst 13CO2 decreases with low pi, thereby increasing 513C. Thus both seasonal WUE and513C increase as a result of stomatal closure in response to drought. The relationshipbetween 513C and WUE would be similar also if CO2 fixation rate and pi were affected byphotosynthetic disturbances of the mesophyll rather than by stomatal behavior.This study uses 813C (an integrated index of WUE) to determine ecotypic variation inWUE for Pinus contorta Dougl. The species has broad edaphic and climatic tolerances, andgrows in a wide range of moisture-limited habitats. Based on Rehfeldt's research (1986,1987) supporting \"specialization\" as the species' mode for survival, I expect P. contortapopulations to be specialized in WUE along moisture gradients. Specialization will beevident in \"ecotypes\" arising in response to local conditions. To accept that such ecotypesexist, differences in WUE must be demonstrated as heritable by appearing when plants fromacross the species' range are grown in \"common gardens\" (Clausen et al. 1948).3Research objectives:1. To develop sampling methods for field trials in this project, as well as for futureprojects;2. To use 513C to test for genetic variation in WUE among seedling and sapling P.contorta provenances;3. To correlate 813C of seedlings with shoot, root and total weight, as well asshoot/root ratio; and to correlate 813C of saplings with diameter, height and volume;4. To correlate 513C of seedlings with that of saplings grown from the same seedsource;5. To correlate 513C of saplings with provenance climate variables: meantemperature of the warmest month, mean annual temperature, mean summer precipitation,mean precipitation of the driest month, mean annual precipitation; and6. To assess the plasticity of populations by examining their ranked 513C valuesamong field sites distinct in available moisture.42.0 Literature Review2.1 Definition of Water-Use EfficiencyIn agriculture or ecosystem science, WUE is defined as the ratio of yield or netprimary production to evapotranspiration (Fisher and Turner 1978; Gardner et al. 1985).Yield is expressed as economic yield (e.g., grain or seed, forage in grasses) or dry weight ofthe whole plant, including roots. The plant parts used depend on the experimenter'sobjectives. If roots or other vegetative organs are not included, the WUE will be lower thanif the entire plant is the basis for measurement. Use of evapotranspiration gives increasedvariation in WUE, since evaporation is influenced by leaf cover and frequency of soilwetting, independently of transpiration (Tanner 1981). The definition which follows is themost useful in agronomy (Kramer 1983):Dry matter or Crop Yield (g)WUE = ^(1)Water used in evapotranspiration (kg)Physiologists often discuss WUE in terms of gas exchange, expressing it as net CO 2uptake per unit of transpired water (Fisher and Turner 1978). Simply defined, at the level ofa single leaf:Net carbon dioxide uptake (mmol)WUE = ^(2)Water used in transpiration (mol)More recently, Hubick et al. (1988) have used the term \"transpiration efficiency\" aswell as water-use efficiency to discuss dry matter production and water use. Theydistinguish the terms according to type of water loss. In the single plant, transpirationefficiency is the ratio of moles of carbon in the plant to moles of water used to accumulatethat carbon. Water-use efficiency, assigned to determinations in the field, is the ratio of drymatter to total water use including losses such as soil evaporation and runoff.5Water-use efficiency is determined by biophysical processes such as CO2 and watervapor gradients between leaf and air, stomatal opening, boundary layer resistance andmesophyll resistance. These effects are accounted for in different ways by Fisher and Turner(1978), Nobel (1980), Osmond et al. (1980) and Farquhar et a/. (1989b). Water lossaccompanies CO2 uptake since water vapor diffuses out of a leaf by the same route that CO2diffuses in. For a given stomatal resistance (rs), more water will be lost than CO2 acquired.The concentration gradient of water vapor is usually greater than that of CO2; water vaporhas a higher diffusion coefficient than CO2; and CO2 diffusion is hindered by an addedresistance (rm or mesophyll resistance) not encountered by water vapor (Nobel 1983). Thetwo gas-exchange processes are:Assimilation =(Conc.0O2ext. - Conc.0O21m.) x diffusivity(3) ra + rs + rm(Conc.H2Oint. - Conc.H20ext) x diffusivityTranspiration \u00E2\u0080\u0094 ^(4)ra + rsThe equations can be combined to give (Fisher and Turner 1978):WUE =Ac x De (ra + rs)(5)Ae x De (ra + rs + rm)where Ac and Ae are the leaf-to-air concentration differences for CO2 and water vapor,respectively; De and De are the diffusivities of CO2 and water vapor, respectively; and ra, rsand rm are the boundary layer, stomatal and internal resistances to diffusion, respectively.Gas exchange through the cuticle is ignored.Researchers interpret mesophyll resistance differently. For many (Nobel 1983),mesophyll resistance represents the diffusion pathway from the substomatal cavity to thechloroplast, and from inside the chloroplast to the site of fixation at ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisC0). For others (Jones 1976; Farquhar et al.61982b), mesophyll resistance comprises several components, of which the diffusion pathwayis only minor since the physical distance between the cell surface and a chloroplast is verysmall. Carboxylation resistance is more important. Kramer (1983) states that in practice, rmincludes biochemical limitations on CO2 fixation as well as physical limitations on diffusionsince rm for CO2 is calculated as the difference between (ra + rs) and the total resistance toCO2 uptake.Farquhar et al. (1989b) use the terms \"conductance\" and \"partial pressure\" to defineWUE. [Conductance, or the reciprocal of resistance, refers to ease of flow (m s-1) of gaseousspecies through a medium or material]. They begin with an approximate expression for theratio of instantaneous carbon assimilation rate (A) and transpiration rate (E):A/E = gc(Pa -^/ gw(ei - ea) (6)= (pa - pi) / 1.6v (7)= Pa(' - Pi/Pa)/ 1.6v (8)where ge and gw are conductances to CO2 and water vapor, respectively; pa and pi are partialpressures of CO2 in the atmosphere and inside the leaf, respectively; and ei and ea are thewater vapor pressures inside and outside the leaf, v being the difference between them. Theratio of the diffusivities of water vapor and CO2 in air equals 1.6.Using the ratio, moles of carbon in the plant to moles of water transpired during thegrowth period, water-use efficiency is defined as:WUE = (1 - \u00E2\u0080\u00A2:1:1c) pa(1 - pi/pa) / 1.6v (9)where 4:Ic is the portion of carbon fixed during the day that is lost to respiration by the leaf atnight and by other plant parts over the whole period.Equations (5) and (9) summarize the important environmental and genotypicinfluences on WUE. Through its effect on v, vapor pressure deficit (air) alters WUE. Insingle leaves VVUE decreases with increasing v, although the opposite may occur in wholeplants (Rawson et al. 1977). Vapor pressure deficit changes daily, seasonally and regionally,7depending on temperature and absolute humidity. Air temperature acts on WUE via itseffect on v. Since v is usually closely coupled to air temperature in the field (Schulze et al.1973), increased air temperature reduces WUE unless leaf temperature is suboptimal forphotosynthesis (Miller and Mooney 1974).Incident irradiance is important, due to the non-linear relationship betweenphotosynthesis and transpiration. Transpiration is always positive, showing an upward linearor curvilinear response with increasing light due to rising leaf temperature. Stomatalconductance may or may not increase. Net photosynthesis plateaus with increased light(reflecting saturation kinetics). Accordingly, there is an optimum irradiance for maximumWUE (Bierhuizen and Slatyer 1965; Downes 1970; Jones 1976).The leaf traits \u00E2\u0080\u0094 boundary layer, stomatal conductance and biochemical capacity forfixation of carbon dioxide \u00E2\u0080\u0094 influence WUE (see Fisher's and Turner's 1978 review).Comparisons between and within genotypes must take place under the same ambientconditions, and genotype x environment interactions must be considered.2.2 Physiology of Pinus contortaPinus contorta is one of the most researched conifers in the Pacific northwest, yetrelatively little is known of its basic physiology (Bassman 1985). Few studies concern itstolerances to crucial, limiting environmental factors such as air and soil temperature, soilmoisture, relative humidity and nutrients.On the other hand, the species has been a frequent subject for ecophysiologicalstudies trying to explain distributional patterns of overstory conifers along elevational andgeographic gradients in montane forests. Its physiological characteristics have beencompared with those of other conifers. Having reviewed earlier work, Lopushinsky (1975)and Smith (1985) characterize the water relations for the species: P. contorta showsmoderate to relatively high transpiration rates, largely decreasing with soil drought; stomatalclosure occurs at a relatively high leaf water potential; the sapwood layer is relatively thick,8conferring large stem water storage capacity; drought resistance is moderate \u00E2\u0080\u0094 less than thatfor P. ponderosa; the species shows high tolerance to saturated soils.Photosynthetic data are hard to interpret, since they are measured under variouscircumstances and expressed in different units. Brix (1979) compared photosynthesis inPseudotsuga menziesii, Picea glauca (Moench) Voss, Tsuga heterophylla (Rafn.) Sarg. andPinus contorta. He measured gas exchange on 4-month-old seedlings (greenhouse grown),following 2 months acclimation in a growth room. Pinus contorta had relatively highphotosynthesis per unit leaf area but the lowest rate per unit leaf dry weight. As well,photosynthesis in P. contorta declined most steeply with drying soil and at a higher thresholdsoil water potential than the other species. Dykstra (1974) observed that 2-year-old P.contorta seedlings (acclimated for 2 months in a greenhouse) attained maximum netphotosynthetic values of about 6.3 larnol CO2 m-2 s-1. Light saturation occurred at close to400 W m-2 of PAR. The temperature optimum for photosynthesis was near 20\u00C2\u00B0C, with afour-fold reduction occurring at about 2\u00C2\u00B0C. Photosynthesis decreased almost linearly frommaximum values to zero when xylem pressure potential decreased from -0.3 MPa to -1.5MPa. Mesophyll resistance to CO2 uptake was greater than rs under all conditions, especiallydrought. Dykstra included carboxylation resistance in rm; by implication, WUE would havedecreased with drought. Photosynthetic rate increased proportionally with decreases in rmand rs.Sweet and Wareing (1968) reported intraspecific variation in net photosynthesis for12 and 19-week-old P. contorta. Photosynthetic rates for the 12-week-old seedlings were inthe order: California-Oregon coast (var. contorta)> Rocky Mountains, Alberta (var. latifolia)> Cascade Mountains, Oregon (var. murrayanallatifolia)> Washington coast (var. contorta).Ranking changed in the 19-week-old seedlings with different leaf morphology. The authorscautioned that large differences in photosynthetic rates might not reflect field performance,since seedling photosynthetic rate, provenance and growth conditions often interacted.Variation among provenance photosynthetic rates could not be related to climate or location9of the seed source (a finding common to studies using non-stressed growing conditions).Correlations with climate/geography have been shown under extreme conditions (Bourdeau1963).Changes in leaf-to-air water vapor pressure difference (v) account for much of thestomatal regulation in P. contorta (Kaufmann 1982); as well, stomatal response is controlledby light intensity. Sandford and Jarvis (1986) observed large variation in stomatal responseto v in current year needles of 3-year-old seedlings of Picea sitchensis (Bong.) Carr., Larix xeurolepis Mill., Pinus sylvestris L. and Pinus contorta. All species' stomates partly closed asv was increased from 0.4 to 2.0 kPa; P. contorta ranked second in degree of stomata! closure.Only the stomates of P. contorta and L. x eurolepis behaved optimally (Cowan 1986),maximizing the exchange rate of CO2 for water during transpiration as v increased, i.e.,change in assimilation/change in transpiration remained constant as v increased. Murdiyaroet al. (1985) found no significant differences in transpiration and stomatal conductanceamong seven provenances of non-stressed 2-year-old P. contorta. After drought stressingnewly grown seedlings from three of the provenances, they found significant provenance xtreatment interactions in stomatal response.On exposed early-successional sites, Pinus contorta saplings had photosyntheticgains similar to Abies lasiocarpa (Hook.) Nutt. and Picea engelmannii Parry, but with moreefficient water use (Carter and Smith 1988). The pines' transpiration decreased greatly withprogression from shaded to open habitats. The authors noted that all three species hadremarkable stomatal and photosynthetic acclimation responses to habitat exposure.2.3 Intraspecific Variation in WUE for Forest Tree SpeciesAlthough considerable work has been done on genetically based, intraspecificvariation in photosynthetic traits and water loss rate, little has been done on variation inWUE. Blake et al. (1984) found genetic variation in WUE for Populus genotypes fromvarious countries. Some clones showed high dry matter production combined with high10WUE. Differences in WUE were related to stomatal resistance (accounting for 40% of thevariation), foliar adaptations such as cuticular ledges or hairs above pore openings, time ofstomatal opening in the morning, and stomatal size and frequency. There was no evidence ofa particular physiological or morphological variable which would consistently explainimproved WUE in every case.Large population differences may occur where dissimilar environments exert intenseselection pressure (Turesson 1922a, 1923). Eickmeier et al. (1975) suggested that Tsugacanadensis has at least two physiological races correlated with environment. Carbon dioxideand water vapor exchange, stomatal resistance, WUE and morphology varied in ecologicallymeaningful ways in first year seedlings from two Wisconsin seed sources. Water-useefficiency was measured as carbon dioxide to water vapor flux ratio.Monson and Grant (1989) found definitive heritable differences in stomatalconductance, maximum photosynthetic rate and intrinsic WUE (measured as pupa) betweenPinus ponderosa progeny from two genetic lines. The higher WUEs in families derivedfrom a coastal x interior cross were manifested in lower pipa values and lower stomatalconductances for any given photosynthetic rate. Results suggested that P. ponderosa hadachieved improved WUE and lower transpiration rates in drier habitats, at the expense ofreduced maximum photosynthetic rates.Smit and van den Driessche (1992) compared WUE in Pseudotsuga menziesii andPinus contorta seedlings. Four provenances were tested \u00E2\u0080\u0094 one from a wet and one from adry site for each species. One-year-old seedlings were planted in wet and dry soil treatmentsand grown outside for 20 weeks. Pseudotsuga menziesii had higher WUE than Pinuscontorta. Pseudotsuga menziesii from the dry site was more water-use efficient than P.menziesii from the wet, but there was little difference between Pinus contorta provenances.Reasons given for P. contorta's lack of response were that origin sites were not sufficientlydifferent or the dry soil treatment was not dry enough. Root exploitation of soil water, notWUE, determined productivity differences between the species. Pinus contorta, with lower11WUE, produced more dry matter than Pseudotsuga menziesii. Pinus contorta seedlings usedwater in their root environment more completely than the Pseudotsuga menziesii.2.4 Intraspecific Variation and Ecotypic DifferentiationThe concept of ecotypic differentiation developed after Turesson's classic studies(1922a, b) on Atriplex L. species. In some cases, diverse forms among groups of individualsin different environments were directly induced by environment, i.e., there was phenotypicplasticity. But in most cases, they were the result of genetic variation among populations.The inherited characteristics were related to distinct native habitats. Believing that \"habitattypes\" arose from genotypic responses to particular environments, Turesson called them\"ecotypes\". The underlying evolutionary process became known as \"ecotypicdifferentiation\".Subsequently, investigators have established that ecotypic differentiation must bemainly responsible for the ability of some species to occupy a wide range of habitats.Ecotypic differences exist with changes in altitude, latitude and precipitation and as a resultof soil type, nutrient availability and metal toxicity. As major habitat factors differ spatiallyin a random or non-random manner, so can ecotypic variation.2.5 The Relationship of WUE to the Ecology of Pinus contortaPinus contorta's features make it an ideal species for studying \"adaptedness\" \u00E2\u0080\u0094 theextent to which individuals are physiologically attuned to their environment (Rehfeldt 1987).It is the most widely distributed conifer in western North America, growing in a variety ofsites and in all but the most adverse conditions. Its range, centered in British Columbia,spans 33 degrees latitude, 35 degrees longitude and more than 3900 m elevation (Wheelerand Critchfield 1985).Three distinct, interfertile, geographic races have evolved, each characterized bymorphological, physiological and biochemical traits different from the others. These12geographic races and affiliated regions are:1. P. contorta var. contorta: Pacific coast, from Alaska to northern California;2. P. contorta var. murrayana: southern Cascades, Sierra Nevada range and themountains of southern Baja California;3. P. contorta var. latifolia: Rocky Mountain and Intermountain regions, northernCascades, Washington; eastern outliers from northern Alberta to SouthDakota.The basis for the species' large distribution is the existence of specialized populationsfor small parts of the species' range. In seedling studies, Rehfeldt (1983, 1986, 1987, 1989)demonstrated that through natural selection, traits have evolved to overcome ecologicalproblems such as changeable frost-free periods, insect infestations and disease epidemics.Research has established within- and among-population variance in drought resistance;however genetic variation in WUE has hardly been tested.2.6 Stable Carbon Isotope Discrimination and WUEIntroduction Plant adaptation and acclimation responses to water shortage have receivedincreasing attention in recent years. During the last decade, agricultural research on WUEhas increased greatly with the development of stable carbon isotope analysis as an effectivescreening tool for variation in WUE (Hubick et al. 1988). The goal of this research has beento select for improved growth in water-limited environments.History The fundamental terms, processes, effects and theory of carbon isotope fractionationor discrimination by plants are discussed by Hayes (1982), Farquhar et al. (1982b, 1984,1989a, b), O'Leary (1981, 1988) and O'Leary et al. (1992). Stable carbon isotopecomposition was initially used to indicate photosynthetic pathways in plants (Bender 1968).13Isotopic surveys have provided information on the distribution of photosynthetic pathwaysamong various phylogenetic groups and ecological zones (O'Leary 1981, 1988). Earlystudies have shown substantial variation in isotopic composition at interspecific andintraspecific levels, as well as variation associated with different environmental growthconditions and dry-matter composition (reviewed by Farquhar et al. 1989a).Units of MeasurementThere are four isotopes of carbon \u00E2\u0080\u0094 14C, 13C, 12C and 11C. The stable isotopes are 13Cand 12C; 14C and 11C are unstable, i.e., \"radioactive\". Terrestrial carbon is 98.9% 12C and1.1% 13C. Green plants contain relatively less 13C than does air, showing their preferentialuse of 12CO2 over 13CO2. Variation in the 13C/12C ratio (R) of plants is due to \"isotopeeffects\" or fractionations, expressed during the formation and destruction of carbon bonds oras a result of processes affected by mass, such as gaseous diffusion. Isotope effects are notobservable but isotopic compositions are altered by them, and from this we infer they exist.Isotopic compositions (513C), expressed as \"del\" values (5), are conveyed in per mil (oho),i.e., parts per thousand. Carbon isotopic composition is calculated relative to the historicstandard, fossil Pee Dee belemnite (PDB):[R(Sample) - R(Standard)]813C -^ x 103^(10)[R(Standard)]Isotope effects, expressed as isotope discrimination (A), are calculated from thedifference in the 513C value between source and product:513Cair - 513CoantA = ^ (11)1 + 513Coant/1000The advantage of reporting A rather than 8 is that it directly expresses theconsequences of biological processes. It is important where isotopic composition of the14source air changes due to antlIropogenic factors or when interpreting results from canopieswhere an isotopic gradient exists with height (Vogel 1978).Since A values are usually positive while those for 5 are negative (when PDB is thereference), confusion can result. Therefore some authors use molar abundance ratios (R) andcompositional deviations (5) as intermediates in determining final isotope effects.Carbon isotopic composition (5), rather than A is reported throughout the thesis. The513C values are the primary data which can be converted to A, knowing the 613C of sourceair. I did not measure 813Cair for the field sites tested or for the greenhouse where theseedlings were grown. Measuring 513Cak introduces its own sampling error into thecalculation of A. Substituting 513Cair reported in the literature (-8.0 Woo) for 5 13Cair in mystudy would have given inaccurate measures of A for seedlings grown in the greenhouse.TheoryPlants discriminate against 13C during photosynthesis. In 1982(b), Farquhar et al.predicted this discrimination would be least in plants (C3) which fixed the most carbon perunit of water transpired, i.e., in those that maintained the greatest WUE. This theory wasconfirmed, in part, from 513C values and short term gas exchange measurements of leavesfrom contrasting species (Farquhar et al. 1982a). In 1984, Farquhar and Richards confirmedthe theory fully by isotope analysis of Triticum aestivum L. genotypes, for which dry matteraccumulation and water use had been measured over the long term. Plants with high WUEhad large ratios of 13C to 12C, meaning that isotope effects were smaller or that the plantsdiscriminated less against the heavier isotope, 13C. These results suggested that carbonisotope analysis could be a tool for seeking differences in WUE and for genotype selection inbreeding programs for increased yield under drought.Current theory states that variation in WUE occurs because WUE is partly dependenton the ratio of CO2 partial pressure inside the leaf to CO2 partial pressure in the atmosphere(pi/pa). This ratio varies and the variation can be seen in the isotopic composition of a plant.15Since the pi/pa ratio varies, we expect a large degree of genetic variation in the physiologicalcomponent of whole plant VVUE. The nature of inheritance in pi/pa is largely unknown,except that it is not under simple genetic control (Farquhar et al. 1989a). Isotopicdiscrimination is likely a polygenic trait as well (Farquhar et al. 1989a). The genetic controlof A appears strong in Arachis hypogaea (Hubick et al. 1988) and in Triticum aestivum(Condon et al. 1987).To increase WUE, a leaf must decrease pi/pa through increased photosyntheticcapacity or decreased stomatal conductance. By rewriting equation (7) to give equation (8),Farquhar et al. (1989a) emphasize that a smaller value of pi/pa leads to an increase in A/E(transpiration efficiency). A simple model for discrimination in leaves of C3 plants is:A = a + (b - a)pi/pa - d, (12)where \"a\" is the fractionation of carbon isotopes due to diffusion in air (4.4 Woo) and \"b\" thenet fractionation due to carboxylation (27.0 oho). The \"d\" represents minor isotopic effectscaused by dissolution of CO2, liquid phase diffusion, respiration and other processes not yetquantified (Masle and Farquhar 1988). Since \"a\", \"b\" and \"d\" are largely constant, A may bea surrogate measure for pi/pa\u00E2\u0080\u00A2 In experiments relating gas exchange properties to short andlong term A (with v kept constant), pi/pa is positively correlated with A (Evans et al. 1986).During whole plant growth, carbon and water losses must be accounted for, requiringanother equation. A portion (0c) of the carbon fixed via the stomates during the day is lost inrespiration. Some water is lost from the plant independently of CO2 uptake, due toincomplete stomatal closure and cuticular openings. Unavoidable evaporative soil water lossoccurs as well. If this \"nonproductive\" water loss is a portion, Ow, of \"productive\" water loss,the A/E equation can be changed to describe the molar ratio of carbon gain by a plant, towater lost:Pa (1 - pupa) (1 - 0c)WUE = (13)1.6v(1 + Ow)16This equation, combined with equation (12), predicts a negative linear dependence of WUEon A:(b - d - A)Pa^ [(1 -(b - a)WUE = ^ (14)1.6v (1 + Ow)Results from pot experiments, using various watering treatments and genotypes, fit thetheory well for a number of species \u00E2\u0080\u0094 Triticum aestivum (Farquhar and Richards 1984;Masle and Farquhar 1988), Arachis hypogaea L. (Hubick et al. 1986, 1988; Wright et al.1988), Lycopersicon esculentum Mill. (Martin and Thorstenson 1988), Coffea arabica L.(Meinzer eta!. 1990) and Vigna unguiculata (L.) Walp. (Hall eta!. 1990).2.7 Carbon Isotope Discrimination and Plant Growth CharacteristicsAlthough Hubick et al. (1988) observed a negative relationship between dry matterproduction and A of field grown Arachis hypogaea cultivars, Condon et al. (1987) found apositive relationship in Triticum aestivum cultivars receiving greater than usual rainfall.Clearly, associations between A and carbon partitioning are important. If strains with largerpi/pa partition more carbon to the shoot, then selecting for large A may result in increasedyield when water is not limiting. Masle and Farquhar (1988) examined relations betweenindividual growth components and A in T. aestivum. The grain was grown in a soil resistantto root penetration. Resistance was increased by augmenting soil density or by lowering soilwater content. Stomatal conductance and shoot growth responses were similar regardless ofwhich method was used. Masle and Farquhar (1988) found a negative relationship between\"partitioning ratio\" (total plant carbon/leaf area) as well as root/shoot ratio and A. Althoughthe relationship was not stable across environments, the authors suggested that it might havecontributed to the positive relationship between plant mass and A reported by Condon et al.(1987). Relative growth rate and A correlated positively with decreasing soil resistance (as Aincreased, there was a relatively greater decrease in partitioning ratio than in CO217assimilation rate). Assimilate partitioning, as well as leaf gas-exchange properties must beconsidered when interpreting the relationship between plant growth and A.2.8 Environmental Effects on Carbon Isotope DiscriminationLightField studies have reported an increase in 8 13C1,1ant with increased irradiance(Farquhar et al. 1989a). It is debatable whether, in the field, carbon isotopic composition ofleaves in different light levels can be interpreted. It is hard to separate effects of light on Afrom correlated effects of source air (813Cthr), both of which affect leaf carbon isotopiccomposition (Farquhar et al. 1989a).Vogel (1978) found a pattern of isotopic variation in canopy leaves, where lightvaried substantially. Foliar 813C decreased by 3.0 Woo between the top (19 m) and bottom (1m) of the canopy. The 813C of soil CO2 was approximately -19.0 oho while 813Cair above thecanopy was only -7.0 oho. [The CO2 released by decomposing plant litter in soil is moredepleted in 13C (ca. -27.0 Woo) than that of atmospheric CO2 (ca. -7.0 Woo), since plantsdiscriminate against 13CO2 in air]. Without measuring isotopic composition of CO2 withinthe canopy and assuming constant physiological discrimination throughout, Vogel attributedthe decrease in 813C of leaves at lower layers to the use of recycled soil CO2.Medina and Minchin (1980) reported 813Coant gradients of 4.7 and 5.6 Woo betweenupper and lower canopy leaves for two tropical forest types. Results were also attributed tothe photosynthetic assimilation of recycled soil CO2 in the lower leaves.In a moist tropical forest (Panama), Jackson et al. (1992) compared shade tolerantand light demanding species in understory and gap environments. They concluded thatdifferent 813C values between plants of the same species growing in gaps and understorywere physiologically based and not solely due to differences in 813C of source air (CO2).Francey et al. (1985) examined 813C as well as 513C1an1 in temperate coniferousforest canopies. They noted a decrease in 5 13Coant with canopy depth but no corresponding18decrease in 813Cair, thus indicating a physiological effect. Deeper in the canopy 813Coaat andpi were negatively correlated. While soil CO2 contributes to the decrease in leaf isotopecomposition, most of the change can be attributed to stomatal and photosynthetic effects(Farquhar et al. 1989a).Fu et al. (1992a) reported on leaf 813C variation in open-grown Quercus virginianaMill. canopies where fixation of respiratory CO2 was likely minimal. Foliar A increasedsignificantly from the upper to lower canopy, attributed to reduced photosynthetic photonflux density (PPFD) in the lower canopy. Cardinal directions showed significant differencesin A values. North facing leaves experiencing lowest PPFD showed highest A. South facingleaves with highest PPFD showed lowest A. Foliar A values were significantly higher duringthe wet sampling period than the dry, suggesting greater WUE during the dry period. Therewere no significant interactions between canopy height, aspect or sample period (wet, dry).Differences in foliar A were attributed to PPFD changes and perhaps to water vapor pressuredifferences between leaves and air. It was suggested that leaves subject to low PPFD hadhigher pi/pa ratios, which might increase quantum yield and give higher photosyntheticcarbon gain under light-limiting conditions. The effect of vapor pressure deficit on A wouldhave been largely due to decreased stomatal conductance, resulting in lower pi/pa ratios.WaterA common physiological response to drought is decreased photosynthesis,transpiration and leaf conductance concurrently (Farquhar and Sharkey 1982). Partialpressure of CO2 inside the leaf will decline if leaf conductance decreases more thanphotosynthetic demand for CO2. This effect will be manifested as an increase in 813Coant ora decrease in A. Carbon isotope analysis reveals that, over the long term, plants under waterstress (induced by low soil moisture) produce leaves with lower pi values. Over the shortterm, without new growth, reduction in pi as a measure of stress can be detected in leafsoluble sugars (Guy and Wample 1984; Brugnoli et al. 1988; Lauteri et al. 1992) or by using19the \"on-line approach to measure variation in 513C of CO2 in air as it flows by a leaf inside astirred cuvette (Hubick et al. 1988). An increase in leaf-to-air vapor pressure difference willcause reduction of pi and A in the short term (Brugnoli et al. 1988; Madhaven et al. 1991)and long term (Farquhar and Richards 1984).Nutrients Water-use efficiency in any genotype increases with more available nitrogen. Thepatterns that occur presumably reflect available water and nitrogen (or phosphorus) in thehabitat (reviewed by Farquhar and Richards 1984).Fu et al. (1992b) reported the effect of soil water and nutrients on A, in relation toleaf developmental stage in two Phaseolus species. In all treatments, the species native towarmer, drier environments had significantly lower A (suggesting greater long term WUE).Both species were more water-use efficient in high nitrogen.2.9 Carbon Isotope Analysis in Woody PlantsCraig (1953) was among the first to document carbon isotope compositions of plantmaterials. He analyzed wood from 22 trees world-wide. Wyoming trees showed a \"slightlocal\" pattern, with samples from the northwest of the state [Abies lasiocarpa, -27.9 0/oo;Pinus contorta, -27.1 Woo; Artemesia tridentata Nutt., -26.6 Voo] being lighter than thosefurther south [Abies lasiocarpa, -23.8 Voo; Artemesia tridentata, -23.7 Woo; Salix bebbianaSarg., -25.0 Woo]. Craig suggested that local environments produced notable 513C variationamong trees of the same species. He supported this idea by referring to his early work on513C differences in one tree, for which he gave no specific causes. He found no correlationsbetween 513C and altitude, or collection date. In P. contorta, 513C values of wood and leaveswere equal, a result unique from the 0.5 to 2.0 Woo difference for the 13 other plants hecompared.20In 1987, Smith et al. reported differences in photosynthetic CO2 fixation and carbonisotope fractionation among subspecies of Artemesia tridentata. The differences wereassociated with varying ploidy levels, growth habit and habitat.De Lucia et al. (1988) used carbon isotope analysis in their study of edaphiccommunities of Pinus ponderosa and Pinus jeffreyi in Great Basin sagebrush and pinyon-juniper communities. They examined the water available to, and the water use of the speciesas potential factors restricting distribution. Intraspecific comparisons were not made sinceinsufficient numbers occupied each soil type. In September, samples of current year foliage,taken at several points around each of five plants per species per site, were analyzed for 513C.Although lighter by 1.5 \u00C2\u00B1 0.3 13/00, whole leaf 513C correlated positively with that forcellulose. Carbon isotope ratios (cellulose) were inversely related to maximum stomatalconductances. Results suggested that Pinus monophylla T. & F. and Juniperus osteosperma(Toff.) Little, followed by Pinus ponderosa and Pinus jeffreyi, functioned at lower internalCO2 concentrations and higher WUEs than the shrubs.Read and Farquhar (1991) reported the first study relating genetic differences in A ofindividual species to their origin climates. They measured leaf A in 22 southern species ofNothofagus, grown in controlled environments. Variation in A was correlated withtemperature range (highest mean monthly maximum temperature minus lowest meanmonthly maximum temperature), mean annual temperature, mean annual precipitation,growing season precipitation (December to March) and latitude (p<0.05). The positivecorrelation between latitude and A for some species, grown under common conditions, wasnot maintained in field samples.Leavitt and Danzer (1992) examined the influence of light, soil moisture, relativehumidity, temperature and CO2 concentration on 813C of Pinus resinosa Ait. seedling growthrings. Treatment and control groups were grown in artificial conditions. Soil moisture andCO2 showed the greatest isotopic effects; however, the true influence of CO2 was masked byinability to regulate the 513C of chamber air. Surprisingly, relative humidity affected 513C to21a minor degree, perhaps due to varying humidities at values only >50%. Control plantsshowed seasonal isotopic changes, suggesting that growth and carbon allocation effectsand/or genetic effects might contribute to tree ring 813C changes.Newberry (1992) studied annual climatic variation effects on WUE of Pinus edulisEngelm. at two elevations (1500 m, 2300 m) with similar water stress. Water-use efficiencywas determined by 813C values of different-aged leaf cohorts, collected from the southeastaspect of each tree at 1 to 2 m height. Lower elevation trees showed similar trends in annualWUE changes. High elevation trees had randomly fluctuating and more variable 813Cvalues. Despite site differences in annual 813C patterns and foliage variation, average site813Cs were similar. Possible explanations for this were that competition, low light andherbivory confounded a simple increased drought \u00E2\u0080\u0094 increased WUE relationship at the moreheavily vegetated, higher elevation site. Influence of soil CO2 was not questioned. At thelower elevation site with open canopy, WUE was constrained only by water availability.Zhang et al. (1993) found genetic differences in A among 25 Pseudotsuga menziesiipopulations in a 15-year-old field trial. Carbon isotope discrimination was related togeographic location of the seed source and was positively correlated with altitude (r=0.76,p=0.01). Intrinsic water-use efficiency (A/g or assimilation/conductance) was negativelycorrelated with altitude (r=-0.63, p=0.0007). Fifteen year heights and diameters were notcorrelated with photosynthetic rate but were negatively correlated with A, pi/pa and g, andpositively correlated with A/g (all at p<0.01). Efficient water use for photosynthesisappeared to be more important than absolute photosynthetic rate in determining growth rates.Results from the physiological and genetic studies on photosynthetic traits and waterloss rate for P. contorta suggest that populations will differ in relative WUE, as indicated by813C. The relationship between 813C and WUE is positive (while that between A and WUEis negative). It is expected that populations from dry habitats will maintain greater WUEs(higher 813C) than those from wet habitats over a range of test sites. On dry test sites all22populations will likely increase in WUE or 813C, while on wet test sites populations willdecrease in WUE.2.10 Variation in 513C Within and Among TreesResearchers recognize that inherent spatial variation and varying proportions oforganic constituents in trees present sampling problems for 513C analysis. Several studieshave examined 513C variation within and among trees. Tans and Mook (1980) found \"good\"correlations between 513C values of whole wood and acid-alkali-acid treated latewood ofmature Quercus rubra L. Treated samples from east and west aspects of wood within a bolecorrelated poorly, with the largest 513C difference for a single year being 4.0 Woo. North andsouth aspects were not discussed. Within the same radial direction, a single ring sampledover a 40 cm vertical distance showed a 613C fluctuation of up to 0.8 oho. When sampledalong the fiber direction, at an angle to the vertical, 513C was nearly constant. Latewood wasjudged no more reliable than earlywood in providing consistent circumferential 513C trends.Leavitt and Long (1982) compared Juniperus monosperma (Engelm.) Sarg. leaf andwood 813C trends over one growing season. The sampled wood was from a south-facingtrunk branching near the ground (noted in Leavitt and Long 1986). Leaves were lighter thanwood for both whole tissue and cellulose by approximately 2.5 Woo and 3.5 Woo, respectively.Leaves followed the same 513C trends as the wood. Although rings and leaves showed intra-annual isotopic changes, the changes were not significant.Leavitt and Long (1984) studied 513C variability in tree rings at one site by examining513C trends in one Pinus edulis, as well as differences among several trees at the site.Cellulose was analyzed. The circumferential range within one tree was about 1.0 to 1.5 0/oo,while the difference among individuals was 2.0 to 3.0 0/00. The authors concluded thatpooling four orthogonal cores from four trees accurately represented site 513C trends andabsolute values. Also, two cores showed a marked improvement over single cores incorrelations between 513C of radii and 513C of the full circumference (taken as the \"true\"23813C value). Orthogonal cores yielded more representive samples than cores taken from oneside of the tree only.In 1986, Leavitt and Long compared 813C values of leaf, twig and wood cellulose inP. edulis from different locations in the southwestern USA. In most cases, P. edulis ringswere heavier than leaves by 1.0 to 2.0 Woo. Leaves were isotopically heavier than their twigsand over all, followed the same directional 813C pattern as the twigs. The authors found nolongitudinal 813C gradients in tree rings, but like previous researchers, they used only singleradii at different levels. Single radii do not accurately represent the trend or absolute 813Cvalues in a ring series representing all radii at a certain level aboveground (Leavitt and Long1986). The range of circumferential variation for ring cellulose was 0.5 to 1.5 0/00; for leavesit was 1.0 to 2.0 0/oo. The authors found a north trunk of f. monosperma yielding a different813C trend over a growing season than a south trunk from the same tree. They suggested thatthe trend in the south trunk might reflect greater sun exposure of its leaves in the early part ofthe season, whereas the north trunk would not receive high exposure until mid season.Yoder et al. (1992) measured 813C of whole wood and needles (taken 1.37 m aboveground level) in Pinus ponderosa spanning different age and size classes. Early wood wassampled from the most recent growth ring. Wood and foliage appeared to be enriched with13C at the rate of 1.0 0/oo for each 20 m increase in tree height. Foliage remained between 1.0to 2.0 Woo more depleted than wood. There was no change statistically in 813C from thebottom to top of individual large trees (actual size of trees not given).243.0 Materials and MethodsThis study involved three experiments: 1. development of sampling methods forprovenance field trials in British Columbia, as well as for future trials; 2. 513C analysis ofwood samples from provenance trials in British Columbia; 3. 513C analysis of seedlingsgrown from the seed used to establish the provenance trials.3.1 Determination of Sampling Methods for Field TrialsTo decide sampling method, 813C variation in foliage and wood was assessed for fiveopen-grown Pinus contorta saplings located at one site (Juliet Creek, lat. 1210 00' N, long.490 43' W, elev. 1010-1067 m). The site had gravelly soil and vegetation suggesting lowmoisture availability (Achillea millefolium, Amelanchier alnifolia, Calamagrostis rubescens,Lonicera utahensis, Lupinus spp., Pachistima myrsinites). The saplings were moderatelyinfected with needle cast. It was assumed that each was infected to the same degree and thateach responded similarly to the disease.The sampling was done on June 11, 1989. Saplings were felled and discs of the boleand needles collected accordingly:Tree No. 1 (Age 21.5 yrs., Ht. 5.3 m)Discs: No. 1 - leaderNo. 2 - at midsection of 1988-89 stemNo. 3 - at 1.2 m above groundNo. 4 - within 10 cm of groundNeedles: 1. Top of treea) needles growing on leaderb) 1989 needles growing on stem and on north, east, south and west-facing branches2. Mid-tree - 1989 needles at four cardinal directions around the crowncircumference at 1.2 m253. Lower tree - 1989 needles at four cardinal directions on bottombranchesTrees No. 2, 3, 4 and 5 (Ages 15, 24.5, 31.5 and 23.5 yrs., respectively; Heights 2.42,3.95, 5.45 and 9.4 m, respectively)Discs: No. 1 - leaderNo. 3 - at 1.2 m above groundNo. 4 - within 10 cm of groundNeedles: 1. growing on leader2. 1989 needles on bottom N, E, S and W-facing branchesFour, 5 mm wide \"cores\" were cut from the air-dried discs, each core taken from thenorth, east, south or west aspect of the sapling. Within each core, rings representing years1983, 1984, 1987 and 1989 were analyzed. Data from six climate stations around JulietCreek showed that 1983 and 1984 had low growing season temperatures with highprecipitation, relative to other years in the 1980s, while 1987 and 1989 had relatively highgrowing season temperatures and low precipitation.Wood was hand filed for homogeneous well-mixed particles. Weighed samples(0.5 mg) were combusted in a fully automatic elemental analyzer (Howarth 1977). Thesamples were introduced into the combustion zone at 1050 \u00C2\u00B0C to coincide with oxygenenrichment of the helium gas. This eased combustion which was completed when thecombustion products passed through the combustion tube. The carbon was convertedquantitatively to carbon dioxide, the hydrogen to water and the nitrogen to oxides ofnitrogen. A secondary reaction tube removed excess oxygen and reduced the nitrogen oxidesto nitrogen before the gases entered the chromatographic column for separation of thecombustion products. For measurement of 1513C, the CO2 was transferred to a VG IsotechPrism triple-collecting mass spectrometer (Middlewich, England) (Nier 1947). Internalprecision of the spectrometer was better than \u00C2\u00B1 0.02 Woo. Samples were corrected for 170contribution to mass 45 abundance.26Needles: After freeze-drying and interim refrigeration, the needles were cleaned,ground in a Wiley Mill (40 mesh), pulverized in a Planetary Micro Mill \"Pulverizette 7\"(Fritsh GmbH) and analyzed as above.3.2 Determination of Genetic Variation in WUE Among Sapling ProvenancesI obtained permission from Dr. Cheng C. Ying (Provenance Forester, BritishColumbia Ministry of Forests) to sample field trials in British Columbia, set up in 1974 toassess genotype x environment effects in Pinus contorta (Illingworth 1978). From 60 testsites, I chose three: 70 Mile House, Holden Lake and Salmon Lake. They formed a north-south transect through the B.C. interior, representing contrasting soil moistures and climates(Table 1, Fig. 1). Of 140 provenances tested, I studied eleven (Table 2, Fig. 1).Each test site comprised two blocks, with nine saplings per provenance per block,planted 2.5 m apart in square plots (Fig. 2). Wood was sampled from every living tree,extracted by coring full diameter (north to south) near ground level. Ground level, ratherthan 1.2 m above ground, was used to avoid missing years. Sapling heights and diameters (at15 years) were obtained from Ministry growth surveys (1988). Volumes were calculatedfrom a formula by Kovats (1977). Features of each sapling and surrounding neighbors whichmight influence carbon assimilation and partitioning were noted (e.g., damaged leaders,physical erectness, sexual precociousness, canopy closure).The sampling extended over two field seasons. During July 1990, Holden Lake wassampled. In July 1991, 70 Mile House and Salmon Lake were sampled.Air-dried cores were microscopically magnified and cleaned of extraneous carbonsources (e.g., black smudges from the increment borer, clinging bark particles) with a razorblade. Ten year segments (1980 - 1989) from each side were combined and ground in aWiley Mill (20 mesh). Following 2 hours freeze-drying, the samples were pulverized forone-half hour in a Planetary Micro Mill and stored in plastic vials.Table 1.^Test site characteristics for 70 Mile House, Holden Lake and Salmon Lake.Test Site Lat Long Elev SMR1 SNR2 Climate Station3 Lat Long Elev MPDM4 MSP5 MAP6 MTWM7 MAT8 Years9(m) (m) (mm) (mm) (mm)^(\u00C2\u00B0C) (\u00C2\u00B0C)70 Mile House 51 17 121 20 1070^3.0^D^70 Mile House 51 18 121 24 1079.9 16.6 167.9 327.0^14.0 2.6^7Holden Lake 51 37 121 31^1160^3.5^C^100 Mile House 51 39 121 16 1059.2 15.0 196.0 386.0^14.4c 3.0\"^10Salmon Lake 54 51 123 55^950^5.0^B/C^Kalder Lalce 54 58 124 10 969.3 23.0 280.4 697.1^12.5 0.2^7'Soil moisture regime (Classes: 0 very xeric; 1 xeric; 2 subxeric; 3 submesic; 4 mesic; 5 subhygric; 6 hygric; 7 subhydric)2Soil nutrient regime (Classes: A very poor; B poor; C medium; D rich; E very rich)3Normalized data from Climatic Data Summaries for the Biogeoclimatic Zones of British Columbia, Version 3, 1991, compiled by D. Meidinger andG. Reynolds (unpublished)4Mean precipitation dry month5Mean summer precipitation (May-Sept. inclusive)6Mean annual precipitation7Mean temperature warmest month8Mean annual temperature9Years station operatingc,c1Temperature adjusted downward 1\u00C2\u00B0C to account for elevation difference between climate station and siteaiiu3sigt plui (unpublished, compiled by Inselberg, MOF 1988/89)70 Mile HouseBlock 51 Slope: 5%, with undulating shape Aspect: variableSoil Depth: 30+ cm^Rooting Depth: 30+ cmTexture: loam with 5% coarse fragments to 20 cm depthloam with 40% coarse fragments from 20 to 30+ cmHorizons: Ae (2 cm)^Humus mor (2 cm)Base Status: highWater: No groundwater table, gleyed horizons or flooding.Indicator Species: Arctostaphylos uva-ursi, Astragalus miser, Calamagrostisrubescens, Galium boreale, Rosa acicularis, Shepherdia canadensis.Block 51 shows low productivity, despite its fine-textured rich soil. Soil depthis shallow and precipitation low. Considerable height variation, even withinprovenances, occurs.Site descriptions (cont'd)70 Mile HouseBlock 52Slope: 12%, with some mounds^Aspect: 50\u00C2\u00B0Soil Depth: 60 cm^Rooting Depth: 60 cmTexture: loam with 5% coarse fragments to 25 cmloam with 35% coarse fragments from 25 to 60 cmHorizons: Ae (2 cm)^Humus mor (2 cm)Base Status: highWater: No groundwater table, gleyed horizons or flooding.Indicator Species: Achillea millefolium, Arctostaphylos uva-ursi,Arnica cordifolia, Astragalus miser, Calamagrostisrubescens, Galium boreale, Rosa acicularis, Shepherdiacanadensis.Block 52 has deeper soil and a slightly cooler exposure than Block 51.Vegetation is only slightly more productive.Holden LakeBlock 53 Slope: 0 - 5%, with mounds^Aspect: flatSoil Depth: 50+ cm^Rooting Depth: 50+ cmTexture: silty loam with 5% coarse fragments to 50 cmsilty loam with 35% coarse fragments to 50+ cmHorizons: Ae (3 cm)^Humus mor (3 cm)Base Status: mediumWater: No groundwater table, gleyed horizons or flooding.Indicator Species: Arctostaphylos uva-ursi, Calamagrostis rubescens,Rosa acicularis, Shepherdia canadensis, Vaccinium scoparium.Block 53 has uniform soil characteristics throughout, with mostprovenances growing well. Exceptions are stunted, multiple leadersamples showing poor vigor.Site descriptions (cont'd)Holden LakeBlock 54Slope: 0 - 5%, with undulating shape^Aspect: flatSoil Depth: 40+ cm^Rooting Depth: 40+ cmTexture: silty loam with 10% coarse fragments to 20 cmsilty loam with 40% coarse fragments from 20 to 40 cmHorizons: Ae (2 cm)^Humus mor (2 cm)Base Status: mediumWater: No groundwater table, gleyed horizons or flooding.Indicator Species: Same as for Block 53.Provenance performance is similar to that on Block 53.Salmon LakeBlocks 99 and 100Slope: 0%, with a straight shape^Aspect: flatSoil Depth: 30+ cm^Rooting Depth: 30+ cmTexture: loamy-sand with 15% coarse fragments to 15 cmsandy-loam with 15% coarse fragments from 16 to 30+ cmHorizons: Ae (10 cm - variable)^Humus mor (5 cm)Base Status: mediumWater: No groundwater table or flooding. Gleyed horizons(light mottles) to 8 cm.Indicator Species: Arnica cordifolia, Aulacomnium palustre,Calamagrostis rubescens, Cornus canadensis, Galiumboreale, Lonicera involucrata, Peltigera apthosa,Petasites palmatus, Polytrichum juniperinum, Rosaacicularis, Ribes lacustre, Rubus pubescens, Spireabetultfolia, Spirea douglasii, Vaccinium caespitosum,Veratrum viride.The blocks have uniform, fine-textured soils which become denseat 30 cm. Mottles at 10 cm suggest a fluctuating water tabledue to poor drainage. Several provenances have reddish olderfoliage. Some provenances are growing slowly; as well, some havebroken or forked leaders suggesting minor snow damage. Most ofthe original forest floor was burned away by a slash fire but mineralsoil was not exposed.STAWYE 100^ CO1 ^200^yArLESKILOynE7RES^100^ 100^203^300 xltOmFIRESDray,b by P. J. Jonce. 1975ell olden Lk70 Milecr 471 e( 1 57130Fig. 1. Map of British Columbia and Alberta showing locations of provenancesand test sites (Salmon Lake, Holden Lake, 70 Mile House). Provenance locationsare represented by the provenance numbers on the map.Table 2. Provenance characteristics and available climate data.Prov Location^Lat^Long Elev Mtwm1 Mat^Msp Map Mpdm Climate Station2 Lat^Long Elev Years3(111)^(\u00C2\u00B0C) (\u00C2\u00B0C)^(mm) (mm)^(mm)^ (m)1 Trapping Creek 49 35 119 01 100613 Horne Lake 51 46 124 44 914 13.2 1.6 133.3 334.2 13.8 Kleena Kleene 51 59 124 56 899 17b30 Lower Post 59 59 128 33 640 15.5 -2.9 222.5 460.8 13.9 Lower Post 59 56 128 30 583 8b44 Marl Creek 51 31 117 11 94557 Inonoaklin Valley 49 54 118 12 579 17.8a 6.4a 254.5 621.8 34.5 Fauquier 49 52 118 04 472 3061 Purden Lake 53 52 121 44 83869 Link Creek (Ab) 49 26 114 25 1379 14.3 3.0 325.1 852.0 39.0 Castle R. S. 49 24 114 20 1364 1471 Fly Hills 50 43 119 27 152495 Petersburg 56 47 132 58 23 13.2 5.2 885.8 2671.3 134.0 Petersburg 56 82 132 95 15 24100 Nina Creek 55 48 124 49 762 13.6 0.3 223.5 493.7 23.7 Germansen Ld. 55 47 124 42 747 27104 Nechako River 54 01 124 32 732 13.9 2.0 258.4 525.2 22.2 Fort Fraser 13S 53 53 124 35 701 5b'Refer to Table 1 for the meaning of Mtwm, Mat, Msp, Map and Mpdm.2Sources: Canadian Climate Normals 1951-1980; Climatic Data Summaries for the Biogeoclimatic Zones of British Columbia, Version 3, 1991, compiled byD. Meidinger and G. Reynolds (unpublished); personal communication, staff at Atmospheric Environment Service (AES), Vancouver, B.C.3Years climate station in service.a Temperature adjusted downward 1 \u00C2\u00B0C to account for elevation difference between climate station and provenance location.b Adjusted normals based on 5-19 years, including those from 1951-1980 and any other data from 1931-1950.Schematised Test Block LayoutCHILCOTIN REGION70 Mile House, Site 1: Block 52Note: NOT PLANTED = \u00E2\u0080\u00A2 ; DEAD = XFig. 2. Schematised test block layout for one field site. The squares representnine-tree plots, the positioning of individuals designated by open circles. Thenumbers inside the squares show the provenances allocated to those plots.3233Whole wood was analyzed, with reproducible analyses to 0.05 0/013 (8 replicates). Thedecision to use whole wood rather than alpha cellulose was based on previous studies and anexperiment on 10 cores representing a range of whole wood 513C values from -23.358 to-27.312 oho. Alpha cellulose was isolated from the wood in Dr. N. Livingston's laboratory,University of Victoria, British Columbia. Alpha cellulose 513C values were regressed on813C values of the wood from which it was isolated. The regression r2 of 0.981 was highlysignificant (p<<0.0001) (Fig. 3).3.3 Seedling ExperimentOne gram of seed per provenance, cold-stored approximately 20 years, was procured.Germination trials showed 60 to 95% viability. In September 1990, the seed was sown in164 ml super cell cone-tainers (Stuewe and Sons, Inc.) containing peat and vermiculite 2:1.To each 113 litre-bag of peat was added 662.25 g Dolomite (WesGrow) and 112.5 g traceelements (WesGrow). Soil pH at planting was 4.7. Seeds were covered with No. 2 GraniteGrit (Imasco). The seedlings were grown in a greenhouse. After emergence they werewatered to field capacity as necessary and after 4 weeks, fertilized every 4 days with 100ppm 20-8-20 (Plant Prod). The 18 hour photoperiod was supplemented by high intensitydischarge sodium lamps. Irradiance ranged from 94 to 513 limo' quanta m-2 s-1 PAR,depending on seedling location and cloud cover. Therefore, seedling positions were rotatedweekly. In full sunlight (which seldom occurred) seedlings were exposed to 1400 lamolquanta m-2 s-1 PAR. Relative humidity varied from 40 to 45%, while air temperature aboveseedlings was 20 \u00C2\u00B1 2\u00C2\u00B0C on average (Vaisala RH/temperature meter, Cole-Parmer). Airabove the seedlings was well mixed by fans. Near the end of December 1990, sulfurdeficiency was corrected by supplementing seedling nutrition with K2SO4 (10 ppm) and acommercial (STEM) micronutrient mix (1 ppm).On February 18, 1991, thirty randomly selected seedlings per provenance wereharvested. Most had set bud and had mature as well as primary needles. Shoots were frozen-27 \u00E2\u0080\u009426 \u00E2\u0080\u009425 \u00E2\u0080\u009424 \u00E2\u0080\u009423 \u00E2\u0080\u009422WHOLE WOOD 6'3C C/00)Fig. 3. Regression of alpha cellulose 813C on whole wood 813C: 813C (cellulose)= -1.357 + 0.903 813C (wood). Regression r2=0.981 (p<<0.0001).3435in liquid nitrogen, freeze-dried and weighed. Roots were washed, frozen, freeze-dried andweighed. An additional five seedlings were prepared for SIT analysis of primary needles,mature needles and stem parts. Since ANOVA showed no significant differences amongparts, total shoots were analyzed for 813C following grinding in a Wiley Mill (40 mesh) andpulverization. Analyses were reproducible to 0.1 0/00 (5 replicates).3.4 Data AnalysisDetermination of Sampling Methods for Field TrialsThree ANOVAs were used to test effects. Analyses were done with SAS/STATsoftware, 6.03 Edition (SAS Institute 1988). The first ANOVA was a three-way analysis of813C values in wood for year, aspect and tree effects as well as interactions in Pinus contortasaplings. The second ANOVA was a three-way analysis, specifically for material effects(needle vs. wood) and material x aspect interactions. The third ANOVA was a three-wayanalysis on one sapling, specifically for needle x position effects. The models were mixedmodels, with aspect fixed and the other factors random. It could be argued that year sampleswere repeated measures and therefore dependent. They were analyzed as if they wereindependent, however, since repeated measures analysis produced similar results to factorialANOVA. Hicks (1982) shows that a repeated measures design is but a special case offactorial and nested-factorial experiments. Numerical results are the same.Determination of Genetic Variation in WUE Among Sapling ProvenancesThree-way ANOVA (using SAS/STAT software) was applied to the replicatedrandomized complete block design of the field trials. All effects were random. The data sethad missing cases (or cells) as well as missing observations. Provenance 57 was not tested atSalmon Lake. Provenance 95 at 70 Mile House was deleted from statistical analyses sinceonly four saplings survived from a possible 18. As well, saplings which were dead, too smallto core or girdled from top to bottom were removed from the data set. Other outliers with no36basis for removal were treated as very unusual measurements and not deleted. To simplifyresults and test all provenances, balanced data sets were analyzed in three ANOVAs. Thedata sets were extracted from the following site-combinations: 1. 70 Mile House, HoldenLake and Salmon Lake (provenances 57 and 95 excluded); 2. 70 Mile House and HoldenLake (provenance 95 excluded); and 3. Holden Lake and Salmon Lake (provenance 57excluded).The design posed a pseudoreplication problem, since sites were not replicated.Pseudoreplication was avoided by not interpreting significant differences among sites as sitetreatment effects (Hurlbert 1984). Inferences were made on provenances within sites.Correlation and regression analyses for 813C and climate variables were done withSYSTAT software, Version 5.01 (Wilkinson 1990). Multiple regression modelled site mean813C values as a function of provenance mean temperature of the warmest month (mtwm),mean annual temperature (mat), mean summer precipitation (msp), mean precipitation of thedriest month (mpdm) and mean annual precipitation (map). Only provenances 13, 30, 57,69, 95, 100 and 104 were used, since nearby climate stations did not exist for the others.Seedling ExperimentStatistical analyses were done with SYSTAT software. The seedling design wascompletely randomized. One-way ANOVAs compared provenance means for 813C values,shoot weights, root weights, total weights and shoot/root ratios. Correlations weredetermined among growth variables and 813C.374.0 Results4.1 Determination of Sampling Methods for Field TrialsTable 3 shows the effects of tree, aspect and year on 513C values of stump wood inPinus contorta saplings. Analysis of variance on the mixed model (trees and years random,aspects fixed) showed significant tree, year and tree x aspect effects (a=0.05). The aspect,aspect x year and tree x year effects were not significant, although the tree x year effect wasnearly so (p=0.057). The tree x aspect x year effect (error variance) could not be tested andwas assumed non-significant. Multiple range tests (Ryan-Einot-Gabriel-Welsch, Tukey)defined two homogeneous tree subsets (trees 1 and 5; trees 2, 3 and 4). As well, twohomogeneous subsets for years were 1987; and 1983, 1984 and 1989.Figures 4 and 5 depict the lack of interaction between tree x year and aspect x yeareffects. Fig. 4 shows that ranking in 513C changed little among saplings for the years tested.Fig. 5 shows that north, south and west aspects maintained ranking over the years tested.During 1987 (a dry year), 513C was greater in all directions.Sampling allowed two cores per sapling in the provenance field trials. For thesaplings at Juliet Creek, it appeared that certain paired aspects would accurately track 813Cyear to year variation among the years tested. The 513C values in stump wood portrayed byaveraging north and south, and east and west aspect 513C values are shown in Figs. 6 and 7.Ranking in 513C among saplings appears to be more stable over the years tested when northand south aspects are used as combined samples rather than east and west aspects.Table 4 shows the effects of tree, aspect and sampling material (needles vs. wood) on513C values near ground level in 1989. Analysis of variance on the mixed model (trees andmaterials random, aspects fixed) showed non-significant tree and aspect effects, as well astree x aspect and tree x material interactions. (Tree 3 was missing an easterly 513C value).Table 3.^ANOVA for five Pinus contorta saplings at Juliet Creek (discs at stump level).Source df SS MS EMS F Value P > FTree 4 11.7332 2.9333 \u00C2\u00B1 4(32,1y^160.2T 22.250 0.0001Aspect 3 1.1313 0.3771 (y2^5(y2Ay^402TA 4. OA 2.192 0.1539Year 3 3.8856 1.2952 42,ry + 20(52y 9.820 0.0015T*A 12 2.0315 0.1693 (y2 + 4452TA 2.530 0.0155T*Y 12 1.5819 0.1318 + 402,ry 1.970 0.0573A*Y 9 0.6262 0.0696 cy2^50.2Ay 1.040 0.4273T*A*Y 36 2.4042 0.0668 (52Satterthwaite Approximation to Test Aspect:Source: AspectError: MS (T*A) + MS(A*Y) - MS(Error)DF^MS^Denominator DF^Denominator MS^F Value^P>F3^0.3771 9.71^0.1721 2.192^0.1539\u00E2\u0080\u009424\u00E2\u0080\u009425\u00E2\u0080\u009426\u00E2\u0080\u00A2 TREE5\u00E2\u0080\u00A2 TREE4\u00E2\u0080\u00A2 TREE3\u00E2\u0080\u00A2 TREE2A TREE!\u00E2\u0080\u009427391983^1984^1987^1989YEARFig. 4. Stump wood 813C values for years 1983, 1984, 1987 and 1989 in fivePinus contorta saplings. Isotopic compositions are the means of fourobservations taken circumferentially per year tested per tree, e.g. sum of north,east, south and west aspect 513C values for 1983/4. Ranking in 813C amongsaplings is relatively stable over the years tested when the average circumferential813C per tree is the measure for comparison. \u00E2\u0080\u00942440\u00E2\u0080\u009425\u00E2\u0080\u009426\u00E2\u0080\u009427\u00E2\u0080\u00A2 WEST^ SOUTH\u00E2\u0080\u00A2 EAST\u00E2\u0080\u00A2 NORTH1983^1984^1987^1989YEARFig. 5. Aspect 813C values (stump wood) for years 1983, 1984, 1987 and 1989.Isotopic compositions are means of five saplings, each sapling contributing one613C value per year per aspect. Ranking in 613C among aspects remains relativelystable over the years tested, when aspect 813C is a measure of several saplings.\u00E2\u0080\u00A2 TREE5\u00E2\u0080\u00A2 TREE4\u00E2\u0080\u00A2 TREE3\u00E2\u0080\u00A2 TREE2\u00E2\u0080\u00A2 TREE141\u00E2\u0080\u009424\u00E2\u0080\u009425\u00E2\u0080\u009426\u00E2\u0080\u0094271983^1984^1987^1989'iNc,-)7.0YEARFig. 6. Stump wood 813C values for years 1983, 1984, 1987 and 1989 asportrayed by averaging north and south aspects in five Pinus contorta saplings.Isotopic compositions are means of two samples per tree. Ranking in 813C amongtrees remains quite stable over the years tested when north and south aspects areused as combined samples.-24\u00E2\u0080\u009425\u00E2\u0080\u009426\u00E2\u0080\u00A2 TREE5\u00E2\u0080\u00A2 TREE4\u00E2\u0080\u00A2 TREE3\u00E2\u0080\u00A2 TREE2A TREE1\u00E2\u0080\u009427421983^1984^1987^1989YEARFig. 7. Stump wood 813C values for years 1983, 1984, 1987 and 1989 asportrayed by averaging east and west aspects in five Pinus contorta saplings.Isotopic compositions are means of two samples per tree. Ranking in 813C amongtrees is not as stable over the years tested when east and west aspects are used ascombined samples rather than north and south aspects.Table 4. ANOVA for 513C comparisons of wood and needles at stump level in five Pinus contorta saplings at Juliet Creek.Source df SS MS EMS F Value P> FTree 4 4.4530 1.1132 c52 + 3752TM + 1.75a2T 2.540 0.1943Aspect 3 1.8463 0.6154 G2 + 4.666702Am + 1.8667(52Th + 0A 0.592 0.6624Material 1 18.8254 18.8254 '32 + 3752TM + 18.462a2m 42.960 0.0028T*A 12 2.2291 0.1858 G2 \u00C3\u00B7 1.9167GTA 0.940 0.5419T*M 4 1.7528 0.4382 c:F2 + 3752TM 2.230 0.1327A*M 3 3.1490 1.0497 02+ 4.66670'2Am 5.330 0.0164T*A*M 11 2.1659 0.1969 G2Satterthwaite Approximation to Test Aspect:Source: AspectError: 0.9739*MS (T*A) + MS(A*M) - 0.9739*MS(Error)DF^MS^Denominator DF^Denominator MS^F Value^P>F3^0.6154 2.89^1.0388 0.592^0.662444Material and aspect x material effects were significant. The influence of aspect on 813C forneedles and wood is shown in Fig. 8. Wood at 1.2 m height is included for interest; it couldnot be compared to wood near ground level in this ANOVA. It appeared that the influenceof aspect was prominent for needles but not wood. Needles had more negative 813C valuesthan wood. The aspect x position effect in wood appeared non-significant (standard errorbars overlap).Figures 9 and 10 depict tree 813C trends with aspect, in needles and wood at stumplevel. The tree x aspect x material effect (error variance) could not be tested and wasassumed non-significant. Overall, wood (base) 813C values were less variable from tree totree than the needles. In 1989, isotope values among trees ranged from -25.267 to -26.840Woo in stump wood (a difference of 1.573 Woo) and from -25.750 to -28.989 Woo in lowerbranch needles (a difference of 3.239 cloo).Analysis of variance (Table 5) shows the effects of aspect, position (base, 1.2 m, top)and material (needles, wood) on 813C values in one sapling during 1989. Materials weresignificantly different; other effects were not. Figure 11 depicts the lack of response ofneedle 813C to aspect and position in the sapling.4.2 Determination of Genetic Variation in WUE Among Sapling ProvenancesTable 6 gives summary statistics for provenance 813C values at 70 Mile House,Holden Lake and Salmon Lake. Block designs with missing observations require unbiasedestimates of treatment means, therefore least squares means (LSMs) are reported. Allprovenances, but 71, showed progressively decreasing 813C values from the driest to wettestsite. Provenance 71 broke the 813C trend from 70 Mile House to Holden Lake by a smallincrease (0.069 0/oo), rather than a decrease. Standard error of the LSM was 0.1 oho for allbut two provenance-site pairs.45- 25 \u00E2\u0080\u009426 -\u00E2\u0080\u009427 -\"co\u00E2\u0080\u009428 -\u00E2\u0080\u009429North East South WestASPECTo 1.2 M WOOD0 BASE WOODA BASE NEEDLESFig. 8. Aspect x material interactions (1989) near ground level and at 1.2 mheight (wood) in Pinus contorta saplings (\u00C2\u00B1SE). Isotopic compositions are meansof five trees (one sample per tree). Needle 613C is influenced by aspect whereaswood 613C is not. Needle 613C is more negative than wood 613C.46\u00E2\u0080\u00A2 TREE5\u00E2\u0080\u00A2 TREE4\u00E2\u0080\u00A2 TREE3\u00E2\u0080\u00A2 TREE2\u00E2\u0080\u00A2 TREE1North East South West- 25\u00E2\u0080\u009426\u00E2\u0080\u009428\u00E2\u0080\u009429ASPECTFig. 9. Needle 513C trends with aspect (1989) in five Pinus contorta saplings.Needles are sampled from the lowest branches. Isotopic compositions representone sample per tree. Sapling 3 is missing its easterly 513C value.47-25\u00E2\u0080\u009426'-'-''s......cP\u00E2\u0080\u009427'Colo\u00E2\u0080\u009428\u00E2\u0080\u009429\u00E2\u0080\u00A2 TREE5\u00E2\u0080\u00A2 TREE4\u00E2\u0080\u00A2 TREE3\u00E2\u0080\u00A2 TREE2\u00E2\u0080\u00A2 TREE1North East South WestASPECTFig. 10. Wood 813C trends with aspect (1989) in five Pinus contorta saplings.Wood is cored at 10 cm above ground level. Isotopic compositions represent onesample per tree.Table 5. ANOVA for sample comparisons in one Pinus contorta sapling at Juliet Creek.Source df SS MS EMS F Value P > FAspect 3 0.2092 0.0697 02 \u00C2\u00B1 302Am + 202Ap + OA 3.496 0.9097Position 2 0.4006 0.2003 02 + 402pm + 802p 0.510 0.6624Material 1 7.4783 7.4783 02 \u00C2\u00B1 402pm + 12a2m 19.030 0.0487A*P 6 0.5233 0.0872 02 4. 202Ap 0.491 0.7958A*M 3 0.3307 0.1102 02 + 302Am 0.621 0.6267P*M 2 0.7859 0.3930 02 \u00C2\u00B1 402pm 2.214 0.1905A*P*M 6 1.0651 0.1775 02Satterthwaite Approximation to Test Aspect:Source: AspectError: MS (A*P) + MS(A*M) - MS(Error)DF^MS^Denominator DF^Denominator MS^F Value^P>F3^0.0697 0.04^0.0199 3.496^0.9097-25\u00E2\u0080\u009426\u00E2\u0080\u009428\u00E2\u0080\u00942949North East South WestASPECT\u00E2\u0080\u00A2 TOP NEEDLES\u00E2\u0080\u00A2 1.2 M NEEDLESA BASE NEEDLESFig. 11. Needle 813C trends with aspect and position in one juvenile Pinuscontorta sapling. Isotopic compositions represent one sample. Needle 813Cresponds little to aspect and position in the tree.Table 6. Summary statistics for 813C in Pinus contorta sapling provenances planted at 70 Mile House, Holden Lake andSalmon Lake.Provenance1 13 30 44 57 61 69 71 95 100 10470 Mile HouseLSMa -24.754 -24.621 -24.867 -24.438 -25.144 -24.264 -24.787 -24.846 -24.518 -24.292S Eb 0.130 0.137 0.130 0.118 0.148 0.110 0.113 0.113 0.113 0.110Tic 13 12 14 16 10 18 17 17 17 18Holden LakeLSM -24.966 -24.942 -25.139 -24.960 -25.415 -24.533 -25.365 -24.777 -24.133 -24.837 -24.726SE 0.128 0.137 0.127 0.132 0.128 0.128 0.125 0.125 0.209 0.132 0.128n 17 15 18 16 17 17 18 18 10 16 17Salmon LakeLSM -25.748 -25.682 -26.120 -25.592 -25.213 -25.912 -25.764 -24.884 -25.467 -25.488SE 0.132 0.152 0.118 0.132 0.132 0.122 0.122 0.127 0.136 0.130n 15 11 18 15 15 17 17 16 14 15aLeast squares mean^ vlbStandard error of the LSM 0cMaximum=1851Table 7 summarizes ANOVA results (all factors random) for the data from 70 MileHouse, Holden Lake and Salmon Lake, where all but provenances 57 and 95 were tested.Site, block (within site) and provenance variances were homogeneous (Bartlett's test). Theanalysis used 431 observations from a total of 486, since 55 trees were missing or dead. Themodel r2 was 0.569. Site and provenance effects were significant (a=0.05), while block(within site), site x provenance and block x provenance (within site) effects were not.Provenance 813C trends across sites are graphed in Figure 12.The literature does not suggest specific tests to distinguish dependent (least squares)means, other than the Bryant-Paulson-Tukey test and conditional Tukey-Kramer test foradjusted means in one-way ANCOVAs (Day and Quinn 1989). I used Fisher's LSD test withan applied Bonferroni inequality to compare provenance LSMs (Miller 1981). The a levelwas decreased to 0.001 because 36 means comparisons were made for the analysis includingall test sites; and 45, for the analyses including two sites. Lowering the comparisonwiseerror rate to 0.001 was an attempt to maintain the overall experimentwise error rate near0.05, e.g., 45 x 0.001 = 0.045. Bonferroni's procedure tested pairwise differences betweensites, each site compared at a reduced a level of 0.01, again to keep the overallexperimentwise error rate near 0.05. Holden Lake and 70 Mile House were different fromSalmon Lake, but not different from each other.Variance components derived according to the format in Table 7 showed that site andsampling error contributed most to the variation in 813C on a per sapling basis. Since the sitex provenance component contributed nothing to the variation in 813C (in fact, it wasnegative), the test term used for means differences among populations was the block xprovenance (within site) mean square (MS) rather than the site x provenance MS. The leastsignificant difference (LSD) using block x provenance (within site) as the test term was0.440 Voo. Table 8 shows provenance means test results. Provenances 61 and 104 weresignificantly different from provenances 69 and 30. As well, provenance 61 was differentfrom provenances 71 and 1.Table 7. ANOVA for provenance trials at 70 Mile House, Holden Lake and Salmon Lake. Variance components (VC) are presented as apercentage of total components.Source df SS MS EMS F Value P > F VC(%)Site 2 80.5247 40.2623 a2+ 7.76324:Y2wpm + 15.5264cr2s.p + 69.868702w) + 139.737402s 68.034 0.0208 47Bl(S) 3 2.2591 0.7530 452+ 7.776402131.pm + 69.9877a2ms) 2.032 0.1375 1Prov 8 20.6984 2.5873 a2+ 7.8192a2m.p(s)+ 15.6384a2s.p + 46.915102p 12.334 0.0001 8S*P 16 3.3536 0.2096 a2+ 7.8517a2131p(s) + 15.70340-2s.p 0.564 0.8809 0Bl*P(S) 24 8.9348 0.3723 0-2\u00C2\u00B17.8828a2B1w) 1.487 0.0673 3Error 377 94.3909 0.2504 a2 41Legend: Bl-BlockSSiteF'=-ProvenanceSatterthwaite Approximations Used to Derive F Values:Source: SiteError: 0.9983*MS[B1(S)] + 0.9887*MS(S*P) - 0.9848*MS[B1*P(S)] - 0.0022*MS(Error)DF^MS^Denominator DF Denominator MS^F Value P > F2^40.2623 1.78^0.5918^68.034 0.0208Table 7. (con't)Satterthwaite Approximations Used to Derive F Values:Source: Block(Site)Error: 0.9865*MS[Bl*P(S)] + 0.0135*MS(Error)DF^MS^Denominator DF Denominator MS^F Value P> F3^0.7530 24.44^0.3706^2.032^0.1357Source: ProvError: 0.9959*MS(S*P) + 0.0041*MS(Error)DF^MS^Denominator DF Denominator MS^F Value P> F8^2.5873 16.16^0.2098^12.334^0.0001Source: Site*ProvError: 0.9961*MS[Bl*P(S)] + 0.0039*MS(Error)DF^MS^Denominator DF Denominator MS^F Value P > F16^0.2096 24.13^0.3718^0.564 0.880954 \u00E2\u0080\u009424\u00E2\u0080\u009425--26\u00E2\u0080\u0094270_ _ PROV104PROV100^ PRO V71^ PROV69^ PRO V61^ PRO V44PRO V30\u00E2\u0080\u0094 \u00E2\u0080\u0094 \u00E2\u0080\u0094 PROV13\u00E2\u0080\u0094 \u00E2\u0080\u0094 - PROV110e^xkoV`el\u00C2\u00B0SITEFig. 12. Changes in provenance 513C least squares means among three sites \u00E2\u0080\u0094 70Mile House, Holden Lake and Salmon Lake. Provenances respond similarly indegree and direction to changes in site moisture.Table 8. Summary of test results for differences in 513C among nine provenances over all sites(70 Mile House, Holden Lake, Salmon Lake).Provenance^61 104 100 44 13 71 1 69 30LSM^-24.670 -24.835 -24.941 -24.997 -25.082 -25.129 -25.156 -25.355 -25.375Note: Underscores show homogeneous subsets (comparisonwise error rate=0.001).Fishers Least Significant Difference=0.440 0/00.56The provenance effect in the ANOVA (Table 7) remained significant at a=0.05 whenthe block x provenance (within site) MS was used rather than the site x provenance MS toderive the F value. This was the case for the other analyses as well (Tables 9 and 11), withchanges to test terms.Least squares means were also compared using the Tukey test since the Fisher's LSDwith a comparisonwise error rate of 0.001 might miss real differences among provenances.Tests based on the studentized range (Tukey) assume that the means are independent. Inunbalanced designs this assumption is violated to some degree. The Tukey test(experimentwise error rate=0.05) found a few more significant differences than the Fisher'sLSD test. The Tukey test determined that provenances 61, 104 and 100 were different fromprovenances 69 and 30. As well, provenance 61 was different from provenances 13, 71 and1.Table 9 summarizes ANOVA results (all factors random) for the data from 70 MileHouse and Holden Lake, where provenance 57 and all other provenances except 95 weretested. The variances of main effects were homogeneous (Bardett's test). The analysis used321 observations from a total of 360 (model r2 = 0.369). Provenance and block x provenance(within site) effects were significant (a=0.05) while site, block (within site) and site xprovenance effects were not. Provenance 813C trends with site are graphed in Figure 13.Variance components (Table 9) showed that site and provenance contributed 11 and15% respectively to the variation in 813C on a per sapling basis. Sampling error contributed65%. The site x provenance component contributed nothing to the variation, therefore thetest term used for means differences among populations was block x provenance (withinsite). The least significant difference using block x provenance (within site) as the test termwas 0.625 No. Table 10 shows means test results. Provenances 61 and 104 were differentfrom provenance 57. As well, provenance 61 was different from provenance 69. The Tukeytest showed a few more differences among provenances. Provenances 61, 104, 100 and 44were different from provenance 57. As well, provenances 61 and 104 were different fromTable 9. ANOVA for provenance trials at 70 Mile House and Holden Lake. Variance components are presented as a percentage of totalcomponents.Source df^SS^MS^ EMS^ F Value^P > F^VC(%)Site^1^7.5938^7.5938 cy2+ 7.768202B1*p(s)+ 15.536002s*p + 77.6820a2B1m+ 155.360002s^9.263^0.1534^11aBI(S)^2^2.0048^1.0024 (524.7.81332Bi*P(s) + 78.1330G2ms)^ 2.208^0.1383^2Prov^9 19.1918^2.1324 (T2+ 7.8634G2wp(s) + 15.7270G2s*p + 31.4540a2p^7.760^0.0027^15SP^9^2.4371^0.2748 0-2+ 7.86340.2wpm + 15.7270a2s*p^ 0.604^0.7783^0Bl*P(S) 18^8.2361^0.4576 G2 + 7.9526G2Bpp(s)^ 1.816^0.0232^7Error^281 70.7883^0.2519 G2^ 65Legend: B1-=B1ockSSiteR--ProvenanceSatterthwaite Approximations Used to Derive F Values:Source: SiteError: 0.9942*MS[B1(S)] + 0.9879*MS(S*P) - 0.9768*MS[B1*P(S)] - 0.0053*MS(Error)DF^MS^Denominator DF Denominator MS^F Value P > F1^7.5938 1.30^0.8198^9.263 0.1534Table 9. (con't)Satterthwaite Approximations Used to Derive F Values:Source: Block(Site)Error: 0.9825*MS[Bl*P(S).] + 0.0175*MS(Error)DF^MS^Denominator DF Denominator MS F Value P> F2^1.0024 18.35 0.4540 2.208 0.1383Source: ProvError: MS(Site*Prov)DF^MS^Denominator DF Denominator MS F Value P> F9^2.1324 9 0.2748 7.760 0.0027Source: Site*ProvError: 0.9888*MS[B1*P(S)] + 0.0112*MS(Error)DF^MS^Denominator DF Denominator MS F Value P > F9^0.2748 18.23 0.4553 0.604 0.778307,10\u00E2\u0080\u009424\u00E2\u0080\u009425\u00E2\u0080\u009426\u00E2\u0080\u00942770 Mile House Holden LakeSITE59PRO V104PRO V100PRO V71PRO V69PRO V61PRO V57PRO V44PRO V30PRO V13PRO ViFig. 13. Changes in provenance 513C least squares means between two sites \u00E2\u0080\u0094 70Mile House and Holden Lake.Table 10. Summary of test results for differences in 513C among ten provenances over two sites(70 Mile House, Holden Lake).Provenance^61 104 100 44 13 71 1 30 69 57LSM^-24.398 -24.509 -24.678 -24.699 -24.782 -24.812 -24.860 -25.003 -25.076 -25.280Note: Underscores show homogeneous subsets (comparisonwise error rate=0.001).Fisher's Least Significant Difference=0.625 0/00.61provenance 69. Provenance 61 was also different from provenance 30.Table 11 summarizes ANOVA results for the data from Holden Lake and SalmonLake (all factors random), where provenance 95 and all other provenances except 57 weretested. The variances of main effects were homogeneous (Bartlett's test). The analysis used315 observations from a total of 360 (model r2 = 0.523). Site, block (within site) andprovenance effects were significant (a=0.05); block x provenance (within site) and site xprovenance effects were not. Provenance 513C trends with site are graphed in Figure 14.The Bonferroni procedure was applied to pairwise comparisons between blocks(within sites). Although the F test found blocks different, pairwise comparisons did not.Huitema (1980) explains this apparent inconsistency. Frequency of false rejections of thenull hypothesis is less with multiple comparison procedures than with the F test (Zar 1984).Variance components showed that site and sampling error contributed most to thevariation in 813C per sapling (43.5 and 41.5% respectively). Provenance contributed 14%.Since the site x provenance and block x provenance (within site) components contributednothing to the variation, sampling error became the test term for means differences amongprovenances. Table 12 depicts means test results. The LSD was 0.399 Woo. Provenance 95was different from all provenances except provenance 61. Provenances 61, 104 and 100were different from provenances 30 and 69. As well, provenance 61 was different fromprovenances 1, 13 and 44. The Tukey test gave nearly the same result as Fisher's LSD test,the only difference being that provenance 61 was different from provenance 71 as well asprovenances 1, 13 and 44.Correlations between growth parameters (diameter, height, volume) and 813C weresite- and provenance-dependent. (Growth measurements were not available for provenance95). Significant correlations (a=0.05) are discussed below. If a reflected the number ofcorrelations tested (87), none would be significant (the \"p\" value required for a significantpairwise correlation would be 0.0006). All correlations are listed in Appendix 1.Table 11. ANOVA for provenance trials at Holden Lake and Salmon Lake. Variance components are presented as a percentage of totalcomponents.Source df^SS^MS^ EMS^ F Value^P > F^VC(%)Site^1 41.0215 41.0215 a2+ 7.3049,52wpm + 14.6100a2s*p + 73.0490a2m(s)+ 14612s 63.954^0.0175^43.5Bl(S)^2^1.2993^0.6497 a2+ 7.30580281.m + 73.058002ms)^ 3.469^0.0501^1Prov^9 26.3096^2.9233 a2+ 7.5547a2E1*p(S) + 15.1090a2s.p + 30.2190a2p^16.594^0.0001^14S*P^9^1.5855^0.1762 a2+ 7.5547a2wp(s)+ 15.1090a2s*p^ 0.954^0.5047^0B1*P(S) 18^3.2983^0.1832 a2 + 7.6880a2Eop(S)^ 0.692^0.8185^0Error^275 72.8135^0.2648 a2^ 41.5Legend: BlaBlockSaSite1)--ProvenanceSatterthwaite Approximations Used to Derive F Values:Source: SiteError: 0.9999*MS[B1(S)] + 0.9669*MS(S*P) - 0.9502*MS[B1*P(S)] - 0.0166*MS(Error)DF^MS^Denominator DF Denominator MS^F Value P > F1^41.0215 1.91^0.6414^63.954 0.0175Table 11. (con't)Satterthwaite Approximations Used to Derive F Values:Source: Block(Site)Error: 0.9503*MS[Bl*P(S)] + 0.0497*MS(Error)DF^MS^Denominator DF Denominator MS^F Value P> F2^0.6497 20.82^0.1873^3.469 0.0501Source: ProvError: MS(Site*Prov)DF^MS^Denominator DF Denominator MS^F Value P > F9^2.9233 9^0.1762^16.594 0.0001Source: Site*ProvError: 0.9827*MS[Bl*P(S)] + 0.0173*MS(Error)DF^MS^Denominator DF Denominator MS^F Value P> F9^0.1762 18.93^0.1846^0.954 0.5047-24-2764Holden Lake Salmon LakeSITE- - - PROV104PROV100PRO V95PRO V71PROV69PRO V61PRO V44PRO V30- PRO V13PRO ViFig. 14. Changes in provenance 813C least squares means between two sites \u00E2\u0080\u0094Holden Lake and Salmon Lake.Table 12. Summary of test results for differences in 813C among ten provenances over two sites(Holden Lake, Salmon Lake).Provenance^95 61 104 100 71 44 13 1 30 69LSM^-24.508 -24.873 -25.107 -25.152 -25.271 -25.276 -25.312 -25.357 -25.629 -25.638Note: Underscores show homogeneous subsets (comparisonwise error rate=0.001).Fishers Least Significant Difference=0.399 o/oo\u00E2\u0080\u00A266At 70 Mile House, all growth variables and 813C were negatively correlated inprovenance 71 (diameter: r=-0.645, p=0.007; height: r=-0.720, p=0.002; volume: r=-0.679,p=0.004). Provenance 104 showed positive correlations between all growth variables and813C (diameter: r=0.574, p=0.013; height: r=0.540, p=0.021; volume: r=0.547, p=0.019). Itappeared that in provenance 71, decreased WUE (measured relatively by 813C) wasaccompanied by increased growth, while the opposite occurred in provenance 104.Interpretations had to be site specific, however. Provenance 71 showed positive correlationsbetween growth and 813C at Holden Lake and Salmon Lake (non-significant).At Holden Lake, height and 813C were positively correlated in provenance 30(r=0.463, p=0.053). All growth parameters correlated positively with 813C in provenance 44(diameter: r=0.512, p=0.043; height: r=0.681, p=0.004; volume: r=0.554, p=0.026), as wellas in provenance 57 (diameter: r=0.745, p=0.001; height: r=0.475, p=0.054; volume:1=0.671, p=0.003). At Holden Lake, it appeared that growth improved with increased WUEin these provenances.At Salmon Lake, diameter and volume correlated negatively with 813C in provenance61 (diameter: r=-0.652, p=0.008; volume: r=-0.596, p=0.019). All growth parameterscorrelated positively with 513C in provenance 30 (diameter: r=0.547, p=0.019; height:r=0.650, p=0.004; volume: r=0.556, p=0.017), as well as in provenance 44 (diameter:r=0.600, p=0.018; height: r=0.699, p=0.004; volume: r=0.672, p=0.006). Decreased WUEwas accompanied by increased growth in provenance 61, while the opposite occurred inprovenances 30 and 44.Provenances 13, 69 and 100 demonstrated no significant relationships betweengrowth factors and 813C at any site.Only one provenance had significant relationships between all growth variables and813C on more than one site. In provenance 44, diameter, height and volume were positivelycorrelated with 813C at Holden Lake and Salmon Lake.67Of interest was whether provenances with more positive 813C values (assumed to bemore water-use efficient) produced higher volumes. Per site, correlations betweenprovenance 813C LSMs and volume LSMs were positive: 1. 70 Mile House (r=0.608,p=0.062, n=10); 2. Holden Lake (r=0.759, p=0.011, n=10); 3. Salmon Lake (r=0.689,p=0.040, n=9). Considering average performance over three sites, the correlation between813C LSMs and volume LSMs was 0.727 (p=0.027, n=9).At the provenance level there were significant correlations (a=0.1) between 813C andannual climate variables at all sites. At 70 Mile House, 813C correlated with meantemperature of the warmest month (mtwm) (r=-0.794). At Holden Lake, 813C correlated withmtwm (r=-0.667), mean summer precipitation (msp) (r=0.737), mean precipitation of thedriest month (mpdm) (r.715) and mean annual precipitation (map) (r=0.730). The coastalprovenance (95) had large influence at this site. Removing provenance 95 eliminated thesignificant correlations between isotope ratios and precipitation variables. At Salmon Lake,813C correlated with mtwm (r=-0.786), mean annual temperature (mat) (r=0.739), msp(r=0.750), mpdm (r=0.783) and map (r=0.753). Removing provenance 95 eliminated thesignificant correlations between 813C and the precipitation variables.At the provenance level, isotopic composition was related to geographic variables attwo sites. At Holden Lake, the correlation between 813C and longitude was 0.822 (p=0.023)while that between 813C and elevation was -0.713 (p=0.072). At Salmon Lake 813Ccorrelated with elevation (r=-0.711, p=0.114). These correlations were high (a=0.1) due tothe influence of provenance 95.Since isotope ratios were influenced by several climate variables, regressionmodelling determined the factors contributing most to the prediction of 813C. Thegeographic variables were not applied to the regression equations, since longitude andelevation were highly correlated with precipitation. Latitude was highly correlated withmean annual temperature. As well, longitude, latitude and elevation were highly correlatedwith each other. Initially a common model was sought for all sites. One model including68mtwm and msp was appropriate for sites with provenance 95, while another including matand msp was suited for 70 Mile House and the other sites without provenance 95. Thecomparison was necessary, since provenance 95 had excessive influence.Collinearity problems were small for the climate factors. Per site, mtwm and msp (atthe provenance level) had a correlation of about 0.318 (p<0.05). The correlation becamenegative when provenance 95 was included. The r value for mat and msp was about 0.311(p<0.05) without provenance 95. Including 95 increased the correlation to 0.477 at HoldenLake and 0.696 at Salmon Lake (both at p<0.05).To find models suitable for the complete set of data points (with and withoutprovenance 95), indicator variables representing site (Si, S2) were included: Si=0, S2=0 for70 Mile House; S1=1, S2=0 for Holden Lake; and S1=0, 52=1 for Salmon Lake. Theadvantage of having replication with regression was the opportunity to test for lack of fit ofthe models (Ott 1988; Bergerud 1990). The best fitting regression model for the data setincluding provenance 95 was 813C = -22.607 - 0.365S1 - 1.133S2 - 0.158MTWM +0.001MSP (r2=0.450; p<<0.0001; df=5, 284). The lack-of-fit test was done by separating theresidual sums of squares into two parts: a lack-of-fit sums of squares (df=14) and a withingroup sums of squares called the pure error (df=271) (see Appendix 3). The test for lack-of-fit was significant (a=0.05); the model was inadequate for predictive purposes. However themodel was useful in describing a relationship between 513C, temperature and precipitation.The best fitting model for the data set without provenance 95 was 513C = -27.970 - 0.380S1 -1.103S2 + 1.756MAT + 0.015MSP - 0.007MAT*MSP (r2=0.476; p<<0.0001; df=5, 258).The test for lack-of-fit was not significant (Appendix 3), indicating that the model wasadequate for predictive purposes. Standardized coefficients from the regression outputshowed that the mat variable had six times the influence on 513C as the msp variable and thatthe interaction between the two had the greatest influence. The equation showed that theslope of the relationship between 813C and mat depends on msp and vice versa. The twovariables interact to negatively affect isotopic composition. For example, depending on what69the msp is, WUE can increase or decrease with increasing temperature. The equationappropriate to 70 Mile House shows that the change in mean 813C per \u00C2\u00B0C increase in mat isgiven by 1.756 - 0.007MSP. For 300 mm msp, the estimated change in 513C per \u00C2\u00B0C increasein mat is -0.344 Woo. For 150 mm msp, the estimated change in 4513C per \u00C2\u00B0C increase in matis +0.706 0/00. Under dry conditions 813C is more positive, indicating higher intrinsic WUE.It is apparent also that as mat decreases, WUE will increase under conditions of increasingmoisture.4.3 Seedling ExperimentSummary statistics (Table 13) show little variation among provenance biomass meansother than shoot/root ratio. The range for shoot weight was 0.186 g; for root weight, 0.076 g;for shoot+root weight, 0.207 g; and for shoot/root ratio, 0.448. The range for 813C was 1.648woo.Shoot variances were heterogeneous (Bartlett x2=27.985, 1)=0.002; Cochran's Cstatistic 0.1947 > Cochran's critical value 0.1678). These heterogeneities were not remediedby transformations (log, square root, 1/3,2, y2). Residuals analysis showed nine outliers.Removing one (1.316 g, prov. 95) decreased variance heterogeneity (Bartlett p=0.01;Cochran's C statistic 0.1663 < Cochran's critical value 0.1678). Since non-parametric(Kruskal-Wallis) analysis of the complete data set gave the same result as ANOVA on thedata set, minus the outlier, no further changes were made to shoot data prior to the F test. Aswell, the outlier was removed from the remaining data sets. Table 14 shows ANOVA resultsfor seedling data. Shoot means were significantly different (a=0.05, p=0.001). Tukey's test(Table 15) found provenance 30, with lowest shoot weight, different from provenances 61, 1,13, 44 and 57. Provenances 69, 71, 104, 95 and 100 formed subsets with provenances 61, 1,13, 44 and 57, as well as provenance 30.Root variances were heterogeneous (Bartlett x2=19.170, p=0.038). The Cochran test,less sensitive to non-normality than Bartlett's test, showed homogeneity (Cochran's CTable 13. Summary statistics for growth parameters of eleven Pinus contorta seedling provenances.1 13 30 44Provenance57^61 69 71 95a 100 104Shoot Weight (g)Mean^0.753 0.753 0.575 0.741 0.740 0.761 0.673 0.654 0.622 0.620 0.654SD^0.189 0.216 0.172 0.249 0.141 0.189 0.158 0.174 0.276 0.180 0.224Root Weight (g)Mean^0.262 0.257 0.247 0.228 0.257 0.235 0.257 0.253 0.186 0.224 0.209S D^0.058 0.078 0.052 0.076 0.061 0.085 0.067 0.075 0.096 0.080 0.089Shoot + Root Weight (g)Mean^1.015 1.010 0.821 0.969 0.997 0.995 0.930 0.907 0.808 0.844 0.863S D^0.220 0.274 0.205 0.314 0.174 0.254 0.199 0.225 0.358 0.246 0.298Shoot I Root RatioMean^2.964 3.056 2.348 3.280 2.984 3.494 2.727 2.756 3.703 2.936 3.391S D^0.736 0.748 0.596 0.655 0.637 0.978 0.760 0.841 1.066 0.678 0.935&BCMean -31.628 -31.861 -31.939 -31.660 -31.870 -31.630 -31.966 -31.882 -30.318 -31.698 -31.803S D 0.868 0.627 0.628 0.848 0.559 0.806 0.647 0.905 0.943 0.809 0.852an=29 from a maximum of 30Table 14. One-way ANOVAs on growth parameters for Pinus contorta seedling data.Source df SS MS F Value P > FShoot WeightBetween Prov 10 1.304 0.130 3.246 0.0010Within Prov 318 12.776 0.040Root WeightBetween Prov 10 0.172 0.017 3.044 0.0010Within Prov 318 1.800 0.006Shoot + Root WeightBetween Prov 10 1.860 0.186 2.824 0.0020Within Prov 318 20.949 0.066Shoot/Root Ratio (Natural Logarithms)Between Prov 10 4.895 0.490 7.525 <<0.0001Within Prov 318 20.688 0.065813CBetween Prov 10 62.058 6.206 10.168 <<0.0001Within Prov 318 194.093 0.610Table 15. Summary of Tukey test results for means differences among seedling shoot weights (g).Prov 61 1 13 44 57 69 71 104 95 100 30Mean 0.761 0.753 0.753 0.741 0.740 0.673 0.654 0.654 0.645 0.620 0.575Note: Underscores show homogeneous subsets (experimentwise error rate=0.05).73statistic 0.1429 < Cochran's critical value 0.1678). Therefore, data were not changed prior toANOVA, which gave a significant result for root means (p=0.001). Provenance 95, withlowest root weight, was different from provenances 1, 13, 57, 69 and 71 (Table 16).Shoot+root variances were heterogeneous (Bartlett x2=26.703, p=0.003; Cochran's Cstatistic 0.1763 > Cochran's critical value 0.1678). Transforming data or removing twoextreme outliers did not correct the problem. Since non-parametric analysis produced thesame result as ANOVA, the data were not changed. Although ANOVA gave a significantresult for shoot+root growth (p4).002), Tukey's test revealed no means differences.Shoot/root ratio variances were heterogeneous (Bartlett x2=20.784, p=0.023).Cochran's C statistic, 0.1623, was less than Cochran's critical value, 0.1678, but thedifference was too small to discount a heterogeneity problem. Homogeneity was achievedby natural log transformation of the data (Bartlett p=0.858). The F test was significant(p\u00C2\u00AB0.0001). The Tukey test (Table 17) showed that provenances 95, 61 and 104 weredifferent from provenances 71, 69 and 30. As well, provenance 95 was different fromprovenances 1 and 100.Isotope variances were homogeneous (Bartlett x2=17.225, p=0.070) and significantlydifferent (p<<0.0001). Tukey's test (Table 18) found two subsets: (provenance 95) and(provenances 1, 13, 30, 44, 57, 61, 69, 71, 100, 104).Correlations between growth variables of all provenances' pooled seedlings (n=329)and 813C were poor: 1. Shoot mass (r=0.014, p=0.807); 2. Root mass (r=0.041, p=0.464); 3.Shoot+Root (r=0.023, p=0.684); 4. Shoot/Root ratio (r4.016, p=0.772). Individually, fewprovenances showed significant correlations (a.05) between growth variables and 813C(see Appendix 2). If a reflected the number of correlations tested (44), none would besignificant (the \"p\" value required for a significant correlation would be 0.001). Shootweight, as well as shoot/root ratio and 613C correlated significantly in provenance 57 (r=-0.364, p=0.048 and r=-0.369, p=0.045 respectively). The shoot/root ratio and 813Ccorrelation in provenance 69 was r=-0.467 (p=0.009). Root weight, as well asTable 16. Summary of Tukey test results for means differences among seedling root weights (g).Pray 1 13 57 69 71 30 61 44 100 104 95Mean 0.262 0.257 0.257 0.257 0.253 0.247 0.235 0.228 0.224 0.209 0.189Note: Underscores show homogeneous subsets (experimentwise error rate=0.05).Table 17. Summary of Tukey test results for means differences among seedling shoot/root ratios (natural logarithms).Prov 95 61 104 44 13 57 1 100 71 69 30Mean 1.270 1.214 1.185 1.168 1.087 1.070 1.057 1.051 0.974 0.969 0.822Note: Underscores show homogeneous subsets (experimentwise error rate=0.05).Table 18. Summary of Tukey test results for means differences among seedling 813C values.Prov 95 1 61 44 100 104 13 57 71 30 69Mean -30.318 -31.628 -31.630 -31.660 -31.698 -31.803 -31.861 -31.870 -31.882 -31.939 -31.966Note: Underscores show homogeneous subsets (experimentwise error rate=0.05).77shoot/root ratio, and 813C correlations were significant in provenance 71 (r=0.405, p=0.026and r=-0.410, p=0.024 respectively). Correlations were positive as well as negative,depending on provenance and growth variable tested. All provenances showed positivecorrelations between root weight and isotope ratio. All but provenances 61 and 104 showednegative correlations between shoot/root ratio and 513C. Correlations between populationaverages of 513C and root weight, as well as shoot/root ratio were significant (r=-0.754,p=0.007 and r=0.680, p=0.021 respectively). Interestingly, the correlations between 513Cand biomass variables (root, shoot/root) for seedlings within populations were opposite insign to correlations using seedling population means. Correlations between populationaverages of 513C and shoot weight, as well as total weight were insignificant (r=-0.188,p=0.580 and r=-0.386, p=0.241 respectively).Of interest was the correlation between seedling and sapling 513C. Over all sites(minus provenances 57 and 95), seedling 513C means and sapling 513C LSMs correlated by0.690 (p=0.040, n=9) (Fig. 15). At Holden Lake, where all provenances were tested, requalled 0.781 (p=0.005, n=11) (Fig. 16). At Salmon Lake and 70 Mile House thecorrelations were 0.812 (p=0.004, n=10) and 0.591 (p=0.072, n=10), respectively.Spearman's rank correlations did not improve any relationships other than the finding for 70\u00E2\u0080\u00A2Mile House, where the correlation coefficient became 0.673 (significant at a=0.05).780c0-31.0\u00E2\u0080\u009431.2\u00E2\u0080\u009431.4\u00E2\u0080\u009431.6PT-1V.T4\u00E2\u0080\u009431.8\u00E2\u0080\u009432.0\u00E2\u0080\u00A2 \u00E2\u0080\u00A2\u00E2\u0080\u00A2 \u00E2\u0080\u00A2\u00E2\u0080\u00A2\u00E2\u0080\u00A2 \u00E2\u0080\u00A2\u00E2\u0080\u009425.5^\u00E2\u0080\u009425.3^\u00E2\u0080\u009425.1^\u00E2\u0080\u009424.9^\u00E2\u0080\u009424.7^\u00E2\u0080\u009424.5SAPLING 8'3C rajFig. 15. Pinus contorta greenhouse seedling 513C means plotted against saplingleast squares means averaged over three sites \u00E2\u0080\u0094 70 Mile House, Holden Lake andSalmon Lake. Provenances 95 and 57 are not included. Correlation coefficientr=0.690 (p=0.040).\u00E2\u0080\u00A21\u00E2\u0080\u00A2\u00E2\u0080\u00A279-30.0\u00E2\u0080\u009430.5\u00E2\u0080\u009431.0\"c4.0:11 \u00E2\u0080\u009431.54.\u00E2\u0096\u00AA 1\u00E2\u0080\u009432.0\u00E2\u0080\u009425.5^\u00E2\u0080\u009425.0^\u00E2\u0080\u009424.5^\u00E2\u0080\u009424.0\u00E2\u0080\u009432.5\u00E2\u0080\u009426.0 \u00E2\u0080\u009423.5SAPLING 4513C (\u00C2\u00B0/00)Fig. 16. Pinus contorta greenhouse seedling 813C means plotted against saplingleast squares means at Holden Lake (where all provenances were tested).Correlation coefficient r=0.781 (p=0.005). The outlier is provenance 95.805.0 DiscussionThis study is based on the assumption that 513C correlates with WUE in trees.Traditional measures of WUE in agricultural crops have been significantly correlated with513C, but these calibrations have not been done for trees. Recently, however, instantaneousWUE (A/E) and intrinsic WUE (A/g) have been positively correlated with 513C in Pinusseedlings and tropical forest shrubs (Elsik eta!. 1992; Jackson eta!. 1992).5.1 Determination of Sampling Methods for Field TrialsIntra- and inter-sapling 513C variation has not been reported. My data show thatcarbon isotope trends in saplings generally follow those for mature trees. This suggests thatsampling methods developed for mature trees may be applied to younger trees. However,saplings usually show less 513C variation within and among individuals. The results ofFreyer and Belacy (1983) and Francey (1981) suggest there is little circumferential variationin the early decades of growth, prior to crown enlargement and the development ofassociated microenvironmental differences.Around the trunk, the mean stump wood variance within saplings is 0.09 Woo (1989samples, n=5 trees). At 1.2 m, the variance is 0.09 Woo. For needles, the mean intra-saplingvariance (lowest branches) is 0.68 Woo (1989 samples, n=5 trees. The circumferentialvariance for 513C in wood of mature trees has not been reported, although a range of 0.5 to1.2 Woo at 1.5 to 2.5 m height is given for Pinus edulis (Leavitt and Long 1986). In thePinus contorta saplings, at about the same height, the range is 0.03 to 1.10 Woo. For needlesof mature trees, Leavitt and Long (1986) cite studies showing variances of 0.38 to 0.46 Woo inJuniperus species and 0.38 Woo in Pinus edulis.For P. contorta saplings, the intra-site variance in stump wood (1989) is 0.12 Woo(four observations per tree, n=5 trees). The variance in needles (lowest branch) is 0.26 Woo(four observations per tree, n=5 trees). Intra-site variance in wood of mature trees has notbeen reported. However, the range for 513C among eight P. edulis has been given as 2.0 to813.0 Woo (Leavitt and Long 1984). Each tree was represented by a pooled sample of fourorthogonal cores comprised of seventeen 5-year ring groups. The range of 513C amongsaplings (16 data points per tree representing four orthogonal cores) is 0.34 to 0.98 Woo. Thevariance in needles for mature Juniperus species (two to four trees per site) is given as 0.38Voo (Arnold 1979, cited in Leavitt and Long 1986).To find a sampling technique to accurately reflect field 513C, I used a statisticalmethod lacking rigor. There were no replicates within years, making the error termirretrievable in all analyses of variance. However, sufficient information was gained about513C variation within and among the saplings at Juliet Creek. The information was adequateto select a technique lending itself to less sampling error than alternative ones (e.g., takingeast-west cores or using needles instead of wood as the sampling material). The chosen fieldsampling technique was to obtain north-south wood stem cores.In Table 3, the assumption of non-significance for the three-way interaction used asthe error term is questionable. However, figures 4 and 5 validate the F tests determined bythat error term. In Tables 4 and 5, the three-way interactions are probably insignificant. Thetwo-way interactions comprising them are insignificant in all but one case; therefore, the Ftests are considered valid.In the saplings, the tree 513C effect (stump wood) as well as tree x aspect interactionswere significant. This finding supported Leavitt's and Long's (1984) recommendation that atleast four orthogonal cores from each of four trees be sampled to accurately represent site513C values.To determine what influenced the tree 513C differences, I tested correlations betweentree 513C and features which might contribute to the variation. Isotopic compositions forwood based on total observations per tree (n=16) had no significant correlations with height,age, percentage canopy cover, number of neighbors within a 3 m radius or average proximityto neighbors. The same lack of correlation was found for 513C means from north and southaspects combined (n=8). Isotopic compositions in both cases were positively correlated with82diameter (for all data points, r=0.891, p=0.043, n=16; for north and south cores, r=0.873,p=0.053, n=8). Needle 513C, derived from averaging 1989 north and south values at stumplevel correlated positively with diameter (r=0.937, p=0.019), height (r=0.941, p=0.017) andaverage proximity to neighbors (r=0.918, p=0.028). From the average of fourcircumferential values (1989), needle 513C correlated positively with proximity to neighbors(r=0.983, p=0.003). Needle 513C at stump level is related to certain features, whereas wood513C is not. The difference may be artificial since wood 513C is estimated from more datapoints than needle 513C. Using wood samples from 1989 only, as was done for needles,shows a significant negative correlation between circumferential wood 513C and percentagecanopy cover (1..41932, p=0.021). The correlations between needle 513C and tree height, aswell as proximity to neighbors, and that between wood 513C and percentage canopy covermay reflect a shading effect. In 1984, Leavitt and Long compared 8 13C trends over 5 yearintervals in mature P. edulis. Using four orthogonal cores per tree, they found norelationship between absolute 513C among trees and age, height, diameter, crown shapes orproximity to neighbors.Tree x aspect effects may have been due to variable shading from neighbors.Saplings had neighbors of different crowns and heights as well as at varying distances anddirections. Measuring duration or intensity of irradiance at each aspect over the growingseason may have illuminated the \"variable shading - variable 513C\" hypothesis. However,the assumption would have had to be a relatively constant irradiance difference amongaspects (within and among trees) over many years. Trees 3 and 4 contributed most to theinteraction effect. They had very suppressed growth relative to the others. Many of theirlower branches had been browsed. Trees 3 and 4 had taller neighbors to the east-south-eastthan the others and therefore, less estimated open canopy. Their south sides were morenegative in 513C than their north sides, a trend opposite to that in the others.Stomatal and photosynthetic effects influence the relationship between 513C and plantirradiance. The sunny side of a plant is likely to be more drought-stressed than the shady83side. Leaf stomates will remain more closed in the sun than in the shade relative tophotosynthetic rate. Intercellular pi/pa in the sunlit leaves will be less than that for shadedleaves. With a lower pi/pa, discrimination against 13C by photosynthetic enzymes will be lessthan that occurring at a higher pi/pa. Increased 813C, positively related to intrinsic WUE, willbe found in leaf samples from the sunlit side. Also, decreased illumination of a plant willresult in decreased photosynthetic capacity. If photosynthetic capacity is relatively higher onthe illuminated side (stomatal control being constant), pi/pa will decrease in the intercellularspaces of sunlit leaves. Leaf samples from more illuminated sides of the plant will recordhigher 513C values than those from less illuminated areas.The saplings at Juliet Creek were relatively open-growing. There appeared to be nosoil respiratory CO2 effect on needle 813C at stump level (Table 5). It was likely, therefore,that wood 513C accurately reflects the physiological functioning of the trees rather thanwithin-canopy gradients in 13CO2. This assumption was applied to the field trials, wheresaplings were also relatively open-grown.It appears that circumferential 813C in sapling stump wood follows no particular trendwith aspect whereas 513C in leaves changes more obviously with aspect. If photosynthatesare laid down spirally in tree trunks, they may reflect 813C values of carbon fixed in leaves atdifferent sides of the tree. Some degree of spiral grain is a normal feature of wood formationin trees (Harris 1989). A spiral consistent in direction and degree throughout a stem isseldom seen. Leavitt's and Long's finding (1982) that needles and wood followed similar813C trends on a monthly basis during one growing season were based on samples from oneaspect only. An explanation for the different needle and wood 513C trends might beSchleser's (1992) finding that large contributions of carbon lower in the tree may originatefrom the top of the tree. Since a 813C gradient in leaves with height is not reflected in trunkwood (Schleser 1992), it is expected that circumferential values of 813C for needles andwood at nearly the same height will not correlate well.84The accuracy of the F test for the aspect effect (wood, Table 3) can be questionedsince one of the test terms (tree x aspect) is significant. Figure 8 supports the finding of anon-significant aspect effect by showing similar trends for 813C values in wood over aspects.Analyses of variance on the tree, aspect and material effects (Table 4) and on the aspect,position and material effects in one year (Table 5) support the finding of a non-significantaspect effect for wood 513C as well.Leavitt and Long (1984) recommend a reliable sampling technique for tree-ringisotope studies reconstructing climate and past atmospheric CO2. In studies like this one,however, extensive sampling may not be possible or necessary. The sampling protocol Iestablished for P. contorta saplings may not apply to all species and environments but isreasonable for population studies on young pines. Wood is a more appropriate samplingmaterial than needles since 813C variation within and among trees is smaller in wood than inneedles. Some studies have used needles from tree tops but in this study the inter-sapling513C variance shown by top needles (1989) was 0.83 Woo (one data point per sapling),compared with a variance of 0.13 oho for 1989 stump wood (north and south aspectsaveraged per sapling).5.2 Determination of Genetic Variation in WUE Among Sapling ProvenancesThere are genetic differences in relative WUE (determined by 513C) among Pinuscontorta populations. This genetic variation validates Rehfeldt's premise that P. contortapopulations are adapted to small parts of the species' range.Geography appears to have little influence on WUE (Fig. 17). Differences in WUEappear related to climate of origin. Leavitt and Long (1983) related 513C in Pinta edulis totemperature. Read and Farquhar (1991) found a significant negative correlation between Aand precipitation, as well as temperature for Nothofagus species. Since temperature andprecipitation were highly correlated (>0.6), they could not regress the two against A.However, they found the negative relationship between A and summer rainfall ecologicallyFig. 17. Geographical trends for 813C in Pinus contorta in British Columbia andAlberta. The 813C values are the least squares means obtained for Holden Lake,where all provenances were tested. The positions of the 813C values on the mapcorrespond to provenance locations.86meaningful since, for some species, the rainfall totals during the dry months exceeded theevaporative load and much soil water remained. This led Read and Farquhar (1991) tosuggest that the high discrimination of species in drier climates with marked seasonaltemperatures resulted from adaptations allowing continued photosynthesis during warm butrelatively dry periods.The regression model for the complete data set in this study shows a similar trendbetween A and summer rainfall as found by Read and Farquhar (1991) (temperatureremaining constant). But higher seasonal temperatures (rainfall constant) give lower 813Cvalues or higher A, the opposite trend found by Read and Farquhar (1991). Read andFarquhar (1991) used a wider temperature range than was used in this study. If bothtemperature and summer rainfall are varied to estimate 813C, the model for this studyindicates that increased temperatures with lower rainfall give lower 813C values. This modelsupports Read's and Farquhar's (1991) suggestion that the relationship between A andtemperature is likely influenced by rainfall regimes. As well, indications are that populationshave adapted to continue photosynthesis during warm but relatively dry periods. The dataare poorly fitted to this model, however, and interpretations are inconclusive.Modelling the relationship between climate and 813C without the coastal influencegives a model with adequate fit. The model predicts that under low summer moistureconditions WUE improves with increasing annual temperature, as expected in dryenvironments. In this case increased WUE may be due to selection for lower stomatalconductance in response to increasing vapor pressure deficit. Less water will be transpiredrelative to CO2 assimilated. Under cool annual temperatures WUE also improves (to a lesserextent) with increasing summer precipitation. Increased WUE may be due to selection forlower stomatal conductance in normally humid environments to avoid dessication inunexpected drought. This model implies that the evolution of improved intrinsic WUE in P.contorta var. latifolia populations in British Columbia is related to complex interactions87between temperature and precipitation and their effect on biophysical and physiologicalprocesses in trees.It appears that populations within the two varieties of the species have evolved highWUEs under very different conditions. The coastal provenance (var. contorta) maintains ahigh relative WUE similar to provenance 61 (var. latifolia) from the wet cool Sub-borealSpruce subzone. The subzone has less than half the precipitation of the coastal location anda slightly higher summer temperature (msp=345 mm, mtwm=14.3 \u00C2\u00B0C, mat=2.6 \u00C2\u00B0C)(Meidinger and Reynolds 1991). To better understand the relationship between climate and813C for the species we would require a class size spanning the species' range, as well asmore appropriate climatic and atmospheric estimates. Coastal provenances in areas ofsummer drought should be sampled and the relative WUEs of interior populations (e.g., 61,104 and 100) should be compared with coastal populations on very wet maritime test sites.Crucial origin site characteristics which would affect WUE (e.g., soil moisture,nutrients, vapor pressure deficit) were not available for regression analyses. Potentialevapotranspiration (PE) was not estimated with climate parameters since Thornthwaite'smethod (1948) applies only to regional analyses. Complex formulas requiring more datahave been devised for local studies. Ideal estimates of PE would require weighing lysimeters(Muller and Oberlander 1984). Atmospheric data such as solar radiation, humidity and windvelocity would have allowed rigorous estimates of PE.The most notable test site factor to elicit changes in 813C was moisture. The test siteswere in biogeoclimatic subzones distinguished mainly by precipitation (Table 1). HoldenLake and 70 Mile House were in the dry cool Interior Douglas-fir subzone while SalmonLake was in the moist cool Sub-boreal Spruce subzone. The drier subzone had highertemperatures than the wetter subzone. Factors such as tree leader damage, number of cones,degree of shading (particularly to the south), disease, girdling or browsing showed noobvious trends with 813C. More than one factor acted on most trees. Signs of severe disease,girdling or browsing were most useful in determining whether to eliminate outliers. Site88nutrient status may have been influential. Although mean 813C was not significantlydifferent between 70 Mile House and Holden Lake, the slightly more positive values at 70Mile House may have been related to its richer soil as well as decreased moisture.Differences in WUE among some populations can be readily linked to climate andgeography while others cannot. Provenance collections 100 and 104 have similar WUEs,differing from that of provenance collection 69. Provenances 100 and 104 are in closeproximity and have similar mean temperatures of the warmest month and summer rainfall.[As seedlings, both populations had 36 stomates mm-2 leaf area (Illingworth 1973). Stomatesare functional and likely to indicate an adaptive response]. Provenance 69 is located furthersouth and has a warmer climate with more precipitation. This population had 35 stomatesmm-2 in the seedling stage. [Mingworth (1973) measured stomatal densities on theprovenance seedlings when they were two years old. Stomatal counts were grouped andaveraged by broad geoclimatic region; e.g., the stomatal density for provenance 69 (in the'southern foothills') was the average of three provenances in that region. An important pointis that average stomatal counts for the provenance collections are likely to have changed overtime in response to environment as well as tree age]. In the context of scant site and climateinformation, provenance 30 appears to be an anomaly. Provenances 30 and 100 have almostidentical summer rainfalls whereas provenances 30 and 69 do not, yet provenances 30 and 69have almost identical WUEs. However provenances 30 and 69 have more similartemperatures (warmest month) than provenances 30 and 100. Provenances 30 and 69 likelygrow in harsh conditions since provenance 69 is a high elevation source while provenance 30is a far northern source. Some research indicates that provenances from the Yukon andnorthern British Columbia are genetically very different from the rest of the var. latifolia(Ying pers. comm. 1993). Morphologically, as seedlings, provenances 30 and 69 had 34 and35 stomates mm-2 leaf area, respectively (Mingworth 1973). The stomatal count forprovenance 100 was 36 mm-2.89In the field trials and seedling study, provenance 95's performance stands out. Thispopulation originates in a very wet coastal environment and yet is the most water-useefficient of the provenances. In the seedling stage, provenance 95 had high stomatal density(45 mm-2) Illingworth (1973) found a significant difference between stomatal densities ofcoastal and inland regions. One explanation for provenance 95's high WUE in differentenvironments may be that this provenance has very responsive stomates to moisture deficits.Relating leaf A to climate parameters in Nothofagus, Read and Farquhar (1991) found thelowest A values (or highest WUE) in species from regions without soil or air moisturedeficits. The reason for this was unclear; they offered KOrner's and Bannister's (1985)suggestion that stomatal sensitivity to humidity would be the most protective mechanismagainst sudden dessication in a normally humid environment. High WUE in wet climates,therefore, might reflect the costs associated with alternative mechanisms for avoidance ofwater stress. Provenance 95 appears to exemplify this.With adequate water, such as in the greenhouse, provenance 95's high WUE mayhave resulted from decreased stomatal aperture instead of increased photosynthetic capacity.In this case, high WUE restricted yield. On the other hand, provenance 61's high WUE mayhave resulted from increased photosynthetic capacity at a given stomatal conductance. Inthis case, high WUE in adequate water led to relatively high yield. This is conjecture, sincephysiological measurements which might have helped explain the differences in yieldassociated with high WUE were not done.In the field, increased 813C is not a reliable predictor of increased growth or yieldwithin P. contorta provenances. But correlating productivity at particular sites withpopulation averages (813C) shows significant correlations between high productivity and highWUE. Carbon isotopic composition may have practical use as a predictor of site yield.Selecting the most water-use efficient individual may not be as important as selecting themost water-use efficient population where long term productivity is concerned. In a P.contorta plantation water is shared among individuals. It is assumed that the more water-use90efficient trees will leave water for those requiring more moisture. Overall growth will bepromoted by the faster growing less water-use efficient individuals. One assumptionunderlying this premise is that reproduction on the plantation prior to harvest will not occur,so the genetic composition of the population will not change. Pinus contorta seedlings donot establish under the canopy. The other assumption is that there will not be an abundanceof other vegetation to compete for the water left by the more water-use efficient pines. In thecase where only the most water-use efficient individuals of several populations are selected,the plantation may or may not ultimately produce a higher yield since high WUE is notalways correlated with high yield in individuals.Stronger relationships between 813C and yield in saplings may have been gained byusing test sites close to the seed sources studied or sites favorable to the provenances'phenological and photoperiodic traits. Some populations were stunted or growing erraticallyon the interior test sites. Provenance 13 from the coast-interior transition region (Ying et al.1989) was very susceptible to frost and foliage disease in the interior environment (Yingpers. comm. 1993). The fast growing populations 44 and 57 (Ying et al. 1985) showedsignificant positive correlations between 8 1 3 C and yield at Holden Lake but different resultsat 70 Mile House. Repeated frost injury to trees in Block 51 at 70 Mile House resulted in anegative correlation between 813C and yield in population 57 (insignificant) while a smallpositive correlation (insignificant) remained in population 44. Fast growing populations areparticularly susceptible to frost injury (Ying pers. comm. 1993).Pest problems may have contributed to the variable correlations between 8 13C andyield. These factors could have increased sampling errors in growth measurements. AtSalmon Lake about 23% of the trees were damaged by porcupines and small mammals. Thevery different correlations between 813C and yield for populations 61 and 104 at this site mayhave been largely due to the severe porcupine damage to population 104 (Ying pers. comm.1993). Although growth in population 44 was slowed by porcupine damage, there was arather high and significant correlation between 8 1 3 C and yield in this population. It appears91that significant, positive correlations can exist despite adverse circumstances. Explanationsremain elusive in these cases. Testing higher numbers of individuals per provenance mayhave given stronger correlations. Alternating provenance individuals systematically in linesor randomizing them throughout blocks may have prevented the rather selective destructionto certain plots.Growth under contrasting field temperature and moisture regimes has led to changesin ecotypic 813C. Evidence of a plastic reponse is shown by a significant site effect(Schlichting's definition 1986). Variation in 813C is similar among ecotypes; i.e., ranking ismaintained from site to site. The similarity in degree and direction of response hasimplications for studies trying to predict the species' response to rapid global warming.Woodwell (1989) suggests that by the middle of the next century, temperatures will rise by0.5 to 1.0 \u00C2\u00B0C per decade in middle and high latitudes. In this study, populations 61, 104, 100and 44 maintained relatively high WUEs as well as yields compared with other populationson climatically different field sites. This pattern could hold over all climates. Thisprediction would require validation by research involving a wide range of climates onreplicated sites. To most accurately assess the species' plasticity in unpredictable conditions,rapidity of response would have to be measured. Response time may differ betweenseedlings and saplings. In the field, especially, short term variation in WUE may be moreeasily determined by gas exchange measurements or by measuring 813C of the leaf solublesugar fraction in needles.Reciprocal transplants could offer useful information about plasticity but anacclimation period following planting would have to be considered in interpreting theoutcome of physiological assessments. Previous site effects, especially drought, coulddetermine the metabolism of trees for several years following transplantation, thusconfounding assessments for plasticity of response to rapid changes in environment.In P. contorta, the relationship between high WUE and survival is not clear. Therelationship is confounded by many factors not measured in this study, e.g., herbivory. Over92all sites, provenances 69 and 30 have the lowest relative WUEs and also the highest and thirdhighest survival rates, respectively. Their performance is in contrast to Populus specieswhere high WUE has been shown to promote survival in moisture-stressed environments(Dickmann et al. 1992). Provenance 61 follows the trend in poplars. With the highest WUEover all sites, it has the highest survival as well.One aspect of the experimental design that could hinder finding true populationdifferences is that provenances were planted in plots rather than randomized throughoutblocks. Certain microsites may be associated with certain plots, making 4513C a measure ofmicrosite rather than provenance performance. This may have resulted in the significantblock x provenance interaction shown in the analysis of variance for 70 Mile House andHolden Lake (Table 9). Yield data supported this as well. Fewer population differenceswere found in this analysis than in the other two. It is also known that block 51 at 70 MileHouse was on a flat site prone to frost (Ying pers. comm. 1993) and would perhaps be bestanalyzed as a separate site. Due to the block x provenance effect, the finding of a non-significant site x provenance effect may not be valid. However, Figure 13 supports a non-significant site x provenance effect. The non-significant block effect can also be questioned.Examining block means with associated standard errors reveals that blocks are different at 70Mile House but not at Holden Lake.5.3 Seedling ExperimentOf 11 seedling populations representing all provenance collections tested, only thecoastal one is different in relative WUE. Favorable growth conditions may have nullifiedexpression of other genetic differences. Alternatively, measuring performance at 5 monthsmay have not allowed time for full expression of genetic differences. Plant developmentmay influence variation in 813C but little research has been done on trees.Read and Farquhar (1991) hypothesized that species in dry summer climates(Nothofagus) have evolved physiological and/or morphological mechanisms to allow93continued water uptake and high stomatal conductance during mild water deficits.Nothofagus seedlings with low WUE had lower shoot/root ratios than the more water-useefficient ones. Pinus contorta seedlings showed the opposite trend in nine out of elevenprovenances, similar to that for Aestivum triticum (Masle and Farquhar 1988). The morewater-use efficient seedlings partitioned more assimilate to roots than the less water-useefficient ones. The positive correlation between root growth and 813C for all provenancessupported this as well. However, provenances 61 and 104, relatively high in WUE,demonstrated the trend found by Read and Farquhar (1991). To accurately interpret theresults of 813C analyses, assimilate partitioning as well as background physiology must bemeasured. Seedling stomatal conductance and photosynthetic measurements were begun tohelp explain these differences but structural and mechanical problems with facilities andequipment precluded their completion.The correlation between 813C and root mass as well as shoot/root ratio for seedlingpopulations followed an opposite trend to that found within populations. For the most part,correlations within populations were very low and not statistically significant, meaning thatindividuals with high or low WUE performed essentially the same. Regardless, aninterpretation to explain the different trends among and within populations might be that thewithin population correlations between high root growth and high WUE largely reflectresponse plasticity to microenvironment while the correlation between high root growth andlow WUE among populations mainly reflects an evolved response to origin habitat. Theseedlings in the greenhouse were not competing for soil water but may have been affected byhumidity changes. The high elevation populations 69 and 71 had the highest within-population correlations between increased root growth and increased WUE. High WUE aswell as increased root growth would provide flexibility in adjusting to possible futuredrought. Either mechanism or a combination of the two could be developed as necessary.On a population basis, it seems reasonable to expect that population 95 would growproportionally the least root mass to shoot mass, having originated in a very wet coastal94climate. Population 61 from a wet, cool environment in the Sub-boreal Spruce zoneperforms similarly. Both have high intrinsic WUEs. The less water-use efficient populations69, 71 and 30 originating in harsh environments (high elevation or northern interior) allocateproportionally more biomass to roots than shoots. Accessing more soil water would be ameans to avoid dessication while achieving growth during a short growing season.Indications are that in seedlings, 513C may detect differences in allocation patterns morereadily than in total biomass.More conclusive relationships between 813C and yield in seedlings may have arisen ifthe seedlings had been grown over two or three seasons out of doors. The illumination in thegreenhouse was far from adequate for the establishment of P. contorta seedlings in theirnatural environment. They set bud earlier than expected, about one month before they wereharvested. This may have been due, in part, to a long photoperiod (18 hours) which has beenreported to sometimes cause early bud set in greenhouse-grown pines (Charleson, pers.comm. 1990).The good correlations between seedling 513C and that of saplings suggest that nurseryperformance may be a useful predictor of field performance. Although the correlations weresignificant, seedling rankings may have been random since only one provenance was clearlydifferent in 513C. Studies with older seedlings over several years under well-defined droughtconditions may give more sound results. Nursery shoot weights as well as total biomasswere very poor (and insignificant) predictors of sapling volumes. Indirectly, 513C ofseedlings shows more potential for predicting sapling yield than seedling growth variables.If seedlings offer a means to predict relative WUE of saplings, the seed may do so aswell. However, the seed would have to be collected from one site in one year. An importantpoint is that the 513C of the seed would give a more accurate measure of the mother plant'srelative WUE than seedling or sapling WUE.956.0 ConclusionsStable carbon isotope analysis has detected genetic variation in WUE amongpopulations of Pinus contorta saplings. The genetic differences are related to provenance(habitat of origin), temperature and precipitation. There are no genetic differences inplasticity among populations. These results suggest that P. contorta provenances mayrespond to environmental changes in British Columbia by maintaining their ranking inrelative WUE.The relationship of WUE to biomass increment in controlled environments as well asin nature, is not clear. High WUE is associated with low or high productivity, depending onthe population, its growing conditions, and its physiological and morphological attributes.However, correlating yield at particular sites with mean population 813C shows a positivecorrelation between high productivity and high WUE. To understand differences within andamong populations, the physiological and morphological bases for high productivity must bemeasured (Dickmann et al. 1992). By testing the WUE response of provenances incontrolled multiple factor experiments, we may improve our understanding of the adaptivenature of high or low WUE in P. contorta seedlings.Early selection for improved WUE as indicated by 813C may be possible in P.contorta. However we do not know whether the significant correlation between 813C inseedlings and juveniles is repeatable.Stable carbon isotope analysis offers a simple means to assess genetic differences inWUE among populations of large trees, as well as genotype x environment interactions.From a yield aspect, it is not clear whether high WUE is more advantageous than low WUEin promoting survival or productivity. It will be important for future investigators toexamine the relevance of using 813C analysis as a selection tool for practical purposes inwoody species. Stable carbon isotope analysis of plant tissue may be more valuable as apredictor of site yield than as a means to understand functional differences among treepopulations.967.0 Recommendations for Further ResearchSeveral goals could be pursued in future studies:1. Improving the models relating environmental factors to the evolution ofdifferences in WUE among wild populations of P. contorta . Precise climatic, atmosphericand site characteristics of origin habitats would be required, the most important being thoserelating to air and soil moisture. Populations from the climatic range of the species should betested. The provenance trials in British Columbia were not set up to assist in modelling theevolution of WUE according to climate. Seed sources were not chosen with particularattention to their vicinity to climate stations, nor were detailed micrometeorological datacollected. Air moisture, soil moisture carrying capacity and nutrients were not documented.Using these trials for further research would require sampling the highest number ofpopulations tested (total= 140) that could be matched with acceptable climate station data.Biogeoclimatic subzone climate data could be used but would introduce more error. Costscould be reduced by pooling cores within populations in each block, giving two samples perpopulation per site for carbon isotope analysis.2. Pursuing the global warming problem. The ability of seedling populations toadjust their WUEs to rapid climate changes in ecologically sensitive areas in BritishColumbia could be assessed. This would require test sites in the boreal forest and thetransition zones, e.g., Boreal White and Black Spruce-Alpine Tundra, Mountain Hemlock-Parkland. Micrometerological data should be recorded. Physiological responses andassimilate partitioning should be measured on reciprocal transplants.3. Determining more conclusively the relationship between improved WUE andyield as well as fitness (surviving to produce viable seed). Stronger correlations between813C and yield for mature trees may be found on sites with optimal growth conditions for thepopulations tested. Individuals from certain populations will show stronger relationshipsbetween 813C and yield than individuals from others. For example, populations selected for97plantation renewal in tree improvement programs may show stronger positive relationshipsbetween 813C and yield than wild populations originating from marginal sites in BritishColumbia. An important distinction must be made between working to understand thevariation in WUE within populations vs. the variation in WUE among populations. Thevariation in WUE within populations is often not addressed in the literature. If seedlingestablishment and productivity as related to WUE should be the focus, 813C vs. yieldrelationships can be measured in multi-factor experiments with regulated moisture,temperature and nutrients. For seedlings, changes in assimilate partitioning in response toenvironment may be more informative than overall growth responses. Improved WUE maybe more strongly related to changes in allocation patterns rather than biomass.4. Continuing to test the correlation between seedling and sapling 813C in otherpopulations and families, as well as in other tree species. At some level, seedling 813C maybe a reliable indicator of long-term performance.98ReferencesBassman, J. H. 1985. Selected physiological characteristics of lodgepole pine. In LodgepolePine \u00E2\u0080\u0094 The Species and Its Management. Proceedings of the Cooperative ExtensionService, ed. D. M. Baumgartner, R. G. Krebill, J. T. Arnott, and G. F. Weetman, 27-43.Pullman: Washington State University.Bender, M. M. 1968. Mass spectrometric studies of carbon-13 variations in corn and othergrasses. Radiocarbon 10:468-472.Bergerud, W. 1990. Biometrics Information. Pamphlet No. 28. Victoria: Ministry of ForestsResearch Branch. 4 p.Bierhuizen, J. F., and R. 0. Slatyer. 1965. Effect of atmospheric concentration of water vaporand CO2 in determining transpiration-photosynthesis relationships of cotton leaves.Agric. Meteorol. 2:259-270.Blake, T. J., T. J. Tschaplinski, and A. Eastham. 1984. Stomatal control of water useefficiency in poplar clones and hybrids. Can. J. Bot. 62:1344-1351.Bourdeau, P. F. 1963. Photosynthesis and respiration of Pinus strobus L. seedlings in relationto provenance and treatment. Ecology 44:710-716.Brix, H. 1979. Effects of plant water stress on photosynthesis and survival of four conifers.Can. J. For. Res. 9:160-165.Brugnoli, E., K. T. Hubick, S. von Caemmerer, S. C. Wong, and G. D. Farquhar. 1988.Correlation between the carbon isotope discrimination in leaf starch and sugars of C3plants and the ratio of intercellular and atmospheric partial pressures of carbon dioxide.Plant Physiol. 88:1418-1424.Carter, G. A., and W. K. Smith. 1988. Microhabitat comparisons of transpiration andphotosynthesis in three subalpine conifers. Can. J. Bot. 66:963-969.Clausen, J., D. D. Keck, and W. M. Hiesey. 1948. Experimental studies on the nature ofspecies. III. Environmental responses of climatic races of Achillea. Carnegie Institutionof Washington Publication No. 581:1-129.Condon, A. G., G. D. Farquhar, and R. A. Richards. 1990. Genotypic variation in carbonisotope discrimination and transpiration efficiency in wheat. Leaf gas exchange andwhole plant studies. Aust. J. Plant Physiol. 17:9-22.Condon, A. G., R. A. Richards, and G. D. Farquhar. 1987. Carbon isotope discrimination ispositively correlated with grain yield and dry matter production in field-grown wheat.Crop Sci. 27:996-1001.Cowan, I. R. 1986. Economics of carbon fixation in higher plants. In On the Economy ofPlant Form and Function, ed. T. J. Givnish, 133-170. Cambridge: Cambridge UniversityPress.Craig, H. 1953. The geochemistry of the stable carbon isotopes. Geochim. Cosmochim. Acta3:53-92.99Day, R. W., and G. P. Quinn. 1989. Comparisons of treatments after an analysis of variancein ecology. Ecological Monographs 59:433-463.DeLucia, E. H., W. H. Schlesinger, and W. D. Billings. 1988. Water relations and themaintenance of Sierran conifers on hydrothermally altered rock. Ecology 69:303-311.DeLucia, E. H., and S. A. Heckathorn. 1989. The effect of soil drought on water-useefficiency in a contrasting Great Basin desert and Sierran montane species. Plant, Cell &Environ. 12:935-940.Dickmann, D. I., Z. Liu, P. V. Nguyen, and K. S. Pregitzer. 1992. Photosynthesis, waterrelations, and growth of two hybrid Populus genotypes during a severe drought. Can. J.For. Res. 22:1094-1106.Downes, R. W. 1970. Effect of light intensity and leaf temperature on photosynthesis andtranspiration in wheat and sorghum. Aust. J. Biol. Sci. 23:775-782.Dykstra, G. F. 1974. Photosynthesis and carbon dioxide transfer resistance of lodgepole pineseedlings in relation to irradiance, temperature, and water potential. Can. J. For. Res.4:201-206.Eickmeier, W., M. Adams, and D. Lester. 1975. Two physiological races of Tsugacanadensis in Wisconsin. Can. J. Bot. 53:940-951.Elsik, C. G., R. B. Flagler, and T. W. Boutton. 1992. Poster 10. Gas exchange and 813CPDB ofloblolly and shortleaf pine as affected by ozone and water stress. In Perspectives of PlantCarbon and Water Relations from Stable Isotopes. Fifteenth Annual Symposium in PlantPhysiology, Riverside, California, 9-11 January 1992.Environment Canada, Atmospheric Environment Service. Canadian Climate Normals 1951-1980. Temperature and Precipitation, British Columbia. Publication of the CanadianClimate Program. U.DC:551.582(711), 268 p.Evans, J. R., T. D. Sharkey, J. A. Berry, and G. D. Farquhar. 1986. Carbon isotopediscrimination measured concurrently with gas exchange to investigate CO2 diffusion inleaves of higher plants. Aust. J. Plant Physiol. 13:281-292.Farquhar, G. D., M. C. Ball, S. von Caemmerer, and Z. Roksandic. 1982a. Effect of salinityand humidity on 813C value of halophytes \u00E2\u0080\u0094 evidence for diffusional isotopefractionation determined by the ratio of intercellular/atmospheric partial pressure of CO2under different environmental conditions. Oecologia 52:121-124.Farquhar, G. D., J. R. Ehleringer, and K. T. Hubick. 1989a. Carbon isotope discriminationand photosynthesis. Ann. Rev. Plant Physiol. Plant Mol. Biol. 40:503-537.Farquhar, G. D., K. T. Hubick, A. G. Condon, and R. A. Richards. 1989b. Carbon isotopefractionation and plant water-use efficiency. In Stable Isotopes in Ecological Research,ed. P. W. Rundel, J. R. Ehleringer, and K. A. Nagy, Ecological Studies 68, 21-40. NewYork: Springer-Verlag.Farquhar, G. D., M. H. O'Leary, and J. A. Berry. 1982b. On the relationship between carbonisotope discrimination and the intercellular carbon dioxide concentration in leaves. Aust.J. Plant Physiol. 9:121-137.100Farquhar, G. D., and R. A. Richards. 1984. Isotopic composition of plant carbon correlateswith water-use efficiency of wheat genotypes. Aust. J. Plant Physiol. 11:539-552.Farquhar, G. D., and T. D. Sharkey. 1982. Stomatal conductance and photosynthesis. Annu.Rev. Plant Physiol. 33:317-345.Fischer, R. A., and N. C. Turner. 1978. Plant production in the arid and semiarid zones.Annu. Rev. Plant Physiol. 29:277-317.Francey, R. J. 1981. Tasmanian tree rings belie suggested anthropogenic 13C/12C trends.Nature 290:232-234.Francey, R. J., and G. D. Farquhar. 1982. An explanation of 13C/12C variations in tree rings.Nature 297:28-31.Francey, R. J., R. M. Gifford, T. D. Sharkey, and B. Weir. 1985. Physiological influences oncarbon isotope discrimination in huon pine (Langarostrobos franklinii). Oecologia66:211-218.Freyer, H. D., and N. Belacy. 1983. 13C/12C records in northern hemispheric trees during thepast 500 years \u00E2\u0080\u0094 anthropogenic impact and climatic superpositions. J. Geophysical Res.88:6844-6852.Fu, Q. A., M. K. Owens, and T. W. Boutton. 1992a. Carbon isotope discrimination in liveoak canopies in a semiarid oak-juniper woodland. Poster 11. In Perspectives of PlantCarbon and Water Relations from Stable Isotopes. Fifteenth Annual Symposium in PlantPhysiology, Riverside, California, 9-11 January 1992.Fu, Q. A., T. W. Boutton, J. R. Ehleringer, and R. B. Flagler. 1992b. Environmental anddevelopmental effects on carbon isotope discrimination by two species of Phaseolus.Poster 12. In Perspectives of Plant Carbon and Water Relations from Stable Isotopes.Fifteenth Annual Symposium in Plant Physiology, Riverside, California, 9-11 January1992.Gardner, F. P., R. B. Pearce, and R. L. Mitchell. 1985. Physiology of Crop Plants. Ames:Iowa State University Press. 327 p.Guy, R. D., and R. L. Wample. 1984. Stable carbon isotope ratios of flooded and nonfloodedsunflowers (Helianthus annuus). Can. J. Bot. 62:1770-1774.Hall, A. E., R. G. Mutters, K. T. Hubick, and G. D. Farquhar. 1990. Genotypic differences incarbon isotope discrimination by cowpea under wet and dry field conditions. Crop Sci.30:300-305.Harris, J. M. 1989. Spiral Grain and Wave Phenomena in Wood Formation. Berlin andHeidelberg: Springer-Verlag. 214 p.Hayes, J. M. 1982. Fractionation, et al.: An introduction to isotopic measurements andterminology. Spectra 8:3-8.Hicks, C.R. 1982. Fundamental Concepts in the Design of Experiments. 3rd ed. Fort Worth:Saunders College Publishing. 425 p.101Howarth, C. J. 1977. The Carlo Erba analyzer: carbon, hydrogen, nitrogen and oxygen. InInstrumental Organic Elemental Analysis, ed. R. Belcher, 75-116. London: AcademicPress Inc.Hubick, K. T., and G. D. Farquhar. 1989. Carbon isotope discrimination and the ratio ofcarbon gained to water lost in barley cultivars. Plant, Cell & Environ. 12:795-805.Hubick, K. T., G. D. Farquhar, and R. Shorter. 1986. Correlation between water-useefficiency and carbon isotope discrimination in diverse peanut (Arachis) germplasm.Aust. J. Plant Physiol. 13:803-816.Hubick, K. T., G. L. Hammer, G. D. Farquhar, L. J. Wade, S. von Caemmerer, and S. A.Henderson. 1990. Carbon isotope discrimination varies genetically in C4 species. PlantPhysiol. 91:534-537.Hubick, K. T., R. Shorter, and G. D. Farquhar. 1988. Heritability and genotype xenvironment interactions of carbon isotope discrimination and transpiration efficiency inpeanut (Arachis hypogaea L.). Aust. J. Plant Physiol. 15:799-813.Huitema, B. E. 1980. The Analysis of Covariance and Alternatives. New York: John Wiley& Sons, Inc. 445 p.Hurlbert, S. H. 1984. Pseudoreplication and the design of ecological field experiments.Ecological Monographs 54:187-211.Illingworth, K. 1973. Lodgepole pine provenance research and breeding in British Columbia.In Management of Lodgepole Pine Ecosystems. Proceedings of the CooperativeExtension Service, ed. D. M. Baumgartner, 47-67. Pullman: Washington StateUniversity.Illingworth, K. 1978. Study of lodgepole pine genotype-environment interaction in BritishColumbia. In Proceedings IUFRO Joint Meeting of Working Parties: Douglas-firProvenances, Lodgepole Pine Provenances, Sitka Spruce Provenances and AbiesProvenances, 151-158. Victoria: Information Services Branch, Ministry of Forests.Jackson, P., G. Goldstein, J. Cavalier, F. Meinzer, and N. Holbrook. 1992. Carbon isotopediscrimination in gap and understory species in a tropical forest. Poster 14. InPerspectives of Plant Carbon and Water Relations from Stable Isotopes. FifteenthAnnual Symposium in Plant Physiology, Riverside, California, 9-11 January 1992.Jones, H. G. 1976. Crop characteristics and the ratio between assimilation and transpiration.J. Appl. Ecol. 13:605-622.Kaufmann, M. R. 1982. Leaf conductance as a function of photosynthetic photon fluxdensity and absolute humidity difference from leaf to air. Plant Physiol. 69:1018-1022.'Omer, Ch. and P. Bannister. 1985. Stomatal response to humidity in Nothofagus menziesii.New Zealand J. Bot. 23:425-429.Kovats, M. 1977. Estimating juvenile tree volumes for provenance and progeny testing. Can.J. For. Res. 7:336-340.Kramer, P. J. 1983. Water Relations of Plants. New York: Academic Press. 489 p.102Lauteri, M., L. Spaccino, and E. Brugnoli. 1992. Carbon isotope discrimination in wholeplant dry matter and leaf soluble sugars and water-use efficiency in sunflower genotypesgrown at different water conditions. Poster 18. In Perspectives of Plant Carbon andWater Relations from Stable Isotopes. Fifteenth Annual Symposium in Plant Physiology,Riverside, California, 9-11 January 1992.Leavitt, S. W., and S. R. Danzer. 1992. Seasonal 513C changes in growth rings of red pineseedlings from in vivo experiments. Poster 19. In Perspectives of Plant Carbon andWater Relations from Stable Isotopes. Fifteenth Annual Symposium in Plant Physiology,Riverside, California, 9-11 January 1992.Leavitt, S. W., and A. Long. 1982. Evidence for 13C/12C fractionation between tree leavesand wood. Nature 298:742-743.^ . 1983. Possible climatic response of 813C in leaf cellulose ofpinyon pine in Arizona USA. Isotope Geoscience 1:169-180.. 1984. Sampling strategy for stable carbon isotope analysis oftree rings in pine. Nature 311:145-147.^ . 1986. Stable-carbon isotope variability in tree foliage andwood. Ecology 67:1002-1010.Livingston, N. J. 1992. Carbon isotope fractionation in tree rings, tree growth and wateravailability. Poster 21. In Perspectives of Plant Carbon and Water Relations from StableIsotopes. Fifteenth Annual Symposium in Plant Physiology, Riverside, California, 9-11January 1992.Lopushinsky, W. 1975. Water relations and photosynthesis in lodgepole pine. InManagement of Lodgepole Pine Ecosystems. Proceedings of the Cooperative ExtensionService, ed. D. M. Baumgartner, 135-153. Pullman: Washington State University Press.Madhavan, S., I. Treichel and M. H. O'Leary. 1991. Effects of relative humidity on carbonisotope fractionation in plants. Bot. Acta 104:292-294.Martin, B., and Y. R. Thorstenson. 1988. Stable carbon isotope composition (813C), wateruse efficiency, and biomass productivity in Lycopersicon esculentum, Lycopersiconpennellii, and the Fl hybrid. Plant Physiol. 88:213-217.Masle, J., and G. D. Farquhar. 1988. Effects of soil strength on the relation of water-useefficiency and growth to carbon isotope discrimination in wheat seedlings. Plant Physiol.86:32-38.Medina, E., and P. Minchin. 1980. Stratification of 513C values of leaves in Amazonian rainforests. Oecologia 45:377-378.Meidinger, D., and G. Reynolds. 1991. Climatic Data Summaries for the BiogeoclimaticZones of British Columbia. Version 3 (unpublished).Meinzer, F. C., G. Goldstein, and D. A. Grantz. 1990. Carbon isotope discrimination incoffee genotypes grown under limited water supply. Plant Physiol. 92:130-135.103Miller, P. C., and H. A. Mooney. 1974. The origin and structure of American arid-zoneecosystems. The producers: interactions between environment, form and function. InProc. 1st Int. Congr. Ecol., 201-209. The Hague.Miller, R. G., Jr. 1981. Simultaneous Statistical Inference. 2nd ed. New York: Springer-Verlag. 299 p.Monson, R. K., and M. C. Grant. 1989. Experimental studies of Ponderosa pine. III.Differences in photosynthesis, stomatal conductance, and water-use efficiency betweentwo genetic lines. A. J. Bot. 76:1041-1047.Muller, R. A., and T. M. Oberlander. 1984. 3rd ed. Physical Geography Today. New York:Random House. 591 p.Murdiyarso, D., J. M. Roberts, and J. R. Milford. 1985. The effects of seed origin on droughtresiistance of lodgepole pine (Pinus contorta Dougl.) seedlings. In Proceedings of the17'n Conference on Agricultural and Forest Meteorology and the ri Conference onBiometeorology and Aerobiology, [preprint volume], 154-156.Newberry, T. L. 1992. The effect of climatic variability on the water use efficiency of pinyonpine (Pinus edulis). Poster 27. In Perspectives of Plant Carbon and Water Relations fromStable Isotopes. Fifteenth Annual Symposium in Plant Physiology, Riverside, California,9-11 January 1992.Nier, A. P. 1947. The mass spectrometer and its application to isotope abundancemeasurements in tracer isotope experiments. In Preparation and Measurement ofIsotopic Tracers, ed. D. W. Wilson, A. 0. C. Nier, and S. P. Reimann, 11-30. AnnArbor: J.W. Edwards.Nobel, P. S. 1980. Leaf anatomy and water use efficiency. In Adaptation of Plants to Waterand High Temperature Stress, ed. N. C. Turner, and P. J. Kramer, 43-55. New York:Wiley. \u00E2\u0080\u00A2 1983. Biophysical Plant Physiology and Ecology. San Francisco: W. H.Freeman and Company. 608 p.O'Leary, M. H. 1981. Carbon isotope fractionation in plants. Phytochemistry 20:553-557.^\u00E2\u0080\u00A2 1988. Carbon isotopes in photosynthesis. BioScience 38:328-336.O'Leary, M. H., S. Madhavan, and P. Paneth. 1992. Physical and chemical basis of carbonisotope fractionation in plants. Plant, Cell & Environ. 15:1099-1104.Osmond, C. B., 0. BjOrkman, and D. J. Anderson. 1980. Physiological processes in plantecology. Toward a synthesis with Atriplex. Ecological Studies 36. Berlin and New York:Springer-Verlag. 468 p.Ott, L. 1988. An Introduction to Statistical Methods and Data Analysis. 3rd ed. Boston:PWS-Kent Publishing Co. 835 p.Rawson, H. M., J. E. Begg, and R. G. Woodward. 1977. The effect of atmospheric humidityon photosynthesis, transpiration and water use efficiency of leaves of several plantspecies. Planta 134:5-10.104Read, J., and G. D. Farquhar. 1991. Comparative studies in Nothofagus (Fagaceae).I. Leafcarbon isotope discrimination. Funct. Ecol. 5:684-695.Rehfeldt, G. E. 1983. Adaptation of Pinus contorta populations to heterogeneousenvironments in northern Idaho. Can. J. For. Res. 13:405-411.^. 1986. Ecological genetics of Pinus contorta in the Upper Snake Riverbasin of Eastern Idaho and Wyoming. Research Paper INT-356. Ogden, UT: U.S.Department of Agriculture, Forest Service, Intermountain Forest and Range ExperimentStation. 9p.^. 1987. Components of adaptive variation in Pinus contorta from the inlandnorthwest. Research Paper INT-375. Ogden, UT: U.S. Department of Agriculture, ForestService, Intermountain Forest and Range Experiment Station. 11p.^. 1989. Genetic variances and covariances in freezing tolerance oflodgepole pine during early winter acclimation. Silvae Genetica 38:133-137.Sandford, A. P., and P. G. Jarvis. 1986. Stomatal responses to humidity in selected conifers.Tree Physiol. 2:89-103.SAS Institute Inc. 1988. SAS/STAT\u00C2\u00AE User's Guide, Release 6.03 Edition. Cary, N.C.: SASInstitute Inc. 1028 p.Schleser, G. H. 1992. 513C pattern in a forest tree as an indicator of carbon transfer in trees.Ecology 73:1922-1925.Schlichting, C. D. 1986. The evolution of phenotypic plasticity in plants. Ann. Rev. Ecol.Syst. 17:667-693.Schulze, E. -D., 0. L. Lange, L. Kappen, U. Buschbom, and M. Evenari. 1973. Stomatalresponses to changes in temperature at increasing water stress. Planta 110:29-42.Smit, J., and R. van den Driessche. 1992. Root growth and water use efficiency of Douglas-fir (Pseudotsuga menziessi (Mirb.) Franco) and lodgepole pine (Pinus contorta Dougl.)seedlings. Tree Physiol. 11:401-410.Smith, W. K. 1985. Western montane forests. In Physiological Ecology of North AmericanPlant Communities, ed. B. F. Chabot, and H. A. Mooney, 95-126. New York: Chapmanand Hall.Smith, B. N., J. H. Nelson, and C. T. Frank. 1987. Differences in photosynthetic carbondioxide fixation and carbon isotopic fractionation in subspecies of Atriplex-canescensand Artemesia-tridentata grown in a common garden. Int. Bot. Congr. Abs. 17:439.Sweet, G. B., and P. F. Wareing. 1968. Comparison of the rates of growth and photosynthesisin first-year seedlings of four provenances of Pinus contorta Dougl. Ann. Bot. 32:735-751Tanner, C. B. 1981. Transpiration efficiency of potato. Agron. J. 73:59-64.Tans, P. P., and W. G. Mook. 1980. Past atmospheric CO2 levels and the 13C/12C ratios intree rings. Tellus 32:268-283.105Tesar, M. B. ed. 1984. Physiological Basis of Crop Growth and Development. Madison:American Society of Agronomy, Inc., and the Crop Science Society of America, Inc.341 p.Turesson, G. 1922a. The genotypical response of the plant species to the habitat. Hereditas3:211-350.^. 1922b. The species and the variety as ecological units. Hereditas 3:100-113.^. 1923. The scope and import of genecology. Hereditas 4:171-176.Vogel, J. C. 1978. Recycling of carbon in a forest environment. Oecol. Plant. 13:89-94.Wheeler, N. C., and W. B. Critchfield. 1985. The distribution and botanical characteristics oflodgepole pine: biogeographical and management implications. In Lodgepole Pine: TheSpecies and Its Management. Proceedings of the Cooperative Extension Service, ed. D.M. Baumgartner, R. G. Krebill, J. T. Arnott, and G. F. Weetman, 1-13. Pullman:Washington State University Press.Wilkinson, L. 1990. SYSTAT: The System for Statistics. Evanston, IL: Systat, Inc. 676 p.Woodwell, G. M. 1989. The warming of the industrialized middle latitudes 1985-2050:causes and consequences. Climatic Change 15:31-50.Wright, G. C., K. T. Hubick, and G. D. Farquhar. 1988. Discrimination in carbon isotopes ofleaves correlates with water-use efficiency of field-grown peanut cultivars. Aust. J. PlantPhysiol. 15:815-825.Ying, C. C., K. Illingworth, and M. Carlson. 1985. Geographic variation in lodgepole pineand its implications for tree improvement in British Columbia. In Lodgepole Pine: TheSpecies and Its Management. Proceedings of the Cooperative Extension Service, ed. D.M. Baumgartner, R. G Krebill, J. T. Arnott, and G. F. Weetman, 45-53. Pullman:Washington State University Press.Ying, C. C., C. Thompson, and L. Herring. 1989. Geographic variation, nursery effects, andearly selection in lodgepole pine. Can. J. For. Res. 19:832-841.Yoder, B. J., J. A. Panek, and A. C. Mix. 1992. The relationship between tree age and 513Cratios in wood: a search for causes. Poster 34. In Perspectives of Plant Carbon and WaterRelations from Stable Isotopes. Fifteenth Annual Symposium in Plant Physiology,Riverside, California, 9-11 January 1992.Zar, J. H. 1984. Biostatistical Analysis. 2nd ed. Englewood Cliffs, N.J.: Prentice-Hall, Inc.718 p.Zhang, J., J. D. Marshall, and B. C. Jaquish. 1993. Genetic differentiation in carbon isotopediscrimination and gas exchange in Pseudotsuga menziesii. Oecologia 93:80-87.Appendix 1. Correlations between 813C and yield and growth variables at each test site(provenance 95 excluded).70 Mile HouseDiameter^Height^VolumeProvenance^1^0.239^0.231^0.12613 -0.449 -0.493 -0.32630^-0.212^-0.342^-0.23144 0.125 -0.045 0.13157^-0.456^-0.596^-0.45661 0.038 0.025 0.06369^0.330^0.386^0.30271 -0.645* -0.720* -0.679*100^-0.002^-0.0001^0.082104 0.574* 0.540* 0.547*106Holden LakeDiameterProvenance^1^-0.02813 -0.13330^0.34544 0.512*57^0.745*61 0.07969^0.26871 0.114100^0.176104 0.273Salmon LakeDiameterProvenance^1^0.31913 -0.14130^0.547*44 0.600*61^-0.652*69 -0.12871^0.233100 0.292104^0.267Height0.037-0.0920.463*0.681*0.475*0.3380.3550.0680.2190.235Volume-0.090-0.0110.3370.554*0.671*0.1510.3750.1190.0980.233Height0.4820.0820.650*0.699*-0.458-0.1180.2680.285-0.027Volume0.324-0.0330.556*0.672*-0.596*-0.0480.2760.3460.191Note: None of the correlations are truly significant.*Correlations significant at a=0.05 (no correction for number of correlations tested).Units of measurement: diameter, mm; height, cm; volume, m3.Appendix 2. Correlations between seedling 613C and biomass variables, as well asshoot/root ratio.Shoot^Root^Shoot+Root Shoot/RootProvenance^1^-0.161^0.200^-0.085^-0.35013^0.021^0.145^0.057^-0.17130^-0.106^0.171^-0.046^-0.21944^-0.048^0.002^-0.037^-0.03057^-0.364*^0.037^-0.281^-0.369*61 0.253^0.059^0.208^0.12069^-0.171^0.341^-0.021^-0.467*71^0.199^0.405*^0.288^-0.410*95^-0.066^0.045^-0.041^-0.096100^0.181^0.206^0.200^-0.107104^0.332^0.306^0.340^0.016Note: None of the correlations are truly significant.*Correlations significant at a=0.05 (no correction for number of correlations tested).Units of measurement: shoot, g; root, g; shoot+root, g.107Appendix 3. Regression analyses of 513C vs. temperature and precipitation (for provenances 13, 30, 57, 69, 95, 100 and 104),including the tests for lack-of-fit of the models.Source df SS MS Calculated F Value F ValueComplete Data Set Including Provenance 95Regression 4 67.725 16.931 58.383 2.37Residual 285 82.698 0.290Lack-of-fit 14a 9.759 0.697 2.591 1.70Within Groups 271 72.939 0.269Complete Data Set Excluding Provenance 95Regression 5 63.816 12.763 46.923 2.21Residual 258 70.242 0.272Lack-of-fit lib 2.921 0.266 0.974 1.80Within Groups 247 67.321 0.273adf=290-4-1-271 (according to Ott 1988, p 369)bdf=264-5-1-247"@en . "Thesis/Dissertation"@en . "1993-11"@en . "10.14288/1.0075197"@en . "eng"@en . "Forest Sciences"@en . "Vancouver : University of British Columbia Library"@en . "University of British Columbia"@en . "For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use."@en . "Graduate"@en . "Population differences in water-use effeciency for Pinus contorta dougl as indicated by stable carbon isotopic composition"@en . "Text"@en . "http://hdl.handle.net/2429/2413"@en .