UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Growth and nutrient relations in black cottonwood in South-Coastal British Columbia McLennan, Donald S. 1993

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

Item Metadata

Download

Media
831-ubc_1993_spring_phd_mclennan_donald.pdf [ 9.42MB ]
Metadata
JSON: 831-1.0075152.json
JSON-LD: 831-1.0075152-ld.json
RDF/XML (Pretty): 831-1.0075152-rdf.xml
RDF/JSON: 831-1.0075152-rdf.json
Turtle: 831-1.0075152-turtle.txt
N-Triples: 831-1.0075152-rdf-ntriples.txt
Original Record: 831-1.0075152-source.json
Full Text
831-1.0075152-fulltext.txt
Citation
831-1.0075152.ris

Full Text

GROWTH AND NUTRIENT RELATIONS IN BLACK COTTONWOODIN SOUTH-COASTAL BRITISH COLUMBIAbyDONALD S. McLENNANB.A., Mt. Allison University, 1973M.Sc., Simon Fraser University, 1981A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE STUDIES(Faculty of Forestry)We accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIAFebruary 1993© Donald S. McLennan, 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.(SignatureDepartment of ^r^`1:-:^SCThe University of British ColumbiaVancouver, CanadaDate^f-Lo ,- I^1DE-6 (2/88)1 1AbstractInitially, the study examined within and among site temporal and spatial variation offoliar nutrients, and spatial variation of soil nutrients to assess the sampling methods employed,and to provide background for the interpretation of nutrient-site index interactions. The studythen examined relationships between the growth of black cottonwood, expressed as site index,and site units, understory vegetation, soil nutrient contents, and foliar nutrient concentrations in29 black cottonwood stands in south-coastal British Columbia. The final phase of the study wasa fertilizer trial in three juvenile black cottonwood stands, with treatments based on used DRISdiagnosis of limiting nutrients.Significant levels of variability in foliar nutrient concentrations were identified withintree canopies, and from tree-to-tree within stands. A protocol was suggested to standardizesampling procedures to reduce spatial variability. Sample size requirements for different levelsof accuracy and precision were presented. Important variation in foliar nutrient concentrationswas also recorded seasonally, and from year to year, in foliage samples collected according tothe same protocol. It was shown that the temporal variability was sufficient to alter theinterpretation of foliar nutrient concentrations for the stands.Spatial variation in soil nutrient concentrations was high and was attributed to order-of-magnitude concentration differences between soil strata in each pedon. Spatial variation of soilnutrient contents (expressed in kg/ha over a 1 m sampling depth) was generally higher than soilnutrient concentrations, because of factors such as bulk density and percent coarse fragmentsthat were used to calculate soil contents, and that are themselves subject to variation. It wasshown that the compositing procedure used to reduce costs approximately doubled the variabilityseen in the intensively sampled sites, and alterations to the compositing procedure wereiiisuggested. It was also argued that sampling over a depth of 1 m, and not over the main rootingdepth, provided the most biologically meaningful estimates of soil nutrients available to blackcottonwood.The ANOVA comparing black cottonwood growth within site units was highlysignificant (p < .001), and explained 87% of the variance in site index within the 29 study sites.This general result suggests that, relative to the ecological requirements of black cottonwood, thesite classification provided an ecologically-meaningful differentiation of the edatopic gradientssampled. For operational purposes, this result predicts that black cottonwood site index can beestimated with considerable accuracy by identifying the site unit on which a stand is located.Growth was best on the high bench of alluvial floodplains (Ss-Salmonberry site association), andon moist upland sites with seepage (Cw-Foamflower site association). Growth was poorest onthe low bench of alluvial floodplains (Ac-Willow site association), and on gleyed, marine siteunits (Cw-Salmonberry and Cw-Black twinberry site associations). About 50% of the variationin site index could be accounted for using understory vegetation from within the stands aspredictors. This relatively low explanatory power was attributed to the fact that blackcottonwood site index changed significantly across the indicative range of many of theunderstory plants.All methods of analysis revealed consistent relationships between measures of sitenutrient status and site index. Sample stands with high pH, high levels of exchangeable Ca andMg, and low levels of soil N, P, and K, had foliar concentrations of N, P, and K diagnosed aslimiting to black cottonwood growth, and had the lowest site index. High site index wasrecorded in stands with more or less opposite soil and foliar properties. Site index was seen todecrease in site units with increasing flooding frequency and duration on alluvial floodplains.The decrease was attributed to the negative impact of flooding on the rate of organic mattermineralization, on nutrient uptake, and on the negative effect of high levels of soil Ca and highsoil pH on the availability of soil P. On upland sites, soil gleying and prolonged rooting zoneivflooding during the growing season was correlated with low site index. Using DRIS analysisbased on foliar norms from the 25 fastest-growing, fertilized trees at the Squamish 23 site, it wasconcluded that black cottonwood stands in the high site index class were limited by K, and thenP.In three juvenile black cottonwood stands, the application of fertilizer based on diagnosisof foliar nutrient concentration using DRIS norms had the following 3 year responses - basalarea increment increased by 65%, and height growth increment by 15% at the Squamish 23 site;basal area increment increased by 65% and height growth increment increased by 30% at theStrawberry site; and basal area increment increased by 27% without a significant height growthresponse at the Soowahlie site. At the Squamish 23 and Soowahlie trials, response wasattributed to fertilization with K and P, as suggested from the foliar nutrient diagnosis of thefast-growing group. Given that relatively low dosages (ca. 100 kg/ha) of P were required toachieve a significant growth response, and acknowledging that, in many forest fertilizationprograms response to P fertilization occurs for a considerable period of time, the results suggestthat the fertilization of fast-growing, juvenile black cottonwood stands in coastal BritishColumbia may be economically justified. Significant correlations between measures of foliarresponse and wood production were not seen in the study, and this finding limits the usefulnessof the graphical procedure for interpretation of the experimental results. Foliar concentrationsfrom the 25 fastest-growing black cottonwoods at the Squamish 23 site are presented as DRISstandards that will be useful in the diagnosis of the nutrient status of black cottonwood stands incoastal British Columbia.Table of ContentsAbstract^ iiTable of Contents^List of Tables viiiList of Figures xiiiAcknowledgements^ xviCHAPTER 1 INTRODUCTION^ 11.1 STUDY RATIONALE 11.2 STUDY OBJECTIVES 41.3 ORGANIZATION OF THE THESIS^ 5CHAPTER 2 DESCRIPTION AND CLASSIFICATION OF STUDY SITES ^ 62.1 STAND SELECTION^ 62.2 BLACK COTTONWOOD SITE INDEX AND STEM ANALYSIS ^ 102.3 ECOLOGICAL DESCRIPTION AND CLASSIFICATION^ 112.3.1 Climate^ 122.3.2 Soil Nutrient Regime^ 122.3.3 Soil Moisture Regime 132.3.4 Vegetation^ 16CHAPTER 3^SPATIAL AND TEMPORAL VARIABILITY OF BLACKCOTTONWOOD FOLIAR NUTRIENT CONCENTRATIONS^ 203.1 INTRODUCTION^ 203.2 METHODS 233.2.1 Site Selection and Description^ 233.2.2. Foliar Sampling^ 243.2.3 Laboratory Analysis 253.2.4 Statistical Analysis 263.3 RESULTS^ 283.3.1 Within-tree Spatial Variation^ 283.3.2 Within-site and Among-site Spatial Variation^ 323.3.3 Seasonal Variation^ 363.3.4 Year to Year Variation 383.4 DISCUSSION^ 423.5 CONCLUSIONS 48CHAPTER 4 SPATIAL VARIABILITY OF SOIL NUTRIENTS IN BLACKCOTTONWOOD STANDS^ 504.1 INTRODUCTION 504.2 METHODS 534.2.1 Descriptions of Study Stands^ 534.2.2 Soil Sampling^ 534.2.3 Laboratory Analysis 564.2.4 Statistical Analyses 574.3 RESULTS AND DISCUSSION^ 584.3.1 Variability in Soil Properties within the Soil Pedon^ 584.3.2 Within-Site and Among-Site Variability^ 724.4 DISCUSSION^ 82vi4.5 CONCLUSIONS^ 87CHAPTER 5 RELATIONSHIPS BETWEEN BLACK COTTONWOOD SITEINDEX, FOLIAR AND SOIL NUTRIENTS, UNDERSTORY VEGETATION,AND SITE UNITS^ 895.1 INTRODUCTION^ 895.2 METHODS 935.2.1 Ecosystem Description and Classification^ 935.2.2 Soil and Foliar Nutrient Sampling and Analysis^ 935.2.3 Stem Analysis and Site Index^ 945.2.4 Statistical Methods^ 945.3 RESULTS^ 955.3.1 Correlations of Black Cottonwood Site Index with CWHSubzones, Site Associations and Soil Nutrient Regime 955.3.2 Principal Component Analysis of Vegetation, Soil Nutrients,and Foliar Nutrients 985.3.3 ANOVAs of Soil and Foliar Nutrients in Black CottonwoodSite Index Classes, Soil Nutrient Regimes, and Site Associations 1105.3.4 Linear Regressions of Vegetation on Site Index 1135.3.5 Linear Regressions of Soil Nutrient Contents on Site Index 1145.3.6 Interaction of Available-P, Exchangeable Ca, and pH. 1195.3.7 Linear Regressions of Foliar Nutrient Concentrations on SiteIndex^ 1215.3.8 Relationships Between Foliar and Soil Nutrients^ 1255.3.9 Identification of Optimal Nutrient Levels for BlackCottonwood 1265.3.10 Diagnosis of Nutrient Limitations in the Site Associations 1325.4 DISCUSSION^ 1345.4.1 Nutrient Availability and Site Index - General Trends^ 1345.4.2 Interpretations of Black Cottonwood Growth in the SiteAssociations^ 136Ac-Willow Site Association (Low Bench)^ 136Ac-Red osier dogwood Site Association (MiddleBench)^ 137Ss-Salmonberry Site Association (High Bench) ^ 138'Gleyed' Site Associations^ 139Cw-Foamflower Site Association^ 140Cw-Swordfern Site Association 1415.5 CONCLUSIONS^ 142CHAPTER 6 GROWTH RESPONSE OF THREE BLACK COTTONWOODSTANDS TO FERTILIZATION BASED ON DRIS DIAGNOSIS^ 1446.1 INTRODUCTION^ 1446.2 METHODS 1496.2.1 Site Selection and Description^ 1496.2.2 Fertilizer Experiments 1516.2.3 Growth Response Measurements 1546.3 RESULTS^ 1566.3.1 Growth Responses and Foliar Nutrient Concentrations ^ 156vi i6.3.1.1 First Year Growth Response^ 1566.3.1.2 Three Year Growth Response 1596.3.2 Changes in DRIS Ratios^ 1666.3.3 Growth Response of the 1988 Terminal Leaders^ 1696.3.4 Determination of Optimal Foliar Levels^ 1716.4 DISCUSSION^ 1716.5 CONCLUSIONS 174CHAPTER 7 SUMMARY AND DISCUSSION^ 176CHAPTER 8 REFERENCES^ 180List of TablesTable 2.1: Site index, site index class, stand age, and relative coverage of trees in theupper canopy at 29 black cottonwood study sites. Sites are listed in order ofincreasing black cottonwood site index and are divided into low (L=8.0-14.9m/15 years), medium (M=15.0-21.9 m/15 years), and high (H=22.0-30.8m/15 years) black cottonwood site index classes 7Table 2.2: Location, Iandform, soil, and ecological classification of 29 black cottonwoodstudy sites^ 8Table 2.3: Actual soil moisture regime (aSMR), soil nutrient regime (SNR) and selectedsoil physical properties of 29 black cottonwood study sites^ 9Table 2.4: Frequency and duration of flooding during the growing season (April 15 toSeptember 30) at selected low bench (Ac-Willow), middle bench (Ac-Red-osier dogwood), and high bench (Ss-Salmonberry). Flooding data have beencalculated for a soil depth of 60 cm and for flooding at the soil surface.Statistics are based on regressions (Figure 2.2) of historical discharge data forthe Fraser River at Mission, and the Squamish River at Power House gaugingstations (Water Survey Of Canada) 16Table 2.5: Presence classl/mean percent cover of common and differentiating speciesfor 5 site associations sampled in the study, using upland and floodplain sitegroups as a primary hierarchical level. The Cw-Swordfern site has beenexcluded from the analysis because of only one site (Chipmunk) in the unit 18Table 3.1: Component loadings, eigenvalues, and % variance explained for PCA axes 1and 2 for 13 foliar nutrients at the Carey II site^ 29Table 3.2: Component loadings, eigenvalues, and % variance explained for PCA axes 1and 2 for 13 foliar nutrients at the Soowahlie site^ 29Table 3.3: Mean foliar nutrient concentrations (n=13) for apical, upper canopy„ andlower canopy foliage, and ANOVA, at the Carey II site. For a given nutrient,figures followed by the same letter are not significantly (p=0.05) different 33Table 3.4: Mean foliar nutrient concentrations (n=15) for upper, middle and lowercanopy foliage, and ANOVA, at the Soowahlie site. For a given nutrient,figures followed by the same letter are not significantly (p=0.05) different 33Table 3.5: Coefficients of variation (CV), mean CVs, and F ratios comparing among-and within-stand variance for foliar nutrient concentrations (n=13 at Carey II:n=15 at all other sites) in upper canopy foliage of black cottonwood at sevenalluvial sites. All F ratios are significant at p < .001 34Table 3.6: Numbers of black cottonwood foliage samples required at various levels ofpercent allowable error, and alpha (a) and gamma (g) significance for 13foliar nutrients 35Table 3.7: Mean foliar nutrient concentrations (n=10 at Soowahlie; 15 at Squamish 23and Strawberry 1) from upper canopy foliage of samples collected in the lastviii2 weeks of August from the same sample trees in 1985, 1986, and 1988. Fora given nutrient, and at the same site, figures followed by the same letter arenot significantly (p=0.05) different 39Table 3.8: Comparisons of DRIS indices at the Soowahlie sites for 1985-1988^46Table 4.1: Descriptions of two alluvial soil profiles from the Carey 1 and Soowahliesample sites^ 59Table 4.2: Ranges of coefficients of variation in bulk density at the intensively sampledsites^ 66Table 4.3: Variability in depth of the Ah layer, and main rooting depth at the intensivelysampled sites^ 71Table 4.4: Coefficients of variation (CVs), mean CVs, and F ratios comparing among-site to within-site variance for soil nutrient concentrations at the 9intensively-sampled black cottonwood stands. CVs and ANOVAs are basedon mean nutrient concentrations of 15 individual soil samples collected overthe upper 1 m of soil 77Table 4.5: Relative sampling errors (RSEs), ranges, and mean RSEs for soil nutrientconcentrations at the 9 intensively-sampled black cottonwood ecosystems.RSEs are based on mean nutrient concentrations of 15 individual soil samplescollected over the upper 1 m of soil 78Table 4.6: Relative sampling errors (RSEs) at alpha=0.90, ranges, and mean RSEs forsoil nutrient concentrations at 16 black cottonwood ecosystems where soilsamples were composited into 4 samples at each site 79Table 4.7: Coefficients of variation (CVs), ranges of CVs, mean CVs and F ratioscomparing among-site to within-site variance for soil nutrient contents(kg/ha) at the 9 intensively-sampled black cottonwood ecosystems. CVs arebased on means of 15 individual samples 80Table 4.8: CVs for coarse fragment content for 6 study sites with a significantcomponent of coarse fragments^ 81Table 4.9: Numbers of soil samples required to estimated soil nutrient concentrations inblack cottonwood ecosystems at different levels of alpha, gamma andpercentage error 81Table 4.10: Numbers of soil samples required to estimate bulk densities and soil nutrientcontents in soils within black cottonwood stands at different levels of alpha,gamma and percentage error 82Table 4.11: Relative order of variability in nutrient concentration in this study comparedto other studies.^ 85Table 5.1: Means and results of ANOVAs for black cottonwood site index (m/15 yrs) in3 subzones, 3 soil nutrient regime groups, and 5 site associations. Valueswith the same letter are not significantly different at p < 0.05 96Table 5.2: Eigenvalues (variance explained), percentage of total variance explained, andtotal cumulative variance explained for PCA axes 1 - 6 of the PCA of 85understory species (presence class > II) in 29 black cottonwood study sites 98ixTable 5.3: List of species significantly (p < .10) correlated with the first and second PCAaxes that have soil moisture and/or soil nutrient regime indicator status(Klinka and Krajina, 1986) 101Table 5.4: Correlations of pH and soil nutrient contents (kg/ha) with the first threeprincipal component axes, eigenvalues, and percentage and cumulativepercentage variance explained by the PCA axes in 29 black cottonwoodecosystems. Bolding indicates significance of the correlations at p < 0.05 103Table 5.5: Correlations of foliar nutrient concentrations with the first three principalcomponent axes, eigenvalues, and percentage and cumulative percentagevariance explained by the PCA axes in 26 black cottonwood stands. Boldingindicates significance of the correlations at p < 0.05 107Table 5.6: ANOVAs of soil nutrient contents (kg/ha), for 29 black cottonwoodecosystems in 3 index classes, 3 soil nutrient regime classes, and 5 siteassociations. The Cw-Swordfern site association had only one site and wasnot included in the ANOVAs. For a given nutrient, in a given class, valueswith the same letter are not significantly different (p < 0.05) 111Table 5.7: ANOVAs of foliar nutrient concentrations of 29 black cottonwood stands in 3site index classes, and 5 site associations. The Cw-Swordfern site associationhas only one site and was not included in the ANOVAs. For a given nutrient,in a given class, values with the same letter are not significantly different at p< 0.05. The number of study stands in each group is given in Table 5.6  112Table 5.8: Models, probabilities, coefficients of determination (R2), and standard errorof the estimate (SEE) for vegetation variables on site index in 29 blackcottonwood stands 113Table 5.9: Models, probability, coefficients of determination (R2), and standard error ofthe estimate (SEE) for univariate regressions of soil nutrient contents on blackcottonwood site index in 29 black cottonwood stands. Only soil nutrientswith significant (p < 0.05 ) regressions are shown 115Table 5.10: Correlation matrices for soil nutrient variables with significant univariatelinear regressions on black cottonwood site index. Bolding indicatessignificant (p<0.05) correlations 116Table 5.11: Probability (p), coefficients of determination (R 2), and standard error of theestimate (SEE) for multiple regressions of soil nutrients on black cottonwoodsite index, using variables with significant univariate regressions on blackcottonwood site index 116Table 5.12: Probability (p), coefficients of determination (R 2), and standard error of theestimate (SEE) for univariate regressions of soil nutrient contents on blackcottonwood site index in the reduced data set (n=22). Only soil nutrients withsignificant (p < 0.05) regressions are shown 117Table 5.13: Correlation matrices for soil nutrient variables with significant univariatelinear regressions on black cottonwood site index for the reduced data set.Bolding indicates significant (p<0.05) correlations 118Table 5.14: Probability (p), coefficients of determination (R 2), and standard error of theestimate (SEE) for multiple regressions of soil nutrients on black cottonwoodxsite index, using variables with significant univariate regressions on blackcottonwood site index in the reduced data set (n=22) 119Table 5.15: Univariate models, probabilities (p), coefficients of determination (R2), andstandard errors of the estimate (SEE) for regressions of foliar nutrients on siteindex in 26 black cottonwood stands. Only foliar nutrients with significant (p< 0.05) regressions are shown 121Table 5.16: Correlation matrix for foliar nutrient variables with significant univariatelinear regressions on black cottonwood site index. Bolding indicatessignificant (p<0.05) correlations 122Table 5.17: Probabilities (p), coefficients of determination (R2), and standard errors ofthe estimate (SEE) for multiple regression models of foliar nutrients on blackcottonwood site index, using variables with significant univariate regressions 123Table 5.18: Probabilities (p), coefficients of determination (R2), and standard errors ofthe estimate (SEE) for univariate regressions of foliar nutrients on blackcottonwood site index in 20 black cottonwood ecosystems (reduced data set).Only foliar nutrients with significant (p<0.05) regressions are shown 123Table 5.19: Correlation matrix for foliar nutrient variables with significant univariatelinear regressions on black cottonwood site index in the reduced data set.Bolding indicates significant (p<0.05) correlations 124Table 5.20: Probabilities (p), coefficients of determination (R2), and standard errors ofthe estimate (SEE) for multiple regressions of foliar nutrients on blackcottonwood site index in the reduced data set (n=20) 124Table 5.21: Univariate models, probabilities (p), coefficients of determination (R 2), andstandard errors of the estimate (SEE) for regressions of foliar nutrients on soilnutrient content of the same nutrient (reduced data set; n=20) 125Table 5.22: Published foliar nutrient critical levels (% dry mass) for Populus spp. andhybrids.^ 132Table 5.23: Comparisons of DRIS ratios in 6 site associations sampled in the study.Norms for the establishment of the ratios are based on mean foliarconcentrations from stands in the high site index class (see Table 5.7).  132Table 6.1: Method of establishment, stand age in 1986, stocking, mean DBH, meanheight, and site index of 3 fertilized stands. Stand age and site index werecalculated using height-age curves from destructive sampling in 1988.Stocking was based on prism data from each experimental block on each site.Mean DBH and height were based on pre-treatment measurements of allexperimental trees 150Table 6.2: Selected site and soil properties 150Table 6.3: Soil pH and soil nutrient contents (using a 1 m sampling depth) 150Table 6.4: Mean foliar nutrient concentrations (% of dry mass) based on 15 samplestaken in August 1985 from the upper third of the canopy at threeexperimental sites 151xiTable 6.5: DRIS indices use to develop 1986 fertilizer prescriptions at the Soowahlie,Strawberry Island and Squamish fertilization experiments^ 151Table 6.6: Summary of fertilizer treatments at the Soowahlie, Strawberry 1, andSquamish 23 sites^ 152Table 6.7: 1986 basal area increment and important foliar concentrations at theSoowahlie site^ 156Table 6.8: 1986 basal area increment (BAI) and important changes in foliarconcentrations at the Strawberry Island site^ 157Table 6.9: 1986 basal area increment (BAI) and important foliar concentration changesat the Squamish 23 fertilizer site^ 158Table 6.10a: Summary of changes in 1987 and 1988 basal area increment, 1988 and1986-1988 height growth increment, 1988 foliar concentrations, and infoliar contents of the 1988 terminal leader at the Soowahlie site 164Table 6.10b: Summary of changes in 1987 and 1988 basal area increment, 1988 and1986-1988 height growth increment,1988 foliar concentrations, and in foliarcontents of the 1988 terminal leader at the Strawberry site 164Table 6.10c: Summary of changes in 1987 and 1988 basal area increment, 1988 and1986-1988 height growth increment, 1988 foliar concentrations, and infoliar contents of the 1988 terminal leader at the Squamish site 165Table 6.11: Changes in N, P, and K DRIS ratios, and basal area increment relative to thecontrol, for the most growth-responsive treatment groups at the Squamish,Soowahlie and Strawberry sites 167Table 6.12: Means, standard deviations, and coefficients of variation (CV) for foliarnutrient concentrations in 25 black cottonwood trees with the highest 1988basal area increment at the Squamish 23 site 172xiiList of FiguresFigure 2.1:^Regressions of the distance of the soil surface below the benchmark onmean daily discharge at three alluvial sites. The horizontal line at theSquamish 23 site shows the soil surface, and indicates the dischargecorrelated with a bank full situation 15Figure 3.1:^Ordinations of within canopy variation of 13 foliar nutrients on the firstand second principal component axes at the Carey II (top) andSoowahlie (bottom) sites^ 30Figure 3.2:^(top)Ordination of 39 foliar nutrient samples + 95% confidence ellipsesfor recently matured, upper canopy leaves (A), newly-formed, apicalupper canopy late leaves (B), and lower canopy late leaves (C) on thefirst and second PCA axes at the Carey II site.(bottom) Ordination of 45 foliar nutrient samples and 95% confidenceellipses for recently matured late leaves at upper (A), middle (B) andlower (C) canopy positions on the first and second PCA axes at theSoowahlie site 31Figure 3.3:^Seasonal fluctuations in concentrations of 13 foliar nutrients in the uppercanopy of a 10 year old black cottonwood stand (Soowahlie) over the1985 growing season^ 37Figure 3.4:^Year to year fluctuations in black cottonwood foliar nutrientconcentrations of P, K, N, S, Ca and Mg. Data are for samples collectedfrom the upper canopies of the same trees, using the same samplingprotocol with all sampling carried out during the last 2 weeks of August 40Figure 3.5:^Year to year fluctuations in black cottonwood foliar nutrientconcentrations of SO4-S, Cu, Zn, Mn, Active-Fe and Fe. Data are forsamples collected from the upper canopies of the same trees, using thesame sampling protocol with all sampling carried out during the last 2weeks of August 41Figure 4.1:^Changes in soil bulk density with depth in the soil profile at the Carey 1and Soowahlie sample sites^ 61Figure 4.2:^Comparisons (lines represent 95% confidence intervals, n=15) betweenmean concentrations of soil nutrients in the Ah and C horizons (to adepth of 1 m) at the Carey 1 sample site^ 62Figure 43:^Comparisons (lines represent 95% confidence intervals, n=15) betweenmean concentrations of soil nutrients in the Ah and C horizons (to adepth of 1 m) at the Soowahlie sample site^ 64Figure 4.4:^Concentrations of soil nutrients by C horizon for the soil profiles shownin Table 4.1 at the Carey 1 site. Concentrations are based on a singlesample for each horizon^ 67xiv^Figure 4.5:^Concentrations of soil nutrients by C horizon for the soil profiles shownin Table 4.1 at the Soowahlie site. Concentrations are based on a singlesample for each horizon^ 69^Figure 4.6:^Comparisons (lines represent 95% confidence intervals, n=15) betweenmean contents (kg/ha) of soil nutrients in the Ah and C horizons (to adepth of 1 m) at the Carey 1 sample site^ 73Figure 4.7:^Comparisons (lines represent 95% confidence intervals, n=15) betweenmean contents (kg/ha) of soil nutrients in the Ah and C horizons (to adepth of 1 m) at the Soowahlie sample site^ 75Figure 5.1:^Box diagrams showing distributions of black cottonwood site index inbiogeoclimatic subzone, soil nutrient regime (M=medium; R=rich;VR=very rich), and site association (1=Ac-Willow; 2="Gleyed" sa's;3=Cw-Swordfern; 4=Ac-Red osier dogwood; 5=Ss-Salmonberry; 6=Cw-Foamflower) groups 97Figure 5.2:^PCA ordination of 85 plant species with presence class > II in 29 blackcottonwood stands. Study sites are labelled by site index class (L=low;M=medium; H=high) and site association (1=Ac-Willow; 2="Gleyed"sa's; 3=Cw-Swordfern; 4=Ac-Red osier dogwood; 5=Ss-Salmonberry;6=Cw-Foamflower) 99Figure 5.3:^PCA ordination of pH and 8 soil nutrient content properties in 29 blackcottonwood stands. Study sites are labelled by site index class (L=low;M=medium; H=high) and site association (1=Ac-Willow; 2="Gleyed"sa's; 3=Cw-Swordfern; 4=Ac-Red osier dogwood; 5=Ss-Salmonberry;6=Cw-Foamflower) 104Figure 5.4:^PCA ordination of 13 foliar nutrient concentrations in 29 blackcottonwood stands. Study sites are labelled by site index class (L=low;M=medium; H=high) and site association (1=Ac-Willow; 2="Gleyed"sa's; 3=Cw-Swordfern; 4=Ac-Red osier dogwood; 5=Ss-Salmonberry;6=Cw-Foamflower) 108Figure 5.5:^Three dimensional scattergram showing the effect of increasing soil pHand content of soil exchangeable Ca on content of soil available P forthe reduced data set (n=22). Study sites are labeled by their blackcottonwood site index class (L=low; M=medium; H=high) 120Figure 5.6:^Regressions of foliar nutrient concentrations on their soil nutrientcontents. Study sites are labeled by their black cottonwood site indexclass (L=low; M=medium; H=high). Linear best-fit lines are shown todemonstrate trends of the different regressions 127Figure 5.7:^Scattergrams of black cottonwood site index and selected foliarnutrients. Study sites are labeled by their black cottonwood site indexclass (L=low; M=medium; H=high). Best fit second order polynomiallines have been drawn using a distance-weighted least squaressmoothing algorithm (McLean, 1974) 129Figure 6.1:^1986-1988 basal area increments, and 1986-1988 and 1988 heightgrowth by treatment at the Soowahlie site. Lines represent 95%confidence limits around group means^ 160XVFigure 6.2:^1986-1988 basal area increments, and 1986-1988 and 1988 heightgrowth by treatment at the Strawberry site. Treatment groups are basedon^1988^basal^area^response^(Low=Treatments^2,5,6,8;Medium=Treatments 13,4,7,12; High=Treatments 3,10,9,11) 161Figure 6.3:^1986-1988 basal area increments, and 1986-1988 and 1988 heightgrowth by treatment at the Squamish 23 site^ 162Figure 6.4:^Comparison of mean P concentrations (n=15) in the upper 10 cm of thesoil profile in fertilized and unfertilized plots at the Squamish 23 site.Lines indicate 95% confidence intervals for P means^ 169Figure 6.5:^Comparisons of mean number of leaves, mean leaf fresh mass, and meantotal leaf fresh mass of the 1988 terminal leaders at the Squamish 23(top), Strawberry (middle), and Soowahlie (bottom) sites^ 170xviAcknowledgementsThis study was something of a marathon in that it was initiated in 1985 and the final draft isbeing completed in the spring of 1993, so a large number of people had input into the project,and it will be difficult to acknowledge all who assisted.In the first place, I would like to acknowledge the constant and patient advisory role played bymy supervisor, Dr. K. Klinka, whose door and ears were always open. It has been my privilegeto have enjoyed such a long and meaningful academic and personal relationship with my truementor. I would also like to thank Dr. T. Ballard for being a sounding board for my ideas overthe course of the research, and the members of my committee, who took the time to read andprovide critical input into the early drafts of the work.This study required a considerable amount of field work, and over the years of the project, I wasassisted in the field, and in the lab, by a small army of folks who are too numerous to mention,but who worked carefully and diligently on the analysis.Many of the methodologies for interpretation and analysis of the data, and inspiration to carry onwith the work came from informal interactions with my friends and fellow forest ecologygraduate students and research associates, and I would like to acknowledge the input of QingliWang, Gordon Kayahara, Reid Carter, and Gaofeng Wang.I would also like to acknowledge the assistance and advise of Peter MacAuliffe and KenStenerson of Scott Paper.Finally, I would like to acknowledge the assistance of my wife Aube, who spent many hourspatiently preparing drafts of the manuscript, and who somehow managed to keep me reasonablysane over the whole process. Merci beaucoup.The bulk of this research was carried out by funding provided through the B.C. Ministry ofForests-Forestry Canada Federal Regional Development Agreement (1985-1990). In addition,Scott Paper Ltd., New Westminster, B.C. provided funds for the analysis of foliar samples.1CHAPTER 1INTRODUCTION1.1 STUDY RATIONALEBlack cottonwood [Populus trichocarpa L. ssp. trichocarpa (Torrey and Gray)Brayshaw] is the largest, and most rapidly-growing broadleaf tree in western North America(Roe, 1958; DeBell, 1990). Given the availability of soil moisture and abundant soil nutrients,the species is capable of very rapid height growth (Smith, 1980) and biomass accumulation(Heilman et al., 1972; Heilman and Peabody, 1981; Heilman and Stettler, 1983, 1985b).Although many studies have been carried out that examine ecological aspects of the growth ofconiferous species in western North America (Carter and Klinka, 1990; Eis, 1962; Kabzems andKlinka, 1987a; Green et al., 1989; Kayahara, 1991; Klinka et al., 1989; Monserud, 1984; Wang,1992), there have been fewer studies on broad-leaved species such as red alder (Harrington,1986; Courtin, 1992), and none on black cottonwood, except for a brief overview by Smith(1957). There have also been a number of evaluations of soil and foliar nutrient status anddiagnosis of conifers (Ballard and Carter, 1986; Courtin et al., 1988; Kabzems and Klinka,1987b; Klinka et al., 1984, 1990a), but none have been conducted in broad-leaved ecosystems,with predominantly Mull humus forms and rich soils. Many of the studies cited have shown thatsite index of the species studied is well correlated with soil moisture, soil nutrient, and regionalclimatic classes of the biogeoclimatic classification, but this work has not been done for blackcottonwood.Within the biogeoclimatic classification, the ecological site quality of a forest site can besummarized by determining its subzone, soil moisture regime, and soil nutrient regime, toaccount for the climatic, soil moisture, and soil nutrient factors affecting site productivity (Pojaret al., 1987). In this study the availability of soil nutrients is assessed both qualitatively, as soilnutrient regime, and quantitatively, through the measurement of soil nutrient contents (kg/ha),and of concentrations of foliar nutrients. The rationale for employing only qualitative measures2of soil moisture and climate is that meaningful quantitative measures of these factors aredifficult to acquire. A model of growing season soil water deficit using an energy-driven modelwith climatic normals and soil physical parameters (Spittlehouse and Black, 1981) has been usedsuccessfully by other workers (Carter and Klinka, 1990; Giles et al., 1985; Wang, 1992), but isless useful in a study such as this where almost all sites receive additional, and relativelyunpredictable inputs of soil moisture from flooding and seepage. The site units identified inBanner et al. (1990) develop special-case classes to account for variation in flooding, and thus,given the limited range of soil nutrient regimes sampled, site unit served as a measure of soilmoisture regime in the present study. The acquisition of meaningful climatic data that woulddifferentiate among the sites sampled would require on-site instrumentation and measurementover a much longer time period that this study. Compared to climate and soil moisture, it isrelatively easier to acquire estimates of the absolute amounts of soil nutrients, using methodsthat have been correlated with productivity of trees on a given site (Curran, 1984; Klinka et al.,1980; Powers, 1980).A precise measure of the plant nutrients available in the soil at a given time is verydifficult to assess directly. A major problem in attempting to measure the availability of soilnutrients for a forest stand is the large spatial variability that exists within the soil, and hence thelarge sampling effort required to acquire accurate estimates of the properties measured (Ball andWilliams, 1968; Carter and Lowe, 1986; Courtin et al., 1983; Mader, 1963; McFee and Stone,1965; Quesnel and Lavkulich, 1980; Troedesson and Tamm, 1969). An important finding fromthis research is the lack of any consistent pattern in the variability of the nutrients measured. Asa result, previous studies are difficult to extrapolate, unless they have been carried out on soilsand stands very similar to the ones being studied (Blyth and MacLeod, 1978; Carter and Lowe,1986; Courtin et al., 1983). For this reason, the nature of spatial variability of soil nutrientcontents in black cottonwood ecosystems is investigated in this study.Evaluations of foliar nutrient concentrations bypass some of the problems associatedwith soil nutrient evaluations by providing direct measures of nutrients that have actually been3taken up by the tree (Ballard and Carter, 1986; Leaf, 1973; van den Driessche, 1974; Weetmanand Wells, 1990). A variety of methods have been developed to use foliar nutrientconcentrations for stand nutrient diagnosis including critical levels (Ballard and Carter, 1986;Everard, 1973; Leyton, 1958; Richards and Bevege, 1972), Diagnosis and RecommendationIntegrated System (DRIS) ratios (Beaufils, 1973; Leech and Kim, 1979a, 1981; Schutz and deVilliers, 1986), nutrient ratios (Ballard and Carter, 1986; Ingestad, 1962), and graphicalinterpretations (Heinsdorf, 1968; Timmer, 1985; Timmer and Stone, 1978; Timmer andMorrow, 1984). Although some foliar nutrient diagnoses and interpretations have been carriedout on other Populus species and on hybrid poplars (Bonner and Broadfoot, 1967; Leech andKim, 1981; White and Carter, 1970), there is a very limited amount of information for blackcottonwood (Heilman, 1985). Evaluation of foliar nutrient status was one of the major toolsused to interpret relationships between black cottonwood growth and soil nutrient status in thisstudy.As for evaluations of soil nutrients, obtaining accurate estimates of foliar nutrientconcentrations for a forest stand is also complicated by spatial variability, both within andamong trees, and from stand to stand (Ellis, 1975; Guha and Mitchell, 1965 a,b; Lea et al., 1979a,b; Mitchell and Chandler, 1939; Verry and Timmons, 1976). In addition to spatial variation,seasonal, and year to year variation has also been documented (Day and Monk, 1977; Hoyle,1965; Lea et al., 1979 a,b; Verry and Timmons, 1976; White, 1954), and this component ofvariability has special importance for foliar sampling in broad-leaved species, because samplesmust be acquired during the growing season. For this reason the temporal and spatial variabilityof black cottonwood foliar nutrients was also investigated in this study.The study examined relationships between ecological site quality of black cottonwoodstands, and the growth of the species along a productivity gradient that included an almost four-fold increase in site index. The approach taken in the study was to use both qualitative andquantitative assessments of ecosystem nutrient status to ascertain relationships between thesefactors and the growth of black cottonwood. A major objective of this study was to better4understand the factors responsible for its observed range of productivity by establishingquantified relationships between site index and measurements of soil nutrient contents and foliarconcentrations. A second major objective was to correlate observations of site index and sitenutrient status with site units so that the information could have operational application.Quantitative relationships between measurements of soil and foliar nutrients and site units willalso help provide a better understanding of the productivity of black cottonwood within the siteunits sampled.A major focus of these analyses is the identification and diagnosis of nutrient limitationin black cottonwood stands. If it is concluded that a certain nutrient or combination of nutrientswere limiting growth in a particular stand, then the validity of the diagnosis can be tested byapplying the nutrients thought to be limiting, and then measuring the response of fertilized trees.Fertilization of fast-growing black cottonwood stands will also provide information on thepotential for increasing growth in stands that are already growing rapidly.1.2 STUDY OBJECTIVESThe rationale for conducting the research, and the general objectives have been discussedabove. The specific objectives for each component of the study are listed in the chapters thatfollow. The overall objectives of the study were:1) to examine spatial and temporal aspects of the variability of foliar nutrients in juvenile blackcottonwood stands;2) to examine spatial aspects of the variability of soil nutrients;3) to develop relationships among site index, foliar and soil nutrients, understory vegetation, andsite units, and to develop diagnoses of nutrient limitations; and,4) to test diagnoses of nutrient limitation through fertilization of three black cottonwood stands.51.3 ORGANIZATION OF THE THESISThe thesis was written so that each chapter is as independent as possible from otherchapters, and so that each represents a distinct component of research. Where methodologiesoverlap they were not repeated and reference is made to where they first appear in the thesis.Chapter 2 provides an overview and ecological description of the 29 black cottonwoodecosystems sampled in the study. Chapter 3 uses foliar data from intensive sampling in a subsetof study sites to examine spatial and temporal variability of foliar nutrients in black cottonwoodtrees. The presentation and analysis of spatial variation in the chapter has been publishedpreviously (McLennan, 1990). Chapter 4 also utilizes intensive sampling in a subset of studysites to examine and evaluate soil nutrient variability. Chapter 5 uses a variety of quantitativeand qualitative ecological variables to assess factors that determine the range of blackcottonwood site index in the 29 sites sampled. Chapter 6 is a fertilization study carried out in 3of the study sites, and examines the response of test trees to the application of fertilizers basedon diagnosis of foliar nutrients. The last chapter briefly discusses some of the more generalresults, and summarizes the major findings of the study.6CHAPTER 2DESCRIPTION AND CLASSIFICATION OF STUDY SITES2.1 STAND SELECTIONTwenty nine stands were selected to represent the range of sites on which blackcottonwood commonly grows in south coastal British Columbia (Tables 2.1, 2.2, and 2.3). Themajority of sites were situated on alluvial floodplains, although upland landforms such asglaciomarine, glaciofluvial, and loess over till landforms were also sampled (Table 2.2).Alluvial floodplain sites were dominated by different Subgroups of Regosol soils, while soils onupland landforms were Gleyed, Sombric, or Orthic Humo-ferric Podzols and Orthic HumicGleysols. Soils were mostly coarse fragment free, although a few sites had a significant amountof coarse fragments (Table 2.3). Soil textures ranged from clay to sand, but generally soils hadpredominantly loamy (silt loam to sandy loam) soil textures. Most sites had Mull humus formsalthough some Moder humus forms were described.Most of the sites selected for sampling supported well-stocked (500-900 stems/ha)deciduous stands dominated by black cottonwood (Table 2.1). However, to sample across theedatopic range of sites on which black cottonwood occurs, it was necessary to include a numberof stands where black cottonwood was not the dominant species. At several sites, blackcottonwoods sampled were scattered among well-stocked plantations of Populus 'robusta' (Table2.1), and these were considered to be ecologically very similar to pure black cottonwood stands.On upland sites, natural stands of black cottonwood do not occur, and black cottonwood iscommon as scattered individuals in a mixture of other deciduous and coniferous species. Standsof this nature were also sampled.7Table 2.1: Site index, site index class, stand age, and relative coverage of trees in the uppercanopy at 29 black cottonwood study sites. Sites are listed in order of increasingblack cottonwood site index and are divided into low (L=8.0-14.9 m/15 years),medium (M=15.0-21.9 m/15 years), and high (H=22.0-30.8 m/15 years) blackcottonwood site index classes.SiteSite Index(m/15 yrs)Site IndexClassStand Age(years)1 Relative % Tree Cover in Main Canopy21 Herrling 8.5 L 18 Ac (100)2. Polygon 19 10.3 L 22 Ac (100)3. Murphy 2 11.5 L 27 Ac (72) / Dr (28)4. Straw 1 11.8 L 23 Ac (100)5. Oyster 12.2 L 49 Dr (60) / Ac (30) / At (10)6. Polygon 20 13.0 L 43 Ac (87) / Dr (13)7. Chilliwack 13.6 L 47 Fd (53)-/ Dr (27) / Ac (20)8. Murphyl 13.9 L 19 Ac (85) / Dr (15)9. Elk 3 14.5 L 49 Ac (75) / Dr (25)10. Elk 1 15.0 M 49 Dr (82) / Ac (9) / Mb (9)11. Chipmunk 163 M 44 Fd (50) / Mb (25) / Ac (13) / Bg (12)12. Elk 2 17.2 M 49 Dr (62) / Ac (38)13. Straw 2 18.5 M 25 Ac (83) / Dr (17)14. Pierce 20.4 M 46 Ac (40) / Dr (25) / Cw (20 / Hw (15)15. Island 12 20.9 M 31 Ac (62) / A rob (30) / Dr (8)16. Squam 38 21.1 M 22 Arob (69) / Ac (31)17. Mercer 21.2 M 38 Arob (65)/ Ac (25 ) / Dr (5) / Mb (5)18. Carey 21.9 M 25 Arob(80) / Ac (18) / Dr (2)19. Salmon 23.0 H 27 Ac (85) / Dr (15)20. Soowahlie 23.0 H 12 Ac (90) / Mb (8) / Dr (2)21. Squam 23 24.4 H 14 Ac (85) / Dr (10) / Mb (5)22. Borden 24.6 H 25 Ac (37) / Dr (60) / Cw (3)23. Tamihi Fan 25.2 H 18 Ac (100)24. Chester 25.7 H 28 Arob (80) / Ac (15) / Dr (3) / Mb (2)25. Tamihi Cr. 26.2 H 15 Ac (83) / Mb (12) / Dr (5)26. Sumas 27.1 H 30 Arob (75) / Ac (15) / (Dr (10) / Mb (5) / Cw (5)27. Squam 29 28.1 H 19 Ac (53) / Dr (35) / Arob (6) / Cw (6)28. Ashlu 28.4 H 21 Ac (60) / Dr (40)29. Ryder 30.8 H 25 Arob (75) / Ac (15) / Ep (8) / Dr (2)1 refers to total age of the stand in 1989 based on the mean age of site index trees2 codes for species are; Ac = black cottonwood; Arob = 'Robusta' hybrid; Dr = red alder; Mb = bigleaf maple; Ep= paper birch; At = trembling aspen; Fd = Douglas-fir; Bg = grand fir; Cw = western redcedar; Hw = westernhemlock8Table 2.2: Location, landform, soil, and ecological classification of 29 black cottonwood studysites.Site LocationElevation(mast) LandformSoilSubgroup'HumusFormeS ubzone/Variant3 Site Association41 Healing Lower Fraser R. 30 floodplain/lb CU.R t.D dm Ac-Willow2. Polygon 19 Lower Fraser R. 30 floodplain/lb CU.R t.D dm Ac-Willow3. Murphy 2 Lower Fraser R. 30 floodplain/mb O.HR J.VL dm Ac-Red osier dogwood4. Straw 1 Lower Fraser R. 20 floodplain/lb CU.R 0.ZL dm Ac-Willow5. Oyster Oyster R. 200 glaciomarine GLHFP J.VL xml Cw-Salmonberry6. Polygon 20 Lower Fraser R. 30 floodplain/mb CU.R t.D dm Ac-Red osier dogwood7. Chilliwack Chilliwack R. 250 glaciofluvial GLHFP J.VL dm Cw-Foamflower8. Murphy' Lower Fraser R. 30 floodplain/Ib O.R O.ZL dm Ac-Willow9. Elk 3 Elk R. 200 glaciomarine O.HG t.D xml Cw-Black twinberry10. Elk 1 Elk R. 200 glaciomarine GLHFP K.VL xml Cw-Salmonberry11. Chipmunk Chilliwack R. 250 glaciofluvial O.HFP O.TD dm Cw-Swordfern12. Elk 2 Elk R. 200 glaciomarine O.HG K.VL xml Cw-Black twinberry13. Straw 2 Lower Fraser R. 25 floodplain/mb O.R O.VL dm Ac-Red osier dogwood14. Pierce Chilliwack R. 250 glaciofluvial GLHFP O.TD dm Cw-Foamflower15. Island 12 Lower Fraser R. 30 floodplain/mb O.R O.VL dm Ac-Red osier dogwood16. Squam 38 Squamish R. 150 alluvial fan O.R t.D dsl Cw-Foamflower17. Mercer Lower Fraser R. 30 floodplain/mb O.HR O.VL dm Ac-Red osier dogwood18. Carey Lower Fraser R. 25 floodplain/mb 0.1-1R K.VL dm Ac-Red osier dogwood19. Salmon Salmon R. 50 floodplain/hb O.HR O.VL xml Ss-Salmonberry20. Soowahlie Chilliwack R. 90 floodplain/hb O.HR K.VL dm Ss-Salmonberry21. Squam 23 Squamish R. 75 floodplain/hb 0.1-1R K.VL dsl Ss-Salmonberry22. Borden Chilliwack R. 100 floodplain/hb O.R tu.L dm Ss-Salmonberry23. Tamihi Fan Chilliwack R. 100 alluvial fan O.R t.D dm Cw-Foamflower24. Chester Lower Fraser R. 15 floodplain/mb O.HR K.VL dm Ac-Red osier dogwood25. Tamihi Cr. Chilliwack R. 100 floodplain/mb 0.1IR K.VL xml Ss-Salmonberry26. Sumas Lower Fraser R. 150 loess/till SM.HFP K.VL dm Cw-Foamflower27. Squam29 Squamish R. 45 floodplain/hb CU.HR tu.L dsl Ss-Salmonberry28. Ashlu Squamish R. 30 floodplain/hb CU.HR tu.L dm Ss-Salmonberry29. Ryder Lower Fraser R. 150 loess/till SM.HFP K.VL dm Cw-Foamflower1 Soil subgroups are identified using Agriculture Canada Committee on Soil Survey (1987); O.R=Orthic Regosol; 0.HR=Orthic Humic Regosol;CU.R=Cumulic Regosol; CU.HR=Cumulic Humic Regosol; O.HFP=Orthic Humo-Ferric podzol; SM.HFP=Sombric Humo-Ferric Podzol;O.HG=Orthic Humic-GI eysol.2 Humus forms are identified using Klinka et al. (1981) with abbreviations from Luttmerding et al. (1990); t.D=tenuic Moder;0.VL=Orthivertnimull; K.VL=Macrovermimull; J.VL=Microvermimull; tu.L=turbic Mull; O.ZL=Orthirhizomull3 CWH subzones and variants identified from Nuszdorfer et al. (1985)4 Site associations were determined from Banner et a!. (1990)9Table 2.3: Actual soil moisture regime (aSMR), soil nutrient regime (SNR) and selected soilphysical properties of 29 black cottonwood study sites.Site aSMR1 SNR2AhDepth(cm)MRD(m)3ARD(m)4SoilVolumeIndex5SoilTextureClass6CoarseFragment(%)Depth toGleying orWater Table7(m)1. Herrling IbF M 0 0.48 0.67 0.67 LS 0 na2. Polygon 19 lbSD M 0 0.22 0.51 0.51 S 0 na3. Murphy 2 mbM R 7 0.60 0.79 0.79 SiL 0 na4. Straw 1 IbM R 3 0.55 > 1.5 1.00 SL 0 na5. Oyster fM R 7 0.45 0.57 0.43 SL 25 0.42g6. Polygon 20 mbM R 0 1.30 > 1.3 1.00 LS 0 na7. Chilliwack M R 7 0.36 0.80 0.78 C 2 0.35g8. Murphyl lb/VM R 3 0.46 1.75 1.00 SL 0 na9. Elk 3 fVM R 1 0.34 0.42 0.42 SiL 0 0.28g10. Elk 1 fM VR 13 0.40 0.51 0.51 SiL 0 0.40g11. Chipmunk SD M 0 0.26 1.03 0.45 S 45 na12. Elk 2 fVM VR 12 0.27 0.35 0.35 SiL 0 0.10g13. Straw 2 mbM R 8 0.63 > 2.2 1.00 SiL 0 na14. Pierce M R 0 0.71 > 1.5 1.00 SL 0 1.00w15. Island 12 mbF R 6 0.59 > 1.5 1.00 LS 0 na16. Squam 38 M R 0 0.49 0.93 0.88 LS 5 na17. Mercer mbM R 9 1.03 1.00 1.00 SL 0 na18. Carey mbM VR 12 0.65 > 1.7 1.00 LS 0 na19. Salmon hbM VR 10 0.83 > 1.5 1.00 L 0 na20. Soowahlie hbF VR 12 0.68 > 2.0 1.00 SL 0 na21. Squam 23 hbM VR 11 0.61 > 2.0 1.00 SL 0 na22. Borden hbF R 0 0.68 > 1.0 0.40 LS 60 na23. Tamihi Fan M VR 2 0.66 > 1.5 0.35 LS 65 na24. Chester mbM VR 17 0.58 > 1.7 1.00 SiL 0 na25. Tamihi Ck. hbM VR 9 0.65 > 1.5 1.00 SCL 0 na26. Sumas M VR 14 0.65 0.91 0.91 Si 0 0.85s27. Squam 29 hbM R 7 0.77 1.13 1.00 LS 0 na28. Ashlu hbM R 1 0.7 > 2.0 1.00 SL 0 na29. Ryder M VR 10 0.55 1.10 1.00 SiL 0 0.95sI Actual soil moisture regime (aSMR) classes are; SD=slightly dry; F=fresh; M=moist, and; VM=very moist and were identifiedusing Banner et a!. (1990) and Green et a!. (1984). For alluvial sites SMRs refer to the moisture conditions when the site is notflooded, and hb, mb, and lb denote flooding regimes for the high, middle, and low bench sites respectively. Sites with poorly-drained, fine textured soils in depressions with winter-summer fluctuating water tables are denoted with an 'f,' and the SMR notedis that during the growing season.2 Soil nutrient regime (SNR) classes are M=nutrient medium; R=nutrient rich, and; VR=nutrient very rich, and were determinedfrom field observations using Banner et aL,(1990) and Green et al. (1984)3 Main rooting depth (MRD) is defined as that depth of soil more or less completely occupied by roots.4 Absolute rooting depth (ARD) is defined as that depth of soil beyond which no roots are found.5 Soil Volume Index is a relative estimate of the soil volume available for rooting, and is based on soil rooting depth, and coarsefragment content. A value of 1 represents an area of 1 ha with unrestricted rooting and no coarse fragments to 1 m, i.e., 10,000m3. Root restricting layers within a depth of 1 m, and coarse fragments reduce the soil volume index relative to this case.6 Soil texture classes are based on laboratory analysis of samples collected over the upper 1 m of soil (or to restricting layer) andhave been identified using Agriculture Canada Committee on Soil Survey (1987); Sand; LS=loamy sand; SL=sandy loam;L=loam; SiL=silty loam; Siilt; SCLandy clay loam; and, C=day.7 w=water table; g=gleyed soil; and, s=seepage water.102.2 BLACK COTTONWOOD SITE INDEX AND STEM ANALYSISBlack cottonwoods selected for stem analysis were canopy dominants or codominants,without physical damage or evidence of disease or suppression. Stem analysis trees were felledat 0.30 m, after which total height of the tree was measured. Based on the difference betweentotal height and breast height (1.3 m), disks were removed at breast height and at 10 equal lengthsegments to the top of the tree. Height of the section above the ground surface was noted for alldisks removed. Stumps were cut off flush with the ground to get an estimate of total age. Thisinvolves some error on alluvial sites because trees may be buried by sedimentation, so that thegermination point can occur somewhere below ground level. Given the rapid juvenile growth ofblack cottonwood, this error was considered to be small.All disks were taken from the field for counting of the annual rings because of thedifficulty in obtaining reliable age estimates from the diffuse porous wood of black cottonwood.All disks were dried in a lumber kiln, sanded with a belt sander, and moistened before countingunder a 10x power binocular stereoscope. All disks were counted until the same age was arrivedat on two separate counts, by two different observers.Height at an index age of 15 years (breast height age) was estimated by first correctingestimated heights to true heights (Carmean, 1972; Dyer and Bailey, 1987), and then using aninterpolation program to calculate total height by 1 year increments. Except for the Strawberry1, Elk 1, Soowahlie, and Squamish 23, all curves were based on the means of three site trees. AtElk 1, only 2 trees were sampled, and at Strawberry 1, Squamish 23, and Squamish 23, meanswere based on the control trees (15 at Strawberry 1 and Squamish 23, and 10 at Soowahlie) usedin fertilizer experiments conducted at those sites.Since a sample size of 15 trees was used at Strawberry 1 and Squamish 23, they can beused to estimate the accuracy and precision of black cottonwood site index estimates where only11three trees were collected. The mean CV for Strawberry 1 and Squamish 23 was 8.3%. Usingan alpha of 0.90, site index means at the sample sites with 3 trees per plot (assuming that thevariances did not differ significantly among sites) were estimated at +/- 15% error.Breast height ages of the stands were distributed fairly evenly between 12 and 49 years(Table 2.1). Site index of black cottonwood showed an almost four-fold increase from 8.5 to30.8 m in 15 years. Estimates of site index for the two stands younger than the index age(Soowahlie and Squamish 23) were based on extrapolation of the distinctly linear height-agecurves that characterizes juvenile height growth of black cottonwood. By dividing thepopulation of study sites approximately by 3, study sites were assigned to low, medium, andhigh site index classes. These groups are used in Chapter 5 to analyze relationships betweenblack cottonwood site index and ecological variables.2.3 ECOLOGICAL DESCRIPTION AND CLASSIFICATIONWithin black cottonwood stands, plots were located by excavating exploratory soil pits toensure sample plots were uniform in general soil properties such as landform, soil subgroup, andhumus form. Plots were also determined to be uniform in the composition and structure of treeand understory vegetation. Such an area delineates a forest ecosystem (Pojar et ca., 1987) andserved as the basic sampling unit for the study. At each sample location a 0.04 ha plot thattypified vegetation, site and soil conditions within the area was used for sampling ecosystemproperties as outlined in Luttmerding et al. (1990). This involved descriptions of site properties(slope, aspect, elevation, landform, mesoslope position, and microtopography), soil properties(soil depth, texture, structure, horizonation, and colour, coarse fragment content, rooting depth,mottling and gleying, and humus form characteristics), and vegetation (percent coverage of allspecies by strata, except epiphytic and epilithic vegetation). Using these observations of site,12soil and vegetation properties, the soil moisture regime (SMR) and soil nutrient regime (SNR)were determined for the plots (Table 2.2) using keys provided in Banner et al., (1990).Estimates of relative soil moisture regime were converted to absolute soil moisture regimefollowing Banner et al. (1990), after which site associations were assigned to each plot (Table2.2).2.3.1 ClimateAll sample sites were located below 250 masl, and across a relatively limited range of theclimatic gradient in coastal British Columbia (Table 2.2). The majority of sites were located inthe CWHdm subzone, with relatively fewer sites in the CWHxm and ds subzones. Also, samplesites in the CWHds and xm subzones were close to the boundary with the CWHdm subzone.Thus, sample sites were all located in cool mesothermal climates with mild, humid winters, andcool, relatively dry summers. In most sample locations sites were level or gently sloping andfew alterations in regional climate due to slope or aspect were anticipated.2.3.2 Soil Nutrient RegimeMedium sites were distinguished from rich and very rich sites by humus forms that wereeither poorly-developed (poor structure) and less than 1 cm in depth (classified as TenuicModers), or well-developed Moders with distinct ecto-organic horizons. Soil colour in nutrientmedium soils was generally light and soil texture coarse. Sample sites assigned to rich and veryrich soil nutrient regimes (SNRs) featured Mull humus forms with different levels of Ah horizondevelopment, and dark, fine-textured soils. Rich sites were distinguished from very rich sitesmostly by having Ah horizons less than 10 cm in depth, or where Ah horizons had been buriedby sedimentation. Sites judged to be nutrient very rich featured Ah horizons deeper than 10 cmand often had an Ah, horizon below the Ah, horizon, where soils were darkly stained by organicmatter, but where the characteristic crumbly structure was not present. Soil textures in nutrientvery rich sites were always loam or finer, at least at the surface. Except for the three nutrient13medium sites, study locations were divided relatively equally between nutrient rich and very richSNRs (Table 2.3). No sites with poor or very poor SNRs supporting suitable black cottonwoodstands were found for sampling. Thus, although sites compared in this study represent only halfof the complete spectrum of SNRs in south coastal British Columbia, the sample represents therange of SNRs on which black cottonwood commonly occurs.2.33 Soil Moisture RegimeAs for SNR, black cottonwood occurs on only a restricted portion of the soil moisturegradient in coastal British Columbia (Table 2.3). Based on qualitative field evaluations of soilphysical properties (Banner et al., 1990; Green et al., 1984; Luttmerding et al., 1990), the SMRfor most sites studied were either fresh or moist (Pojar et al., 1987). Only the Chipmunk sitehad a SMR that was estimated to include a period of soil drought. Most soils sampled alsoreceive flooding or laterally-moving, sub-surface seepage water that provides additional inputsof soil moisture. The site classification of Banner et al. (1990) provides special-caseclassification units to identify sites that receive additional moisture inputs that complicate anevaluation of soil moisture regime based solely on soil physical properties. Sites located onalluvial floodplains (Ac-Willow, Ac-Red-osier dogwood, and Ss-Salmonberry site associations)were inundated either annually, or much more infrequently, depending on their elevation relativeto the flooding characteristics of the river on which they are located. Study sites identified asCw-Salmonberry or Cw-Black twinberry site associations were situated in low-relief, poorlydrained upland landscapes that were subjected to an annually-fluctuating water table. WinterSMRs in these site associations were very moist and wet, and summer SMRs are fresh andmoist, respectively.To provide information on the differences in flooding characteristics on differentbenches within alluvial landforms, on several occasions, water levels were surveyed relative to abench mark established within study sites on the Fraser River (Carey 1 and 2, Strawberry 1 and142), and at Squamish 23 on the Squamish River. These observations were then regressed againsthistorical discharge records from Water Survey of Canada gauging stations (Figure 2.1) near thestudy sites. Because the number of observations was low, the regressions should be consideredas preliminary. However, the very strong linear relationships (R 2s > 0.97) suggested that thismethod provided valid information on flooding parameters within the sites studied. A furthervalidation of the analysis was provided by reference to Hicken and Sichingabula (1988), whoestimated that bankful discharge in the Squamish River occurred at approximately 1,200 m 3/sec,and this corresponded almost exactly with overbank flows at the Squamish 23 site, as calculatedfrom the regression shown in Figure 2.1.Using the regression models shown in Figure 2.1, the discharges that corresponded toflooding at the soil surface, and at a depth of 60 cm in the soil, were calculated. Soil water at 60cm was considered as an index, above which prolonged flooding may be biologically significantin reducing the amount of rooting volume for black cottonwood. For the period of the growingseason (April 15 to September 30) all discharges in excess of these amounts were tabulated, andflooding parameters were summarized for the period of record (Table 2.4).Although the flooding duration at the soil surface for low and middle bench sites wassimilar, flooding frequency was much higher in the low bench sites. On average, the 2 lowbench sites have been flooded above the surface at least once every two years, whereas themiddle bench sites have been flooded above the surface only once every 4 to 6 years. Theduration of soil flooding above 60 cm soil depth has been much higher at the low bench sites,with durations up to a month, on average over the last 24 years. The high bench site sampled(Squamish 23) has had a similar flooding frequency as middle bench sites, but the floodingduration has been much shorter. The high bench site has been flooded above the surface for anaverage of only 1.3 days during the growing season, compared to 17 days for middle bench sites.Strawberry 1 Squamish 23(r2 = .998)( r2 = .969)543sn75 201585'2 4C^3.132100 5000^10000^15000^ 0^200 400 500 8C0 1000 1200Mean Daily Discharge (comic meters/second) Mean Daily Disc.narge (comic meters/second)Carey 1(r2 = .995)Mean Daily Discharge (cam meters/second)Figure 2.1: Regressions of the distance of the soil surface below the benchmark on mean dailydischarge at three alluvial sites. The horizontal line at the Squamish 23 site showsthe soil surface, and indicates the discharge correlated with a bank full situation.16Table 2.4: Frequency and duration of flooding during the growing season (April 15 toSeptember 30) at selected low bench (Ac-Willow), middle bench (Ac-Red-osierdogwood), and high bench (Ss-Salmonberry). Flooding data have been calculatedfor a soil depth of 60 cm and for flooding at the soil surface. Statistics are based onregressions (Figure 2.2) of historical discharge data for the Fraser River at Mission,and the Squamish River at Power House gauging stations (Water Survey OfCanada).Low Bench^Straw berry 1^ Carey 2(Ac-Willow sa.)soil surface 60 cm. soil surface 60 cm.Years of record 24 24 24 24Years flooded 13 20 9 12Frequency (flood/ x yrs) 1.2 1.85 1.4 2Mean Duration (days) 17 27 17 30Middle Bench^Straw berry 2^ Carey 1(Ac-Red osier dogwood sa.)soil surface 60 cm. soil surface 60 cm.Years of record 24 24 24 24Years flooded 6 13 4 9Frequency (flood/ x yrs) 1.85 4 2.6 6Mean Duration (days) 17 17 17 19High Bench^Squamish 23(Ss-Salmonberry sa.)soil surface 60 cm.Years of record 39 39Years flooded 6 16Frequency (flood/ x yrs) 2.4 6.5Mean Duration (days) 1.3 2.12.3.4 VegetationThe black cottonwood communities sampled in this study represented primary, as well aspost-logging secondary successional stages on mostly fresh and moist, nutrient rich to very richsites. The plant species that occurred with black cottonwood reflected this range of site quality -most had nutrient rich to very rich indicator values, and fresh to moist, or wetter, soil moisture17regime indicator values (Klinka et al., 1989b). Alnus rubra, Cornus sericea, Lonicerainvolucrata, and Symphoricarpos albus occurred with black cottonwood on almost all study sites(Table 2.5). Table 2.5 shows the species that can be used to differentiate the sites, given thehierarchical structure of the site associations into upland and floodplain groups. The differentialspecies listed in Table 2.5 had presence class values of III or greater, and were at least 2 classesgreater than that of the group to which they were compared (Mueller-Dombois and Ellenberg,1974). Given the small number of sample plots for each unit, no attempt has been made todevelop a formal diagnostic table for the vegetation data.Floodplain site associations were distinguished from upland site associations by a groupof species indicating nutrient medium (Mahonia nervosa, Rosa gymnocarpa, Rubus ursinus) andrich (Achlys triphylla group) SNRs, and a range (moderately dry to very moist-wet) of SMRindicator status. The floodplain group at this hierarchical level was weakly represented by 3species typical of middle and low bench site associations. For the two upland site associationsan Acer circinatum group, almost all of which had fresh to very moist soil moisture indicatorstatus, differentiated the Cw-Foamflower site association from the 'Gleyed' site association, thatare differentiated by species that have very moist to wet (Spiraea douglasii, Viola glabella,Maianthemum dilatatum), and wet to very wet (Malus fusca, Angelica genuflexa, Carexobnupta) soil moisture indicator status, and reflect the winter flooding that characterizes thesesite units. The Ss-Salmonberry s.a. is differentiated from the other floodplain site associationsby a list of species indicative of nutrient rich SNR status, and primarily fresh to very moist, orvery moist to wet, soil moisture status. The Ac-Red osier dogwood s.a. was poorly-differentiated by semi-agricultural species such as Rubus discolor, Rubus laciniatus, andPopulus robusta, which suggest a history of agricultural land use on Fraser River, middle benchsites. Species that differentiated the Ac-Willow s.a. were weedy, annual herbs that occupyexposed mineral surfaces created by frequent flooding and sedimentation in low bench sites.18Table 2.5: Presence classl/mean percent cover of common and differentiating species for 5 siteassociations sampled in the study, using upland and floodplain site groups as aprimary hierarchical level. The Cw-Swordfern site has been excluded from theanalysis because of only one site (Chipmunk) in the unit.Site Association^A.1^A.2^B.1 B.2 B.3Number of Plots^n=6^n=4^n=7 n=5 n=6ALL STANDS Common SpeciesTrees Anus rubra V/10^5/38^V/17 4/3 IV/8Populus trichocarpa V/30^5/13^V/37 5/26 V/40Shrubs Cornus sericea III/1^4/13^111/5 4/10 V/3Lonicera involucrata IV/5^4/11^IV/3 4/10 V/3Symphoricarpos albus V/7^4/10^IV/5 5/17 V/5A. UPLAND SITES Differential SpeciesShrubs Rosa gymnocarpa HA^3/6Herbs Galium triflorum IV It^4/t III/t la IAMycelis muralis IVA^5/t MA l/t IAPolystichum munitum V/5^5/11 V/4 2/1Pteridium aquilinum III/t^5/3Stachys cooleyae HA^3/3 III/tMosses Plagiominum insigne III/1^4/1 IV/1 2/1A.1 Cw-Foamflower Differential SpeciesTrees Acer circinatum V/9 111/2 1/1Acer macrophyllum 5/2 2/1^V/6 2/1 HAThuja plicata IV/6 V/6 IATsuga heterophylla 111/3 111/2Shrubs Sambucus racemosa V/7 V/6 1/2 l/tCorylus cornuta III/t IA 3/5 11/2Herbs A thyrium filix-femina V/2 3/t^V/4 4/tCarex deweyana V/2 3/t^V/1 3/4 HADryopteris expansa 1111t IIItGeranium robertianum 111/4 III/1Geum macrophyllum MA WitTellima grandiflora 111/2 III/t l/tTolmiea menziesii V/5 11/1 1/3Urtica dioica  III/1 HA 1/1A.2^Fluctuating Water Table('Gleyed') Sites Differential SpeciesTrees Malus fusca 4/3Shrubs Holodiscus discolor^lit 3/1Mahonia nervosa I/1 4/2 IARubus ursinus 5/21 11/2Spiraea douglasii 3/7 1/1Herbs Achlys triphylla^1/t 5/4 l/tAngelica genuflexa 3/4Carex obnupta 4/13Maianthemum dilatatum^lilt 41t 11111 l/tPteridium aquilinum III/t 5/3Viola glabella 4ItMosses Isothecium stoloniferum^IA 4/t IV 11 4It 11/419Table 2.5 (continued):Presence classi/mean percent cover of differentiating species for 5 siteassociations sampled in the study, using upland and floodplain sites as a primaryhierarchical level. The Cw-Swordfern site has been excluded from the analysisbecause of only one site (Chipmunk) in the unit.Site AssociationNumber of PlotsA.1n=6A.2^B.1^B.2n=4^n=7^n=5B.3n=6B. FLOODPLAIN SITES Differential SpeciesShrubs Rubus discolor 2/7 l/t^5/13 III/1Herbs Equisetum arvense 2/t III/t^4/t IV/2Mosses Isothecium stoloniferum l/t 4/t IV/1^4.t 11/4B.1 Ss-Salmonberry (High Bench) Differential SpeciesTrees Acer circinatum V/9 111/2 1/1Acer macrophyllum V/1 2/1 IV/6 2/1 II/tThuja plicata IV/6 V/6 l/tTsuga heterophylla 111/3 III/t l/tShrubs Oplopanax horridusRibes bracteosum11/3litIII/3I11/2Rubus parviflorus IV/9 4/10 V/6 2/2 11/3Sambucus racemosa V/5 V/6 1/2 l/tHerbs Carex deweyana V/2 3/t V/1 3/4 IIItDisporum hookeri III/t 2/t IBAElymus glaucus 11/1 2/t III/t 1/t litCallum triflorum IVA 4/t III/t l/t l/tGeranium robertianum 111/4 III/1Geum macrophyllum III/t IVAMaianthemum dilatatum lilt 4/t III/1 l/tMycelis muralis IVA 5It III/t lit I/tPolystichum munitum V/5 5/11 V/4 2/1Smilacina stellata IIIt III/tStachys cooleyae Hit 3/3 III/tTellima grandiflora III/t MA litMosses Plagiomnium insigne I11/1 4/1 IV/1 2/1B.2 Ac-Red osier dogwood(Middle Bench) Differential SpedesTrees Populus robusta 11/13 11/1 4/30Shrubs Rosa nutkana 2/1 3/2 litRubus discolor 2/7^I/t 5/13 III/1Rubus laciniatus IIIt 3/tB.3 Ac-Willow (Low Bench) Differential SpeciesHerbs Agrimonia striata lit IV/2Agrostis stolonifera 2/t^I/t IV/2Aster hesperius III/tDactylis glomerataHypericum perforatum I/tIII/tIVAMelilotus alba III/tPlantago lanceolata III/tPresence class codes: I = 0-20%; II = 2140%; III = 41-60%; W = 61-80%; V = 81-100%. Roman numerals are used only when the numberof plots for the group is > 5. Mean percent cover < 1 denoted by t ("trace")20CHAPTER 3SPATIAL AND TEMPORAL VARIABILITY OF BLACK COTTONWOODFOLIAR NUTRIENT CONCENTRATIONS3.1 INTRODUCTIONThe use of foliar nutrient analysis for the determination of forest stand nutrient status iscomplicated by spatial variability in foliar nutrient concentrations both between trees in the standand within the canopy of individual trees. Although spatial variability of many hardwoodspecies has been investigated by a number of workers in North America (Baker and Russell,1975; Blackmon and White, 1972; Ellis, 1975; Guha and Mitchell, 1965a,b; Hoyle, 1965; Lea etal., 1979a,b; Mitchell, 1936; Mitchell and Chandler, 1939; Morrison, 1972, 1974, 1985; Tamm,1951; Wallihan, 1944; Woodwell, 1974) there is little information on black cottonwood.Compared to samples from the lower canopies, Heilman (1985) found significantly higherconcentrations of foliar N in the upper canopies of 6 year-old black cottonwood trees.Blackmon and White (1972) showed similar differences between foliage from the upper andlower crowns for foliar N in Populus deltoides, although foliar P values did not varyappreciably. Guha and Mitchell (1965a) found lower concentrations of Co, N, Fe, V, Ti, Cr, Pb,and Al in upper canopy foliage of Acer pseudoplatanus, Aesculus hippocastanum, and Fagussylvaticum, but for Mo, Zn, Ca, Mn, B, Si, Cu, Sr, Ba, Mg, and P intra-canopy differences wereinsignificant and rarely exceeded 20-30%. In Betula alleghaniensis only Ca was significantlyhigher in the upper canopy foliage, while in Acer saccharum Ca, K, Mg, Fe, Zn, and Na weresignificantly higher in the lower canopy (Morrison 1985). Similar within-canopy variationtrends in independent studies of Acer saccharum (Ellis 1975; Morrison 1985) show thatvariation may show some consistent trends within the same species on different sites but morecomparative studies are needed to confirm this. As a result of within-canopy variation in foliarnutrient concentrations, most workers standardize their sampling methodologies so that foliagesamples are collected from the same canopy position and the same types of leaves.21A second component of spatial variability important for standardizing foliage samplingprocedures is the determination of the number of samples required to obtain estimates ofpopulation parameters that meet desired accuracy and precision criteria. Coefficients ofvariation for the different foliar nutrients have been published for a number of hardwood species(Ellis, 1975; Guha and Mitchell, 1965a; Morrison, 1985). Morrison (1985) recommended that30 Acer saccharum trees be sampled for estimates of macronutrient concentrations, and 40-70trees for micronutrients, at an allowable error of 10% with a 0.95 significance level. Aconsistent trend in these findings is that foliar nutrient concentrations for macronutrients areconsiderably less variable than for micronutrients, but that, even in the least variable nutrients, amajor sampling effort is required to attain high levels of statistical accuracy and precision.Foliage nutrient concentrations of both coniferous and hardwood trees have beenreported to fluctuate over the growing season as a result of dilution effects as leaves enlarge,internal translocation of mobile nutrient elements, leaching of foliage nutrients, andenvironmental factors (Day and Monk, 1977; Guha and Mitchell, 1965a,b; Knight, 1978; Lea etal 1979a,b; McHargue and Roy, 1933; Mitchell, 1936; Sampson and Samish, 1935; Tamm,1951; Wells and Metz, 1963; White, 1954; Woodwell, 1974). In conifers, growing seasonfluctuations in foliar nutrient concentrations have led most researchers to restrict foliar samplingto the fall or winter months (Ballard and Carter, 1986; Leaf, 1973; Lavender, 1970; van denDriessche 1974), although the diagnostic precision of foliar data collected outside of the growingseason has been questioned (van den Driessche, 1974).In hardwoods, foliage collection must occur during the growing season and this has ledto a number of investigations that attempt to document seasonal changes and use them todetermine the best time for sampling (Lea et. al, 1979a,b; Guha and Mitchell, 1965a,b; Leaf,1973; Leyton, 1948; Mitchell, 1936; Tamm, 1951). A similar pattern in many of these studies isfor N, P, and Mg concentrations to decrease in the early part of the growing season as leavesenlarge, remain fairly stable over the growing season, and then decline sharply at the end of thegrowing season as mobile macronutrients are translocated out of foliage before abscission (Day22and Monk, 1977; Lea et al. 1979 a,b, 1974; White, 1954). Considerable fluctuations in K havebeen reported during all periods of the growing season (Tamm, 1951; Day and Monk, 1977).Non-mobile nutrients such as Ca and many micronutrients, gradually increase in concentrationand show a rise in concentration towards the end of the season as mobile nutrients are removed(Lea et al. 1979b; Tamm, 1951). Heilman (1985) documented a significant decrease in foliar Nconcentrations in Populus trichocarpa after the middle of August. Based on these trends mostresearchers have recommended sampling hardwood foliage during the latter period of thegrowing season but before yellowing begins (Lea et al. 1979a,b; Leyton, 1948; Mitchell, 1936;Tamm, 1951), as this is the period of highest stability for most of the important macronutrients.A related aspect of temporal variability in hardwood foliage concentrations is the amountof variation that can be expected from one year to the next in foliage samples collected in thesame manner, from the same trees, and during the same seasonal period (Atterson, 1965 to 1970,reported in van den Driessche, 1974; Bickelhaupt et al, 1979; Hoyle, 1965; Leaf et al, 1970;Verry and Timmons, 1976). Significant year to year changes in foliar N concentrations havebeen reported in the first six years of a black cottonwood plantation by Heilmann (1985) but it isnot clear how this applies to older trees. Variation from year to year in foliage concentrationshave been attributed to internal reactions to external factors. For example, Miller (1966)correlated a number of climatic variables with foliage nutrient concentrations and found thataverage mean and maximum daily temperatures in the period preceding sampling wereconsistently correlated with fluctuations in foliage nutrient concentrations. Soil conditions havealso been implicated as a factor influencing foliage nutrient concentrations (Hoyle, 1965; Pharisand Kramer, 1964; Walker, 1962). Compared to well-watered clones, Broadfoot and Farmer(1969) documented significantly higher N and slightly higher P foliage concentrations for youngPopulus deltoides grown under conditions of moisture stress. Based on a review of theliterature, van den Driessche (1974) concluded that the reports of significant year to yearfluctuations in foliar nutrient concentrations were to be expected since the factors whichdetermine these variables also fluctuate from year to year.23Clearly, knowledge of spatial variability of the different foliar nutrients within thepopulation being studied is a prerequisite for drawing reliable conclusions from foliar nutrientdata (Morrison 1985; Woodwell 1974). Also, since sampling of black cottonwood foliage mustbe carried out during the growing season when considerable temporal variation in foliarconcentrations may occur, seasonal and year to year fluctuations in foliar nutrient concentrationsshould also be studied (Tamm, 1951; van den Driessche, 1974). The specific objectives of thisstudy were;1) to evaluate the magnitude and nature of within tree, among tree, and among site variation infoliar nutrients of black cottonwood, in stands of various ages and from a range of locations;2) to evaluate the magnitude and nature of seasonal and year to year temporal variation at someof the study sites to test the assumption of relative stability of foliar nutrient concentrations inthe latter half of August, and;3) to utilize these observations to recommend the most efficient sampling strategies forevaluating the nutrient status of black cottonwood stands using foliar nutrient sampling.3.2 METHODS3.2.1 Site Selection and DescriptionSites were selected to represent a range of different-aged black cottonwood stands onalluvial sites in several locations in coastal British Columbia. Stands ranged in age from 2 to 43and represented both naturally-regenerated stands and plantations. Black cottonwood standsfrom a wide geographic range encompass considerable genetic variation and Heilman (1985) hasshown how clone effects can influence concentrations of foliar nutrients. By including geneticand age variation in the sample design, estimates of foliar variability from this study can be usedto estimate sample size requirements for a wide range of black cottonwood stands in coastal24British Columbia. All sites sampled had fresh to moist soil moisture regimes (Pojar et al. 1987)with variations in the frequency and duration of flooding, medium to rich soil nutrient regimes(Pojar et al. 1987), and were located within a cool, mesothermal climate (see Tables 2.1 - 2.3).3.2.2. Foliar SamplingAt each of the seven locations the stand was divided into 15 (13 at the Carey II site)approximately even-area plots and a random process was used to select a sampling point withineach; the closest healthy, dominant or codominant black cottonwood was selected for foliarsampling. Foliage samples were collected between August 23 and 27, 1985 by a variety ofmethods (clipping with a pole pruner, tree felling, and shooting) depending on stand height andcanopy characteristics, and followed recommendations of Mitchell (1936). Black cottonwood ischaracterized by heterophyllous foliage so that two types of leaves, preformed, early leaves andlate leaves, are found within the same branch. Critchfleld (1960) has shown that the preformedleaves are formed in the bud in the previous year and expand rapidly with spring growthinitiation. The first late leaves develop from arrested primordia in the bud, while those formedlater in the growing season are initiated from the apical meristem as internodes elongate. Lateleaves continue to develop as long as growing conditions remain favorable Critchfield (1960).In this study, black cottonwood late leaves were easily distinguishable from early leaves by theirlarger size and darker green colour. For comparisons among the seven sites only the mostrecently matured late leaves were sampled and this meant avoiding both the early leaves and thenewly-formed, apical late leaves. Using these sample selection criteria, 30 g fresh weightfoliage samples were collected from lateral branches within the upper one third of the canopy atall locations. To compare variation within individual trees, samples of recently matured, lateleaves were also collected from the middle one third and the lower one third of the canopy at theSoowahlie site, and from the lower one third of the canopy at the Carey II site. At Carey II,foliage samples were also collected from newly formed, apical late leaves for comparison withrecently matured, late leaves.25Samples for the within-year analysis were collected in 1985 from 15 sample trees at theSoowahlie site on June 4, July 5, August 1, August 25, September 28, and October 15. Thesesamples were collected from the upper canopy using the protocol described above. All samplesexcept the August 25 sample were composited into one sample for analysis, and thus an estimateof sample variability is only possible for the August sample. Samples for the year to yearcomparisons were based on foliar analyses of control trees at the three sites (Strawberry 1,Squamish 23, Soowahlie) used for the fertilizer experiments (Chapter 6). All 1985, 1986, and1988 concentrations are means of 15 (Strawberry 1 and Squamish 23) or 10 (Soowahlie)individual samples. In 1987 a composited sample was collected, so no estimates of variabilityare available for the 1987 samples.All foliage samples were placed in paper bags and air-dried briefly until they could beoven-dried at 70°C for 24 hours, and then ground to pass a 20-mesh screen.3.2.3 Laboratory AnalysisFoliar concentrations of N, P, K, S, SO 4-S, Ca, Mg, Cu, Zn, Fe, active-Fe, Mn, and Bwere determined using the following procedures. One-gram samples were wet ached followingParkinson and Allen (1975), followed by colorimetric analysis for N (phenol-hypochloritemethod) and P (unreduced vanadomolybdate complex), and atomic absorptionspectrophotometry for K, Ca, Fe, Mg, Mn, Zn and Al. Copper was determined by digestion innitric acid and hydrogen peroxide followed by atomic absorption spectrophotometry. Boron wasdetermined by dry ashing followed by colorimetric analysis by the azomethine H method(Gaines and Mitchell, 1979). Active-Fe was extracted by a modification of the method ofOserkowsky (1933) using 1 M HCl and analyzed using atomic absorption spectrophotometry.Sulphur was analyzed using a Fisher Sulphur Analyzer, as described by Guthrie and Lowe(1984). The method of Johnson and Nishita (1952) was used to assess concentrations of SO 4-S.Macronutrients were expressed as percentage concentration and micronutrients as parts permillion (ppm) of oven-dry mass.263.2.4 Statistical AnalysisPrincipal component analysis was used as an exploratory technique to reveal variancetrends in within-canopy foliage concentrations and to reduce the overall complexity in thevariables measured so that predominant patterns and potential variable groupings could beidentified (Gauch 1984; Pielou, 1975). For within-canopy foliage samples 95% confidenceellipses (Jolicoeur and Mosimann, 1960) around centroids of group PCA scores wereconstructed to evaluate relationships among the different canopy locations.At both the Carey II and Soowahlie sites the single-factor, one-way ANOVA model wasused to test the hypothesis of no significant difference among canopy strata for each of the 13nutrients. The foliar nutrient concentration data met the criteria of being normally-distributedvariables from a random sample and, in most cases, homogeneity of variance was achievedthrough logarithmic transformations of those samples that did not satisfy the Bartlett test. Sincethe treatment effect was random and quantitative, and the objective was to compareconcentrations of individual nutrients among canopy strata, Duncan's multiple range test wasused to compare means (Mize and Schultz, 1985).The SASCAL program (Marshall 1987) was used to compute the numbers of samplesrequired at several levels of accuracy and precision. Required sample sizes were calculatedusing both alpha (a) and gamma (g) levels of significance since workers may want to use thisinformation to decide on the numbers of samples to composite for desired levels of precision(Ballard, 1985; Marshall and Jahraus, 1987; Marshall et al., 1992). Gamma significance levelsdetermine the probability that the confidence limit of the sample taken does not exceed thepercentage error term used to determine the number of required samples. When g levels are notspecified, a value of 0.50 is assumed and this may be an undesirably high probability forcomposited samples. When a levels alone were considered (g = 0.50) the following formulafrom Freese (1962) was used to calculate required sample sizes;[2] N2—^PE 2CV 2 x F(1-a)(1,n-1) X F(1-g)(n-1,0)272t (1_a/2, ni _o X CV 2[1]^N1 =^PE 2where, N 1 = predicted sample size, t = t statistic for desired a and sample n, CV = coefficient ofvariation, n1 = size of pilot sample 1, and PE = percentage error. Where both a and g wereconsidered in the determination of sample size, the following equation (Harris et al. 1948) wasused;where N2 = the predicted sample size adjusted by g, CV and PE are the same as above, F = thevalue of the F statistic for desired levels of a and g, and Q = the appropriate degrees of freedomassociated with the estimate of the CV. Both equations used assumed that the population ofpotential foliar samples in the stand was sufficiently large so that finite population correctionswere not necessary.283.3 RESULTS3.3.1 Within-tree Spatial VariationPrincipal components analysis (PCA) using concentrations of 13 foliar nutrients fromdifferent locations within the canopies of black cottonwood stands at the Carey II and Soowahliesites showed similar trends (Tables 3.1 and 3.2, Figure 3.1). The first two principal componentsaccounted for 66% of the variation in the data set at the Carey II site (Table 3.1) and 63% atSoowahlie (Table 3.2). At both locations the first principal component revealed a contrastbetween foliar nutrients with high positive (N, S, P, Cu, K, SO 4-S) and those with high negative(Mn, Ca, Zn) loadings. Magnesium and B, although positive at both sites, were much lower atSoowahlie than at Carey II. The similar relationships of nutrients along the first PCA axis atboth sites demonstrates a consistent gradient from leaves relatively high in N, S, P, Cu, K, andSO4-S, and low in Mn, Ca, and Zn, to foliage where the relative levels of these two nutrientgroups are reversed. At both sites Fe and active-Fe had the highest values on the second PCAaxis, suggesting that variation in the concentrations of these nutrients are controlled by differentprocesses than those that determine foliar concentrations of the other nutrients.The positions of foliage samples with respect to the first and second axes of the PCAs ofsamples from different locations within the canopies at the Soowahlie (45 samples) and Carey II(39 samples) sites is shown using 95% confidence ellipses in Figures 3.2. At the Soowahlie site(Figure 3.2) ellipses from the upper and lower canopies had very little overlap while the ellipsefor the middle canopy samples overlapped both. This pattern shows the distinctness of foliarnutrient concentrations in the upper and lower canopy at this site. The overlapping of middlecanopy foliar samples with both upper and lower canopy samples suggests a gradient ofchanging foliar nutrient concentrations within the canopies of the black cottonwood treesstudied. A similar but more pronounced separation of data swarms occurred at the Carey H site(Figure 3.2) where 95% confidence ellipses for the three groups of samples (upper canopy,lower canopy, apical foliage from the upper canopy) did not overlap.29Table 3.1: Component loadings, eigenvalues, and % variance explained for PCA axes 1 and 2for 13 foliar nutrients at the Carey II site.Nutrient PCA 1 PCA 2N 0.940 0.022P 0.901 0.000K 0.735 0.194Ca -0.522 0.470Mg 0.690 0.314S 0.913 0.159SO -S 0.691 0.0644Cu 0.800 0.142Zn -0.676 0.280Mn -0.786 0.312B 0.544 0.373active-Fe -0.196 0.874Fe -0.031 0.886Eigenvalues 6.242 2.271% Variance 48.01 17.47Table 3.2: Component loadings, eigenvalues, and % variance explained for PCA axes 1 and 2for 13 foliar nutrients at the Soowahlie site.Nutrient PCA 1 PCA 2N 0.718 0.202P 0.946 0.088K 0.829 0.064Ca -0.827 -0.008Mg 0.123 -0.267S 0.949 0.072SO -S 0.901 -0.0534Cu 0.870 0.020Zn -0.448 0.455Mn -0.765 0.118B 0.316 -0.240active-Fe -0.014 -0.894Fe -0.118 -0.903Eigenvalues 6.141 2.020% Variance 47.24 15.541.0active—Fe^Fe0.5 CaMa znBMgP.,v1. K CUge7C-I0.0SOp4CV-.ItC,a.—0.5—1.0-1.0^—0.5^0.0^0.5^1.0PCA 1 — Carey II21.00.5 Zn.Z. N;0ria010.0Mn-Ca^ - K_-Cu-SO:fcv lit^BE—0.5active—FeFe—1.0—1.0^—0.5^0.0^0.5^1.0PCA 1 — SoowahlieFigure 3.1: Ordinations of within canopy variation of 13 foliar nutrients on the first and secondprincipal component axes at the Carey II (top) and Soowahlie (bottom) sites.3031PCA 2Figure 3.2: (top) Ordination of 45 foliar nutrient samples and 95% confidence ellipses forrecently matured late leaves at upper (A), middle (B) and lower (C) canopy positionson the first and second PCA axes at the Soowahlie site.(bottom) Ordination of 39 foliar nutrient samples + 95% confidence ellipses forrecently matured, upper canopy leaves (A), newly-formed, apical upper canopy lateleaves (B), and lower canopy late leaves (C) on the first and second PCA axes at theCarey 11 site.32ANOVA on non-transformed and log-transformed data, and Duncan's multiple range testfor 13 nutrients at different positions within black cottonwood canopies at the Carey H andSoowahlie sites, showed that there were significant (p < 0.05) differences in concentrations of allnutrients except for Fe at Carey II and Fe, active-Fe, and Mg at Soowahlie (Tables 3.3 and 3.4).A similar trend was observed at both sites in that N, P, K, S, SO 4-S, Mg, and Cu weresignificantly higher in the upper canopy foliage and Ca, Fe, active-Fe, Mn, and Zn weresignificantly higher in foliage collected from the lower canopy. Boron concentrations followeda different pattern with highest concentrations in upper canopy foliage at the Carey H site and inthe middle canopy samples at Soowahlie. At the Soowahlie site middle canopy foliar nutrientconcentrations were intermediate between the upper and lower canopy extremes for all elementsexcept Fe and B. The pattern showed increasing concentrations of N, P, S, SO4-S, Mg and Cuand decreasing concentrations of Ca, Fe, active-Fe, Mn, and Zn with increasing height in thecanopies of black cottonwood trees at the two sites. At Carey II apical foliage containedsignificantly higher concentrations of N, P, S, and Cu and significantly lower concentrations ofZn and Mn compared to mature late leaves on the same branch in the upper canopy.3.3.2 Within-site and Among-site Spatial VariationMean coefficients of variation (CVs) for 13 elements in black cottonwood foliage atseven sites were lowest for N, P, K, S, Mg, and Ca (12-17%), intermediate for active-Fe (22%),and highest for SO4-S, Cu, Zn, Mn, B, and Fe (26-37%) (Table 3.5). In general, macronutrientconcentrations were characterized by relatively low levels of variation, and variation of foliarmicronutrients was generally considerably higher. Foliar SO 4-S was the outstanding exceptionwith a mean CV of 37.4%, the highest of all nutrients studied.F ratios shown in Table 3.5 compare among-site to within-site variance for the sevensample sites and all were highly significant (p < .001). These comparisons suggest that, eventhough there is considerable variation of foliar nutrient concentrations within sites, variationamong sites is significantly and consistently higher.Table 3.3: Mean foliar nutrient concentrations (n=13) for apical, upper canopy„ and lower canopy foliage, and ANOVA, at the Carey I Isite. For a given nutrient, figures followed by the same letter are not significantly (p=0.05) different.Position^ N2^I'^K^Ca^Mg^S^SO4-S^Cue^Zn2^Mn^B^Active Fe^Fe(%)^(%)^(70^(70)^(%)^(%)^(ppm)^(ppm)^(ppm)^(ppm)^(ppm)^(ppm)^(Pr")Apical leader^2.47a^0.3311a^I.88a^1.15a^0.219a^II.305a^1162a^13.0a^97.7a^64.5a^33.6a^91.0a^156aUpper canopy 1.79b^0.218b^1.89a^1.22a^0.215a^0.279b^1406b^9.39b^122.7b^84.4b^34.3a^89.3a^168aLower canopy^1.33c^0.165c^1.54b^1.466^0.182b^0.211c^660c^6.85c^170.8c^I05.8c^26.8b^106.11b^I85aSignificance' •••^••• ••• ••• • ••• * ••• ••• ••• • •^NSI significance of the ANOVA at p = .05 ('), p = .01 ("), and p = .001 (***)2 variables for which ANOVA was carried out on log-transformed data to stalisfy requirements for homogeneity of variance or normalityTable 3.4: Mean foliar nutrient concentrations (n=15) for upper, middle and lower canopy foliage, and ANOVA, at the Soowahlie site.For a given nutrient, figures followed by the same letter are not significantly (p=0.05) different.Position^ N^I'22 ^Ca^Mg^S2^$04 -S^Cue^Zn2^Mn2^B^Active Fe e^Fee(%)^(%)^(%)^(%)^(%)^(%)^(ppm)^(Pr")^(PPI")^(PPnl)^(PPrn)^(ppm)^(ppm) Upper canopy^2.42.i^0.239a^2.04a^0.881a^II.227a^0.335a^1648a^14.4a^102a^27.5a^36.7a^88.4a^124aMiddle canopy 2.01b^0.172b^1.33b^1.45b^0.205a^0.225b^752b^8.13b^130ab^43.7b^26.7b^90.0a^112aLower canopy^1.8313^0.142c^1.1313^1.60b^0.219a^0.192c^341c^7.47b^153a^46.3b^35.9a^95.5a^I58aSignificance' •••^••• ••• ••• NS ••• ••• ••• • ••• ••• NS^NSI significance of the ANOVA at p = .05 (*), p = .01^and p = .001 (***)2 variables for which ANOVA was carried out on log-transformed data to stalisfy requirements for homogeneity of variance or normalityTable 3.5: Coefficients of variation (CV), mean CVs, and F ratios comparing among- and within-stand variance for foliar nutrientconcentrations (n=13 at Carey II: n=15 at all other sites) in upper canopy foliage or black cottonwood at seven alluvial sites. All F ratiosare significant at p < .001.Site N(%)I'(70)K(%)Ca(70)Mg(%)S(%)SO4 -S(P13m)Cu(PPni)Zn(PPm)Mn(PP(33)13(PPHI)Active Fe(ppm)Fe(ppm)S00%vahlie 13.6 16.3 17.0 21.3 15.4 15.4 38.9 38.2 14.0 18.6 20.4 19.7 66.4Carey I 15.7 8.3 7.7 9.1 12.8 16.1 31.7 19.5 14.3 19.1 14.0 18.7 18.9Carey II 12.7 11.5 9.5 14.3 10.2 13.0 7A.0 42.5 22.9 17.3 26.7 18.4 28.6Scott Nursery 15.1 25.1) 17.1 15.3 12.4 16.6 51.8 17.5 51.1 43.6 11.0 23.9 27.9Strawberry 1 10.8 9.11 18.7 9.6 7.5 9.3 30.9 13.7 32.4 13.8 32.3 15.1 16.6Squainish 23 12.4 19.4 16.7 27.9 16.3 12.7 45.5 26.2 24.8 43.8 49.4 16.8 22.8I lornalliko 5.8 13.1) 11.5 20.1 30.4 10.0 39.1 28.0 49.9 22.0 32.2 43.1 44.7Mean CV 12.3 14.6 14.0 16.8 15.0 13.3 37.4 26.5 29.9 25.5 26.6 22.2 32.3SD 3.1 5.6 4.0 6.3 6.9 2.7 8.7 9.9 14.3 11.8 12.1 8.9 16.3F Ratio 20.3 18.1 25.3 12.1 53.2 29.5 15.5 86.4 17.4 33.8 71.3 22.4 13.235Since the mean CV for each nutrient was used to calculate generalized sample sizerequirements (Table 3.6), the wide ranges of CVs shown in Table 3.5 for all nutrients hasimportant implications for composited samples, where the CV cannot be calculated. Compositedsamples collected from sites where the mean CV is exceeded would not meet the expected levelsof precision for the nutrient considered and the size of the sample collected.Table 3.6: Numbers of black cottonwood foliage samples required at various levels of percentallowable error, and significance for 13 foliar nutrients.1 - a = 0.951 - g = 0.501 - a = 0.901 - g = 0.501 - a = 0.951 - g = 0.951 - a = 0.901 - g = 0.80Nutrient 5% 10% 5% 10% 5% 10% 5% 10%N 26 8 18 6 38 13 23 8P 35 11 25 7 51 17 31 10K 33 8 23 7 47 16 29 9Ca 46 13 32 10 65 21 40 12Mg 37 11 26 8 53 17 32 10S 30 9 21 7 43 15 26 9SO4-S 217 56 153 40 288 79 178 48Cu 110 29 78 21 149 43 92 26Zn 140 37 99 26 187 53 116 32Mn 102 27 72 19 139 40 86 24B 111 30 78 20 150 43 93 26Active-Fe 78 21 55 15 107 32 66 19Fe 163 43 115 30 217 61 135 37The numbers of samples required for different levels of accuracy and precision for the 13nutrient elements analyzed in this study are given in Table 3.6. Patterns in the table are similarto patterns in the coefficients of variation (Table 3.5) from which the sample numbers werecalculated. In general, for a given level of accuracy and precision, lower numbers of samplesare required for N, P, K, Mg, S, and Ca and higher numbers for the other elements. Samplesizes required for high levels of accuracy and precision are large and would be expensive to36collect and analyze, even for the less variable nutrients. For example, a significance level of0.95 with an allowable error of 5% would require at least 26 samples for N, the least variableelement, and 217 samples for SO 4-S, the most variable element. Sample size requirements for Nand SO4-S increase to 38 and 288 respectively if the a and allowable error are kept the same andg significance raised from 0.50 to 0.95. Changing the allowable error to 10%, and keeping the asignificance level at 0.95 (g = 0.50), would mean that the least variable elements (N, P, K, S,Mg, Ca) could be reliably estimated with less than 15 samples (Table 3.6). However, a sampleof 15 trees would be well under the sampling requirements of the other elements, which rangefor 21 for active-Fe to 56 for SO4-S at this level of accuracy and precision. For compositedsamples the level of g significance should be considered; 20 samples would provide estimateswithin 10% of the mean with an a and g significance of 95% for N, P, K, Mg, and S. This mayrepresent a desirable level for operational purposes, where these common nutrients are mostoften considered and where costs of sampling are important.3.3.3 Seasonal VariationThe foliar concentrations of the 13 nutrients analyzed showed considerable fluctuationsover the 1985 growing season at the Soowahlie site (Figure 3.3). Foliar concentrations of N, P,K, and S exhibited an identical pattern in that concentrations were relatively constant from earlyJune to early August, a major concentration peak occurred on the August 25 sampling date, afterwhich concentrations declined rapidly as leaves aged before abscission. The pattern exhibitedby foliar concentrations of Mg, Fe, and Cu was to increase up to August 25, decline until mid-September, and then increase in concentration just prior to abscission. SO 4-S, concentrationspeaked during the middle of August, then declined, and finally rose during leaf abcission. Caconcentrations showed little change over the growing season, and an increase just prior toabscission. Zn, active Fe, and Mn had only small variations in foliage concentration over the1985 growing season.120 16003June 4 July 5 Aug 2 Aug 25 Sept 28 Oct 150.35 Ai ‘^/ ^\0.3 ^ /^ A /A^ \^Mg2 /:a •■^.^i ^♦^•‘...>, 0.25 ^ -L -a i ,^ •.,',.-.. ,,,r^‘,•^,,,,_^,,,,,^s....^,^. -..:-..^.... , •0,^0.2*^1....reak.... A^. •0.15*0.1z0.05June 4 July 5 Aug 2 Aug 25 Sept 28 Oct 15370.51400 ^ /1,—,..-.•^8 0...,-.:Tz=Z 60 ''QA"=0,...E 7..1  V^^- ^-.............7; 800-0■...=,-, 1000-=7///-/IL II \100^ \\ \ T, !1 ///5041200-^ t^ it'=^ =0 73^ iC.) Mn^:...1/.".t..,. = 6C0- /^.7-7 40* —leU/•z:^- ...^..^,,. —^/S•.. ...-♦., 1Z ;t---4<.^♦ . -•^• rt 400* r^ ^•^.... B.  -...-^...-•• -•^ \^. .. 20+m2001- Fe•CuZn0June 4^July 5^Aug 2 Aug 23 Sept 28 Oct 15Figure 3.3: Seasonal fluctuations in concentrations of 13 foliar nutrients in the upper canopy of a10 year old black cottonwood stand (Soowahlie) over the 1985 growing season.0 ,June 4 July 5 Aug 2 Aug 25 Sept 28 Oct 1538None of the nutrients examined exhibited a pattern of steep concentration decline in theearly part of the season (Day and Monk, 1977; Guha and Mitchell, 1965a; Hoyle, 1965,Mitchell, 1936; ) and it is assumed that, because leaves emerged on the Soowahlie site in lateMarch, by June 4 there was little dilution effect due to expanding leaf size. The steepconcentration decline in N, P, K, and S at the end of the growing season on the October 15sampling date is well documented for hardwoods (Day and Monk, 1977; Guha and Mitchell,1965a; Hoyle, 1965; Lea et aL 1979 a; Mitchell, 1936) and is interpreted as showing thetranslocation of important mobile macronutrients out of the foliage for winter storage prior toleaf abscission. Increases in foliar concentrations of Ca, Cu, and Fe are interpreted as a result ofthe translocation of mobile macronutrients such as N, P, K, and S out of the foliage at that time.Foliar concentrations of N, P, K, S, Mg, Fe, Cu all demonstrated a pronounced seasonalpeak on the August 25 sampling date, and this is almost the same group of nutrients (B and SO 4-S are absent) for which canopy concentrations were significantly higher in the upper canopyfoliage at the Soowahlie and Strawberry sites. Given the ability of this group of nutrients to bemobile in the phloem, it appears that their foliar nutrient concentrations may fluctuate in arelatively unpredictable manner over the growing season in black cottonwood stands.3.3.4 Year to Year VariationSignificant (p < .05) fluctuations in the concentrations of P, K, Ca, S, SO4-S, and Mnoccurred at the Soowahlie, Strawberry 1, and Squamish 23 sites from year to year in 1985, 1986,and 1988 in foliar samples collected from the same trees, using the same sampling protocol, withall sampling carried out during the last two weeks of August in each year (Table 3.7). In 1987,samples from the control trees were composited into one sample for analysis and were thereforenot used for the ANOVAs shown in Table 3.7. Year to year patterns for all nutrients are shownin Figures 3.4 and 3.5. Significant year to year differences in foliar concentrations of Mg, Cu,Table 3.7: Mean foliar nutrient concentrations (n=1(1 at Soowahlie; 15 at Squamish 23 and Strawberry 1) from upper canopy foliage ofsamples collected in the last 2 weeks of August from the same sample trees in 1985, 1986, and 1988. For a given nutrient, and at the samesite, figures followed by the same letter are not significantly (p=0.(15) different.Location YearN(70)l'(70)K(70)Ca(70)Mg(70)S(%)SO4 - 5;(PPril)Cu(P1")Z n(PPm)Mn(PPrn)B(PP111)Active Fe(ppm)Fe(Ppm)Soowahlie 1985 2.42a 0.'7,1a 2.04a 0.88a 0.23a 0.34a 1648a 14.4a 102a 27.4a 36.7a 88.4a 124a1986 2.47a 0.22a 1.67b 1.136 0.21a 0.23b 452b 10.0a 107a 30.9a 28.6b 94.5a 128a1988 2.4Ia 0.34b 2.47c 0.69c 0.22a 0.28c 960c 13.3a 95.6a 22.3b 49.5c 84.2a IllaSignificance' NS ••• ••• ••• NS2 ••• ••• NS2 NS2 •2 •• NS2 NSSquamish 23 1985 2.38a 0.2Ia 1.76a 0.84a 0.25a 0.31a 994a 15.7a 96.9a 36.8a 17.7a 75.2a 141a1986 2.46a 0.19a 1.59b 1.25b 0.22b 0.21b 352b 9.93b 115b 37.4a 15.7a 62.0b 80.5b1988 2.39a 0.29b 2.50c 0.51c 0.21b 0.28a 603c 16.7a 87.1a 18.1b 18.9a 61.7b 74.3bSignificance' NS ••• ••• ••• •• ••• ••• ••• •2 ••• NS2 •• •••Strawberry 1 1985 I.95a 0.18a 1.14a 1.15a 0.33a 0.22a 557a 7.67a 96.3a 66.8a 31.3a 83.2a 148a1986 2.09a 0.21b 1.256 1.07a 0.30a 0.20ab 444ab 7.47a 96.1a 58.9a 26.5a 85.3a 130a1988 1.616 0.19a 1.66c 0.55b 0.24b 0.18b 373b 8.60b 61.76 15.2b 31.8a 61.6b 73.3bSignificance' ••.2 •• ••• ••• ••• ••• • •2 .••2 ••• NS ••• •'^significance of the ANOVA at p = .05 (•), p = .01 ••), and p = .001 (••)2 variables for which ANOVA was carried out on log-transformed data to statisfy requirements for homogeneity of variance or normality0.4:1 az020.1275a, 5-1.851.10•• e z^/2.75 ^ 3.402 -,!!!!1.85.01.40 0.4U2Ae 084c ass- 0290.2 -0.00 0.0 ^1985^1988^1987^1988^ 1985^1988^1987^19884040.001985^1988^188'7^19e8^ :985^1988^1987^:988YEAR YEA R0.01985^1988^1987^1988^ 1985^19e6^1987^:988YEAR YEARYEAR YEARFigure 3.4: Year to year fluctuations in black cottonwood foliar nutrient concentrations of P,N, S, Ca and Mg. Data are for samples collected from the upper canopies of thesame trees, using the same sampling protocol with all sampling carried out durinv,the last 2 weeks of August.41175014®1050I^_--.........2015- .-IJ. .io700 _o „,..350__.. . :...-, --.. 0. _...-a--.....--0^-50 i 01985 1988 1987 1988YEAR165i, 1001132r 80I 0 =\=66 \b^1I^.24033IIts.ii 200I 01985 1988 1987 1988YEAR100 18080 - \ -.--..=120uU..8040a—O1:.4020 -0  ' ' , 01985^1988^1987^1988YEARL%1985^1988^1987^1988YEAR° ''''''''''''1985^1988^1987^1988^ 1985^1988^1987^1988YEAR YEARFigure 3.5: Year to year fluctuations in black cottonwood foliar nutrient concentrations of SO4-S, Cu, Zn, Mn, Active-Fe and Fe. Data are for samples collected from the uppercanopies of the same trees, using the same sampling protocol with all samplingcarried out during the last 2 weeks of August.42Zn, active-Fe, and Fe were measured at the Squamish 23 and Strawberry 1 sites (Table 3.7).Concentration differences for B were significant only at the Soowahlie site and foliar Nconcentrations fluctuated significantly only at the Strawberry 1 site. At the Soowahlie andSquamish 23 sites P, K, and Cu demonstrated a similar year to year pattern - their concentrationsdecreased from 1985 over the 1986-1987 period, and then rose quite sharply in 1988. The trendof Ca was the exact opposite, S fell in 1986 and then rose in 1987 and stayed the same in 1988,and N showed no significant change during the 4 year period at these two sites . Other macro-and micronutrients fluctuated in different ways from year to year.3.4 DISCUSSIONPrincipal components analyses of within-canopy variation in foliar nutrientconcentrations in two black cottonwood stands revealed two relatively distinct groups of foliarnutrients; the ANOVAs conducted on the same data showed similar trends. On the basis ofthese observations foliar nutrients in this study are divided into two groups - Group 1 includesN, P, K, S, SO4-S, Cu, and possibly Mg and B (although the last two are less stronglyassociated) and had highest concentrations in the upper canopy; Group 2 includes Mn, Zn, andCa with highest concentrations in foliage of the lower canopy. Due to their distinct patterns inthe ordinations at the Carey II and Soowahlie sites, Fe and active-Fe could be considered as athird group with different patterns of variability than all other nutrients studied.Group 1 nutrients are all highly mobile in the phloem (Devlin, 1966; Fife and Nambiar,1982; 1984; Ostman and Weaver, 1982; Switzer and Nelson, 1972) and were included in a'translocated group' by Kumata et al. (1988), where, with leaf senescence, translocation of thesenutrients from the foliage to the twigs was observed. In this study N, K, P, and S demonstrateda concentration decline at the end of the 1985 growing season and Mg, Fe, Cu, and SO 4 -S didnot. Concentrations of all Group 1 nutrients (except SO 4-S) showed a pronounced peak on theAugust 25 sampling data. Group 2 elements are all micronutrients (except Ca) that are relativelyimmobile in the phloem so that, once metabolized, are not easily transported to other areas43within the canopy (Attiwill, 1986; Kumata et al., 1988). Group 2 nutrients demonstrated muchlower levels of seasonal changes, although most showed significant year to year differences.Differences in element mobility within the phloem has been used to interpret within-canopy variation in other hardwood species, and in conifers (Fife and Nambiar, 1982, 1984;Guha and Mitchell 1965 a,b; Morrison, 1985; Sheriff et al., 1986; Wallihan, 1944). Lack ofsignificant within-canopy variation in macronutrient foliar concentrations, lower coefficients ofvariation for nutrients in the lower canopy, and ease of sampling, have led some investigators torecommend sampling from the mid-crown of hardwood trees to estimate stand nutrient status(Ellis, 1975; Morrison, 1985). Guha and Mitchell (1965 a,b) stated that canopy samplinglocation is only important when sampling for immobile micronutrients since they did notobserve significant differences in macronutrients within the canopy. None of these studiesobserved the pattern found for black cottonwood in this study where foliar nutrientconcentrations of all macronutrients were higher in the upper canopy, and there is some evidenceto suggest that this pattern of within-canopy variation is characteristic of fast-growing Populusspecies. Concentrations and concentration differences in foliar N between the upper and lowercrowns of 6-year old black cottonwood trees published by Heilman (1985) are very similar tothose revealed in this study. White and Carter (1970), working with Populus deltoides, showeda similar pattern for mobile and immobile nutrient groups and recommended sampling bothupper and lower crowns for the determination of stand nutrient status of mobile nutrients such asN, P, and K. White and Carter (1970) also recommended that only upper foliage need besampled for determination of Ca status since the element is highly immobile and thusdeficiencies will appear in the youngest foliage.Higher concentrations of Group 1 nutrients in the upper crown of black cottonwood maybe the result of translocation from lower canopy foliage, differential allocation of absorbednutrients, or both. Whatever the process, variability in foliar nutrient concentrations between theupper and lower crown of black cottonwood trees has important interpretative implications sincevery different conclusions about stand nutrient status could be drawn from the Carey II and44Soowahlie data, depending on which concentration is considered. For example, Heilman (1985)has suggested that, for black cottonwood, foliar N concentrations below 2.5% indicate acondition of N deficiency (see Table 5.20). Using 2.5% as a critical value, samples collectedfrom the lower canopy at the Carey II and Soowahlie sites would indicate nitrogen deficiency,while upper canopy concentrations are very close to the critical level of 2.5% (Tables 6.4 and6.5). The significantly lower concentrations of mobile nutrients in foliage of the upper canopymay indicate a condition of nutrient stress in that mobile macronutrients are translocated to morerapidly-growing areas of the tree crown (Devlin 1966; White and Carter, 1970a). However, it isdifficult to draw this conclusion from the findings of this study since the 15 black cottonwoodtrees sampled at the Soowahlie site were growing rapidly and had a mean site index of 23.0 m in15 years (see Table 2.1). In the absence of more precise information on the diagnosticusefulness of sampling and comparing upper and lower canopy nutrient concentrations, andgiven the higher costs of foliage sampling and laboratory analysis, it is believed that uppercanopy foliage samples from black cottonwood will provide the most useful and economicalinterpretations of black cottonwood stand nutrient status.This study revealed significant changes in foliar nutrient concentrations during thegrowing season at the Soowahlie site, and from year to year at three sites. The most significantaspect of changes in nutrient concentration over the 1985 growing season is the pronounced peakof many Group 1 nutrients (N, P, K, S, Fe, Cu, and to a lesser extent Mg and B) in late August.Foliar concentrations of these nutrients decreased significantly with decreasing canopy position(upper, middle and lower thirds) at the Soowahlie site. A similar gradient of significantlydecreasing concentrations of this group of nutrients in apical, upper canopy and lower canopyfoliage samples was shown for the Carey II site. Guha and Mitchell (1965b) felt that seasonalvariations in foliar concentrations of nutrients appeared to be related to the physiologicalimportance of the nutrients, and that seems to be the case in this study as well. In a review ofvariability in foliar nutrient concentrations within the crowns of forest trees, van den Driessche(1974) attributed within-canopy variation to auxin-controlled apical dominance, and stated that,in response to competition or environmental stress, it is likely that the apical region of the plant45will maintain a relatively higher foliar nutrient concentration, at the expense of more distalregions. This may be especially true for a highly shade-intolerant species such as blackcottonwood, where maintenance of canopy dominance or codominance is a prerequisite forsurvival.The relatively high concentrations of the group of mobile nutrients on the August 25sampling date is contrary to the accepted view that foliar nutrient concentrations in hardwoodspecies remain relatively stable over the latter part of the growing season (Day and Monk, 1977;Lea et al. 1979a,b; Mitchell, 1936), and thus is the best period for sampling foliage and applyingand interpreting critical foliar standards. The observed year to year changes are to be expectedgiven the significant seasonal fluctuations discussed above and shown in Figures 3.3. Foliarconcentrations fluctuated seasonally in response to poorly understood environmental andphysiological factors, thus foliar concentrations taken at the same time in different years mayshow considerable variation.The year to year fluctuations observed at the three sites become an important aspect ofvariation in foliar nutrient analysis if they are of sufficient magnitude to alter interpretationsbased on the data, i.e., if they fluctuate across boundaries of critical limits or alter nutrient ratiossignificantly. Using the critical levels for P in natural Populus stands and plantations (see Table5.20), the critical value is between 0.17 and 0.24%, which means that, at the Soowahlie site, thefoliar analysis interpretations would range from barely sufficient or slightly deficient in 1987, tomore than sufficient in 1988. A similar situation arises for K for the same site using the samestandards. At the Strawberry 1 site, foliar N concentrations are just above a criticalconcentration of 2.0 % from 1985 to 1987, and then fall below this critical level in 1988.A comparison of DRIS indices (Beaufils, 1973; Schutz and de Villiers, 1986) using the1985 to 1988 foliar concentrations for N, P, K, Mg, and Ca for the Soowahlie site is shown inTable 3.8. Criteria for the DRIS analysis utilized the greenhouse standards of Leech and Kim(1981) given in Table 5.20. Interpreting the indices, in 1985 only P is limiting, in 1986 N, P,and Ca are about equally limiting, 1987 is similar to 1985 in that P is the primary limiting46nutrient, and in 1988 N and P, are interpreted as limiting. In all cases P is seen as a limitingnutrient, but this is mostly because of the high greenhouse standard for the nutrient used in thisanalysis, so that concentrations never attain the optimal level. Clearly, using a DRIS approach,very different interpretations of black cottonwood nutrient status would be made, depending onthe year in which the samples were collected. The observed changes in the DRIS indices (andhence in their interpretation) are a result of relatively independent fluctuations of the differentnutrients from year to year, and this altered the nutrient ratios from which the DRIS indices werecalculated. These results contradict the assumption that the ratios used to calculate DRIS indicesare constant (Leech and Kim, 1979b, 1981; Schutz and de Villiers, 1986; Sumner, 1978, 1979),even though nutrient concentrations may fluctuate.Table 3.8: Comparisons of DRIS indices at the Soowahlie sites for 1985-1988.Year N P K Ca Mg1985 8 -112 -3 44 631986 -26 -28 47 -28 351987 -2 -88 -6 46 501988 -14 -78 24 9 60Using observations of within-year fluctuations in foliar nutrient concentrations in beech,sycamore, and horse chestnut, Guha and Mitchell (1965b) found that there were few elementsfor which stable periods could be utilized for diagnostic analysis. Based on his review of thefoliar analysis literature, van den Driessche (1974) stated that annual variation of foliarconcentrations were of sufficient magnitude to contribute substantially to the imprecision of themethod. The results of this study support this conclusion for black cottonwood. Accurateinterpretations of the nutrient status of black cottonwood using foliar nutrient information must47attempt to account for the fact that significant changes in critical foliar nutrient concentrationsmay occur seasonally, and from year to year on a given site. Observations of foliage nutrientconcentrations over a number of years will be required to accurately evaluate stand nutrientstatus using foliar nutrient concentration alone. Studies of the physiological ecology of blackcottonwood, in conjunction with the monitoring of important environmental factors, are requiredto attempt to understand, and thus be able to predict, the rapid changes in foliar concentrationswithin a given tree.Estimates of the numbers of foliar samples necessary for desired levels of accuracy andprecision in this study are comparable to other studies (Ballard, 1985; Ellis, 1975; Guha andMitchell, 1965a,b; Heilman, 1985; Lavender 1970) for both macronutrients and micronutrients.Estimates of black cottonwood foliar nutrient variability presented in this study are based onsamples of most recently matured, late leaves collected from lateral branches within the upperthird of the canopies of dominant or codominant trees. Thus the sample size requirementspresented here are valid only for samples collected according to this protocol and any deviationfrom this procedure can be expected to alter sample variability in an unpredictable manner.Samples of 15 trees were used in this study, and the results of within-stand variation show that asample of this size would estimate the sample mean at an a significance level of 0.95, with anallowable error of 10% for the macronutrients N, P, K, Ca, Mg, and S.Turner et al. (1977) have shown the usefulness of SO 4-S in predicting stand response toN fertilization. Foliar 504-S concentrations were the most variable (mean CV = 37.4%) of allnutrients studied in this report and this variability should be considered when interpreting foliarSO4-S data. Both total Fe and active-Fe were measured in this study because of the betterability of active-Fe to diagnose iron deficiencies (Ballard, 1981). In this study active-Fe wasshown to be much less variable than Fe so that, at an a significance level of 0.95, with anallowable error of 10%, Fe would require 43 samples and active-Fe only 21.A sample size of 15 trees has been recommended to estimate macronutrients by severalworkers (Ballard, 1985; Ellis, 1985; Mitchell, 1936) and the results of this study support this48number as a compromise between desired levels of accuracy and precision, and practical aspectsof foliage sample collection. Furthermore, it is these nutrients that are usually of the mostinterest to forest managers. Micronutrient concentrations (Cu, Zn, Fe, Mn, and B) are muchmore variable and between 20 and 60 samples would be required to attain the accuracy andprecision stated above for macronutrients. These estimates for micronutrients are similar tothose published by Ellis (1975) and are lower than many other studies.3.5 CONCLUSIONS1) The study revealed significant spatial and temporal variability of foliar nutrients in blackcottonwood stands. Group 1 nutrients (N, P, K, S, SO4, and possibly Mg and B) hadsignificantly higher concentrations in the upper canopy at 2 sites, and were observed to fluctuatethe most both seasonally, and from year to year. Group 1 nutrients are mostly macronutrientsthat are mobile within the tree, and thus their foliar concentrations may change unpredictably.2) Foliar concentrations of several Group 1 nutrients was observed to increase in the third weekof August which contradicts literature reports that foliar nutrient concentrations in broad-leavedtrees are relatively stable at that time. Also, temporal fluctuations of individual foliar nutrientswere often independent of other nutrients which was seen to alter the foliar nutrient ratios usedto establish DRIS ratios. Observations of temporal variability in foliar nutrient concentrationsreported here support those of other workers, and may seriously complicate the direct and simpleinterpretation of foliar nutrient concentrations.3) Given poor knowledge of the physiological determinants of variable foliar concentrations indifferent areas of the canopy, it is recommended that foliar samples in black cottonwood becollected from the upper one third of canopy, according to the protocol described in the report.If this protocol is followed then the estimates of variability for the foliar nutrients studied will bevalid.494) Levels of variability and the required sampling effort in this study are similar to other studies,and support the established procedure of collecting 15 samples to capture spatial variability at agiven moment in time. F ratios comparing variability among stands and within stands werehighly significant for all nutrients and suggest the potential for establishing relationshipsbetween levels of foliar nutrient concentrations and black cottonwood site index.50CHAPTER 4SPATIAL VARIABILITY OF SOIL NUTRIENTS IN BLACKCOTTONWOOD STANDS4.1 INTRODUCTIONThe principal objective of soil nutrient sampling in forest productivity studies is toquantify levels of important soil nutrients within a given stand or other area of interest. Theresults of such sampling represent estimates of the quantities of nutrients available at a giventime within the soil. These are often used as independent variables to investigate the relation togrowth of a particular tree species over a range of sites (Broadfoot, 1969; Carmean, 1970, 1972;Carter and Klinka, 1990; Kabzems and Klinka, 1987b; Kayahara, 1991; Wang, 1992), or toquantify qualitative assessments of soil nutrient status (Courtin et aL, 1988; Kabzems andKlinka, 1987a; Klinka et al., 1984). These objectives require an understanding of the spatialvariability of the soil properties investigated to develop effective sampling strategies, and todetermine statistical levels of accuracy and precision that can be associated with the estimates ofmean values for the various nutrients.If the objective of soil nutrient sampling is to correlate levels of soil nutrients with someproductivity measure, such as the site index of a tree species, then the variability associated witha given nutrient within the site must be less than its variability among sites, if meaningfulconclusions are to be drawn from the measurements (Ball and Williams, 1968; Carter and Lowe,1986; Mader, 1963). For example, if spatial variability within sites is high, then significantrelationships between soil nutrients and tree productivity may not be discernible at a givensampling intensity, even though these relationships do exist (Blyth and MacLeod, 1978). Manyresearchers have shown high spatial variability in soil nutrients within areas relatively uniform insoil and site properties, and, as a result, have recommended that an investigation of thisvariability should precede any attempts at correlating these assessments with measures of tree or51ecosystem productivity (Ball and Williams, 1968; Blyth and MacLeod, 1978; Carter and Lowe,1986; Courtin et al., 1983; Mader, 1963; Quesnel and Lavkulich, 1980; Troedsson and Tamm,1969).The work that has been carried out on spatial variation of soil nutrient values hasfocussed either on the forest floor under coniferous stands (Grier and McColl, 1971; Lowe,1972; Mader, 1963; Mader and Lull, 1968; McFee and Stone, 1965; Quesnel and Lavkulich,1980; Youngberg, 1965), or on the spatial variability of mineral soil properties (Blyth andMacLeod, 1978; Courtin et al. 1983; Drees and Wilding, 1973; Hart et al., 1969; Lewis, 1976;McFee and Stone, 1965; Slavinsky, 1977; Troedsson and Tamm, 1969). None of these havebeen in hardwood stands with medium to very rich soil nutrient regimes, and Moder and Mullhumus forms. Such a study was considered a necessary prerequisite to understanding therelationships between soil nutrient levels and black cottonwood site index.Several workers have shown that the variability of soil nutrients is often very local innature, i.e., as much as half of the spatial variability within a stand is contained in any squaremeter area of the stand, and that increasing plot size has little effect on sample variability(Beckett and Webster, 1971; McFee and Stone, 1965; Troedsson and Tamm, 1969). However,Robertson (1987), and Robertson et al. (1988) showed that, for mineralizable-N, samples within20 m were highly correlated, and recommended that samples be placed at least 20 m apart tomost efficiently capture site soil variability. The variability of soils in the vertical dimension,i.e., with depth, is in large part responsible for the large spatial variability over short horizontaldistances, especially the continuity, depth, and character of soil organic horizons (Binkley andHart, 1989). A number of factors create variability in surface organic layers: windthrown treesdisturb the continuity of, and mix mineral soil into surface organic layers; understory andcanopy species deposit leaf litter irregularly over the site, and each has litter that may havedifferent properties for mineralization; large organic debris, such as branches, twigs and fallentrees, are distributed unevenly over the site; and, on alluvial sites, different parts of the stand willreceive varying amounts and kinds of sediment deposition after flooding events. Mineral soil52layers are also subject to factors which affect the variability of soil nutrient concentrations:changes in soil texture affect the ability of a soil to hold nutrients; differences in physicalconditions within the soil, such as the presence of soil water tables or poorly-aerated areas, canaffect the rates at which minerals are weathered; differences in soil physical properties affect thetotal amount of nutrients available in the soil, e.g., the bulk density of the fine fraction and thepercentage of coarse fragments, and both can vary considerably within the site. Soil samplescollected to a given depth within any location within the stand will therefore incorporate bothvertical and lateral spatial variability.Temporal variability is seldom considered in studies of soil nutrients. Binkley and Hart(1989) stated that seasonal changes in nutrient concentrations in a given location are small,compared to spatial variability. However, Peterson and Rolfe (1982, 1985), and PetersonHammer (1986) found significant seasonal differences of available N and P in a floodplain soil,and related these changes to the interactions of seasonal flooding and nutrient uptake. Importanttemporal changes in the levels of soil nutrients in other soils have also been documented (Hainesand Cleveland, 1981; Harrison, 1979; Mollitor et al., 1980; Weaver and Forcella, 1979).Seasonal variation in the availability of soil nutrients is not considered in this study.This study investigates the spatial variability of soil nutrient concentrations and contents,and of some soil physical properties, both within a given soil pedon, and within and among sites,in 30 black cottonwood stands in south-coastal British Columbia. The within-pedon variabilitywas assessed in an attempt to understand some of the factors that are responsible for thevariability of nutrient determinations within study stands. Based on the analysis of 15 individualsamples at 9 of the study sites, estimates of the numbers of soil samples required to attainvarious levels of accuracy and precision are presented. These variability estimates werecompared to those from 21 sites where soil nutrient concentrations were estimated fromcomposited samples. The results of the investigation are used to assess the sampling procedureused to quantify soil nutrient levels in black cottonwood ecosystems. Given the spatial53variability that exists, the usefulness of quantitative measures of soil nutrient concentrations andcontents as independent variables to predict black cottonwood productivity is also discussed.The specific objectives of the study were;1) to analyze the sources and magnitudes of variability in soil nutrient concentrations andcontents that occur within a soil pedon and contribute to overall variability in estimating soilproperties for the site;2) to quantify the level of variability in each of the soil nutrients measured within and among the30 sites sampled;3) to compare the effectiveness of the compositing technique used for sampling at 21 sites withthe intensive sampling procedure used at 9 of the sites; and,4) to evaluate the sampling procedures used, based on the variability observations and, ifnecessary, suggest improvements to the sampling methodology.4.2 METHODS4.2.1 Descriptions of Study StandsStudy stands used for the analysis of soil chemical variability were the same as thosedescribed in Chapter 2. Landforms, subzone, soil and humus form descriptions, and siteassociation designations are given in Table 2.1. General properties of the soils are summarizedin Table 2.2.4.2.2 Soil SamplingSoil sampling for chemical and physical properties was carried out at two levels ofintensity. The intensive sampling, carried out at 9 sites, was used to evaluate within-sitevariability of soil nutrient concentrations and contents, and some other soil physical parameters.54Soil nutrient concentration is an estimate of the amount of a nutrient, expressed as a percentage,or as parts per million, of the dry mass of the soil fine fraction (< 2 mm diameter). Soil nutrientcontent attempts to estimate the total amount of a nutrient, and is expressed as kg/ha for a givensoil depth. Estimates of soil nutrient concentrations and contents were made at a less-intensivelevel using compositing of soil samples at an additional 21 sites.Each of the 9 black cottonwood stands sampled for the intensive analysis was dividedinto 15 approximately even-area plots, and a random process was used to select a soil pitlocation within each. For chemical analysis a 5 cm x 5 cm column of soil was excavated fromthe side of a pit, starting at the top of the mineral soil to a depth of 1 m. Estimates of mainrooting depth, absolute rooting depth, depth of the Ah horizon, and changes in the texture of thevarious C horizons were carried out in each of the 15 pits used for the soil chemical sampling.Main rooting depth was defined as that depth of soil that is more or less completely occupied byroots. Absolute rooting depth was defined as that level beyond which no additional roots couldbe found. The accurate determination of absolute rooting depth was impractical given the deepnature of many of the soils studied. In many cases, absolute rooting depth was described simplyas greater than the maximum depth of the soil pit excavated. Surface organic horizons (L, F, orH layers) were either absent or too thin to be included in the soil chemical sampling in all of theintensively sampled sites. At the Soowahlie and Carey I sites, 15 separate samples werecollected from the Ah horizon, and from the underlying C horizons, to assess variability innutrient concentrations with depth in the soil. At four of the sites, samples from individual Chorizons were collected to evaluate changes in soil nutrient concentration with increasing depthbelow the Ah horizon.In the 21 less intensively sampled study sites, each black cottonwood stand was dividedinto 4 equal areas, and a random procedure used to select a soil sampling location within each.In each quadrat a soil pit was excavated to a depth of at least 1 m (or to a restricting layer), andsoil samples were removed from each of the 4 walls, using the same procedure described abovefor the intensively-sampled plots. These 4 samples were then placed in one sample bag to make55up a composite soil chemical sample. In some cases the less intensively sampled ecosystems hadforest floors, and, in these cases, separate mineral soil and forest floor samples were collected inthe following manner. At each of the 4 pits, 4 forest floor samples were cut with a knife so thatthe undisturbed dimensions of the rectangular section of forest floor removed could be measuredand the volume calculated. Each forest floor sample was bagged separately for laboratoryanalysis, and later composited to get one sample for each of the 4 pits sampled.Mineral soil bulk densities were measured in the field using two methods. In the firstmethod a cylindrical hole was excavated to 30 cm soil depth and all material placed into a plasticbag and labelled. The volume of the hole was measured by inserting a thin, plastic bag into thehole, filling the bag with water to the soil surface, and then measuring the volume of waterwithin the plastic bag in a graduated cylinder. This method was used to estimate soil bulkdensity at all 15 pits in the intensively sampled sites. The second method utilized a coringdevice in the side of the excavated soil pit where a 7 cm long cylinder of known volume wascarefully pressed into the soil, after which the soil was removed and placed into a plastic bag.To coincide with the soil chemical sampling, this procedure was repeated until bulk densitymeasures were made over the same soil depths as the soil chemical samples. Bulk densities ineach of the 4 pits at the less-intensively sampled locations were made using the second method.Bulk density measurements at three to four soil pits at each of the intensively-sampled sites werealso determined over the depth of soil nutrient sampling.Coarse fragment content within the pits was evaluated by separating and weighing, in thefield, all mineral fragments larger than 2.5 cm diameter, and by carefully excavating the soil pitto a known dimension so that the volume of the soil pit could be calculated. Using an averagesolid particle density conversion factor of 2.65, the total mass of coarse fragments >2.5 cm forthe pit was converted to volume and expressed as a percentage of the soil volume. All mineralfragments greater than 2 mm diameter were removed by sieving soils in the laboratory,converted to a volume measure using the average solid particle density factor, and then added tothe >2.5 cm coarse fragment fraction to get a total coarse fragment percentage.564.2.3 Laboratory AnalysisMineral soil samples were transported in plastic bags to the laboratory, where they werethoroughly mixed, air-dried, passed through a 2 mm sieve to remove coarse fragments, and thensubsampled for analysis. Forest floor soil chemical samples were air-dried to constant mass,ground in a Wiley mill, and then composited for analysis. Mineral soil pH was measured with apH meter using a 1:2 soil:0.01 M CaC1 2 suspension, as described by Peech (1965). Forest floorpH was measured with a pH meter using a 1:5 suspension in distilled water. Total carbon wasdetermined using a Leco Induction Furnace (Bremner and Tabatabai, 1971). Total nitrogen wasdetermined by semi-microKjeldahl digestion (Bremner and Mulvaney, 1982), followed bycolorimetric analysis of ammonium using a Technicon Autoanalyzer (Anonymous, 1966).Mineralizable nitrogen was determined from incubated samples for 14 days at 30°C using theanaerobic incubation method of Wareing and Bremner (1964), as modified by Powers (1980),using a Technicon Autoanalyzer to measure released ammonium. The Mehlich extractionmethod (Mehlich, 1978) was used to measure extractable P, as suggested by Curran (1984).Available sulphate-sulphur was determined by ammonium acetate extraction (Bardsley andLancaster, 1965), reduction to sulphide, followed by colorimetric determination of the reducedsulfide (Kowalenko and Lowe, 1972). Extractable K, Ca, and Mg were determined byextraction with Morgan's solution of sodium acetate with a pH of 4.8 (Grewelling and Peech,1960), as recommended by Klinka et al. (1980). All soil nutrient measurements were expressedas percent or parts per million of soil dry mass.Subsamples of the soil chemical samples were used to determine the percentage of clay,silt, and sand in the samples using the pipette method (Anonymous, 1974). Measurements ofsoil texture were carried out on samples composited over the entire depth used for the soilchemical sampling.Coarse fragment free bulk densities were determined by measuring the mass of samplesof known volume after oven-drying at 105°C to constant mass, and passing the samples througha 2 mm sieve to remove the coarse fragments. Mass of soil <2 mm in diameter was then divided57by the volume of soil <2 mm diameter (corrected for coarse fragments >2 mm using the averagesolid particle density factor) to arrive at coarse fragment free bulk density.Soil nutrient measurements were expressed as concentrations (% or ppm) of soil drymass based on the analytical procedures. Using soil nutrient concentration, coarse fragment freebulk density, and a measure of soil volume (coarse fragment corrections and main rootingdepth/root restricting layer measurements), soil nutrients were expressed on a mass per unit area(kg/ha) basis.4.2.4 Statistical AnalysesThe SASCAL program (Marshall, 1987) was used to compute the numbers of samplesrequired at several levels of accuracy and precision, using the procedures and approach outlinedin Chapter 3. Analysis of variance was carried out using assumptions and methods of themultivariate general linear model (Cohen, 1968; Knapp, 1978), as outlined in Chapter 3.Relative sampling error (RSE) was calculated to compare the confidence intervals obtained fromthe composited samples with those from the intensively sampled sites using the followingformula;RSE —NR./71 x to-0] x 100xwhere; s2 = variance of the sample; n= number of soil samplesanalyzed; to _ o= t value associated with the n of samples at therequired accuracy; and x = mean of the sample584.3 RESULTS AND DISCUSSION4.3.1 Variability in Soil Properties within the Soil PedonTable 4.1 presents descriptions of two alluvial soil profiles (Carey I and Soowahlie)where sampling by depth was carried out. Ecto-organic surface horizons were largely absentfrom the soil profile in black cottonwood ecosystems, except in the fall after the period of annuallitterfall. Litter materials are rapidly incorporated into the upper mineral soil forming an Ahhorizon of varying depths, depending on the age of the stand and the nature of the floodingregime. The A horizons described in Table 4.1 are typical of the well-developed Ah horizonscommonly found under black cottonwood stands. Soil colour in the Ah horizon was variable;dark brown and brown hues with values between 3 and 5 and chromas of 2 to 3 were common.Soil structure in the Ah horizon was distinctly granular, very loose and friable, and earthwormswere common in most profiles observed in this study. Ah horizons under cottonwood stands hadabundant roots of very fine to medium size. Characteristically, relatively large lateral rootsspread out in the interface between the bottom of the Ah layer and the top of the mineralhorizons below.A region of darkly-stained mineral soil commonly underlaid the well-developed Ahhorizon, and was designated as an Ah2 horizon (Agriculture Canada, 1987). In alluvial soils, theorganic-enriched A horizons on the surface usually overlaid mineral soils at depth that had beenfluvially-deposited in distinct, sorted horizons. Structure in these horizons was single-grained(Luttmerding et al., 1990) and coarse fragments were generally absent. The Borden site, locatedon a cobbly and bouldery alluvial surface along the high gradient portion of the ChilliwackRiver, was the only exception. The Carey 1 profile described in Table 4.1 is typical of many ofthe alluvial sites sampled; sandy loam at the surface overlaid layers of sand and loamy sand,which in turn overlaid channel gravels at depth. The coarse-textured loamy sand layer directlyunder the Ah horizon at Soowahlie pit 14 (Table 4.1) was relatively uncommon in the sitesstudied.59Table 4.1: Descriptions of two alluvial soil profiles from the Carey 1 and Soowahlie samplesites.Carey IHorizon Depth (cm)LF 2-0Ahl 0-12Ah2 12-22C/ 22-42C 11 42-81C1II 81-144Civ 144-170+Horizon Depth (cm)LF 2-0Ah 1 0-10C1 10-34CH 34-52CIII 52-65CIV 65-81Cv 81-100+Horizon Descriptionpartially decomposed leaf and twig litterdark brown (10 YR 4/3); loam; coarse, granular structure; roots very fine tomedium, abundant; pH=6.3; earthwormsbrown (10YR 5/3) sandy loam; single grained structure; roots very fine tocoarse, abundant; pH=6.3light brownish-gray (2.5Y 6/2) sandy loam; single-grained structure; rootsfine to coarse, plentiful; pH=7.0white (2.5Y 8/2) sand; single-grained structure; roots medium to coarse,very few; pH=6.3pale yellow (2.5Y 7/4) loamy medium sand; single-grained structure; rootsmedium to coarse, plentiful; pH=7.9yellow (2.5Y 7/6) coarse sand; single-grained structure; roots coarse, veryfew; pH=7.0SoowahlieHorizon Descriptionpartially decomposed leaf and twig litterdark grayish brown (10 YR 4/2) loamy sand; coarse, granular structure;roots fine to medium, abundant; pH=5.7; earthwormspale yellow (2.5Y 7/4) sand; single-grained structure; roots coarse, few;pH=5.8pale yellow (2.5Y 7/4) sandy loam; single-grained structure; small layers ofpure, coarse sand; roots medium to coarse, plentiful; pH=6.0light gray (2.5Y 7/2) sand; roots coarse, few; pH=5.6pale yellow (2.5Y 7/4) loamy fine sand; roots medium to coarse, few;pH=5.6light gray (2.5Y 7/2) sand; roots coarse, very few60At both sites, roots in the mineral soil were distributed according to the texture of thestratum, so that very coarse-textured soil horizons had very few roots, while loamy strata atgreater depth had abundant roots of all sizes (Table 4.1). This pattern often made themeaningful determination of main rooting depth very difficult. At most sites, some roots werefound throughout the profile down to, and into, the gravel layer, so that absolute rooting depthwas also difficult to determine.Bulk density data presented in Figure 4.1 are based on samples collected to a depth of 1m in the soil pits from Carey I and Soowahlie shown in Table 4.1. Changes in bulk densitywithin the soil profile paralleled soil morphology. Bulk density in the organic-enriched Ahhorizon was about 0.95 mg/cm 3 , and increased in the C horizons to between 1.1 and 1.4 mg/cm 3(Figure 4.1) in both profiles. Coefficients of variation for mineral soil bulk density within agiven profile ranged as high as 42% (Table 4.2), and show that vertical heterogeneity in soiltexture and porosity add a significant component of variability to estimates of soil nutrientcontents.Figures 4.2 and 4.3 compare pH and mean concentrations of soil nutrients in 15 samplescollected from the Ah horizon with 15 samples collected from the underlying C horizons (to adepth of 1 m) at the Carey 1 and Soowahlie sites. Nutrient concentrations were significantlyhigher in the Ah horizon for all nutrients except extractable-P at the Carey I site (Figure 4.2). AtSoowahlie P concentrations were significantly higher in the Ah horizon. Soil pH changed littlethroughout the profile at both sites. The higher organic content of the Ah horizon resulted inhigher concentrations of total-C, as well as total-N, mineralizable-N, and SO 4-S. Organicsurfaces in the Ah horizon provide exchange sites for nutrient cations such as K, Ca, and Mg andmay account for the higher concentrations of these cations. For most nutrients there were order-of-magnitude differences between concentrations in the Ah and C horizons within a given site.MEE=EMEEMEMESEEEMEEMEEM -MEMENEEEI -gEZEMME^-61Pit 3 — Carey I^Pit 15 — Soowahlie0-77-1414-2121-2828-36C)a) 35-4242-49Ci49-65+.)63-7070-7777-8484-9191-980-77-1414-2121-285v 35-424049-68Q68-83A 83-7070-n77-8484-1991-9828-3542-490.5^0.7^0.9^1.1^1.3^1.5 0.5^0.7 0.9^1.1^1.3^1.5Bulk Density (gm/cm3) Bulk Density (gm/cm3)Figure 4.1:^Changes in soil bulk density with depth in the soil profile at the Carey 1 andSoowahlie sample sites.62Figure 4.2:^(Overleaf) Comparisons (lines represent 95% confidence intervals, n=15) betweenmean concentrations of soil nutrients in the Ah and C horizons (to a depth of 1 m)at the Carey 1 sample site.AnSoil Ft armSi t-braon17570950SW t-brzoi14010626I^1 1 121^113I IDOD052DAnSd Nartron'5co023468ASot riortralAr,Soli Floc Iron64Figure 4.3:^(Overleaf) Comparisons (lines represent 95% confidence intervals, n=15) betweenmean concentrations of soil nutrients in the Ah and C horizons (to a depth of 1 m)at the Soowahlie sample site.65Ahsou 1 .13drcn Sol Itortron10Sc. Horton Sou Hzekancia HortonAhSol HortonAhSol Hart=8Table 4.2: Ranges of coefficients of variation in bulk density at the intensively sampled sites.Site No. of Pedons C.V. RangeSoowahlie 3 13-23Carey I 4 8-18Squamish 23 3 11-16Strawberry I 4 8-11Chester 3 6-14Squamish 38 3 11-21Sumas 3 20-42Salmon 3 22-29Significant differences in soil nutrient concentrations between the Ah and C horizonsimplies that the nutrient concentrations determined from samples collected over a 1 m depth willbe affected to a large extent by the ratio of the depth of the Ah to the C horizon within a givensample. Table 4.3 presents means and CVs for the depth of the Ah horizon, and for mainrooting depth in the intensively-sampled sites. CVs for Ah depth range from 17-51%. Ifsamples were collected over the main rooting depth distinguished in the soil profile, then thiswould introduce another component of variability (CV range 15-53%) that would affect theestimation of both mean soil nutrient concentrations and contents.Concentrations of soil nutrient concentrations in the CI to Civ horizons at the Soowahlieand Carey I soil profiles described in Table 4.1, appear to be vary with the soil textures of thehorizons, and observations of root abundance within the horizons (Figures 4.4 and 4.5).6667Figure 4.4:^(Overleaf) Concentrations of soil nutrients by C horizon for the soil profilesshown in Table 4.1 at the Carey 1 sample site. Concentrations are based on asingle sample for each horizon.69Figure 4.5:^(Overleaf) Concentrations of soil nutrients by C horizon for the soil profilesshown in Table 4.1 at the Soowahlie sample site. Concentrations are based on asingle sample for each horizon.0 CS)01 (SL)CV (S)ov LS) 0V LS)700 15 30 45 60 75^ o 500 1000 1500 2000^ 0 20 40 80 00 100ElcfsegrableiC (ppm) Escftwyeatile-Ca Om) Extargestio-t.42 (Own)71Table 4.3: Variability in depth of the Ah horizon, and main rooting depth at the intensivelysampled sites.SiteAh DepthMean (m) / CVMain RootingDepthMean (m) / CVSoowahlie 0.13 / 38 0.62 / 38Carey 1 0.12 / 22 0.64 / 15Squamish 23 0.12 / 28 1.01 /24Strawberry I 0.03 / 28 0.55 / 27Homathko 0.02 / 33 0.39 / 53Chester 0.17 / 17 0.58 / 25Squamish 38 0.06 / 51 0.49 / 20Sumas 0.14 / 22 0.68 /31Salmon 0.10 / 23 0.83 / 23Concentrations shown in Figures 4.4 and 4.5 are based on only one sample for each horizon, sono statistical significance can be assigned to the differences between C horizons within a soilprofile. Concentration values should be interpreted cautiously given the high variability of soilnutrient concentrations reported below. However, at both sites higher concentrations of total N,total C, and exchangeable Ca and Mg, occurred in the finer-textured horizons, and thesehorizons consistently had a higher abundance of roots (Table 4.1). Mineralizable Nconcentrations followed those of total N closely at the Carey I site but were less correlated at theSoowahlie site. Exchangeable K followed the same pattern as the other exchangeable cations atthe Carey site, but not at the Soowahlie site. At the Soowahlie site the clay content decreasedfrom 6.8% in the CH (silt loam texture) to 1.2% in the Cm layer below (sand texture).Similarly, at the Carey 1 site, clay content of the soil decreased from 8.5% in the C 1 horizon (siltloam texture) to 1.6% in horizon CH. (sand texture). Higher concentrations of exchangeable Caand Mg in these layers may be the result of a greater surface area for cation exchange in soilhorizons with higher clay contents, and to weathering of clay minerals in these layers. Also, thehigher concentrations of total N and total C may be due to the decomposition of roots in thesehorizons. Differences between horizons were small for soil nutrients with very lowconcentrations, such as available P and SO 4-S.72The pattern of having much higher nutrient concentrations in the Ah compared to the Chorizons (Figures 4.2 and 4.3) is reversed when soil nutrient contents are compared (Figure 4.6and 4.7). Except for mineralizable N at the Carey I site (Figure 4.6), soil nutrient contents wereconsistently much higher in the C horizons (over a depth of 1 m) than in the Ah horizon. Thehigher value for the content of mineralizable N in the Ah horizon at the Carey site results fromthe much higher concentration of mineralizable N in the Ah, compared to the C horizons at thatsite. Soil nutrient contents were based on a mean Ah depth of 0.12 m at the Carey I site and0.13 m at the Soowahlie site. Thus C horizon soil contents were based on a depth of 0.88 m and0.87 m respectively, and this factor, along with the higher bulk density for the C horizons,outweighed concentration differences and accounted for the reversal in the relationship betweenthe two horizons. These comparisons show that, even though nutrient concentrations were muchhigher in the organic-rich A horizons in black cottonwood ecosystems, mineral subsoils alsoprovided an important source of available nutrients.4.3.2 Within-Site and Among-Site VariabilityCoefficients of variation (CVs) for soil nutrient concentrations for 9 intensively sampledsoils under cottonwood stands demonstrated high variability for concentrations of all soilnutrients within sites (Table 4.4). Mineralizable N (mean CV=52%) and available SO4-S (meanCV=42%) had the highest, and exchangeable K (mean CV=25%) and available P (meanCV=26%) had the lowest mean within-site variabilities. The relatively higher within-sitevariability in the concentrations of mineralizable N and available 50 4-S is to be expected, sincethe availability of these nutrients is a function of microbiological activity, and will thus varyaccording to soil moisture, soil temperature, and other biotic factors that do not affect theconcentrations of the other nutrients.73Figure 4.6^(Overleaf) Comparisons (lines represent 95% confidence intervals, n=15) betweenmean contents (kg/ha) of soil nutrients in the Ah and C horizons (to a depth of 1m) at the Carey 1 sample site.C3500300325032000I 150310005000'1201015(to60046C302150040?1. 3°202D151D05ODCtircrt75Figure 4.7:^(Overleaf) Comparisons (lines represent 95% confidence intervals, n=15) betweenmean contents (kg/ha) of soil nutrients in the Ah and C horizons (to a depth of 1m) at the Soowahlie sample site.g 1°5fi7030700600; 600;400w 3002001030Haim86; 39o.2e0SoI Homan002C10000'PSca6040; 302000077The wide range of CVs for a given site show that the variability of soil nutrientconcentrations was high and relatively unpredictable from site to site. This has importantimplications for the determination of the optimal sampling effort required for assessing soilnutrient concentrations within a given site, because the formula used to calculate the number ofrequired samples at various levels of accuracy and precision is based on the mean CV valuesgiven in Table 4.4.ANOVAs were carried out on the soil nutrient concentrations to compare varianceswithin and among the 9 intensively sampled black cottonwood stands (Table 4.4). All F ratioswere highly significant (p < .001) and show that, for the 9 intensively-sampled sites, the majorvariation is among, rather than within sites.Table 4.4: Coefficients of variation (CVs), mean CVs, and F ratios comparing among-site towithin-site variance for soil nutrient concentrations at the 9 intensively-sampledblack cottonwood stands. CVs and ANOVAs are based on mean nutrientconcentrations of 15 individual soil samples collected over the upper 1 m of soil.Total C Total N Min-N Av-P Ex-Ca Ex-Mg Ex-K SO4-SSite (%) (%) (%) (%) (%) (%) (%) (%)Soowahlie 56 52 79 24 30 31 30 58Carey I 24 23 68 20 11 8 20 18Squamish 23 31 27 31 14 33 29 19 47Strawberry I 34 15 65 19 17 17 8 13Homathko 59 46 40 16 36 38 25 38Chester 31 20 58 20 19 17 21 42Squamish 38 48 56 61 25 73 64 35 56Sumas 30 25 28 63 72 41 35 60Salmon 22 20 43 29 23 26 28 47Range of CVs 22-59 15-56 28-79 14-63 11-73 8-64 8-35 13-60Mean CV 37 32 52 26 35 30 25 42F Ratio 21.9 32.9 11.1 152 86.6 239 48.5 37.678Patterns of relative sampling errors (RSEs) shown in Table 4.5 are similar to thevariability data described for the 9 intensively sampled sites. The mean RSEs for SO 4-S andmineralizable N were the largest (+/- 36% and +/- 26% of the mean respectively at an alpha ofTable 4.5: Relative sampling errors (RSEs), ranges, and mean RSEs for soil nutrientconcentrations at the 9 intensively-sampled black cottonwood ecosystems. RSEs arebased on mean nutrient concentrations of 15 individual soil samples collected overthe upper 1 m of soil.Total C Total N Min-N Av-P Ex-Ca Ex-Mg Ex-K SO4-SSite (%) (%) (%) (%) (%) (%) (%) (%)Soowahlie 19 18 28 8 10 11 10 20Carey 1 8 9 24 7 4 3 7 6Squamish 23 11 9 11 5 12 10 7 16Strawberry I 12 5 23 7 6 6 3 5Homathko 20 16 14 6 13 13 9 13Chester 11 7 20 7 13 6 7 15Squamish 38 17 20 21 9 25 22 12 19Sumas 10 9 10 22 25 14 12 21Salmon 8 7 15 10 8 9 10 16Range of RSE 8-20 5-20 10-28 5-22 4-25 3-22 3-12 5-20Mean RSE 13 11 18 9 13 10 9 150.90), and those of exchangeable K (+/- 17%) and Ca (+/- 18%) the smallest.Calculation of RSE provided the opportunity to compare the relative variability of theintensively sampled sites with that of the composited sites (Table 4.6). Four of the 16composited sites have been excluded from Table 4.6 because some of their mean nutrientconcentrations had no variability. This occurred when concentrations of a nutrient were at thelow end of the detection range for a given analytical procedure, so that all concentrations aregiven as the minimum detectable value. For all nutrients considered, the mean RSEs in thecomposited sites were approximately double that of the same nutrient in the intensively-sampledplots.79Table 4.6: Relative sampling errors (RSEs) at alpha=0.90, ranges, and mean RSEs for soilnutrient concentrations at 16 black cottonwood ecosystems where soil samples werecomposited into 4 samples at each site.Total C Total N MM-N Av-P Ex-Ca Ex-Mg Ex-K SO4-SSite (%) (%) (%) (%) (%) (%) (%) (%)Borden 29 14 18 44 26 25 6 11Chilliwack 28 28 25 28 20 21 27 25Island 12 54 58 63 64 33 29 33 120Mercer 8 13 31 16 9 7 15 18Murphy high 6 13 20 20 19 14 12 36Oyster 21 25 19 39 24 22 14 10Squamish 28 26 21 11 8 19 4 19Tamihi Fan 36 35 53 16 7 20 27 103Ryder Lake 11 14 10 23 10 23 29 27Chipmunk Creek 46 19 47 17 15 26 12 28Pierce Creek 8 15 19 6 17 16 25 12Elk3 17 21 8 44 21 25 12 31Elk 1 35 28 36 21 48 55 28 55Elk 2 18 21 26 17 15 15 21 44Ashlu 20 1 12 25 5 12 4 8Tamihi 16 13 13 8 11 25 8 26Range of RSEs 6-54 1-58 8-63 6-64 5-48 7-48 4-33 8-120Mean RSEs 24 22 26 25 18 22 17 36Coefficients of variation for the 9 intensively sampled sites were consistently higher forsoil contents (Table 4.7) than for concentrations (Table 4.4). The order of variability for thedifferent nutrients was much the same; mineralizable N (mean CV=60%) and SO4-S (meanCV=46%) had the highest variability, and exchangeable K (mean CV=34%) and available P(mean CV=34%) the lowest. The variability in nutrient contents for the intensively sampledsites was primarily a function of variability in bulk density (mean CV=23%) and nutrientconcentrations from pit to pit. At all sites, a depth of 1 m, and not main rooting depth (see Table4.3), was used to calculate the depth factor of the volume multiplier used to calculate soilcontents, so this aspect of variability was not included in the estimates shown in Table 4.7.80Also, soils in the intensively sampled plots contained very few coarse fragments. Presumably,the variability would have been higher if soils in the intensively sampled plots had contained anycoarse fragments. As for soil nutrient concentrations (Table 4.4), variation in soil nutrientcontents from site to site was significantly higher than within-site variability, as shown by thehighly significant F ratios for all nutrients (Table 4.7).Table 4.7: Coefficients of variation (CVs), ranges of CVs, mean CVs and F ratios comparingamong-site to within-site variance for soil nutrient contents (kg/ha) at the 9intensively-sampled black cottonwood ecosystems.^CVs are based on means of 15individual samples.SiteBulkDensity Total C Total N MM-N Av-P Ex-Ca Ex-Mg Ex-K 504-SSoowahlie 21 49 47 63 25 32 25 34 25Carey I 21 46 49 112 22 29 28 25 24Squamish 23 18 36 31 36 28 37 41 30 65Strawberry I 24 48 30 61 26 34 35 24 26Homathko 30 82 52 44 36 29 56 41 48Chester 26 42 32 67 36 31 30 35 45Squamish 38 9 47 55 59 23 74 64 35 54Sumas 24 43 38 41 65 78 47 39 65Salmon 30 42 29 58 42 39 37 46 60Range of CVs 9-30 36-82 29-55 36-112 22-65 29-78 28-64 24-46 24-65Mean CV 23 48 41 60 34 43 40 34 46F Ratio 10.2 66.9 87.0 14.1 125 81.8 230 28.6 24.6Table 4.9 shows the numbers of samples required for the determination of soil nutrientconcentrations at the sites for this study, at different levels of accuracy and precision. The orderof sample requirements for the nutrients measured are similar to the patterns of variability shownin Table 4.4; mineralizable N and available a:14-S would require the largest sampling effort, andexchangeable K the smallest, at a given level of accuracy and precision. However, for allnutrients, very intensive soil sampling would be required to obtain highly accurate and preciseestimates of soil nutrient concentrations. For example, at an alpha and gamma level of 0.95%81with an error of 10%, the most variable nutrient, mineralizable-N, would require 186 individualsamples, and the least variable nutrient, exchangeable K, would require 52 individual samples.Table 4.8: CVs for coarse fragment content for 6 study sites with a significant component ofcoarse fragments.SiteCV of Coarse FragmentVolumePolygon 19 14Ryder 52Tannin Fan 48Oyster River 83Chipmunk Creek 35Borden Creek 57Table 4.9: Numbers of soil samples required to estimated soil nutrient concentrations in blackcottonwood ecosystems at different levels of alpha, gamma and percentage error.AlphaGammaNutrient^ Error0.950.5010%0.950.5020%0.900.5010%0.900.5020%0.950.9510%0.950.9520%0.900.8010%0.900.8020%pH 3 3 3 3 5 3 3 3Total C (%) 52 15 37 11 111 31 55 15Total N (%) 37 11 26 8 79 23 39 12Mineralizable N (ppm) 87 24 62 17 186 50 91 25Available P (ppm) 30 9 22 7 65 19 32 10Exchangeable Ca (ppm) 61 17 43 12 130 36 64 18Exchangeable Mg (ppm) 44 13 31 9 94 27 47 14Exchangeable K (ppm) 25 7 17 6 52 16 26 8Extractable SO4 (ppm) 73 20 52 14 157 43 77 21Numbers of samples required for the same levels of accuracy and precision were, formost nutrients, close to twice as high for the determination of soil contents (Table 4.10) as forsoil concentrations (Table 4.9). This is to be expected given the extra variability involved in theestimation of soil nutrient contents, as discussed above for Table 4.7.82Table 4.10:Numbers of soil samples required to estimate bulk densities and soil nutrientcontents in soils within black cottonwood stands at different levels of alpha, gammaand percentage error.AlphaGammaNutrient^ Error0.950.5010%0.950.5020%0.900.5010%0.900.5020%0.950.9510%0.950.9520%0.900.8010%0.900.8020%Bulk Density (gm/cm3) 22 7 16 5 54 17 23 8Total C (%) 92 25 65 18 229 61 96 26Total N (%) 68 19 48 13 170 46 72 20Mineralizable N (ppm) 141 37 100 26 352 92 148 39Available P (ppm) 46 13 33 10 114 32 48 14Exchangeable Ca (ppm) 72 20 51 14 180 49 76 21Exchangeable Mg (ppm) 65 18 46 13 161 44 68 19Exchangeable K (ppm) 48 14 34 10 118 33 50 15Extractable SO4 (ppm) 83 23 59 16 207 56 87 244.4 DISCUSSIONThe magnitude of variability for soil nutrient concentrations sampled in this study washigh for all measures except soil pH, and is thus comparable to results from similar studies ofmineral soils (Ball and Williams, 1968, 1971; Binkley and Hart, 1989; Courtin et. al., 1983;Mader, 1963). Mean CVs ranged from 25% for exchangeable K to 52% for mineralizable N,and, for any nutrient, there was a wide range of CVs among the different study stands. Thisrange in variability has also been observed in other studies (Blyth and MacLeod, 1978; Courtinet al., 1983).High variability in nutrient concentrations within a site has been attributed to changes innutrient concentration with depth in the soil profile (Binkley and Hart, 1989). In this study,nutrient concentrations in the Ah horizon were higher than underlying C horizons by an order ofmagnitude for most nutrients. There is also some evidence to suggest that soil nutrientconcentrations varied as a function of soil texture within the C horizons of the alluvial soils83sampled. These observations suggest that, within a given pedon, soil nutrient concentrationswill vary as a function of the nature (texture, organic matter content, aeration) and depth of thevarious soil horizons, relative to the depth of sampling.Variability in the estimation of soil nutrient contents in this study was consistently higherthan that for soil concentrations, and was of the same magnitude as that found by Courtin et al.(1983). Variability in soil nutrient contents is higher than soil nutrient concentrations because,in addition to the variability introduced by soil nutrient concentrations, the calculation of soilnutrient content requires multiplication by factors that are themselves subject to considerablevariation. Variation in soil bulk density, coarse fragment content, and rooting depth allcontributed considerable variability to the accurate estimation of soil nutrient contents.To develop the most meaningful relationships between measures of black cottonwoodsite index and soil nutrients, the optimal depth of soil nutrient sampling should be, withinpractical limits, that depth over which black cottonwood collects the majority of its nutrients.Most absorption of soil nutrients is through the fine roots (Bowen, 1984) and concentrations offine roots are well correlated with soil mineral nutrient concentrations within the soil profile(Kimmins and Hawkes, 1978; Powers, 1984). The results of this study were consistent with thispattern in that fine root concentrations were highest in the upper soil layers, and these layerswere seen to have significantly higher concentrations of almost all soil nutrients. Given thisinformation, one sampling approach would be to restrict sampling to the main rooting depth,since the majority of fine roots were located within that depth (Kayahara, 1991). Compared tosampling over a fixed depth such as a meter, this approach has the advantage of requiringconsiderably less sampling effort in each pedon. However, in the alluvial soils sampled in thisstudy, determination of the main rooting depth often proved problematic because the distributionof roots in the soil profile varied with soil texture. In some cases, sandy mineral soil layers nearthe surface had very few roots, after which more fine-textured layers at depth were completelyoccupied by roots. Also, in this study, fine-textured subsurface horizons appeared to havehigher concentrations of nutrients than coarse-textured layers near the surface. White and Wood84(1958) showed that fine-textured, subsurface layers were important sources of K in K-limitedred pine systems in New York. Also, although subsurface layers had lower concentrations ofsoil nutrients, their greater depth and bulk density resulted in much higher nutrient contents, andthey thus represented an important source of nutrients within the soil profile. Other argumentsagainst sampling over the main rooting depth are; - 1) the main rooting depth varies from pedonto pedon and thus becomes another source of variability that reduces the accuracy and precisionof determinations of soil contents, and - 2) some of the soils in the study had root restrictinglayers that reduced rooting volume, so that, if samples had been collected over the main rootingdepth, the effects of reduced soil volume would not be expressed in the estimates of soilcontents. This would provide inaccurate comparisons of soil nutrient contents available forblack cottonwood growth at different sites. Given the presence of nutrients at considerabledepth in the soil profile, the presence of roots to utilize these nutrients, the variability anddifficulty of accurately determining the main rooting depth, and the potential for including theeffects of restricted rooting depth, it was concluded that the 1 m depth of sampling utilized inthis study represents the most practical and meaningful method for estimating soil nutrientsavailable to black cottonwood.The order of variability of the different nutrients found in this study differs from thosereported by Courtin et al., (1983), Ike and Clutter (1968), Lewis (1976), and Slavinsky (1977)(Table 4.11). Blyth and MacLeod (1978) studied several soils in which the order of variabilityalso varied significantly from area to area. These differences in the order of nutrient variability,and the observed high range in variability from site to site for all nutrients, means that variabilityestimates obtained from previous soil studies have limited value as pilot studies. As stated by anumber of workers (Ball and Williams, 1968; Blyth and MacLeod, 1978; Carter and Lowe,1976; Courtin et al. 1983), serious studies of the relationships between soil nutrients and treeproductivity or other variables, should be preceded by variability studies that provideinformation specific to the soils being sampled.85Table 4.11:Relative order of variability in nutrient concentration in this study compared to otherstudies.Study   Relative Order of VariabilityThis study^Min-N, SO4-S > %C, Ca, %N, Mg > P > K > pHSlavinsky (1977) Ca > Mg > %N, %C, K > pHLewis (1976) Ca, Mg> K > %C, %N> pHCourtin et. al (1983)^P, SO4, Ca> Min-N, K > Mg > %C, %N > pHIke and Clutter (1968) Ca > P, K, MgThe high variability inherent in soil nutrient concentrations and contents in forest soilsmeans that precise estimates based on individual samples, even for areas that appear to haverelatively uniform soils, will be too expensive to obtain (Courtin et al., 1983; Mader, 1963). Toincrease the efficiency and decrease the cost of soil nutrient sampling, a compositing procedurewas utilized for 21 of the 30 sites sampled. Other studies (Carter and Lowe, 1986; Mader, 1963;Slavinsky, 1976) have shown that soil nutrient values obtained from composited samples fallwithin the confidence interval established for intensively-sampled estimates of the mean for thesame site, and have recommended a compositing procedure to increase sampling efficiency.Ball and Williams (1968) however, showed that for the level of sampling employed, the valuesobtained from composited samples fell outside the confidence interval of the mean for somenutrients. In this study, a direct comparison of the relative efficiencies of the two methods wasnot possible, because the two sampling approaches were never carried out at the same site.However, comparisons of the relative sampling errors for the intensive and compositingprocedures for each of the nutrients measured, showed that the confidence intervals for thecomposited sites were consistently higher than those for the intensively sampled sites, for allnutrients considered. This result means that the compositing sampling procedure used wasconsiderably less successful than the intensive sampling procedure in capturing soil variabilitywithin a given site.86The rationale for collecting four samples from each of four pedons in the composited soilsamples in this study is based on the idea that up to one-half of the variability in soil nutrientconcentration for a given site is included in any square meter of the site (Beckett and Webster,1971; Troedsson and Tamm, 1969). Given this, four samples collected from each of the fourwalls of a meter square soil pedon, and then thoroughly mixed, should incorporate up to one halfof the variability of the site. If soil nutrient concentrations within each soil pedon wereindependent of each other, then the four composite samples should account for the same amountof variability as 16 individual samples, and the resulting mean should have about the sameprecision as the mean estimated by 15 individual samples. However, the fact that the confidenceinterval provided by the compositing sampling procedure was close to double that of theintensive sampling procedure, suggests that the samples collected from a single soil pedon werehighly correlated. In retrospect, this correlation is to be expected given the observed uniformityof soil strata within a sample pedon. This result supports the observations of Robertson (1987),who showed a high degree of correlation in soil samples collected over a small area, andrecommended that sampling locations be separated by at least 20 m to provide the most efficientsampling design. In this study, the intensive sampling procedure, where soil pits were always atleast 40 m apart, estimated soil nutrient concentrations with almost twice the precision of thecomposited approach. For these reasons it is argued that the compositing sampling procedureused in the study should be altered so that fewer samples are collected from within each pedon,and that more soil pits, located at least 20m apart, should be excavated.Although levels of within-site variability of concentrations and contents were high for allnutrients, F ratios comparing variances among- to within-sites were all highly significant(p<.001). This suggests that, given the sampling intensity employed, it may be possible tocorrelate changes in black cottonwood site index with changes in soil nutrients. It can beexpected, however, that, high within-site variability will tend to obscure these relationships.Also, because variability differs among the nutrients measured, correlations between soilnutrient levels and black cottonwood site index will be more discernible for some nutrients thanfor others.874.5 CONCLUSIONS1. For two Humic Regosol soils typical of the study sites sampled, concentrations of soilnutrients were higher by an order of magnitude in the Ah horizon, than in to the underlying Chorizons. As a result, variation in soil nutrient concentrations was due principally to variabilityin the depth of the Ah layer, relative to the overall depth of the sample.2. For the same two soils, soil nutrient contents (kg/ha) were significantly higher for allnutrients in the C horizons than the overlying the Ah horizon. Higher soil content in the Chorizon was a function of their higher bulk density, and much greater volume. Variability ofsoil nutrient content was consistently higher than concentration variability. This was attributedto variability in factors such as bulk density, and coarse fragment content, that, in addition to soilnutrient concentration, are used to calculate soil nutrient contents. For the nine soils sampledintensively and used for estimating coefficients of variations in this study, coarse fragmentcontent was very low, and soils were sampled to a depth of 1 m. Variability estimates for soilnutrient contents can be expected to increase where sampling is carried out over the main rootingdepth, or in soils with high and variable coarse fragment contents.3. Coefficients of variation for soil nutrient concentrations and contents in this study were high,and were similar to other studies for each of the nutrients. Mineralizable N and SO4-S were themost variable, and exchangeable K and available P the least. Between 25 and 87 samples wouldbe required to estimate soil nutrient concentrations at an alpha of 0.95 with 10 percent error. Fornutrient contents the required number of samples would be 46 to 141 for the same accuracy andprecision. At the intensively sampled sites, where 15, well-spaced samples were collected, analpha of 0.90 with 20 percent error was achieved for mineralizable N, the most variable nutrient.Despite high within-site variability, variability was higher among than within sites, and thismakes it possible to correlate changes in soil nutrient levels with black cottonwood site index.884. Compared to the intensive sampling procedure, the compositing procedure used in this studyresulted in much higher within-site variation for all nutrients. This was attributed to the highcorrelations of four samples from within the same soil pit. It was suggested that the use of morewidely-spaced sample pits, with fewer subsamples from each, would more effectively capturewithin-site variability. It was also argued that the 1 m sampling depth used in the study was themost effective sampling depth for estimating nutrient availability for black cottonwood in thesites sampled.89CHAPTER 5RELATIONSHIPS BETWEEN BLACK COTTONWOOD SITE INDEX,FOLIAR AND SOIL NUTRIENTS, UNDERSTORY VEGETATION, ANDSITE UNITS5.1 INTRODUCTIONResearch that has attempted to correlate ecological variables with the site index of treespecies within a given geographic area has been summarized into 'factorial' and 'holistic' (Jones,1969) approaches. The objective of the factorial approach has been to measure and identifyecological (principally physiographic and soil) factors that limit the productivity of the species,usually through the use of multivariate models (Carmean, 1954, 1965; Cox et al. 1960; Eis,1962; Schmidt and Carmean, 1988; Wilde, 1970). The approach has been widely applied in theUnited States, and studies have been reviewed by Carmean (1975), Hagglund (1981), Spurr andBarnes (1980), and Coile (1952). In general, the percentage of variance explained by the linearmodels developed through the factorial approach seldom exceeds 60% (Covell and McClurkin,1967; Hagglund, 1981; Jones, 1969), unless the sampling universe is stratified, as in the studiesof Myers and van Deusen (1960) and Carmean (1965). In British Columbia the factorialapproach has been used to correlate environmental variables such as potentialevapotranspiration, average water balance as derived from energy-driven models (Spittlehouseand Black, 1981), and soil nutrient contents, with site index of Douglas-fir (Carter and Klinka,1990), lodgepole pine (Wang, 1992), and western hemlock (Kayahara, 1991). Wang (1992) forexample, was able to explain more than 80% of the variation in lodgepole pine site index using acombination of soil nutrient and soil moisture measures.90Holistic or integrative ecological approaches are based on correlating site index withtaxonomic units such as those derived from classification of soils (Agriculture Canada ExpertCommittee on Soil Survey, 1980; Soil Survey Staff, 1975) or sites (Pojar et al., 1987; Pfister etal. 1977). By correlating site index of tree species with the taxonomic units of a landclassification system, an estimate of the productivity of the species on the site can be generated,even if the species is not occupying the site. Also, the units are generally used for more generalmanagement purposes, so that species' productivities can be related to other aspects of forestmanagement. Attempts to correlate soil taxonomic units with measures of species productivityhave been largely unsuccessful, and this has been principally attributed to the wide range ofecological conditions that are included within a single soil taxon (Carmean, 1970,1975; Jones,1969). Correlations with site units have met with higher success (Green et al., 1989; Kayahara,1992; Klinka et al., 1989a; Klinka and Carter, 1990; Monserud, 1984; Wang, 1992). Monserud(1984) found differences in site index of Douglas-fir, and different height growth patternsamong habitat types. Green et al. (1989) explained 86% of the variation in Douglas-fir siteindex using site units of the biogeoclimatic ecosystem classification (Pojar et al., 1987). Usingsimilar site units Kayahara (1991) explained 71% of the variation in western hemlock site index,and Wang (1992) explained 81% of the variation in lodgepole pine site index. These resultssuggest that site units can be very effective in differentiating ecologically-significant segmentsof regional edatopic gradients, as measured by the site index of the species examined. Theproblem with utilizing the holistic approach alone to predict the productivity of a species is thatthe correlation does not provide functional explanations for the growth of a particular species ona given site unit.No studies employing either a factorial or a holistic approach have been carried out onblack cottonwood, although considerable work has been done on the eastern cottonwood (P.deltoides) by Broadfoot (1960), and Baker and Broadfoot (1976, 1979). Broadfoot (1969)criticized the factorial approach for it's difficulty in establishing and measuring site and soilproperties that can be reliably correlated with site index of the species in question, and for theinapplicability of the results outside of the restricted geographic area or climate-landform91conditions in which the study was carried out. On the other hand, he felt that subjective systems,such as those provided by assessment of indicator plants or landform-soil classes, did notprovide a sufficiently accurate assessment of site index (Baker and Broadfoot, 1976). Theapproach taken by Baker and Broadfoot (1976, 1979) combines both subjective and objectiveapproaches by using quantitative measures of site properties to define the characteristics of 4subjectively-derived factors important for cottonwood growth on all sites - soil physicalcondition, moisture availability, nutrient availability, and soil aeration. Harrington (1986) useda similar approach employing stepwise linear regression to identify major environmental factorsaffecting site index of red alder in western Washington and Oregon.The approach taken in this study combines elements of the holistic and factorialapproaches by establishing inter-relationships between black cottonwood site index, qualitativemeasures of ecological site quality as assessed using biogeoclimatic ecosystem classification,and quantitative estimates of soil nutrient contents and foliar nutrient concentrations.Correlations of black cottonwood site index with subzone, soil nutrient regime, and siteassociation are used as a starting point for assessing the factors that determine black cottonwoodproductivity in the present study. Qualitative estimates of soil nutrient regime describe nutrientavailability for a site in general terms, but cannot account for the fact that the availability of soilnutrients for a particular species will differ from other species because of different physiologicaladaptations and nutrient requirements (Chapin et al. 1986). By establishing relationshipsbetween black cottonwood site index, quantities of soil-available nutrients (soil nutrientcontents), and measures of nutrients taken up by the target trees (foliar nutrient concentrations),the particular nutrient or nutrients that are limiting can be identified (Attiwill, 1986; Chapin etal. 1986; White and Carter, 1970a,b). Relationships within groups, such as soil nutrientinteractions, and between soil and foliar nutrient levels, can also aid in the interpretation of themeasurements as they may affect the productivity of black cottonwood. The determination offoliar nutrient status also permits the application of analytical methods such as critical nutrientlevels (Ballard and Carter, 1986; Lavender, 1970; Weetman and Wells, 1990), and assessmentsof nutrient balance through the determination of DRIS indices (Beaufils, 1973; Leech and Kim,921979, 1981; Schutz and de Villiers, 1986), that are based on foliar nutrient concentrations. Bysummarizing soil and foliar nutrient measures over black cottonwood site index classes, theoptimal amounts of nutrients can be assessed, and these can be compared to nutrient measures ineach of the site associations. This will provide quantitative information for interpreting thenature of black cottonwood nutrient limitation within the site associations, and will express theresults of the analysis in a format that has operational application.The success of vegetation classification as a method of assessing ecological site quality isbased on the fact that certain species have relatively narrow ecological amplitudes along climate,soil moisture, and soil nutrient gradients (Daubenmire, 1976; Klinka et al., 1989b; Mueller-Dombois and Ellenberg, 1974). In British Columbia the indicative value of many plant specieshas been summarized by Klinka et al. (1989b), and several workers have attempted to use eitherindividual species, or indicator species classes as predictive variables to evaluate siteproductivity, and to predict site index of tree species (Kayahara, 1991; Klinka et al., 1989a;Klinka and Carter, 1990; Klinka and Krajina, 1987; Wang, 1992). The usefulness of understoryvegetation in predicting site index of black cottonwood is assessed in the present study, usingestimates of cover of understory vegetation species within the sample plots, and total cover of allspecies in several indicator species groups.The rate of height growth of black cottonwood in the 29 sites used for the study shows agreater than three-fold increase from 8.5 m/15 yrs in a low bench alluvial site to 30.8 m/15 yrson an upland loess soil with seepage (see Table 2.1). This range in site index implies that thereis a range of ecological conditions that parallels the increase in height growth. The generalobjective of this study was to begin to understand the changes in nutrient availability that occuras ecological factors change along this range of black cottonwood height growth in coastalBritish Columbia. The specific objectives of the study were:1) to correlate black cottonwood site index with taxa of the biogeoclimatic ecosystemclassification (Pojar et al, 1987) - especially site association, subzone, and soil nutrient regime;932) to assess the usefulness of understory vegetation in assessing black cottonwood site index;3) to characterize black cottonwood site index classes, and site association in terms of soil andfoliar nutrient quantities;4) to establish relationships among foliar nutrients, soil nutrients, and black cottonwood siteindex so that limiting nutrients and optimal foliar ratios can be established; and,5) to use foliar nutrient levels, DRIS indices, and soil contents to interpret the potential cause ofnutrient limitation or sufficiency in the different site units.5.2 METHODS5.2.1 Ecosystem Description and ClassificationA summary of ecological characteristics for the 29 sites used in the study has beenprovided in Chapter 2 and in Tables 2.2 and 2.3.5.2.2 Soil and Foliar Nutrient Sampling and AnalysisSampling protocols and analytical methods for foliar nutrients are described in Chapter3, and for soil nutrients in Chapter 4.945.2.3 Stem Analysis and Site IndexThe stem analysis procedure used to estimate black cottonwood site index is described inChapter 2. Details of stand age and composition are summarized in Table 2.1. The sites wereranked from low to high site index in Table 2.1, and divided into 3 site index classes: Low=8.5-14.5 m/15 yrs; Medium=15.0-21.9 m/15 yrs; and High=23.0 to 30.8 m/15 yrs. These classeswere used as grouping variables in this study to correlate black cottonwood site index with siteunits and ecological variables.5.2.4 Statistical MethodsBox diagrams (Titus, 1987; Tukey, 1977; Velleman and Hoaglin, 1981) were used in thisstudy to show the general distribution of the soil nutrient and foliar nutrient data within thedifferent qualitative classes. The central line of each box is the median of ordered values, andthe ends of the box ('hinges') split each half of the data again. The 'Hspread' is the absolutevalue of the length of the box, and the 'whiskers' (the lines extending from either end of the box)show the range of data within 1.5 Hspreads of the hinges. 'Outside' values (asterisks) are thosegreater than 1.5 Hspreads from the hinge, and 'far outside' values (open circles) are greater than3 Hspreads from the hinge.Approaches and procedures for testing the assumptions of ANOVA and regressionanalysis are described in Chapter 3. Ecological variables used to predict black cottonwood siteindex are seldom independent and thus violate a major assumption of linear regression(Chatterjee and Price, 1977). For this reason a series of univariate regressions were used todemonstrate relationships between dependent and independent variables, and were combinedinto complete multiple regression models only for variables where intercorrelations were notsignificant. Where intercorrelations between predictive variables were significant, only thepercentage of variance explained was used to evaluate the models.955.3 RESULTS5.3.1 Correlations of Black Cottonwood Site Index with CWH Subzones, Site Associationsand Soil Nutrient RegimeThe relatively low influence of stand location within the 3 CWH subzones on blackcottonwood site index is shown by the Box diagrams in Figure 5.1. The ANOVA comparingmeans for the 3 groups (Table 5.1) was not significant (p=.593) and suggests that the limitedrange of climates in which study sites were located had no significant effect on the heightgrowth of black cottonwood.As shown in the Box diagrams and ANOVAs (Figure 5.1; Table 5.1), black cottonwoodsite index in the 6 site associations sampled falls into two main groups - a group withsignificantly lower site index that includes study sites within the Ac-Willow, 'Gleyed' (Cw-Salmonberry/Cw-Black twinberry), and Cw-Swordfern site associations, and a second group,with significantly higher site index, that includes the Ac-Red osier dogwood, Ss-Salmonberry,and Cw-Foamflower site associations. Because of the small number in each class and similarblack cottonwood site index, study sites within the Cw-Salmonberry/Cw-Black twinberry siteassociations were considered together as the 'Gleyed' s,a.'s in this study. The ANOVA on siteassociations was highly significant (p < .001) and explained 87% of the variance in blackcottonwood site index (Table 5.1).Black cottonwood site index increased in a linear fashion along a gradient of increasingsoil nutrient availability, as shown by the Box diagrams (Figure 5.1) and the ANOVA results(Table 5.1). The results of the ANOVAs show that black cottonwood site index within themedium and rich SNR groups was not significantly different, but were significantly less thanblack cottonwood site96Table 5.1: Means and results of ANOVAs for black cottonwood site index (m/15 yrs) in 3subzones, 3 soil nutrient regime groups, and 5 site associations. Values with thesame letter are not significantly different at p < 0.05.Black cottonwoodSite IndexGroup/Subgroup^n^(m/15 yrs)SUBZONECWHdm^ 17^19.3aCWHds 6 21.8aCWHxm 6^18.1aSignificance)^ NSSOIL NUTRIENT REGIMEMedium^ 3^11.7 aRich 15 18.3 aVery Rich 11^23.6 bSignificance)^ **SITE ASSOCIATIONAc-Willow^ 6^11.5 a'Gleyed' 4 14.7 aAc-Red osier dogwood^5^21.6 bSs-Salmonberry 7 25.4 bCw-Foa mil ower 6^23.5 bSignificance1 Statistical significance of the ANOVA; NS = p > 0.05; * = 0.05 > p> 0.01; ** = 0.01 > p > 0.001; *** = p < 0.001index in the very rich SNR group. As for site associations, the ANOVA on SNR groups washighly significant (p=.003), but the SNR ANOVA explained only 36% of the variance in blackcottonwood site index.The ANOVAs of black cottonwood site index within site association or soil nutrientregime did not isolate the effect of soil moisture or soil nutrients because the site associationgroupings included medium to rich soil nutrient regimes, and the soil nutrient classesencompassed the range of flooding regimes and soil moisture conditions. The ANOVA of blackcottonwood site index by soil nutrient regime within site associations explained 88% of theFigure 5.1: Box diagrams showing distributions of black cottonwood site index inbiogeoclimatic subzone, soil nutrient regime (M=medium; R=rich; VR=veryrich), and site association (1=Ac-Willow; 2="Gleyed" sa's; 3=Cw-Swordfern;4=Ac-Red osier dogwood; 5=Ss-Salmonberry; 6=Cw-Foamflower) groups.9798variance within the rich class, and 74% of the variance within the very rich class. Because of thesmall numbers in each group it was not possible to assess the effect of nutrient regime within siteassociation.5.3.2 Principal Component Analysis of Vegetation, Soil Nutrients, and Foliar NutrientsThe first six axes of the PCA of 85 understory species in the 29 study sites explained85% of the variation in the vegetation cover data (Table 5.2). The first two PCA axes accountedTable 5.2:^Eigenvalues (variance explained), percentage of total variance explained, and totalcumulative variance explained for PCA axes 1 - 6 of the PCA of 85 understoryspecies (presence class > II) in 29 black cottonwood study sites.Axis EigenvaluesPercent of TotalVariance ExplainedTotal CumulativeVariance ExplainedPCA 1 441.58 29.8 29.8PCA 2 343.80 23.3 53.1PCA 3 180.58 12.2 65.3PCA 4 147.97 10.0 753PCA 5 88.22 5.96 80.0PCA 6 69.78 4.72 85.0for 53.1% of the variation in the vegetation data, and an ordination of plot locations of these twoaxes is shown in Figure 5.2. To show inter-relationships among the groupings, study sites in theordination are labelled by site index class, and site association (s.a.). Those species significantlycorrelated with the first two PCA axes are given in Table 5.3, and their SNR and SMR indicatorstatus are also given in that table to assist with the interpretation of the ordination in Figure 5.2.Species positively correlated with PCA axis 1 (Equisetum arvense group) are common onflooded sites, both on the alluvial Ac-Red osier dogwood and Ac-Willow s.a.'s, and on the'Gleyed' s.a.'s. Species negatively correlated with PCA axis 1 included a group of speciescommon in the99Figure 5.2: (Overleaf) PCA ordination of 85 plant species with presence class > H in 29 blackcottonwood stands. Study sites are labelled by site index class (L=low;M=medium; H=high) and site association (1=Ac-Willow; 2="Gleyed" sa's; 3=Cw-Swordfern; 4=Ac-Red osier dogwood;  5=Ss-Salmonberry; 6=Cw-Foamflower). (a) Site Index Class100101Table 5.3: List of species significantly (p < .10) correlated with the first and second PCA axesthat have soil moisture and/or soil nutrient regime indicator status (Klinka andKrajina, 1986). The first group for each axis is negatively correlated with the axis.r2Species (PCA 1) SNR1 SMR1Rubus spectabilis 0.887 R VM-WAthyrium felix-femina 0.710 R VMWSambucus racemosa 0.701 R F-VMRibes bracteosum 0.490 R VM-WPolystichum munitum 0.446 RTellima grandiflora 0.425 R F-VMGeum macrophyllum 0.378 R F-VMPlagiomnium insigne 0.363 R VM-WCarex deweyana 0.320 R F-VMEquisetum arvense 0.310 MCrataegus douglasii 0.325 R VM-WCornus sericea 0.332 R VM-WSymphoricarpos albus 0.398 RLonicera involucrata 0.553 R VM-W12Species (PCA 2) SNR1 SMR1Fragaria virginiana 0.440 MAdenocaulon bicolor 0.408 R MD-FClintonia uniflora 0.392 P MD-FAcer circinatum 0.381 R F-VMRhytidiadelphus triquetrus 0.370 MHylocomium splendens 0.368 PSmilacina stellata 0.318 RAsarum caudatum 0.315 R F-VMAngelica genuflexa 0.335 R W-VWStachys cooleyae 0.379 R VM-WRubus spectabilis 0.401 R VM-WAchlys triphylla 0.409 RCarex obnupta 0.429 R W-VWSymphoricarpos albus 0.462 RCornus sericea 0.591 R VM-WRubus parviflorus 0.612 RLonicera involucrata 0.627 R VM-W1 SMR and SNR codes are as shown in Table 2.2102Cw-Foamflower and Ss-Salmonberry s.a.'s, and indicate a gradient along the axis from sitesregularly flooded, to those sites where flooding is either infrequent, as in the Ss-Salmonberrys.a., or sites within the Cw-Foamflower s.a. that do not experience flooding. A gradient ofincreasing black cottonwood productivity is evident from right to left along PCA axis 1 - all low(except one) and medium (except two) site index plots occur at the positive end of the axis. Onthe second PCA axis, species positively correlated species had very moist to wet (VMW) andwet to very wet (WVW) SMR indicator status, and rich SNR status (Table 5.3), and werecommon on gleyed upland sites (Table 2.5). This group can be contrasted with speciesnegatively correlated with PCA axis 2, which had moderately dry to fresh, and fresh to verymoist, SMR status, and poor to rich SNR indicative values (Table 5.3). A gradient from nutrientpoorer and drier sites, to those with very moist to wet soil moisture regimes is also demonstratedby the distribution of site associations along the axis. All Gleyed s.a.'s occurred at the positiveend, and the Cw-Swordfern, the driest of the s.a.'s sampled in the study, is situated at thenegative extreme of PCA axis 2. Also, all sites with medium SNR status are located at theextreme negative end of PCA axis 2, and are associated with a group of sites with low blackcottonwood site index. The PCA of vegetation indicates that vegetation within study sites isaffected by combined flooding and soil nutrient gradients, although the effects of flooding aremore evident. The site index of black cottonwood is weakly associated with gradients of soilmoisture and nutrients, as indicated by the vegetation on the sites.The first three axes of the PCA of soil pH and soil nutrient contents for 29 blackcottonwood ecosystems explained a cumulative total of 80% of the variation in the data (Table5.4, Figure 5.3). PCA 1 explained 33.8% of the variance in the data, and presents a gradientfrom sites with relatively high organic matter (total N, total C, mineralizable N), to those withlow organic content and high pH. PCA 2 explained 28.5% of the variance and represents agradient from sites with high pH and high levels of exchangeable Ca, to those with relativelyhigher levels of mineralizable N, total N, and available P. The third PCA axis explained 17.6%of the variance in the data and contrasts sites with high available P and exchangeable K, with103Table 5.4: Correlations of pH and soil nutrient contents (kg/ha) with the first three principalcomponent axes, eigenvalues, and percentage and cumulative percentage varianceexplained by the PCA axes in 29 black cottonwood ecosystems. Bolding indicatessignificance of the correlations at p < 0.05.Nutrient PCA 1 PCA 2 PCA 3Total N 0.969 -0.068 0.079Mineralizable N 0.560 -0.230 0.087Total C 0.927 0.197 0.000Exchangeable Ca 0.250 0.915 0.019Exchangeable K 0.106 0.238 0.577Available P 0.041 -0.132 0.968Available 804-S -0.026 0.207 0.010Exchangeable Mg -0.050 0.243 -0.188pH -0.328 0.766 -0.251Eigenvalue 3.04 2.57 1.58% of Variance Explained 33.8 28.5 17.6Cumulative^%^of^Variance 33.8 62.3 79.9Explainedthose that have higher pH and exchangeable Mg. The general trend of variation summarized inthe three PCA axes is to contrast sites with a high organic matter content, available P, andexchangeable K, with sites with high pH and contents of exchangeable Ca and Mg.Figure 5.3 uses labels for black cottonwood site index class and site association tocompare characteristics of the sites on PCA axes 1 and 2 of the soil nutrient-pH ordination. Theupper left quadrant of the ordination includes mostly Ac-Willow (and 2 Ac-Red osier dogwood)sites that have low black cottonwood site index, and medium to rich soil nutrient regimes.Given the correlations shown in Table 5.4, this distribution suggests that Ac-Willow and Ac-Redosier dogwood sites with high pH and exchangeable Ca levels were relatively poor sites forblack cottonwood growth. These sites are contrasted with those in the lower right quadrant ofthe ordination, which includes most of the sites with very rich soil nutrient regimes, mostly104Figure 5.3: (Overleaf) PCA ordination of pH and 8 soil nutrient content properties in 29 blackcottonwood stands. Study sites are labelled by site index class (L=low;M=medium; H=high) and site association (1=Ac-Willow; 2="Gleyed" sa's;3=Cw-Swordfern; 4=Ac-Red osier dogwood; 5=Ss-Salmonberry; 6=Cw-Foamflower).(a) Site Index Class105106medium to high black cottonwood site index, and which were included in a variety of siteassociations. This quadrant includes sites with relatively high organic matter content andavailable P, and correspondingly low pH and exchangeable Ca (Table 5.4). This comparisonsuggests that, in general, sites with high organic matter and available P, and lower pH andexchangeable Ca contents, were relatively good sites for black cottonwood growth. Theinclusion of a few sites with low black cottonwood site index in the lower right quadrant (Figure5.3) shows that this generalization does not apply to all sites. Both of the sites with low blackcottonwood site index in the lower right quadrant of the ordination have gleyed horizons withinthe rooting zone, so that uptake of nutrients that would otherwise be available may be impededby poor soil aeration. A weak gradient of increasing flooding frequency from right to left alongPCA axis 1 is discernible in the ordination (Figure 5.3). Two Cw-Foamflower sites occur at theextreme right end of the axis, and lower bench alluvial floodplain site associations increase,from right to left along the axis.The first three PCA axes of 13 foliar nutrients in 26 black cottonwood ecosystemsaccounted for about 64% of the variance in the foliar nutrient data matrix (Table 5.5). PCA axis1 accounted for about 35% of the variability in the foliar nutrient data, and contrasts sites whereblack cottonwood foliage had relatively high foliar concentrations of N, K, Cu, B, S, S-SO 4, andactive-Fe, with those with relatively higher concentrations of Ca, Mg, and Mn. The gradientalong PCA axis 1 is correlated with decreasing black cottonwood site index, and a change fromsites representing the Cw-Foamflower and Ss-Salmonberry s.a.'s, to those representing the Ac-Willow and Gleyed s.a.'s (Figure 5.4). This gradient can be interpreted as one of increasingrooting zone flooding associated with a reduction in black cottonwood site index. PCA axes 2and 3 each accounted for about 14% of the variation in the foliar data, and demonstrate gradientsfrom foliage high in N, Mg, and Mn, to that high in Zn, and from foliage high in Zn and B, tothat high in P, Fe, and active -Fe, respectively.107Table 5.5: Correlations of foliar nutrient concentrations with the first three principal componentaxes, eigenvalues, and percentage and cumulative percentage variance explained bythe PCA axes in 26 black cottonwood stands. Bolding indicates significance of thecorrelations at p < 0.05.Foliar Nutrient PCA 1 PCA 2 PCA 3N (%) .617 .608 -.026P (%) .111 .113 -.420K (%) .837 -.094 -.320Ca (%) -.540 .214 .142Mg (%) -.384 .682 .140Cu (ppm) .820 .337 .155Zn (ppm) .223 -.396 .481Fe (ppm) .313 -.346 -.527Mn (ppm) -.495 .594 -.143B (ppm) .490 -.294 .639S (%) .879 .381 .145SO4 (ppm) .837 .146 .243active-Fe (ppm) .466 -.065 -.693Eigenvalue 4.528 1.903 1.843% of Variance Explained 34.8 14.6 14.2Cumulative % of Variance Explained 34.8 49.4 63.6The contrast along PCA 1 of the foliar ordination between sites with high concentrationsof foliar N and K, and those with high foliar Ca and Mg concentrations, is similar to PCA 1 ofthe soil nutrient ordination - sites with high soil contents of mineralizable N are contrasted withsites of relatively higher exchangeable Ca and Mg. In both the soil and foliar nutrientordinations, black cottonwood site index decreased along this gradient.108Figure 5.4: (Overleaf) PCA ordination of 13 foliar nutrient concentrations in 29 blackcottonwood stands. Study sites are labelled by site index class (L=low; M=medium;H=high) and site association (1=Ac-Willow; 2="Gleyed" sa's; 3=Cw-Swordfern;^4=Ac-Red osier dogwood; 5=Ss-Salmonberry; 6=Cw-Foamflower). (a) Site Index1091105.3.3 ANOVAs of Soil and Foliar Nutrients in Black Cottonwood Site Index Classes, SoilNutrient Regimes, and Site AssociationsANOVAs comparing the statistical significance of the differences in mean soil nutrientcontent in 3 site index classes, 3 soil nutrient regime classes and 5 site associations are presentedin Table 5.6. Only mineralizable N and total N demonstrated significant (p < 0.05) changesamong site index classes. Mineralizable N, total N, total C, available P, and exchangeable K allincreased with black cottonwood site index. Exchangeable Ca and Mg decreased withincreasing black cottonwood site index. This pattern of soil content of exchangeable Ca and Mgbeing negatively related to black cottonwood site index, and opposite in trend to the other soilnutrients was observed in the PCA for soil nutrient contents shown in Figure 5.3. Sulphate Sshowed no trend with regard to site index class.Only total N and total C had significant differences in soil contents among soil nutrientregime classes (Table 5.6). Soil nutrient contents increased from medium to rich to very richsoil nutrient regimes mineralizable N, total N, and total C. Considering that the differentiationof soil nutrient regime classes in this study was based to a large extent on the depth anddevelopment of the Ah layer, it is to be expected that nutrients associated primarily with themineralization of organic matter are higher in the rich and very rich classes. For exchangeableMg and K, and for available P, soil contents decreased from nutrient rich to very rich soilnutrient regimes, although the decreases were not statistically significant (Table 5.6).Significant (p < .05) differences among 5 site associations were demonstrated for all soilnutrients (Table 5.6). For total N, total C, mineralizable N, and available P, the significance ofthe ANOVAs was the result of much lower soil nutrient contents in the Ac-Willow s.a. Soilnutrient contents for mineralizable N, total C, exchangeable K, and available P all increasedalong a gradient of decreasing flooding frequency, from Ac-Willow to Ss-Salmonberry s.a.'s.Exchangeable Ca and Mg (and pH - not shown) decreased along the same gradient. High total Nand total C contents in upland s.a.'s (Cw-Foamflower, 'Gleyed') are the result of the presence ofModer humus forms, unique to those sites.111Table 5.6: ANOVAs of soil nutrient contents (kg/ha), for 29 black cottonwood ecosystems in 3index classes, 3 soil nutrient regime classes, and 5 site associations. The Cw-Swordfern site association had only one site and was not included in the ANOVAs.For a given nutrient, in a given class, values with the same letter are not significantlydifferent (p < 0.05).SITE INDEX CLASS^n^Total N^Total C^Min-N^Av-P^Ex-Ca^Ex-Mg^Ex-K^SO4-SLow^ 9^5,506a^95,001a^94a^34a^12,912a^1,769a^453a^20aMedium 9^9,964ab^135,998a^155ab^46a^8,257a^1,506a^705a^18aHigh 11^11,019b^145,364a^187b^60a^7,666a^768a^755a^23aSignificance'^ * NS * NS'^NS^NS'^NS'^NS'SOIL NUTRIENT REGIMEn^Total N^Total C^Min-N^Av-P^Ex-Ca^Ex-Mg^Ex-K^SO4-SMedium^ 3^3,859a^71,648a^90a^16a^11,277a^723a^573a^21aRich 15^8,124b^108,961b^144a^57a^10,106a^1,605a^674a^22aVery Rich 11^11,784c^169,724c^177a^43a^8,231a^1,091a^568a^20aSignificance'^ ** ** NS NS^NS^NS NS^NS'SITE ASSOCIATIONn^Total N^Total C^Min-N^Av-P^Ex-Ca^Ex-Mg^Ex-K^SO4-SAc-Willow^ 6^4,177a^72,282a^58a^16a^15,907a^2,084a^448ab^25abAc-Red osier dogwood^5^10,216ab^114,947ab^126b^34ab^14,412a^3,025a^815c^42bSs-Salmonberry 7^8,673ab^122,603ab^146b^66ab^6,263b^754b^789c^22ab'Gleyed'^ 4^11,017b^181,362b^180b^38ab^2,985b^674b^221a^6.5aCw-Foamflower^6^14,227b^201,903b^220b^84b^6,675b^849b^719bc^17abSignificance' ** ** ** *2 ***2 *5* ***2 *2' statistical significance of the ANOVA; NS = p > 0.05; ' = 0.05 > p > 0.01; ** = 0.01 > p > 0.001; *** = p < 0.0012 variables transformed to natural logarithms to satisfy model requirements for homogeneity of variance or normality.ANOVAs comparing the statistical significance of group differences in 13 foliarnutrients are presented in Table 5.7. Foliar concentrations of N, P, K, S, Cu, SO 4, and active Feall increased with increasing black cottonwood site index. The increases were statisticallysignificant for all of these nutrients except P (Table 5.7), and the recurrent pattern was for foliarconcentrations in the high site index class to be significantly higher than concentrations in boththe low and medium site index classes. Differences in foliar N concentrations were significantfor all site index classes. The nutrients demonstrating this trend of increase as black cottonwoodsite index increases, are part of a group of nutrients positively correlated with the first axis of the112PCA of foliar nutrient data, where there occurrence was correlated with the high site index class(Figure 5.4).Relationships between foliar nutrient concentrations and site associations were morecomplex than for site index class, and many of the differences were statistically significant(Table 5.7). Foliar concentrations of N, K, and S demonstrated similar relationships within siteassociations - they all increased with increasing bench height on alluvial floodplains, followedby concentration decreases in upland site associations. Foliar P concentrations showed the sametrend of increasing with increasing bench height on alluvial floodplains, but maintained highconcentrations in upland site associations. Concentrations of foliar Ca, Mg, B, Mn, and SO 4 ,demonstrated irregular differences among site associations, that appeared to be unrelated toTable 5.7: ANOVAs of foliar nutrient concentrations of 29 black cottonwood stands in 3 siteindex classes, and 5 site associations. The Cw-Swordfern site association has onlyone site and was not included in the ANOVAs. For a given nutrient, in a given class,values with the same letter are not significantly different at p < 0.05. The number ofstudy stands in each group is given in Table 5.6.SITE INDEX^N^P^K^Ca^Mg^Cu^Zn^Fe^Mn^B^S^SO4 act-FeCLASS (%)^(%)^(%)^(%)^(%) (ppm) (ppm) (ppm) (ppm) (ppm) (%) (ppm) (ppm) Low^ 1.81a 0.21a^1.10a^1.30a 0.28a^6.8a^93a^91a^47a^26a^0.19a 411a^68aMedium 2.05b 0.21a^1.28a^1.38a 0.27a^7.8a^81a^97a^56a^28a^0.21a 420a^69aHigh 2.45c 0.24a^1.76b^1.21a 0.21a 10.9b^99a^91a^34a^26a^0.27b 679b^71aSignificance'^***^NS^** NS^NS^***^NS^NS^NS^NS^***^*43^NSSITE^ N^P^K^Ca^Mg^Cu^Zn^Fe^Mn^B^S^SO4 act-FeASSOCIATION^(%)^(%)^(%)^(%)^(%) (ppm) (ppm) (ppm) (ppm) (ppm) (%) (ppm) (ppm) Ac-Willow^1.71a 0.19a^1.18a,b^1.16a 0.24a 7.3a,b^94a^100a 41a,b^32b 0.20a,b 476a^71aAc-Red osier dogwood^2.02b 0.19a^1.31b,c 1.19a,b 0.31a 10b,c^96a^100a 45a,b^38b 0.21a,b 445a^69aSs-Sal mon berry^2.55c 0.26b^1.90d^1.12a 0.21a^11c^99a^82a^37a^24a,b 0.27b 646b^66aCw-Foamflower 2.04b 0.28b 1.50b,c,d 1.42a,b 0.21a^6.8a^97a^102a 66a,b 23a,b 0.23b 566a,b^77a'Gleyed'^2.11b 0.21a^0.82a^1.75b 0.43b^6.8a^76a^99a^66b^13a^0.20a 234a^74aSignificance' ***^**a^*** *^***^***^NS^NS **^**^*2 NS' statistical significance of the ANOVA; NS = p > 0.05; * = 0.05 > p > 0.01; *• = 0.01 > p > 0.001; •** = p < 0.001variables transformed to natural logarithms to satisfy model requirements for homogeneity of variance or normality113black cottonwood site index. Foliar Fe, active Fe, and Zn showed slight changes among siteassociations and no significant differences were found (Table 5.7).5.3.4 Linear Regressions of Vegetation on Site IndexThe linear model fit by forward stepwise regression analysis using the first 10 axes of thevegetation PCA utilized the first 4 PCA axes, and accounted for 51% of the variation in siteindex (Equation 1 - Table 5.8). PCA axis 2 had the highest partial F value, and the coefficientwas positive, and reflected the clustering of sites in the low site index class at the negative end ofthe axis (Figure 5.2). Similarly, PCA axis 1 had the second highest partial F value, had anegative regression coefficient, and had a preponderance of sites in the high site index class atthe negative end of the axis in the vegetation ordination (Figure 5.2). The regression suggeststhat about half of the variation in black cottonwood site index can be explained by the presenceor absence of the different understory species growing in association with black cottonwood onthe study sites.Table 5.8: Models, probabilities, coefficients of determination (R 2), and standard error of theestimate (SEE) for vegetation variables on site index in 29 black cottonwood stands.MODEL p R2 SEE (m)(1) SlAc = 19.597 + 0.096 (PCA2) - 0.201 (PCA1) - 0.158 (PCA4) - 0.139 (PCA3) .001 0.51 4.73(2) Sim = 15.835 - 0.152 (M) + 0.044 (R) .007 0.32 5.35(3) SlAc = 11.214 - 0.126 (FVM) + 0.067 (VMW) .003 0.36 5.20(4) SIAc = 19.603 - 0.114 (PCA2) + 0.043 (PCA1) .002 0.39 5.07The regression of the frequency of species in three soil nitrogen indicator species groups(N-poor, N-medium, and N-rich) on black cottonwood site index was less successful than thespecies themselves, and explained 32% of the variation (Equation 2 - Table 5.8). N-mediumspecies had a negative regression coefficient, and N-rich sites had a positive regression114coefficient, and showed the respective negative and positive correlations of these speciesindicator groups with black cottonwood site index. The regression of soil moisture indicatorspecies groups on black cottonwood site index explained a slightly higher amount of variation(36%) than the soil nitrogen indicator species groups (Equation 3 - Table 5.8). The negativeregression coefficient of the fresh to very moist indicator species group showed the reduced siteindex on sites supplied with relatively lower amounts of soil moisture, and can be compared tothe positive effect of moister sites with a preponderance of species with very moist to wet SMRindicator status.Given the high intercorrelations of soil moisture and soil nitrogen indicator speciesgroups, a stepwise linear regression analysis combining both indicator groups was not carriedout. The forward stepwise regression on the first 5 axes (which explained 85% of the totalvariation in the PCA) of the PCA using soil moisture and nitrogen indicator species group dataidentified PCA axes 1 and 2 as the most important variables correlated with black cottonwoodsite index, and explained 39% of the variation in the model (Equation 4 - Table 5.8).5.3.5 Linear Regressions of Soil Nutrient Contents on Site IndexThe results of univariate regressions for each of the soil nutrients on site index (Table5.9) followed the same trends as those discussed for Table 5.6. Log-transformed variables arelisted where their regressions explained a higher percentage of the variance in the model. Logtotal N (ltotN), log total C (ltotC), log mineralizable N (lminN), log available P (lavP), and logexchangeable K (lexK) all had significant linear relationships with black cottonwood site indexfor the 29 study sites. The percentage of variance explained by the univariate regressions rangedfrom 42% for log total N and 37% for log mineralizable N, to 16% for log exchangeable K(Table 5.9). The standard error of the estimate for the models was large and varied from 4.76 mfor log total N to 5.91 m for log exchangeable K. The univariate models suggested that a totalN/mineralizable N/total C group were the most important nutrients determining the height1 15growth of black cottonwood across all 29 study sites. Available P and exchangeable K were alsoimportant but appeared to play a secondary role.Table 5.9: Models, probability, coefficients of determination (R2), and standard error of theestimate (SEE) for univariate regressions of soil nutrient contents on blackcottonwood site index in 29 black cottonwood stands. Only soil nutrients withsignificant (p < 0.05) regressions are shown.MODEL p R2 SEE (m)(1) SIAc = -42.64 + 6.900 (ItotN) .000 0.42 4.76 m(2) SIAc = -38.11 + 4.956 (ltotC) .043 0.15 5.90 m(3) SIAc = -3.67 + 4.192 (lminN) .001 0.37 5.13 m(4) SIAc = 8.56 + 3.156 (lavP) .027 0.18 5.91 m(5) SIAc = -8.84 + 4.478 (lexK) .033 0.16 5.91 mThe correlation matrix of variables that had significant linear regressions on blackcottonwood site index (Table 5.10) shows that the significant soil nutrients fell into 2 correlatedgroups - total N (totN), total C (totC), and mineralizable N (minN) were significantly correlatedwith each other, as were available P (avP) and exchangeable K (exK). In the multipleregressions shown in Table 5.11, mineralizable N was used as the measure of nitrogenavailability, because it was highly correlated with both total N and total C. Exchangeable K andavailable P were also used in the multiple regressions, both individually, and together withmineralizable N. Equations 1 and 2 in Table 5.11 included all 3 soil nutrients and explained60% of the variance in site index for the log-transformed variables, and 58% for theuntransformed measures of soil nutrient contents. Substituting total N for mineralizable N(Equation 3 - Table 5.11), and including soil pH (Equation 4 - Table 5.11) did not increase theexplanatory power of the multiple linear regression of soil nutrients on black cottonwood siteindex. Forward stepwise analysis of log transformed soil nutrient variables resulted in a modelthat included only mineralizable N and available P, and accounted for 53% of the variation inthe model (Equation 5 - Table 5.11). The same procedure for untransformed variables resulted116in Equation 6 in Table 5.11, and accounted for 42% of the variation in the model. Models 5 and6 in Table 5.11 are presented as the 'best' complete equations relating soil nutrient contents toblack cottonwood site index for the 29 study sites.Table 5.10: Correlation matrices for soil nutrient variables with significant univariate linearregressions on black cottonwood site index. Bolding indicates significant (p < 0.05)correlations.minN^exK^avP^totC^IminN lexK^lavP^ItotC exK^.045 lexK^.194avP^.228^.447 lavP^.299^.418totC^.695^-.221^-.098^ItotC^.845^-.046^.062totN^.640^.074^.014^.859^ltotN^.838^.167^.195^.916Table 5.11:Probability (p), coefficients of determination (R 2), and standard error of the estimate(SEE) for multiple regressions of soil nutrients on black cottonwood site index,using variables with significant univariate regressions on black cottonwood siteindex.SOIL NUTRIENTS^p^R2^SEE (m)(1) IminN, lavP, lexK^ .000^.60^4.60 m(2) minN, avP, exK .001^.58^4.71 m(3) totN, avP, exK .005^.49^5.01 m(4) pH, minN, avP, exK^ .002^.59^4.79 mMODEL p^R2^SEE (m)(5) Sim = -10.24 + 4.41 IminN + 2.15 lavP^.000^.53^4.75 m(6) Sim = 9.23 + 0.041 minN + 0.007 exK^.002^.42^5.12 mParameters and summary statistics for univariate regressions of soil nutrients on siteindex for a reduced data set that includes only those stands where cottonwood predominated areshown in Table 5.12. In sites deleted from the full data set, black cottonwood occurred as117Table 5.12: Probability (p), coefficients of determination (R 2), and standard error of the estimate(SEE) for univariate regressions of soil nutrient contents on black cottonwood siteindex in the reduced data set (n=22). Only soil nutrients with significant (p < 0.05)regressions are shown.MODEL p R2 SEE (m)(1) SIAc = 11.58 + 0.001 (totN) .000 0.52 4.64(2) Sim = -93.19 + 9.810 (ItotC) .000 0.57 4.45(3) SIAc = -8.14 + 6.16 (1minN) .000 0.62 4.12(4) Sim = 6.12 + 4.198 (lavP) .009 0.31 5.68(5) Sim = 26.62 - 0.003 (exCa) .013 0.30 5.76(6) SIAc = 24.88 - 0.001 (lexMg) .030 0.22 5.99as scattered individuals among other deciduous and coniferous species (see Table 2.1). The 22sites included in the reduced data set provided the opportunity to examine soil nutrient - blackcottonwood site index relationships in more uniform soil conditions that were typical of well-stocked black cottonwood stands. Univariate regressions of soil nutrient contents on blackcottonwood site index for the reduced data set were similar to those shown for the full data set,except that exchangeable K was not significant, and exchangeable Ca and Mg demonstratedsignificant and negative linear relationships. For all nutrients except exchangeable K, thevariance explained by the linear models was about twice as high for the reduced data set,compared to the complete data set. The stronger relationships demonstrated for soil nutrients inthe reduced data set were due primarily to the reduction in humus form variability, and in thedeletion of upland sites where soil drainage was impeded, so that uptake of nutrients presentwithin the soil was impaired. As for the complete data sets, coefficients for exchangeable Caand Mg in the reduced data set were negative, and show that increasing amounts of thesenutrients were significantly correlated with decreasing site index.Correlation matrices for both the log-transformed and the non-transformed values of soilnutrients in the reduced data set in Table 5.13 showed the same pattern as the full data set (Table1185.10). Total N, total C and mineralizable N formed a significantly, positively correlated group,as did exchangeable Ca and Mg. Available P had a significant negative correlation with bothexchangeable Ca and Mg.Table 5.13: Correlation matrices for soil nutrient variables with significant univariate linearregressions on black cottonwood site index for the reduced data set. Boldingindicates significant (p < 0.05) correlations.totN^totC^minN^avP^exCa^ ItotN^ItotC^IminN^IavP^lexCa totC^0.870 ItotC^0.920minN^0.662^0.728 IminN^0.850^0.865avP^0.070^-0.070^0.350^ IavP^0.226^0.057^0.407exCa^-0.407^-0.396^-0.471^-0.600^ lexCa^-0.294^-0.279^-0.389^-0.473exMg^-0.118^-0.174^-0.278^-0.527^0.745^IexMg^0.046^0.009^-0.119^-0.416^0.881Multiple regression models of soil nutrients significantly correlated with site index(Model 1 - Table 5.14) explained about 80% of the variance in site index for the reduced dataset. Complete models with regression coefficients are not given because of significantcorrelations among the explanatory variables with significant univariate probabilities (Table5.12). As for the full data set, mineralizable N was used to represent soil nitrogen availability.Given the high positive correlation of exchangeable Ca and Mg, a new variable was created (exCa+Mg) that summed the 2 values. Using this variable did not increase the explanatory powerof the models (Model 2 - Table 5.14). Dropping exchangeable Mg from the model did notreduce the percentage of variance explained (Model 3 - Table 5.14), although, whenexchangeable Ca was omitted, the percentage of variance explained dropped to 69% (Model 4 -Table 5.14). The model using the log transformed values explained about the same percentageof variance as the non-transformed soil nutrient values (Model 5 - Table 5.14).119Table 5.14: Probability (p), coefficients of determination (R 2), and standard error of the estimate(SEE) for multiple regressions of soil nutrients on black cottonwood site index,using variables with significant univariate regressions on black cottonwood siteindex in the reduced data set (n=22).Soil Nutrients p R2 SEE(1) minN, avP, exCa, exMg .000 .79 3.59(2) minN, avP, ex (Ca+Mg) .000 .79 3.38(3) minN, avP, ex Ca .000 .80 3.34(4) minN, avP, ex Mg .000 .69 4.08(5) IminN, lavP, lex(Ca+Mg) .000 .80 3.465.3.6 Interaction of Available-P, Exchangeable Ca, and pH.The solubility and availability of soil P is determined to a large extent by soil pH and theconcentrations of Ca available to fix P as calcium phosphate concretions (Boishot et al., 1950;Cole et al., 1953; Griffin and Jurinak, 1973; Russell, 1974). The significant (p < 0.05) negativerelationship between available P and exchangeable Ca (Table 5.13) in the reduced data suggeststhat this effect may be responsible for the low amounts of available P in study site soils withhigh exchangeable Ca contents. This relationship is well demonstrated for the 22 sites in thereduced data set (Figure 5.5). In Figure 5.5, available P is much lower when pH is in excess of6.5, and when exchangeable Ca contents increase over about 15,000 kg/ha (see hatched lines inFigure 5.5). Data labels in Figure 5.5 refer to high, medium, and low site index classes, and thetrend is for high site index class sites to have high amounts of available P, relatively lowamounts of exchangeable Ca, and relatively low pH. Although an indication of total P in thesesoils is required to confirm it, one explanation for this relationship is that soil contents ofavailable P are fixed and thus made unavailable in black cottonwood stands with high pH andexchangeable Ca contents.Figure 5.5: Three dimensional scattergram showing the effect of increasing soil pH andcontent of soil exchangeable Ca on content of soil available P for the reduced dataset (n=22). Study sites are labeled by their black cottonwood site index class(L=low; M=medium; H=high).1201215.3.7 Linear Regressions of Foliar Nutrient Concentrations on Site IndexFoliar nutrients with significant univariate linear regressions on site index (Table 5.15),were the same as those with significant ANOVAs among site index classes in Table 5.7. Theregressions of foliar N, K, S, SO4, and Cu were all highly significant. The percentage ofvariance in site index explained by the regressions ranged from about 70% for foliar S to 20%for foliar SO4 . Standard errors of the estimate ranged from 3.23 to 5.34 m.Table 5.15: Univariate models, probabilities (p), coefficients of determination (R 2), and standarderrors of the estimate (SEE) for regressions of foliar nutrients on site index in 26black cottonwood stands. Only foliar nutrients with significant (p < 0.05 )regressions are shown.MODEL p R2 SEE (m)(1) SIAc = 13.325 + 132.68 (folS) .000 .695 3.232(2) Sim = -8.834 + 12.788 (folN) .000 .630 3.481(3) Sim = 5.552 + 9.576 (folK) .000 .462 4.372(4) SIAc = 4.58 + 1.640 (folCu) .000 .455 4.398(5) Sim = 14.703 + 0.007 (folSO4) .023 .198 5.337Almost all of the foliar nutrients that had significant regressions on site index werehighly correlated (Table 5.16). As for soil nutrients, this creates problems of collinearity whenconstructing multiple regression models of the foliar nutrients on site index. Variouscombinations of those foliar nutrients with significant univariate regressions on blackcottonwood site index are shown in Table 5.17, where the percentage of variance explainedranged from 58 to 77%, and the standard errors of the estimate ranged from 3.0 to 3.6 m. Model1 in Table 5.17 includes all five foliar nutrients with significant univariate regressions, and hasthe highest R2 (0.77) of all models. Model 2 excludes foliar S, and assumes that S is notlimiting growth, but rather is taken up in proportion to the amount of N taken up. The very high122correlation between foliar S and N shown in Table 5.16 supports this interpretation. Model 2explains 73% of the variation in black cottonwood site index with a standard error of theestimate of 3.12 m. Combining highly correlated variables into one variable that wasTable 5.16: Correlation matrix for foliar nutrient variables with significant univariate linearregressions on black cottonwood site index. Bolding indicates significant (p < 0.05)correlations.Cu^SO4^S^NSO4^.320S^.714N^.712K^.484.669.263.680.814.696 .451the sum of the concentrations of all of the nutrients (N+Cu+S and N+Cu+S+K) did notsignificantly increase the explanatory power of the models (Models 3 and 4 - Table 5.17).Forward stepwise regression analysis of all significant nutrients identified foliar N and SO 4 asthe most important nutrients (Model 5 - Table 5.17), although the percentage of varianceexplained was lower than for Models 1 and 2 in the same table. Using just foliar N, K, and Cu(Model 6), forward stepwise regression identified foliar N and K as the most important indetermining site index. The percentage of variance explained for this model was about the sameas Model 2, where all three variables are included.Table 5.18 presents model parameters for foliar nutrients with significant regressions onblack cottonwood site index for the reduced data set (n=20) described above for the soil nutrientregressions. The list of nutrients was essentially the same as for the complete data set, exceptthat the regression with foliar P was significant, and that for SO4 was not. In general, thepercentage of variance explained was higher (32-70%) for the reduced data set, and the standarderrors of the estimate about the same. Foliar N alone accounted for about 70% of the variationin site index, followed by foliar K, Cu, P and S, in decreasing order of variance explained.123Table 5.17: Probabilities (p), coefficients of determination (R2), and standard errors of theestimate (SEE) for multiple regression models of foliar nutrients on blackcottonwood site index, using variables with significant univariate regressions.Foliar Nutrients^ p^R2^SEE(1) N, K, S, SO4, Cu^ .000^.774^2.967(2) N,K,Cu^ .000^.730^3.115(3) (N+Cu+S), K .000^.642^3.425(4) (N+Cu+S+K) .000^.584^3.598(5) forward stepwise (N, K, S, SO4 , Cu) - N, SO4^.000^.635^3.536(6) forward stepwise (N, K, Cu) - N, K^ .000^.729^3.048Table 5.18:Probabilities (p), coefficients of determination (R 2), and standard errors of theestimate (SEE) for univariate regressions of foliar nutrients on black cottonwood siteindex in 20 black cottonwood ecosystems (reduced data set). Only foliar nutrientswith significant (p<0.05) regressions are shown.Foliar Nutrient^p^R2 S^EE (m)(1) SIAc = -7.937 + 12.767 (folN)^.000^.701^3.362(2) SIAc = 2.772 + 11.506 (folK)^.001^.458^4.660(3) Sim = 274.66 + 446.19 (folCu)^.004^.381^4.979(4) SIAc = 2.550 + 80.442 (folP)^.010^.315^4.762(5) SIAc = 4.903 + 61.107 (folS)^.010^.315^5.237As for the complete data set, the correlation matrix presented in Table 5.19 shows a highdegree of intercorrelation among foliar nutrients that have significant regressions on site index.The multiple regression model that included all foliar nutrient variables with significantunivariate regressions on black cottonwood site index (Model 1 - Table 5.20) had the highestexplanatory power with an R 2 of about 80%. Model 2, which excluded S, had a slightly lower124R2, as did Model 3, which included only foliar N, P, and K. Forward stepwise regression thatbegan with all foliar variables with significant univariate regressions identified foliar N and P asthe most important nutrients determining site index in the reduced data set (Model 4). Thismodel explained 78% of the variation in black cottonwood site index, an had a standard error ofthe estimate of 3.07 m.Table 5.19: Correlation matrix for foliar nutrient variables with significant univariate linearregressions on black cottonwood site index in the reduced data set. Boldingindicates significant (p < 0.05) correlations.N^P^K^S P .573K .817^.687S .824^.272^.648Cu^.823^.287^.650^.804Table 5.20: Probabilities (p), coefficients of determination (R 2), and standard errors of theestimate (SEE) for multiple regressions of foliar nutrients on black cottonwood siteindex in the reduced data set (n=20).Foliar Nutrients^ p^R2^SEE(1) N,P,K,S,Cu^ .001^.795^3.28(2) N,P,K,Cu .000^.788^3.20(3) N,P,K .000^.779^3.24(4) Forward stepwise (N,P,K,S,Cu) - N,P^.000^.775^3.071255.3.8 Relationships Between Foliar and Soil NutrientsUnivariate regressions (Table 5.21) show that concentrations of P, N, K, and Mg in blackcottonwood foliage were well correlated with measures of soils contents (kg/ha) for the samenutrients. Scattergrams for all nutrients listed in Table 5.21 are shown in Figure 5.6, where best-fit lines are shown. Regressions and scattergrams are based on the reduced data used foranalysis in both the soil and foliar nutrient analyses above. Measures of soil available P, asmeasured by the new Mehlich method, accounted for 83% of the variation in foliageconcentration of black cottonwood trees on those sites, and had a standard error of the estimateof .018% dry mass. Increases in foliar P were also associated with increasing black cottonwoodsite index, as shown by the site index class labels for the sites (Figure 5.6).Table 5.21: Univariate models, probabilities (p), coefficients of determination (R 2), and standarderrors of the estimate (SEE) for regressions of foliar nutrients on soil nutrientcontent of the same nutrient (reduced data set; n=20).MODEL p R2 SEE1) Foliar P = 0.175 + .001 (Soil Available P) .000 .831 .0182) Foliar N = 1.642 + .004 (Soil Mineralizable N) .001 .497 .2823) Foliar Mg = 0.190 + .00002 (Soil Exchangeable Mg) .011 .323 .0424) Foliar K = 1.042 + .004 (Soil Exchangeable K) .018 .272 .3175) Foliar S = 0.210 + .0005 (Soil Available SO 4) .365 .051 .0356) Foliar Ca= -1.197 + .0004 (Soil Exchangeable Ca) .922 .001 .293Although the percentage of variance explained was not as high (Table 5.21), similarrelationships existed for foliar N and foliar K concentrations (Figure 5.6). The increasingconcentrations of foliar Mg were significantly correlated with measures of soil exchangeable Mgcontents, but the increase was not associated with increases in site index (Figure 5.6). Most sitesin the high site index class had relatively low foliar Mg concentrations and soil Mg contents.Regressions for foliar S and Ca on measures of soil SO4 and Ca contents were not significant,126and suggested that soil contents of these nutrients were sufficient and did not limit their uptakein the sites studied.The results of the regressions shown in Table 5.21 and illustrated in Figure 5.6 support atrend that is evident from analysis of both the soil and foliar nutrient data for the stands studied.The availability of N, P, and K in soils within the study sites were the principle nutrient factorsthat determined black cottonwood site index. Foliar S was associated with increases in siteindex, but was not considered to be causative. This conclusion is supported by the regressionsshown in Figure 5.6, where increases in foliar S are unrelated to the availability of SO4 in studyarea soils. As suggested by several authors (Dijkshoorn and van Wijk, 1967; Kelly andLambert, 1972; Turner et a/., 1977), S is present in proportion to the uptake of N, and in a ratiorequired for the synthesis of plant proteins.5.3.9 Identification of Optimal Nutrient Levels for Black CottonwoodScattergrams relating foliar concentrations of N, P, K, and Cu to site index are shown inFigure 5.7. Mean values for foliar nutrient concentrations for the high site index class (Table5.7) are shown as vertical dashed lines in Figure 5.7, and are included for reference. A distance-weighted least squares smoothing algorithm (McLain, 1974) has been used to fit a second orderpolynomial line through the data points to show the general trend of the data. For foliar N, P,and Cu the trend is for site index to increase through the low and medium site index classes, asthe concentrations of the foliar nutrients increase, and then to taper off in the high site indexclass as foliar concentrations increase. These trends can be interpreted as a 'deficiency tosufficiency, or critical levels curves (Ballard and Carter, 1986; Chapin et al., 1986; Leyton,1948; Lavender, 1970; Weetman and Wells, 1990). The trend for foliar K differs from the otherthree nutrients in that no levelling of the curve is apparent, and the relationship is more or lesslinear. This may mean that, given the concentrations of the other nutrients, the K sufficiencylevel has not been reached, and thus that higher levels of foliar K127Figure 5.6: (Overleaf) Regressions of foliar nutrient concentrations on their soil nutrientcontents. Study sites are labeled by their black cottonwood site index class(L=low; M=medium; H=high). Linear best-fit lines are shown to demonstratetrends of the different regressions.1280350.30025u_020150Soil MIneralizable N (<p/ha)60^100Soil Available P Ge4/ha)3 0.4HHLMHLL M LLL0.1 L0010 20 30 40 50 80 70Soil Available SO4 (kg/ha)15006000 1000Sol Exchangeable K (kg/he)2.0HL^LMMIkMHLLH0.5 10 10000 20000 30000 40000 50000 80000Sol Exchangeable Ca (kg/ha)Soil Exchangeable Mg (kg/he)129Figure 5.7:^(Overleaf) Scattergrams of black cottonwood site index and selected foliarnutrients. Study sites are labeled by their black cottonwood site index class(L=low; M=medium; H=high). Best fit second order polynomial lines have beendrawn using a distance-weighted least squares smoothing algorithm (McLain,^ 1974). 2.0^2.5Foliar N (X dry weight)03050.15^020^025^030Polar P (X dry weight)0.351 152520301030113,c) 25t..I 2015010AAH1--P "H11H50 1^2^3Folier K (X dry weight)54^6^8^10^12^14Folier Cu (ppm dry weight)1613013 1will result in higher black cottonwood site index. Although foliar Cu concentrations followed analmost identical trend to foliar N and P, it was difficult to attach a critical level to it, sincerequirements for Cu are normally very low (Ballard and Carter, 1986). The increase in foliar Cuconcentration may reflect higher uptake of the element, as more rapid growth occurs incottonwoods on sites well supplied with other limiting nutrients such as P, N, and K.Table 5.22 presents published critical foliar nutrient concentrations for black cottonwoodand other Populus species. Mean foliar concentrations for the high black cottonwood site indexclass (Table 5.7) are included in the table to compare foliar levels in this study. Compared to astudy of similar-aged P. deltoides in natural stands in Mississippi (White and Carter, 1970a),foliar concentrations in the fastest-growing trees in this study are higher for all nutrients exceptCa and Mg. Foliage concentrations of N and P in young hybrid poplars grown in greenhouseculture are much higher than those for older, native trees in this study. The value of 2.5% forfoliar N, as reported by Heilman (1985) is for a 6 year-old plantation of black cottonwood andwas very close to the value of 2.45 % measured in this study. For all 5 nutrients considered,foliage concentrations measured for the high site index class in this study were most similar tothose reported by Leech and Kim (1981) for plantations of hybrid poplars in Ontario. Nocritical values have been published for the other macro- and micro-nutrients measured in thisstudy. Given the paucity of other data for black cottonwood, and the relatively good correlationswith the data for other Populus species and hybrids that are available, the mean foliarconcentrations measured in the 11 stands in the high site index class (site index > 22 m/15years), were considered to be optimal foliage levels for the species, and were used in thecalculation of DRIS ratios for comparing the nutrient status of the different site associations.132Table 5.22: Published foliar nutrient critical levels (% dry mass) for Populus spp. and hybrids.FoliarNutrient 1 2 3 4 5 6ThisStudyN 2.00 2.20 3.00 2.50 3.78 2.45 2.45P 0.17 - 0.30 - 0.57 0.24 0.24K 1.30 1.40 1.20 - 2.64 1.40 1.76Ca 2.30 - - - 1.21 0.68 1.21Mg 0.18 0.20 - - 0.26 0.15 0.211. White and Carter (1970a) for P. deltoides2. van der Meiden (1960) for Populus spp.(cited in White and Carter, 1970a3. Bonner and Broadfoot (1967) for P. deltoides in greenhouse culture4. Heilman (1985) for P. trichocarpa5. Leech and Kim (1981) for P deltoides clone D38 in greenhouse culture6. Leech and Kim (1981) for P deltoides clone D38 in field plantation5.3.10 Diagnosis of Nutrient Limitations in the Site AssociationsDRIS (Beaufils, 1973; Leech and Kim, 1981) ratios for the 6 site associations sampled inthis study are presented in Table 5.23, and utilized foliar nutrient concentrations in the high siteindex class (Table 5.7) as norms for comparison among the site associations. The DRIS processTable 5.23: Comparisons of DRIS ratios in 6 site associations sampled in the study. Norms forthe establishment of the ratios are based on mean foliar concentrations from stands inthe high site index class (see Table 5.7).SITEASSOCIATION nN(%)P(%)K(%)Ca(%)Mg(%)Cu(ppm)Zn(ppm)Fe(ppm)Mn(ppm)B(ppm)S(%)Ss-Salmonberry 7 1 3 3 -3 0 0 0 -4 3 -3 0Ac-Red osier dogwood 5 -8 -10 -10 -1 17 -4 -2 4 12 17 -11Ac-Willow 6 -11 -5 -5 3 10 -13 2 8 13 13 -8Cw-Foamflower 6 -8 6 -7 6 0 -22 -1 5 34 -5 -1'Gleyed' 4 -4 -3 -41 24 49 -21 -10 8 45 -35 -10Cw-Swordfern 1 -2 5 -1 18 -7 -23 -8 -20 9 34 -8133assumes that not only are the concentrations of foliar nutrients optimal for natural stands, butalso that the relative concentrations among species in the high site index class represent acondition of nutrient balance (Shear et al., 1946; 1948).The Ss-Salmonberry s.a. came closest to representing optimal nutrient intensity andbalance for black cottonwood growth, as shown by very low DRIS indices (Table 5.23). This isto be expected because it was principally Ss-Salmonberry sites that made up the high site indexclass (see Table 2.1) and thus were used to calculate the norms.Sites in the Ac-Red osier dogwood s.a. were diagnosed as deficient in the order S > P =K > N > Cu > Zn. It was concluded from the regression of foliar S on soil S, that S was notlimiting, but was likely taken up in proportion to N (see Section 5.3.9). DRIS indices for Cuand Zn were very low and thus not considered critical. From this analysis, site index in Ac-Redosier dogwood sites appears to be limited principally by P and K, and to a lesser degree N.According to the DRIS indices given in Table 5.23, Ac-Willow sites are limited in theorder Cu > N > P = K. Foliar Cu concentrations were never below 6.8 ppm in any of the siteassociations, and Ballard and Carter (1986) suggest that, in conifers, Cu deficiencies do notoccur until at least 4 ppm. Thus for Ac-Willow sites, it is assumed that N is the most limitingnutrient, followed by P and K, which are about equally deficient.DRIS indices for the Cw-Foamflower s.a. indicate deficiency in the order Cu > N > K.Using the same reasoning for Cu deficiency level, it can be argued that N, and then K limit blackcottonwood growth in this site association.Gleyed s.a.'s demonstrated a more complex pattern of nutrient deficiency, and arediagnosed as nutrient-limited in the order K > B > Cu > Zn = S > N > P. The highly negative KDRIS index is considered to be very important in limiting black cottonwood growth on these siteassociations. The mean foliar concentration of 0.82% for the Gleyed s.a.'s was well below allcritical levels listed in Table 5.20 for this nutrient. The mean B concentration for the gleyedgroup of site associations was 13 ppm, which is diagnosed as 'possibly deficient' by Ballard and134Carter (1986) based on observations of conifers. It is possible therefore that black cottonwoodon sites within the gleyed group are also limited by low B levels. As discussed above for Cu, Znconcentrations are well above deficient levels proposed by Ballard and Carter (1986) forconifers, and are not considered to be limiting. Thus, black cottonwood growth on the gleyedsites is considered to be limited principally by K and possibly B, and only slightly by P and N.The Cw-Swordfern was represented by only one study site, which was diagnosed aslimited in the order Cu > Fe > S > Mg > N > K. As discussed above, Cu and S at the foliarconcentrations measured are not considered to be limiting. Fe and active-Fe concentrations were53 and 38 ppm respectively, neither of which was considered limiting in conifers by Ballard andCarter (1986). The foliar concentration of Mg was 0.16 % which was considerably lower thanthose of the other site associations, and was below the critical level proposed by van der Meiden(1960; cited in White and Carter, 1970a) for hybrid poplars, and by White and Carter (1970a)for P. deltoides (see Table 5.20). Based on this reasoning, this site is interpreted as having amoderate Mg deficiency, and a slight N and P deficiency.5.4 DISCUSSION5.4.1 Nutrient Availability and Site Index - General TrendsIn the study, inter-relationships among black cottonwood site index, classes of thebiogeoclimatic ecosystem classification, soil nutrient contents, and foliar nutrient concentrationswere consistent in the PCAs, ANOVAs and linear regression analyses. In the PCA of soilcontents, sites with low soil organic matter (total N, total C, and mineralizable N), high pH, highsoil contents of exchangeable Ca and Mg, and low site index were correlated with one another,and were contrasted with sites with high soil organic matter, lower pH, relatively high contentsof available P and exchangeable K, and overall higher site index. The PCA of foliar nutrientscontrasted trees with high foliar N, P, and K on sites in the high site index class, to trees withhigh Ca and Mg, on sites with low site index. In both the soil and foliar PCAs, a gradient of135increasing flooding severity was associated with the low site index group of sites, both in theupland and alluvial site associations. In the ANOVAs of black cottonwood site index classes,and linear regressions of soil and foliar nutrients on black cottonwood site index, measures ofsoil and foliar N, P, and K were consistently associated with increasing black cottonwood siteindex, while pH, and foliar soil measures of Ca and Mg had negative relationships.The high concentration of Ca, high pH, and low P contents found in this study have beenreported by Peterson and Rolfe (1982, 1985, 1986) in alluvial soils subjected to periodic annualflooding. They observed a decrease in soil P concentrations and an increase in pH and Cacontent following flooding in 2 years of measurements, and attributed the increase in Caconcentration and pH to soil reducing conditions. In that study, the pH increased above 6.5,after which solubility of P decreased rapidly, and a higher concentration of Ca resulted in theprecipitation of P as insoluble calcium phosphates (Peterson and Rolfe, 1982). In this study ithas been demonstrated that high pH and high soil Ca content were negatively correlated withsoil P content, and with site index. If soils on the frequently flooded sites sampled in this studydo become anaerobic, then the mechanism suggested by Peterson and Rolfe (1982) may beresponsible.Analysis of soil nutrient - foliar regressions, and black cottonwood site index classprovided the opportunity to identify levels of foliar nutrients that are considered to be optimalfor black cottonwood growing in unmanaged stands in coastal British Columbia. The levels aresimilar to critical levels published by other workers for Populus spp. growing in plantations ornatural stands. The mean foliar concentrations were used as DRIS norms for comparing nutrientdeficiencies among the different site associations. Interpretations of black cottonwood site indexin the context of the ecological processes operative within the various site associations aredescribed below.Nutrient limitations within the different site associations sampled were compared tothose found in the high site index class, since, under unmanaged conditions, these were observedto be growing the most rapidly. However, black cottonwood stands in the high site index class136may also be nutrient-limited. DRIS norms generated in Chapter 6 from the 25 fastest-growing,fertilized trees at the Squamish 23 site can be used to assess nutrient limitation in the high siteindex class. Trees used to develop the norms were fertilized with a balanced fertilizer whichincluded all macro- and micro-nutrients. Using DRIS, nutrient deficiencies for trees in the highsite index class were identified as B (-24) > K (-18) > P (-13) > S (-6). The role of B in limitinggrowth of black cottonwood growth is difficult to assess (Carter and Brockley, 1990), and thefoliar concentrations of trees in the high site index class were well above that required forconifers (Ballard and Carter, 1986). Also, B deficiency should be expressed in apical areas ofthe trees (Carter and Brockley, 1990), and there was no indication of B deficiency symptoms inthe fast-growing population of cottonwood studied. It was apparent from regressions of foliar Son soil S that sufficient soil S was available, but was not taken up by the trees. For this reason itis assumed that S is not limiting to black cottonwood in the stands studied. The regression offoliar K on soil K was the only relationship that was more or less linear, and it was suggestedthat increasing the uptake of K may result in growth increase for rapidly-growing blackcottonwood stands. The high negative K index from the DRIS analysis supports this conclusion.Based on the DRIS analysis, and the data presented in this study, it is concluded that, inunmanaged stands, the fastest-growing black cottonwoods are limited by the uptake of K, andthen P.5.4.2 Interpretations of Black Cottonwood Growth in the Site AssociationsA c-Willow Site Association (Low Bench)The Ac-Willow s.a. represents sites located on the lowest elevations of alluvialfloodplains, and thus are the most frequently flooded. The data collected in this study indicatesthat these site units are flooded more or less annually for 2 to 3 weeks above the surface duringthe growing season. Ac-Willow sites had considerably lower soil contents of total C, total N,mineralizable N, and available P, and higher exchangeable Ca, than all ecosystems studied.137Only the gleyed sa's had lower levels of soil exchangeable K. Comparisons of foliarconcentrations from black cottonwoods on these sites with those in the high site index classsuggested serious limitation by the availability of N, P, and K, in that order.It appears that growth of black cottonwood on sites representing the Ac-Willow sa. islimited by low availability of soil nutrients, and possible impedance of uptake of those nutrientsthat are available, due to frequent and prolonged inundation during the growing season.Frequent inundation erodes surfaces and disrupts decomposer communities and reducesmineralization of soil organic matter. Nitrogen will be leached from the soil if flooding isprolonged enough to create reducing conditions. Soil P may be less available because offlooding-related interactions with high soil pH and content of soil Ca. Low soil K may be theresult of leaching, where soils have high Ca concentrations so that K is displaced from theexchange complex. All of these factors may reduce nutrient availability and uptake and severelylimit black cottonwood growth on Ac-Willow sites.A c-Red osier dogwood Site Association (Middle Bench)Sites classified within the Ac-Red osier dogwood s.a. occur in the middle of the floodinggradient on active alluvial floodplain surfaces, and are inundated above the surface during thegrowing season about once every 5 years, for a period of about 2 weeks. Soil contents on Ac-Red osier dogwood sites reflect the intermediate flooding position in that, compared to Ac-Willow sites, they have much higher levels of organic matter (total C, total N, and mineralizableN), and exchangeable K, but have comparable soil contents of exchangeable Ca and Mg.Available P is almost double that of Ac-Willow sites, but is half that of Ss-Salmonberry sites,located on the highest areas of alluvial floodplains. According to the DRIS analysis, MiddleBench ecosystems were limited mainly by P and K, and to a lesser extent N.Black cottonwood site index on sites of the Ac-Red osier dogwood s.a. is mostly in theupper half of the medium site index class. This significant increase in productivity over Ac-138Willow sites is attributed primarily to the reduced flooding frequency and reduced physical andsoil chemical effects of flooding. The reduced flooding permits more active decomposition, andthus promotes nutrient cycling and the availability of N, and other important nutrients that havebeen correlated with black cottonwood growth in this study. The infrequent flooding that doesoccur has a relatively long duration (about 2 weeks) which may impede the uptake of nutrientsduring the warmest part of the growing season. The flooding may also limit the development ofhumus layers, and, given the high soil Ca and Mg contents and relatively low available P, mayalso limit the availability of P and K, as discussed above for Ac-Willow sites.Ss-Salmonberry Site Association (High Bench)Sites located within the Ss-Salmonberry s.a. are inundated less frequently, and for ashorter duration, than all other site units on alluvial floodplains. It is estimated that these sitesare flooded about as frequently as sites of the Ac-Red osier dogwood s.a, but for a much shorterduration. Whereas Ac-Red osier dogwood sites can be expected to flood for 2-3 weeks, Ss-Salmonberry sites are inundated above the surface, during the growing season, for several daysat the most. Compared to the Ac-Red osier dogwood, Ss-Salmonberry sites have approximatelyequal amounts of total C, total N, mineralizable N, and exchangeable K, much lower contents ofsoil Ca and Mg, and about twice the available P. Based on the DRIS analysis, the nutrient statusof this site association is optimal, and, relative to stands growing on sites representing other siteassociations, no nutrients are limiting growth.The mean site index for the Ss-Salmonberry sites was 25.4 m/15 yrs, and was higher thanall other site units. All study sites for this site association fell within the high site index class.The high productivity of these sites is attributed to reduced flooding effects, especially floodingduration during the growing season, which permits the development of a deep Mull humus thatis a dynamic centre for cycling of nutrients within the ecosystem. Because flooding is of shortduration the main effect is to recharge soil water and provide optimal conditions for nutrientuptake and black cottonwood growth. Higher availability of soil P may be related to lower139levels of soil Ca. Although soil K is higher in Ac-Red osier dogwood sites, foliar K isconsiderably higher in Ss-Salmonberry sites, and this may also be a result of shorter duration offlooding.'Gleyed' Site AssociationsThe 4 sites considered together as the 'gleyed' group are comprised of 2 sites belongingto the Cw-Black twinberry s.a., and 2 sites classified within the Cw-Salmonberry s.a. All 4 siteswere located in landscape depressions, where relatively well-drained marine sands overlaycompact, gleyed marine silt and clay at various depths. The depth to the underlying compactlayer is the basis for site differentiation, so that sites classified within the Cw-Salmonberry sa.had at least 35 cm above the gleyed horizon, and the Cw-Black twinberry between 15 and 35 cm(Banner et al., 1990). A third site association, the Cw-Slough sedge is defined where the gleyedlayer is less than 15 cm from the soil surface, and the Elk 2 site is transitional to this unit. Soilmoisture in these ecosystems fluctuates seasonally, so that in the winter there may be standingwater to various depths (wet and very wet SMRs), while in the summer, the most elevated siteunit (Cw-Salmonberry s.a.) may achieve a fresh SMR, which implies that there is no soilmoisture in excess of that required for uptake (Pojar et al., 1987). Soil contents of total C, totalN, mineralizable N and available P were comparable, but contents of exchangeable Ca, Mg, andK were lower than all other sites. The 'Gleyed' site associations were diagnosed as havingserious B and K deficiencies, with slight deficiencies of P and N.Black cottonwood site index in the gleyed group of site associations was in the low, orlow half of the middle site index class, and ranged from 12.2 m/15 years at the Oyster site to17.2 m/15 years at Elk 2. This poor growth is interpreted to be a function of reduced volumeabove compact, gleyed horizons, and to nutrient deficiencies particular to the marine soils onwhich the sites were located. The reduced rooting depth decreased the volume of soil availablefor supplying nutrients and probably impeded uptake where soils are anaerobic. Thedeficiencies of K can be related to very low soil K contents, and may be a result of the type of140mineral present for weathering in the marine soils in which all of the sites occur. Although nodata on soil B was collected, B deficiencies have been diagnosed in coastal British Columbia onsorted sandy soils (Carter and Brockley, 1990), such as those that occur over the compactdeposits in the soils sampled.Cw-Foamflower Site AssociationSites in the Cw-Foamflower s.a. had a moist to very moist soil moisture regime, whichmeans that available soil moisture ranges from soil water being in excess of that which can beutilized, to soils where a water table is present at greater than 30 cm depth (Pojar et al., 1987).As a result soil moisture is available for nutrient uptake over the entire growing season. Forblack cottonwood, an important differentiation of sites with very moist SMRs is whethersubsurface water is freely-flowing and oxygenated, as in the case of seepage sites, or whetherwater is slow-moving and anaerobic conditions develop, as evidenced by gleyed soil horizons.In this study seepage sites were sampled on alluvial fans (Tamihi Fan, Squamish 38) and wheredeep loess blankets overlie impermeable basal till (Ryder, Sumas). Gleyed sites were sampled inlevel terrain, where fine-textured glaciofluvial materials have been deposited over compactlayers at depth (Chilliwack, Pierce) so that soil drainage is impeded. Soils sampled in the Cw-Foamflower sa. had the highest total N, total C, mineralizable N, and available P contents of allof the site associations sampled. Levels of exchangeable Ca, Mg, and Ca were comparable tosites of the Ss-Salmonberry s.a. As a group, sites within this site association were diagnosed asdeficient in N and then K, using DRIS analysis.Productivity of black cottonwood in the Cw-Foamflower s.a. must be broken down in tothe seepage and gleyed types described above to assess the range of site index that this site unitencompasses. Black cottonwood site index on one of the two seepage sites sampled for thisstudy was the highest of all study stands (Ryder - 30.8 m/15 years), and the other was the fourthhighest (Sumas - 27.1 m/15 years). Both of these sites had Ah horizons in excess of 10 cm, siltyloess soils, and permanent seepage - all features which are interpreted as providing optimal141growth conditions for black cottonwood. The Tamihi fan (25.2 m/15 years) and Squamish 38(21.1 m/15 years) sites were located on alluvial fan landforms where seepage is presentthroughout the year, but soils were considerably coarser, and Ah horizons thinner. Under theseconditions mineralization processes were probably somewhat reduced (although foliar N washigh at both sites) and the surface area for cation exchange and nutrient retention considerablylower. Given the coarse soil textures and thin humus layers, the high productivity of alluvial fansites is somewhat anomalous, and requires further investigation. Two of the 6 sites in the Cw-Foamflower were gleyed within 60 cm of the surface, and thus the sites experience anaerobicconditions within the rooting zone. The Pierce site (20.4 m/15 years) had a permanent watertable at a depth of about a meter, with gleyed horizons beginning at a depth of 65 cm. TheChilliwack site (13.6 m/15 years) had a compact silty soil with pronounced mottles and gleyingthat start at 10 cm and increases with depth. The negative influence of soil gleying on blackcottonwood site index is suggested by this comparison, and by the low site index of the gleyedgroup of site associations described above.Cw-Swordfern Site AssociationOnly 1 study site was sampled in the Cw-Swordfern sa., so few general conclusions canbe drawn about the productivity of black cottonwood within the unit. On the site sampled, blackcottonwood had a site index of 16.3 m/15 years, and was located on relatively coarseglaciofluvial materials, with a Moder humus form. The site was diagnosed as having a moderateMg deficiency, and slight N and P deficiencies. The medium productivity of the species isattributed to a relatively short moisture deficit, relatively slower mineralization rate, and soilmineralogy that is somewhat deficient in content of soil Mg.1425.5 CONCLUSIONS1) Site index differed insignificantly among the 3 subzones sampled, and it was concluded thatthe limited climatic range of study sites was insufficient to significantly affect growth of blackcottonwood. Membership in site association and soil nutrient regime classes explained 87% and36% of the variation in black cottonwood site index, respectively. This showed that blackcottonwood site index was highly predictable, if the site association was known, and much lesspredictable, based on soil nutrient regime alone. Much of the poor predictive capability of soilnutrient regime can be attributed to the fact that the three soil nutrient regime classesincorporated a range of soil moisture regime classes and flooding regimes. Soil nutrient regimeexplained 88% of the variance in site index when stratified within site association, which can beused as a surrogate for soil moisture regime class. Also, it was demonstrated in the study thatsoil nutrient regime was principally a gradient of increasing N availability, and that theavailability of other important nutrients, such as P and K, did not increase along this samegradient in the stands studied. P and K were diagnosed as limiting nutrients on some sites,especially in the high site index class, and this may also help explain the poorer predictive powerof soil nutrient regime.2) In general, about 50% of the variation in site index was accounted for by the understoryvegetation growing in the sample stands. Many understory species have ecological amplitudesthat covered a range of soil moisture and/or soil nutrient regime classes, and it was observed thatblack cottonwood site index changed significantly along these ecological gradients. Forexample, salmonberry indicates a very moist to wet soil moisture regime range, which meantthat it was abundant on sites belonging to both the Ss-Salmonberry s.a., where site index washigh, and to the 'Gleyed' s.a.'s, where site index was much lower. Also, sample standsrepresented juvenile stages of forest succession following different types of disturbance, and the143presence of many 'weedy' species that reflected disturbance rather than ecological conditionsreduced the predictive capacity of the vegetation models.3) All methods of analysis revealed consistent relationships between measures of site nutrientstatus and site index. Sample stands with high pH, high levels of exchangeable Ca and Mg, andlow levels of soil N, P, and K, had foliar concentrations of N, P, and K diagnosed as limiting toblack cottonwood growth, and had the lowest site index. High site index was recorded in standswith more or less opposite soil and foliar properties.4) Site index was seen to decrease in site units with increasing flooding frequency and durationon alluvial floodplains. The decrease was attributed to the negative impact of flooding on therate of organic matter mineralization, on nutrient uptake, and on the negative effect of highlevels of soil Ca and high soil pH on the availability of soil P. On upland sites, soil gleying andprolonged rooting zone flooding during the growing season were correlated with low site index.5) Optimal foliar levels for 13 foliar nutrients based on mean foliar concentrations from the highsite index class were used as a 'field standard' (Leech and Kim, 1981) for DRIS interpretations ofblack cottonwood nutrient status. Using DRIS norms from the fastest-growing, fertilized treesin Chapter 6, it was concluded that black cottonwood stands in the high site index class arelimited by K, and then P.144CHAPTER 6GROWTH RESPONSE OF THREE BLACK COTTONWOOD STANDS TOFERTILIZATION BASED ON DRIS DIAGNOSIS6.1 INTRODUCTIONThe impressive response of many Populus species and hybrids to nutrient additions hasbeen demonstrated in North America (Aird 1962; Bowersox and Ward, 1976 a,b; Cannell andSmith, 1980; Crist and Dawson, 1975; Dawson et al., 1976; Ek and Dawson, 1976; Isebrands etal., 1983; Palmer, 1991; Switzer et al. 1976) and Europe (Anderson and Zsuffa, 1975; Cannell,1980; FAO, 1958; Kolster and van der Meiden, 1979). Much of this fertilization work has beencarried out on hybrid poplars under intensive silvicultural regimes that optimized growthconditions so that the full growth potential of the trees could be realized (Cannell and Smith,1980). In Washington, the productivity of black cottonwood and its hybrids in short-rotationintensive culture has been examined both alone (Heilman et al. 1972; Heilman and Peabody,1981; Heilman and Stettler, 1985b; Stettler et al., 1988), and in association with Alnus rubra(DeBell and Radwan, 1979; Harrington and DeBell, 1984; Heilman and Stettler, 1983; Heilmanand Stettler, 1985a; Radwan and DeBell, 1988). Overall, it has been shown that young standsand plantations of Populus species and hybrids respond to fertilization with increased growth,even in temperate climates, if nutrient balance is maintained (Leech and Kim, 1979, 1981;Schutz and deVilliers, 1986) and site conditions are optimal.Little fertilization work has been carried out in unmanaged Populus stands, althoughfertilization of unmanaged stands of other hardwood species has been carried out (Auchmoodyand Filip, 1973; Czapowskyj and Safford, 1979; Ellis and von Althen, 1973; Safford and Filip,1974; van Cleve, 1973; von Althen, 1973). In addition to examining how much growth responsecan be expected, the addition of nutrients also provides an opportunity to test hypotheses ofnutrient limitation (Chapin et al., 1986; Timmer and Ray, 1988; White and Carter, 1970a), to145establish optimal or critical foliar nutrient levels for a species (Leech and Kim, 1981), and to testthe effectiveness of several techniques of foliar diagnosis.The objective of the diagnosis of stand nutrient status is to determine what nutrients arelimiting growth, and has been based primarily on an analysis of nutrient concentrations infoliage (Ballard and Carter, 1986; Lavender, 1970; Morrison, 1974), although other planttissues, such as xylem (Barnes, 1962, 1963) or phloem (White et al., 1972; Will, 1965), havealso been used. It has been generally accepted that, of all of the alternatives (soil analysis,bioassays, analysis of different plant tissues) foliar diagnosis, combined with a knowledge of soilnutrient levels and site factors, provides the most practical approach for evaluating the nutrientstatus of forest trees (Ballard and Carter, 1986; Leaf, 1973; Morrison 1974; Weetman and Wells,1990). Foliage samples are collected according to an established protocol for the species,analyzed for concentrations of plant nutrients and compared to established critical levels for thespecies to determine the relative sufficiency of the various nutrients. The critical level for agiven nutrient is the foliar concentration above which little growth response is obtained if thesupply of the nutrient is increased (Ballard and Carter, 1986; van den Driessche, 1974). Thecritical level is often associated with a second order polynomial curve in which growth responseis linear until the critical value is reached, levels off as the requirement for that nutrient issatisfied, and then declines, because of a 'toxic' effect (Everard, 1973; Leyton, 1958; Richardsand Bevege, 1972).Simple interpretations of critical foliar nutrient levels are complicated by observationsthat foliar nutrient concentrations can be effected by climate, season, aspect, altitude, geneticvariation, competition, stress, plant part sampled, age of tissue, moisture content, position on theplant, and time of day, as reviewed by Schutz and de Villiers (1986). Although many of thesefactors can be controlled by standardizing sampling procedures and local interpretations (Ballardand Carter, 1986), additional problems, such as dilution effects and nutrient balanceconsiderations, limit the general usefulness of the critical levels approach. The critical levels146approach has been most successful when one nutrient has severely limited the growth of foreststands (Ballard and Carter, 1986; Morrison, 1974a,b).Given the problems associated with utilizing critical levels of individual nutrients, foliardiagnosis methodologies that compare ratios of nutrients are commonly used (Ballard andCarter, 1986; Schutz and de Villiers, 1986; van den Driessche, 1972; Weetman and Wells,1990). The use of nutrient ratios recognizes that nutrients required for the metabolism of planttissue must be available in the correct proportions (Ingestad, 1962; Leech and Kim, 1981; Shearet al, 1946, 1948), and thus acknowledges the importance of nutrient balance. Ballard andCarter (1986) present interpretations of important nutrient ratios for conifers of western NorthAmerica. Ingestad (1962) has used fertilization methods to develop optimal ratios betweenfoliar nitrogen and other nutrients, and has shown that these ratios are very consistent amongconifer species for macronutrient concentrations (reviewed by van den Driessche, 1974; andWeetman and Wells, 1990).A diagnostic procedure that uses the ratios of all nutrients simultaneously is theDiagnosis and Recommendation Integrated System (DRIS), originally used by Beaufils (1973),and later by many others (Beverly et al., 1984, 1986; Leech and Kim, 1979a,b, 1981; Letzschand Sumner, 1983; Sumner 1977a, 1977b, 1978, 1979). Using the DRIS method, a series ofequations, based on ratios of all pairs of nutrients measured in the analysis, are used to computeindices that compare the nutrient balance within the stand being assessed to DRIS 'norms' -nutrient concentration levels from rapidly-growing populations of the species being tested. Foragricultural crops, DRIS norms have been based on very large data sets from a wide geographicsample (Sumner, 1977b). The DRIS indices indicate both the most limiting nutrient, and theorder in which other nutrients measured are either limiting or sufficient. DRIS methodology hasbeen successfully applied to hybrid poplar plantations by Leech and Kim (1981), based onnorms developed from greenhouse experiments for the hybrids used. As pointed out byWeetman and Wells (1990), the major drawback in using DRIS for applied forestry is the lack of147appropriate norms on which to base the diagnosis, and uncertainty concerning theappropriateness of applying norms derived from greenhouse tests on seedlings to forest stands.The success of the diagnosis of stand nutrient status can be ascertained by adding thenutrients thought to be limiting and measuring the response of treated trees or stands to controls.Direct measurements of absolute growth response (height, diameter, basal area, or volume) offorest trees to fertilizer treatments can lead to incorrect conclusions about the effectiveness ofthe treatment if pretreatment size or rate of growth of test trees is not accounted for in theassessments of growth response (Auchmoody, 1985; Ballard and Majid, 1985; Gagnon, 1975;Lipas, 1979; Miller and Tarrant, 1983; Salonius et al., 1982; Whyte and Mead, 1976; Woolonsand Whyte, 1988). Salonius et al. (1982), and Ballard and Majid (1985) developed arithmeticindices to account for the pre-treatment growth rate of treated trees. Woolons and Whyte (1988)have analyzed this approach and concluded that, in the case of Salonius et al. (1982), valuableinformation was lost and improper conclusions drawn from the relatively low sensitivity of theapproach, compared to covariance analysis using pre-treatment rate of growth as a covariate.The approach taken in this study is to assess the correlations of a variety of pre-treatment sizeand growth rate variables on response variables, and, where the relationships are significant, touse the most significant measure as a covariate to adjust all response estimates (Woolons andWhyte, 1988).An assessment of differences in foliar nutrient levels between treated and control trees isoften used to help interpret observed responses to fertilizer additions. A frequent anomalyencountered is that foliar concentrations of applied nutrients are often seen to decrease in treesshowing a growth response, and this has been attributed to a 'dilution' effect, where increasedfoliar mass decreases the concentration, but not the absolute amount of the nutrient added(Ballard and Carter, 1986; Leaf, 1973; Morrison, 1974; van den Driessche, 1974; Weetman anWells, 1990). Heinsdorf (1968) developed a 3-axis graphical procedure, where response offoliar mass, and foliar nutrient concentration and content are used to interpret conditions ofluxury consumption, toxicity, and other effects. The method is based on correlations between148foliage response and bolewood production, and has been applied with success in coniferfertilization (Carter and Brockley, 1990; Timmer and Morrow, 1984; Timmer and Ray, 1988;Timmer and Stone, 1978). Timmer (1985) has applied the graphical procedure to young hybridpoplar in Ontario, based on observed correlations between foliar response and wood productionas reported by Larson and Isebrands (1972). The application of the graphical method to Populusspp. is problematic, because leaf growth is indeterminate, and thus may respond to increases innutrient status by producing more, rather than larger leaves. For this reason, Timmer (1985)used measures of total leaf mass of 1 year old saplings as the foliar mass response variable. It isnot known whether there is a close relationship between wood production and foliage responseto fertilizer treatment in juvenile black cottonwoods. The possibility of utilizing the graphicalprocedure to assess changes in foliar and tree response in juvenile stands of black cottonwoodwill be assessed in this study.The specific objectives of the study were:1) to use the DRIS method (Beaufils, 1973; Leech and Kim, 1981; Schutz and de Villiers, 1986)to diagnose potential nutrient deficiencies in rapidly-growing and poorly-growing natural standsof black cottonwood;2) to apply fertilizers based on these diagnoses, and measure response in height and basal areaincrement over a 3 year period;3) to measure foliar concentrations annually, and re-apply fertilizers based on changes innutrient ratios as assessed by the DRIS method;4) to assess and compare the magnitude of growth response;5) to test the assumptions of the graphical procedure (Timmer, 1985) by assessing relationshipsbetween wood production and measures of foliar response; and,6) utilizing the most rapidly-growing trees, to establish optimal foliar values for blackcottonwood that can act as a 'field standard' (Leech and Kim, 1981), and can be used as149preliminary DRIS norms for evaluating nutrient status of black cottonwood stands in coastalBritish Columbia.6.2 METHODS6.2.1 Site Selection and DescriptionThree young (8 to 19 years) black cottonwood stands, that were neither over- norunderstocked (Table 6.1), were selected for fertilization. The three stands included two siteswith high site index (Soowahlie, Squamish 23) and one with a low site index (Strawberry 1).Soil characteristics and other relevant information is summarized in Tables 2.1-2.3, and inTables 6.1-6.3. All sites were on alluvial floodplains, and had Regosols with relatively coarsertextures occurring with increasing depth. Soil texture was either sandy loam over loamy sand,or silt loam over sandy loam. Humus form was Mull at all three sites, and the Ah horizon wasdeeper, and had more granular structure at the Squamish and Soowahlie sites, compared to theStrawberry site (Table 2.2). Soil chemical sampling was carried out as described in Chapter 4(Table 6.3).Foliage samples were collected in the last two weeks of August in 1985, 1986, 1987, and1988 from the upper canopy of experimental trees following sampling protocol outlined inChapter 3. In 1985, 1986, and 1988, individual, 30 g foliage samples were collected from allexperimental trees. In 1987 a composited foliar sample was collected for each treatment group.Foliage concentrations of N, P, K, Ca, Mg, S, SO4-S, Cu, Zn, Fe, active Fe, Mn and B weredetermined using the procedures described in Chapter 3.S-SO4^Avail-P^Ex-Ca(kg/ha^(kg/►a) ^ (kg/ha)Ex-Mg^Ex-K(kg/ha)^(kg/ha)SoilpHSiteMM-N(kg/ha)150Table 6.1: Method of establishment, stand age in 1986, stocking, mean DBH, mean height, andsite index of 3 fertilized stands. Stand age and site index were calculated usingheight-age curves from destructive sampling in 1988. Stocking was based on prismdata from each experimental block on each site. Mean DBH and height were basedon pre-treatment measurements of all experimental trees.1986^ Mean^MeanStand Age^Stocking^DBH^Height^Site IndexSite^Stand Establishment^(years)^(stems/ha)^(cm)^(m)^(m/15 yrs)Strawberry natural - sprouts 19 650 11.7 12.1 11.8Soowahlie planted - rooted whips 8 545 14.0 13.5 23.0Squamish 23 natural - sprouts 14 750 21.8 22.2 24.4Table 6.2:^Selected site and soil properties.Site Watershed LandformSiteassociationSoilsubgroupSoiltextureStrawberrySoowahlieSquamish 23Fraser RiverChilliwack RiverSquamish Riveralluvial - low benchalluvial - high benchalluvial - high benchAc-WillowSs-SalmonberrySs-SalmonberryOrthic RegosolHumic RegosolHumic RegosolSL/LSSL/LSSiLiSLTable 6.3: Soil pH and soil nutrient contents (using a 1 m sampling depth)Strawberry 6.9 118 61 16 14,112 3,575 635Soowahlie 6.0 136 48 62 14,148 1,455 905Squamish 23 5.6 205 53 90 4,308 432 8881516.2.2 Fertilizer ExperimentsFertilizer treatments applied in March of 1986 were determined from site-specific DRISanalyses of foliar samples collected in 1985 (Table 6.4). In the absence of data for Populustrichocarpa, DRIS norms used to develop the DRIS indices in Table 6.5 were based on thoseTable 6.4: Mean foliar nutrient concentrations (% of dry mass) based on 15 samples taken inAugust 1985 from the upper third of the canopy at three experimental sites.Site^N^P^K^Mg^CaStrawberry 1.95 0.18 1.14 0.33 1.15Soowahlie 2.42 0.24 2.04 0.23 0.88Squamish 23 2.38 0.21 1.76 0.25 0.84Table 6.5: DRIS indices use to develop 1986 fertilizer prescriptions at the Soowahlie,Strawberry Island and Squamish fertilization experiments.Site N P K Mg CaStrawberry -16.1 -55.8 -37.1 26.4 81.6Soowahlie -6.2 -47.8 13.7 4.5 35.8Squamish 23 03 -57.2 -0.2 4.8 53.0presented by Leech and Kim (1981) for N, P, K, Ca, and Mg in Populus clone D-38. The normsproposed by Leech and Kim (1981) were very similar to those published for Populus deltoidesgrown under controlled conditions (Bonner and Broadfoot, 1967), and supported theirapplication to other Populus species. DRIS diagnoses suggested that phosphorus was the mostlimiting nutrient in all three stands, and that nitrogen and potassium were the next most limiting,depending on the site. The Strawberry site differed from the other two in that N, P, and K wereall determined to be about equally deficient. 1986 fertilizer prescriptions were developed so thatN, P, and K were added in the same ratios as their DRIS indices. Fertilizers applied for the 3year duration of the experiment are shown in Table 6.6.152Table 6.6: Summary of fertilizer treatments at the Soowahlie, Strawberry 1, and Squamish 23sites.Site Treatment 1986 Treatment 1987 TreatmentSoowahlie control1234567no treatment177 kg/ha P22 kg/ha N177 kg/ha P + 22 kg/ha Nbrushingbrushing + 177 kg/ha Pbrushing + 22 kg/ha Nbrushing + 177 kg/ha P + 22 kg/ha Nno treatmentno treatment200 kg/ha N200 kg/ha Nbrushingno treatment200 kg/ha N200 kg/ha NStrawberry 1 control123456789101112no treatment150 kg/ha NPK* (22N+77P+52K kg/ha)150 kg/ha NPK (22N+77P+52K kg/ha)150 kg/ha NPK (22N+77P+52K kg/ha)300 kg/ha NPK (44N+154P+104K kg/ha)300 kg/ha NPK (44N+154P+104K kg/ha)300 kg/ha NPK (44N+154P+104K kg/ha)450 kg/ha NPK (67N+231P+156K kg/ha)450 kg/ha NPK (67N+231P+156K kg/ha)450 kg/ha NPK (67N+231P+156K kg/ha)600 kg/ha NPK (89N+308P+207K kg/ha)600 kg/ha NPK (89N+308P+207K kg/ha)600 kWha NPK (89N+308P+207K kg/ha)no treatmentno treatment200 kg/ha N400 kg/ha Nno treatment200 kg/ha N400 kg/ha Nno treatment200 kg/ha N400 kg/ha Nno treatment200 kg/ha N400 kg/ha NSquamish 23 control1234no treatment75 kg/ha P150 kg/ha P225 kg/ha P300 kg/ha Pno treatment750 kg/ha 'complete fertilizer"750 kg/ha 'complete fertilizer" + 200 kg/ha N750 kg/ha 'complete fertilizer" + 400 kg/ha N750 kg/ha 'complete fertilizer" + 600 kg/ha N1 'complete fertilizer composition - 0% N, 19.6% P (triple superphosphate), 10.5% K (KCI, K-MgSO4), 0.60% Fe, 0.20% Zn, 0.26% Mn, 0.10%Cu, 0.10% B, 0.006% Mo.Stands selected for fertilization were divided into equal area sections, and a randomprocess was used to select dominant or codominant experimental trees within the blocks. AtStrawberry and Squamish 23 there were 15 blocks and 5 trees within each. At the Soowahliesite there were 10 blocks with 8 trees in each. Fertilizer treatments were surface-applied evenlyin March of 1986 and 1987 over 5 m radius circular plots surrounding each experimental tree.The fertilizer trial at the Soowahlie site employed a randomized complete block designwith a 2x2x2 factorial arrangement of 8 treatments randomly applied within 10 blocks. This153design was used to examine interactions and main effects of phosphorus and nitrogenfertilization, and to test the effects of removing competing vegetation on black cottonwoodgrowth and response to fertilizers. These plots were brushed with a mechanical brush sawmonthly, during the growing season, for the three years of the experiment. Based on analysis of1986 foliar data, an additional 200 kg/ha of N was added in March 1987 to those trees thatreceived N in 1986 (Table 6.6). The block effect was not significant during any year of analysis.At the Strawberry 1 site a randomized complete block design was employed in 1986,where five treatments were applied randomly within 15 equal-area blocks. Fertilizer treatmentswere equidistant and quantitative so the method of orthogonal polynomials could be utilized toanalyze growth response (Hicks, 1982). All 1986 responses were tested using this model, andthe block effect was not significant. The Strawberry site was flooded in May of 1986, after thefertilizer was applied, and, because there was no uptake of N at that site, it was decided to testthe effect of flooding on fertilization with N by applying fertilizer before and after the flood in1987. However, no flooding occurred in 1987, so only some of the blocks had N applied, andthis changed the design of the experiment. Application of 0, 200, and 400 kg/ha of nitrogenacross blocks (excluding controls) changed the experimental design from a one-way ANOVAwith orthogonal polynomials, to one where 12 treatment groups of 5 trees each were comparedwith 5 randomly selected control trees using orthogonal contrasts (Table 6.6).At the Squamish 23 site, a randomized complete block design was used, with 5treatments applied randomly over 15 blocks. In 1986, increasing amounts of P were applied in75 kg/ha increments from 75 to 300 kg/ha. In 1987 a 'complete fertilizer' was applied (seefootnote to Table 6.6 for composition) in combination with increasing amounts of nitrogen todetermine whether or not greater growth performance could be achieved by supplying allnecessary macro- and micronutrients. As at the other sites, t The block effect was notsignificant during any year of analysis.he block effect was not significant during any year ofanalysis.154Covariance analysis was used to estimate the effects of variance in initial tree size on theamount of response measured (Lipas, 1979; Miller and Tarrant, 1983; Woollons and Whyte,1988). At each site, regressions of measures of pretreatment rate of growth (basal areaincrement) or size (basal area, height) on response variables (1986, 1987, and 1988 basal areaincrement, 3 year height increment) showed that the 1983-1985 basal area increment was themost highly correlated with basal area treatment response for the 1987 and 1988 measurements.The basal area of the trees (1985 basal area) at the beginning of the experiment was the mostsignificant covariate for 1986 basal area increment. Generally, neither height nor basal areacovariates was significantly correlated with height response, so in most cases height incrementresponses were not corrected for pre-treatment size. For all models, covariates were tested forhomogeneity of slopes to ensure that the influence of the covariate was consistent acrosstreatment groups. Covariates were then introduced into the ANOVA model for the variablesassessed, and adjusted means and variations for each of the treatment cells were calculated.The approach taken to assess linear model violations followed the procedure outlined inChatterjee and Price (1977). For all models estimated, the assumption of homogeneity ofvariance was tested by examining plots of studentized residuals and estimated values. Wherepatterns of increasing variance were observed, transformations were used to re-estimate themodels. This was only necessary in a few cases and the results of the ANOVAs were notsignificant for those models. Normality of residuals was assessed by probability plotting of themeasured values against those expected from a normal population, and noting any deviationsfrom a linear relationship. Deviations from that expected from a normal distribution were notseen for any of the models estimated. Outliers were deleted from the model if they were morethan 2.5 studentized residuals from the mean.6.2.3 Growth Response MeasurementsThe DBH and total height of all experimental trees was determined in March 1986,before the application of fertilizers (Table 6.1). DBH and height were remeasured each year to155assess year-to-year growth response and, with foliar analysis to prescribe additional fertilizeradditions if it was considered necessary.In October of 1988 all experimental trees were destructively sampled to measure basalareal growth at breast height and 7 m, to provide access to the 1988 terminal leaders ofexperimental trees, and for another study. All measurements at 7 m were highly correlated withbreast height measures, so only the results for breast height measures are reported. 1988 leaderlength, fresh mass and diameter, number of leader leaves, and total fresh mass of leader leaveswere measured for all experimental trees felled. Foliage from a sample of at least 30 leaders ateach site was dried to constant mass and used to develop wet weight-dry weight regressions foreach of the sites. Using the foliar nutrient concentration data, the regression equations were usedto estimate the total content of foliar nutrients in the 1988 leaders of all sample trees.Three-year height increments were determined for all trees by subtracting 1988 treeheights (determined from felled trees) from pretreatment heights (estimated in March 1986 usinga clinometer and tape). Disks were removed from sampled trees at 1.2 and 7 m, kiln-dried, andsanded in preparation for measuring ring width. To account for variation in ring width,measurements of ring increment were made along a radius that was the mean of the longest andshortest radii. Basal area at breast height was calculated for the six years prior to the end of theexperiment in 1988, and thus including the three years of the experiment (1986, 1987, 1988) andthe three year period before the experiment (1983, 1984, 1985). The increment of basal area foreach year was then calculated by subtraction from the previous year and the calculated 'annualbasal area increment' was used as a growth response variable.1566.3 RESULTS6.3.1 Growth Responses and Foliar Nutrient Concentrations6.3.1.1 First Year Growth ResponseMeasurements of 1986 basal area increment showed either little change or a significantreduction for fertilized trees compared to controls at the three fertilized sites (Tables 6.7, 6.8,and 6.9). The basal area increment data shown in Tables 6.7 to 6.9 are from the destructivesampling that was carried out in 1988, after disks were removed and measured. Based on DBHmeasurements made on trees after the 1986 growing season, a mean diameter growth incrementwas calculated for all treatment groups, and these measurements were used to estimate growthresponse to 1986 treatments, and to prescribe new treatments for 1987. 1986 mean diametergrowth increments were highly correlated with 1986 basal area increments shown in Tables 6.7to 6.9.Table 6.7: 1986 basal area increment and important foliar concentrations at the Soowahlie site.1986 P S-SO4 B S NTreatment BAI (%) (ppm) (ppm) (%) (%)(cm2)No P 39.89 .224 446 32.6 .240 2.45177 kg/ha P 39.08 .259 588 36.5 .256 2.47Probability .596 .002 .010 .057 .059 .767At the Soowahlie site the effects of 22 kg/ha N or monthly brushing were not significantfor any of the foliar nutrient concentrations or for 1986 basal area increment. For this reasononly the effect of adding 177 kg/ha P is shown in Table 6.7. Although basal area increment didnot change, the foliar concentrations of P and SO 4 were significantly higher in the P-fertilized157plots. Concentrations of S (p=.059) and B (p=.057) were also higher in the P-fertilized plots,while foliar N concentrations did not change. Since P was apparently taken up by the fertilizedtrees, and this uptake did not affect growth response, it appeared that either P was not limiting,trees did not have time to respond to the treatment, or that the P added was not completelyavailable to the trees due to very slow movement of P in the soil. Based on the lack ofobservable growth response to P, only N was added in 1987 at the Soowahlie site. Given thenative level of 136 kg/ha of mineralizable N in the soil, the addition of 22 kg/ha in the treatmentwas probably not enough to affect uptake, so an extra 200 kg/ha was added to all plots thatreceived N in 1986.Table 6.8: 1986 basal area increment (BAI) and important changes in foliar concentrations atthe Strawberry Island site.Treatment1986BAI(cm2)N(%)P(%)K(%)Mg(%)Zn(ppm)Control 15.84 a 2.13 a .202 a 1.245 a .304 a 96.1 aNPK150 12.79 a 2.07 a .212 a 1.331 a .288 a,b 85.1 a,bNPK300 13.99 a 2.13 a .220 a 1.324 a .303 a 83.8 a,bNPK450 12.68 a 2.21 a .217 a 1.358 a .282 a,b 80.9 a,bNPK600 16.64 a 2.19 a .221 a 1.423 a .267 b 77.6 bProbability .090 .182 .090 .081 .016 .036Although not significant (p=0.05), 1986 basal area increments were lower than controltrees for all treatment groups except NPK600 at the Strawberry site (Table 6.8). Foliar P(p=.090) and K (p=.081) concentrations increased as fertilizer levels increased, while foliar Nconcentrations showed a slight increase in the two highest treatments. Concentrations of foliarMg and Zn decreased significantly as levels of fertilization increased. Since increases in P andK concentrations were not correlated with basal increment response, it was decided to add158different levels of N before and after flooding (as discussed in Section 6.2.1) across blocks in1987.Table 6.9: 1986 basal area increment (BAI) and important foliar concentration changes at theSquamish 23 fertilizer site.Treatment1986BAI(cm2)SO4(ppm)Cu(ppm)S(%)P(%)N(%)Control 47.04 a 290 a 9.5 a .205 a .182 a 2.46 aP 75 40.62 a,b 505 b 11.4 b .231 b .199 a 2.56 aP 150 32.50 b 360 a,b 9.9 a .217 a,b .194 a 2.44 aP 225 35.6 a,b 406 a,b 9.4 a .213 a,b .200 a 238 aP 300 36.20 a,b 498 b 9.4 a .224 b .201 a 2.41 aProbability .026 .001 .001 .005 .075 .339Compared to controls, 1986 basal area increment was lower in all fertilized plots, andsignificantly lower in the P150 treatment, at the Squamish 23 site (Table 6.9). Foliar SO 4 and Sconcentrations increased significantly with increasing amount of P fertilization. Cu foliarconcentrations were significantly higher for the P75 treatment. Foliar P increased slightly, whilefoliar N did not show any change. Thus, in response to very high additions of P fertilizer, thereis some evidence for P uptake, but there is a negative response in black cottonwood basal areaincrement. The results suggested that either P was not limiting growth, or that the uptakeproduced a negative growth response. Increasing concentrations of foliar SO 4 suggested that Pfertilization may have aggravated N supply (Turner et al., 1977), although foliar Nconcentrations do not support this interpretation (Table 6.9). To determine the effect ofproviding all trees with a balanced supply of all the macro- and micronutrients, a 'completefertilizer' (see Table 6.4 for composition) was applied in conjunction with increasing amounts ofN in 1987.1596.3.1.2 Three Year Growth ResponseOver three years, all sites showed a similar growth response trend - compared to controls,treatment group basal area response was low or negative in 1986, even or slightly positive in1987, and significantly positive in 1988 (Figures 6.1, 6.2, and 6.3). Height growth response oftreated trees over three years was generally non-significant. Treatment group means shown inFigures 6.1, 6.2, and 6.3 have been adjusted through covariate analysis where applicable.At the Soowahlie site, both 1988 basal area increment and 1988 height growth weresignificantly higher than controls for the P and N treatments (Table 6.10a). Table 6.10 comparesthe direct effects of the three treatments, based on orthogonal contrasts. Comparisons whereinteraction terms were significant are noted. 1987 basal area increments were significantlyhigher for the N treatment only, although trees receiving P fertilizer had a larger basal areaincrement than those that did not. No significant change occurred in three year height responsefor all treatments or in any variable with the brushing treatment. Foliar concentrations of P, N,S, Ca, and Zn (not shown), and foliar contents of P, were also significantly increased in the P-treated trees. Only the foliar concentration of N increased for the N treatment, and Cuconcentration was significantly reduced. For the brushing treatment, all foliar concentrationswere lower in brushed plots compared to unbrushed plots, the difference being significant for P,N, S, and SO 4. All significant interactions involved N or P fertilization with a brushingtreatment - in all cases foliar concentrations and growth response measures in fertilized plotswere lower in combination with brushing treatments.2 km,  Mi^kliftNkkO KV tSCP*‘;'s.M^• • 0000CN‘, L00 0000Cd tlf)aLi=Oa)a)1.-O0 0 0 0 0 0 0CO^LO^Tr^(1^CV^•-(than) luatnaran BOA/ IRS% L861.Sa(w0) tilmonD litiGH 886 tI'-r0 00-00 0 0CO^N.^coH( &JO) IWMueruui 138N peseg goat0000t--,00O,c2 8^CO) 8CO^14-- ^CO^10^tt^N^t-041PACU9 1440H 886l -988l(Zur) luementil u9NIR9e9 988L47,16150 3020310t 2. a a a la 4. 7 2 a d43'Treatment504030E-t20pst ,,,ne‘,E00CAScocorn806040200C014-Cm-1751401057035cool to3'Isi,E2Q1.4‘Got,'Figure 6.2: 1986-1988 basal area increments, and 1986-1988 and 1988 height growth bytreatment at the Strawberry site. Treatment groups are based on 1988 basal arearesponse (Low = Treatments 2,5,6,8; Medium = Treatments 13,4,7,12; High =Treatments 3,10,9,11).2C5.;//y)/,653g 43cococo— 2cat2 150_ 100co(160EU2W.;sip %.??.50403020100iiij80c3 50ao3020coca— 10itis607 502 40CO• 302 20cocr, 10986-1988163Although the ANOVA was not significant, a gradient of increasing 1988 basal areagrowth response to NPK fertilization was observed at the Strawberry Site (Figure 6.2). Basedon these results the 12 treatments were divided into 4 groups - a control group, and 3 responsegroups - low (treatments 2,5,6,8), medium treatments (4,7,12,13) and high (treatments3,9,10,11), depending on their position along the gradient displayed in Figure 6.2. Three of thefour treatments in the low productivity group were only fertilized in 1986, and generally hadlower levels of nitrogen fertilizer additions (see Table 6.4 for treatments). The medium and highproductivity groups included the NPK600 group from the 1986 treatment, and trees that hadreceived varying degrees of additional nitrogen fertilization in 1987. Orthogonal contrasts wereused to compare the significance of differences in growth response and foliar nutrients betweenproductivity groups and controls (Table 6.10b). The medium and high productivity groups hadsignificantly higher basal area increments in 1987 and 1988 and showed larger 3 year heightgrowth than controls. Significant increases in foliar N and SO 4 concentrations occurred in themedium group, foliar K was higher in the high productivity group, and foliar Zn concentrationand foliar SO4 contents were higher in both the medium and high productivity groups. Foliar Pconcentration increased with productivity groups but differences from controls were notsignificant.Compared to control trees, all treatments showed a significant increase in 1989 basal areaat the Squamish 23 site (Table 6.10c, Figure 6.3). Response in 1988 basal area incrementincreased up to treatment 4 (P 225 + complete fertilizer + 400 kg/ha N) after which responsedeclined for treatment 5 (P 300 + complete fertilizer + 600 kg/ha N). There were no significantdifferences over controls for 1987 basal area increment, or for 1986-1988 height increments.Although not significant, 1988 height growth showed a similar pattern to 1988 basal areaincrement response. Increases in 1988 basal area increment were not correlated with significantincrease in N or P foliar concentrations, although concentrations of both of these nutrientsincreased with increasing levels ofTable 6.10 a: Summary of changes in 1987 and 1988 basal area increment, 1988 and 1986-1988 height growth increment, 1988 foliarconcentrations, and in foliar contents or the 1988 terminal leader at the Soowahlie site.Trcilinient'1988Basal AreaIncrement(cm-) )1987Basal AreaIncrement(em2)1988HeightGrowth(cnI)1986-1988heightGrowth(cm)P(%)N(%)Foliar ConcentrationsS^Ca(%) (ppm)SO4(print)Cu(ppril)FoliarContentsP( mg)No P 43.26 50.53 1.32 5.64 0.33 2.42 0.271 0.690 788 12.1 0.23I' 177 kg/ha 53.17• 55.39 1.46• 5.83 0.39•" 2.59• 0.301• 0.767• 855i 11.2 0.29•No N 42.57 49.71 1.28 5.93 0.36 2.44 0.285 0.744 834 12.5 0.26N 222 kg/ha 53.86• 56.20• 1.50• 5.55 0.36 2.58• 0.286 0.712 810 10.7• 0.26No brushing 45.83 54.46 1.38 5.82 0.38 2.58 0.305 0.733 917 12.0 0.2613rushed monthly 50.60 51.44 1.39 5.66 0.34° 2.44• 0.267•• 0.723 727• 11.4 0.26Interactions N•lirP•13r P•lir P•BrN•P•lirsignificance is denoted as •, • •, and • •• to indicate significance between 2 treatments at p < 0.05, p < 0.01, and p < 0.001, respectivelyTable 6.10 b: Summary of changes in 1987 and 1988 basal area increment, 1988 and 1986-1988 height growth increment,1988 foliar concentrations, and in foliar contents of the 1988 terminal leader at the Strawberry site.Foliar ConcentrationsFoliarContents1988 1987 1988 1986-1988Basal Area Basal Area lleighl Height P N^K Zn SO4 SO4Treatment' Increment Increment Growth Growth (%) (%) (%) (PP110 (ppm) (mg)(cm2) (cm2) (cm) (cin)Control 14.82 16.02 1.20 3.05 0.194 1.53^1.58 68 512 25.98Low Productivity 18.51 19.81 1.04 3.50 0.210 1.61 1.72 66 419 17.28Medium Productivity 23.34• 26.56• 1.22 3.86• 0.212 1.82•^1.65 48• 327• 15.45•IlighProduciivity 26.11••• 29.71• 1.19 3.97• (1.215 1.68 1.80• 51• 364 15.28•' significance is denoted as •, • •, and • •• 10 indicate significance between treatment and control al p < 0.05, p < 0.01, and p < .001, respectivelyTable 6.10 c: Summary of changes in 1987 and 1988 basal area increment, 1988 and 1986-1988 height growth increment,1988 foliar concentrations, and in fo liar contents of the 1988 terminal leader at the Squamish site.1988Basal Area1987Basal Area1988Height1986-1988HeightFoliar Concentrations Foliar ContentsIncrement Increment Growth Growth P N Mn I) K Mn B SO4Treatment' (c1112)2 (cm2) (cm) (cut) (%) (%) (ppm) (ppm) (mg) (mg) (nig)  (mg)Control 27.86a 45.27a 1.37a 4.19a 0.284a 2.39a 18.1a 18.9a 1122a 0.885a 0.944a 28.80aI'75+9(' 42.14b 44.42a 1.37a 4.19a 0.308a 2.48a 23.1a 36.8b 958a 0.924a 1.381ab 27.34a1'15tJt'lT+N20N) 41.87b 43.87a 1.45a -I.18a 0.315a 2.49a 27.16 43.7b 1090ab 1.117a 1.983a8 36.40a1'225 t 'T1.' t N400 45.836 45.93a 1.53a 4.37a 0.294a 2.45a 24.0ab 44.3b 14176 1.221ab 2.135c 37.57ab1'300+1F+N600 40.77b 43.57a 1.50a 4.29a 0.322a 2.50a 35.6c 42.2b 1372b 1.634b 2.189c 44.32bSignificance' •• • 0.98 0.12 0.66 0.65 (1.84 •• • •• • • • •• • •••' codes for the treatments are as in Table 6.6significance is denoted as •, ••, and • •• to indicate significance at p <0.05, p <0.01, and p <0.001 respectivelyfigures followed by the same letter are not significantly different at p=0.05 using Tukey's test166fertilization. Significant increases occurred in foliar B and Mn concentrations and contents andfor foliar K and SO4 contents (Table 6.10c).6.3.2 Changes in DRIS RatiosDRIS indexes in Table 6.11 are based on the same Leech and Kim (1981) greenhousestandards used to make the original nutrient diagnoses, and were used to demonstrate theresponse of foliar nutrients to fertilizer additions for the most responsive treatments at each site.Changes in basal area increment in Table 6.11 are expressed relative to the control basal areaincrement. Addition of 225 kg/ha of P at the Squamish 23 site in 1986 did not change the Pindex, and had only a slight effect on N and K indexes. In 1987, a complete fertilizer (thatincluded more P and K) and 400 kg/ha of N was added, and this had an immediate effect on theN index, with little change in the others. There was little response in basal area to these changesin 1986 and 1987. Although no fertilizer was added in 1988, DRIS indexes for all 3 nutrientsshow large changes with N becoming negative, P changing from -105 to -43 and K changingfrom -13 to +72. Basal area increment in this treatment was 1.5 times that of the control in1988.Addition of 600 kg/ha NPK in 1986 at the Strawberry site increased DRIS indexes for all3 nutrients, although basal area increment was slightly lower than controls (Table 6.12). Inresponse to the addition of 400 kg/ha N in 1987, the N index changed from -35 to +8, whileindexes for P and K decreased, and basal area showed a 2-fold increase over controls. In 1988,with no fertilizer additions, the N index decreased, DRIS indexes for P and K increased, andbasal area increment was about double that for controls.Addition of P and N fertilizer at the Soowahlie site in 1986 raised both the N and Pindexes, but decreased the K index, while basal area increment was about the sameTable 6.11: Changes in N, P, and K DRIS ratios, and basal area increment relative to thecontrol, for the most growth-responsive treatment groups at the Squamish, Soowahlieand Strawberry sites.Site^Year^N^P^K^BAI^TreatmentSquamish 23^1985^-10^-103^4^na^pre-treatment1986^-2^-103^-5 0.75^225 kg/ha P1987^26^-105^-13^1.01^'CF + 400 kg/ha N1988^-12^-43^72 1.51^no treatmentStrawberry^1985^-67^-142^-106^na^pre-treatment1986^-35^-88^-38 0.95^600 kg/ha NPK1987^8^-124^-88^2.17^400 kg/ha N1988^-60^-70^31 2.05^no treatmentSoowahlie^1985^-13^-79^25^na^pre-treatment1986^-9^-52^-5 1.01^177 kg/ha P+22 kg/ha N1987^38^-83^-51^1.39^400 kg/ha N1988^-10^-10^9 2.81^no treatment167168as controls (Table 6.11). Addition of 400 kg/ha N in 1987 increased the N index, decreased theP and K index, and may have caused a slight increase in basal area increment over controls.With no further additions of fertilizer in 1988, the N index decreased, K and P indexesincreased, and basal area increment was almost 3 times that of the control group.A consistent trend that is evident at all 3 sites is the immediate response of N indexes tothe addition of N fertilizer, and the relatively slow or small response of P indexes to additions ofsimilar levels of P fertilizer. The change from a positive N index in 1987 following addition of400 kg/ha of N fertilizer, to a negative N index at all 3 sites in 1988 when no fertilizers wereadded, suggests that the effect of the N fertilization, as expressed by the N index, was short-lived. Over the 3 year period of the experiment, P indexes demonstrated an overall increase andsuggest a gradual decrease in P deficiency. This trend paralleled the trend in basal arearesponse. The lower P index in 1988 may have been partially brought about by the increase in Pconcentration for all trees in 1988 (see Section 3.3.4). However, P concentrations in treatedtrees were significantly higher than in controls, which suggests higher uptake of applied P intreated trees.The low solubility and slow movement of P fertilizer in the soil system may explain therelatively slow foliar response to P additions at the 3 sites. Figure 6.4 compares changes in themean concentrations of Mehlich-available P with depth in 15 control plots, and 15 plots thatreceived 300 kg/ha P in 1986, and additional P in the 'complete fertilizer' applied in 1987.Samples were collected in September, 1988. The steep P concentration gradient that occurredover a short soil depth increment, 2.5 years after the P was applied, illustrates the very slowmovement and strong fixation of P in the soil at the Squamish site.^ CONTROLO FEAT1696.3.3 Growth Response of the 1988 Terminal LeadersMeasurements of the mean number of leaves, mean leaf fresh mass, and mean total leaffresh mass in the 1988 leaders of experimental trees demonstrated few significant changesbetween controls and treatments, and did not follow patterns in basal0-3-goCcc 3-6.c0O(i)6-90^100^200Menlich-extractacle P (ppm)Figure 6.4:^Comparison of mean P concentrations (n=15) in the upper 10 cm of the soilprofile in fertilized and unfertilized plots at the Squamish 23 site. Lines indicate95% confidence intervals for P means.area increment for the same treatment groups (Figure 6.5). At the Squamish site the meannumber of leaves was relatively constant, while mean and total leaf fresh mass decreased belowcontrols at low levels of fertilization, and then increased to those of control trees at higherfertilization levels. The Soowahlie and Strawberry sites had a similar pattern, with mean leafnumber slightly decreased, mean leaf fresh mass varied among treatments, and total leaf freshmass decreased compared to controls (Figure 6.5).17025 22520 180135 HILI 9°5 45Treatrnent Treatment2527620 H220 H1 55 165 H10 H5H55 H0012 3251020 0Treatment Treatment Treatment0cc#61 6'1.403 A410) .seci)vi700.1P2261300.-cfGoo6 ,ror,..gro''BaØ'r°Control Low MedanProducevIty Clroup6040,01,`,Psoo's ,ergi"cor‘V6/5.1 .0203 ,040°,0°te2:2-5otIP v;300rIPTreatmentContra Low MedianProdx9rIly260 H195ILj 130 H-a650coeV-16*1 .02°° 444°C) 1 5°°P egi0-f .15.1Pcco6001',,scro' ol"Contra Low Mezlikrn I-110ProckrctIvey GroupFigure 6.5:^Comparisons of mean number of leaves, mean leaf fresh mass, and mean total leaffresh mass of the 1988 terminal leaders at the Squamish 23 (top), Strawberry(middle), and Soowahlie (bottom) sites.1716.3.4 Determination of Optimal Foliar Levels1988 basal area response at the Squamish site (Figure 6.3, Table 6.10c), increased from27 cm2 to a high of 45 cm 2 in the P225+CF+N400 treatment, after which basal area responsedecreased at the higher rate of nutrient addition. This response pattern could be interpreted asrepresenting a 'deficiency to sufficiency' response curve (Everard, 1973; Leyton, 1958), wheregrowth response increased until nutrient limitations were overcome. Such a response might beexpected if all of the nutrient requirements of the trees were being met. It should be noted forthis data that treatment responses were not significantly different from each other and suggestedthat addition of 75 kg/ha P and the complete fertilizer resulted in a similar level of response asmuch higher levels of P and N additions. Table 6.12 lists means and variance statistics for 1988foliar concentrations of macro- and micronutrients of the 25 most rapidly-growing trees underthe complete fertilizer treatments at the Squamish site. If it is assumed that, by 1988, fertilizedblack cottonwoods at the Squamish site were supplied with a complete and balanced nutrientsupply, then their foliar nutrient concentration ratios can be utilized to develop foliar norms forDRIS analysis.6.4 DISCUSSIONAlthough pronounced responses in basal area and height growth were observed at all 3sites, the relationships between these responses, the fertilizers added, changes in foliar nutrientconcentrations, and the responses of foliar mass and nutrient contents of the 1988 terminalleader, were less clear. Analysis of year to year foliar response and changes in nutrient balanceusing DRIS norms showed that N concentrations responded directly to additions of N fertilizer,K response was also fairly rapid, but that P response was either minimal or very slow. The slowresponse of the trees to additions of surface-applied P was probably a function of the slow172movement and P-fixing potential of the soils. This has been discussed and reviewed by manyworkers (Ballard 1980; Bengston, 1968; Brendemuenl, 1968; Cole et al., 1974; Russell, 1974).Analysis of soil P concentrations over very small depth increments in treatment plots thatreceived large amounts of surface-applied P as super triple phosphate almost 3 years earliershowed that the added P was only very slowly incorporated into subsoils.Table 6.12: Means, standard deviations, and coefficients of variation (CV) for foliar nutrientconcentrations in 25 black cottonwood trees with the highest 1988 basal areaincrement at the Squamish 23 site.N(%)P(%)MACRONUTRIENTSK^Ca(%)^(%)Mg(%)S(%)Mean 2.50 0.33 2.70 0.56 0.23 0.32Standard deviation 0.30 0.08 0.55 0.16 0.05 0.05CV 0.12 0.24 0.20 0.29 0.22 0.16MICRONUTRIENTSCu^Zn^Fe^Mn^B^SO4^active Fe(PPrn)^(PPm)^(PPin)^(PPIn)^(PM)^(PPm)^(PPm)Mean 17 85 79 26 44 850 67Standard deviation 2 22 14 9 11 257 11CV 0.12 0.26 0.18 0.35 0.25 0.30 0.16The factorial arrangement of treatments in the experiment at the Soowahlie site permittedan evaluation of the effect of adding P fertilizer alone on growth response and foliarconcentration. Both the P and N treatments resulted in a significant increase in 1988 basal arearesponse and height growth and in foliar concentration of the nutrient applied. However, theincrease in P concentration with application of P fertilizer was paralleled by significant increases173in the foliar concentrations of N, S, Ca, and in P content. By comparison, the addition of Nresulted in a significant increase in N only. These results can be interpreted as indicating a Pdeficiency that has been alleviated, because, when the concentration of foliar P was increased,the foliar concentrations of other available nutrients were also increased to maintain a state ofnutrient balance.The consistently lower growth and foliar nutrient response to the brushing treatment atthe Soowahlie site was unexpected, and largely unexplained by the information collected. It wasexpected that, by removing competition from understory vegetation, surface applied fertilizerswould be more available for uptake by the test trees, and this would result in an increasedgrowth response that would provide some estimate of the importance of competition on uptakeof applied nutrients. One explanation may be that repeated traffic at the base of sample treescompacted the upper soil horizons and altered soil structure enough to reduce nutrient uptake.Larson and Isebrands (1972) showed good correlations between total shoot leaf mass andwood production in young hybrid poplars, and Timmer (1985) used the graphical method tointerpret the effects of pH on nutrient availability in a hybrid poplar nursery. The results of thepresent study indicate that the responses of the foliage mass of the terminal leader were poorlycorrelated with wood production, as expressed in basal area increment or height growth. Thesefindings suggest that the response of juvenile trees to nutrient additions is much more complexthan in first or second year hybrid poplar saplings, and that the utilization of the graphicalprocedure to interpret growth response to fertilization is of limited value for black cottonwoodtrees of this age.Ballard (1978) showed that an application of 224 kg/ha P to the first rotation of P.radiata was still measureable 20 years later in the second rotation. Ballard (1980) cited anumber of studies carried out on P-deficient conifer plantations where responses to Pfertilization lasted for 15-20 years. Ballard (1980) attributed the long response to P fertilizationto the fact that nutrients were generally applied in excess of requirements, and to the ability oftrees to recycle nutrients. Given the significant increases in basal area increment in response to174as low as 75 kg/ha of P with a complete fertilizer, and the potential for the effect to be long-lived, the operational fertilization of juvenile black cottonwood stand may be economicallyjustified.6.5 CONCLUSIONS1. In three juvenile black cottonwood stands, the application of fertilizer based on diagnosis offoliar nutrient concentration using DRIS norms established for greenhouse-grown hybridpoplars, resulted in little growth response in the first year, and considerable growth response inthe third year following fertilization.2. Compared to controls in the highest 3 year treatment response group, basal area incrementincreased by 65%, and height growth increment by 15% at the Squamish 23 site; basal areaincrement increased by 65% and height growth increment by 30% at the Strawberry site; andbasal area increment increased by 27% without a significant height growth response at theSoowahlie site.3. Although it is not clear from the observations of growth and foliar response, there is someevidence to suggest that the relatively slow response to P fertilization at the Squamish 23 andSoowahlie sites was due to high rates of fixation and very slow movement of soil surface-applied P.4. Given that relatively low dosages (ca. 100 kg/ha) of P were required to achieve a significantgrowth response, and acknowledging that in many forest fertilization programs response to Pfertilization occurs for a considerable period of time, the results suggest that the fertilization offast-growing, juvenile black cottonwood stands in coastal British Columbia may beeconomically justified.1755. Significant correlations between measures of foliar response and wood production were notseen in the study, and this finding limits the usefulness of the Heinsdorf (1968) graphicalprocedure for interpretation of the experimental results.6. DRIS norms for the 25 fastest-growing black cottonwoods at the Squamish 23 site arepresented, and are based on the idea that the trees used for the norms were supplied with allrequired macro- and micronutrients.176CHAPTER 7SUMMARY AND DISCUSSIONObservations from this study support the previously published conclusion (Smith, 1957;DeBell, 1990, Roe, 1958) that deep, loamy soils, with high nutrient status, circum-neutral pH,and which are abundantly supplied with well-oxygenated soil moisture over the entire growingseason, are optimal for black cottonwood growth. These soil requirements are very similar tothose reported for eastern cottonwood (P. deltoides) in the southern United States (Baker andBroadfoot, 1979; Demeritt, 1990). The ANOVA comparing black cottonwood growth withinsite units was highly significant (p < .001), and explained 87% of the variance in site indexwithin the 29 study sites. This general result suggested that, relative to the ecologicalrequirements of black cottonwood, the site classification provided an ecologically-meaningfuldifferentiation of the edatopic gradients sampled. For operational purposes, this result predictsthat black cottonwood site index can be estimated with considerable accuracy by identifying thesite unit on which a stand is located. Growth was best on the high bench of alluvial fioodplains(Ss-Salmonberry s.a.), and on moist upland sites with seepage (Cw-Foamflower s.a.). Growthwas poorest on the low bench of alluvial floodplains (Ac-Willow), and on gleyed, marine siteunits (Cw-Salmonberry, Cw-Black twinberry).A general objective of the study was to examine the nature of nutrient limitation inunmanaged black cottonwood stands in south coastal British Columbia. Nutrient availabilityand uptake is interwoven very closely with the availability and characteristics of the soilmoisture, and it is often very difficult to isolate either factor (Cole et al., 1990). In coastalBritish Columbia, black cottonwood appears to be more or less restricted to those sites withoutseasonal drought, and all study sites except one were assessed as having no water deficit duringthe growing season. Differences in soil moisture regime among the site units was due mostly to177the behaviour of soil moisture, i.e., to the nature of flooding, or degree of aeration or gleying inthe soil. Site association explained a slightly higher percentage of the variation in blackcottonwood site index when only rich sites were included (R2=.88), and a lower percentage foronly very rich sites (R2=.74). Because soil nutrient regime was constant for these two models,the main effect was that of soil moisture regime on black cottonwood site index.Although soil moisture regime classes were more highly correlated with blackcottonwood site index, many of the mechanisms through which soil moisture regime effectsblack cottonwood site index are related to soil nutrient regime. For example, one of the majoreffects of flooding on alluvial floodplains is to decrease soil oxygen levels, and thus impedenutrient uptake by trees (Kozlowski, 1982; Greenwood, 1969; Epstein, 1972). For most treespecies, inundation of soil for a few weeks or more during the growing season reduces treegrowth (Kozlowski, 1982). Regehr et al. (1975) showed an immediate reduction inphotosynthesis following rooting-zone flooding of P. deltoides, and, after 28 days,photosynthesis was reduced by 50%. The rate at which soil oxygen is depleted will depend onthe activity of microorganisms, soil characteristics, and the nature of flooding, and in many soilsmicro-organisms consume much of the soil oxygen within a few hours of inundation(Ponnamperuma, 1972). Prolonged flooding and anaerobic conditions will also result in a broadrange of changes in the soil chemical status of many soil nutrients, and in the activities of thedecomposer community. For example, even at relatively low redox levels, nitrification isreduced (Kramer, 1979), and much of the soil nitrate can be denitrifed and leached from the soil(Scott-Russell, 1977; Kozlowski, 1982). Peterson and Rolfe (1982) showed an increase in pHand in Ca concentrations, and a decrease in P availability following seasonal inundation in abroad-leaved floodplain ecosystem in Illinois. Thus, although soil moisture regime was bestcorrelated with black cottonwood site index in the study, it is suggested here that soil moistureregime influences black cottonwood growth primarily through its influence on the availabilityand uptake of soil nutrients.178Measurements of soil nutrient contents attempt to provide estimates of the amount of thenutrients in readily-available form present in the soil at the time of sampling. In spite of theconsiderable variability measured in this study, the overall results of this study support theapplicability of the soil analytical tests used to determine the availability of soil nutrients(Waring and Bremner, 1964; Curran, 1984; Klinka et al., 1980). Foliar P was especially wellcorrelated with the availability of soil P as determined by Mehlich (1978). The estimation ofsoil mineralizable N using the anaerobic procedure developed by Waring and Bremner (1964)resulted in good correlations with black cottonwood growth, but it is worth noting that thismethod measures only the ammonium component of available N (Binkley and Hart, 1989).Black cottonwood ecosystems are characterized by Mull humus where mineralization is veryrapid. In Mull humus nitrates often provide an important component of soil N availability(Bobcock and Gilbert, 1957; Aber and Melillo, 1980; Melillo and Aber, 1982; Flanagan and vanCleve, 1983). Estimates of soil nitrate may have provided a more relevant estimate of soilavailable N for black cottonwood in the soils studied.The interpretation of foliar nutrient concentrations using DRIS (Beaufils, 1973)methodology provided a useful tool for evaluating and comparing stand nutrient status, usingnorms derived from published greenhouse standards for hybrid poplar (Leech and Kim, 1981),and from the fertilizer study. These analyses were possible in spite of the considerable temporaland spatial variability in foliar nutrient concentrations shown for black cottonwood.Using foliar concentrations from the most rapidly-growing trees in the fertilizerexperiment as norms, P and K were determined to be limiting to black cottonwood growth in thefastest-growing, unmanaged stands. One reason for this limitation in fast-growing trees may bethe manner in which the nutrients are available in the soil solution, and the way they are takenup by the tree. P, for example, is present in very low concentrations in the soil solution and is inrapid equilibrium with P absorbed on soil surfaces (Russell, 1974). Whereas nitrates, sulphatesand calcium ions move to the root by mass flow, the majority of K and P ions travel by diffusionalong concentration gradients (Barber, 1977). It is possible that, because P and K travel to the179root primarily by diffusion, they may limit growth because they cannot move to the root rapidlyenough meet the nutrient requirements of the trees during peak growth periods.The addition of N, P, and K resulted in pronounced basal area and some height responsesin the third year of the fertilizer experiment, but the growth response was not easily correlatedwith the nutrients added. Much of this uncertainty may be due to the complexity of internalnutrient cycling, and to the way in which black cottonwood responds to higher levels of growth-limiting nutrients. Some evidence was used to suggest that the growth response observed was aresult of uptake of P, and that the relatively slow growth reaction was the result of P fixation andslow movement within the soil of surface-applied P fertilizer. The behaviour of nutrients oncethey entered the black cottonwood trees is a large unknown identified by this study, and a largeportion of tree to tree and within tree variability in foliar nutrient concentrations has beenattributed to this source. Attiwill (1986) identified physiological studies as an area whereconsiderable research needs to be conducted to understand the growth of forest trees in responseto nutrient availability, and the results of this study confirm his statement. The translocation ofnutrients within cottonwood, storage and overwintering, and the physiological role of internally-fixed atmospheric N as demonstrated by van der Kamp (1979) are all examples of areas wheredetailed information is required to more scientifically interpret the ecological factors thatdetermine black cottonwood productivity.180CHAPTER 8REFERENCESAber, J.D., and J.M. Mellilo. 1982. Nitrogen immobilization in decaying hardwood leaf litter asa function of initial nitrogen and lignin content. Can. J. Bot. 60:2263-2269.Agriculture Canada Expert Committee on Soil Survey. 1987. The Canadian system of soilclassification. 2nd Ed. Agric. Canada Public. 1646. Supply and Services Canada,Ottawa, Ontario. 164 pp.Aird, P.L. 1962. Fertilization, weed control, and the growth of poplar. For. Sci. 8(4): 413-428.Anderson, H.W., and L. Zsuffa. 1975. The yield and wood quality of Euroamerican poplar. Int.Poplar Comm., FAO, 15th Session, Rome, Italy. FO/CIP/75.Anonymous. 1966. Technicron Autoanalyzer II.^Methodology: individual/simultaneousdetermination of nitrogen and/or phosphorus in BD acid digests. Industrial MethodNo. 329-74W/A.Anonymous. 1974. Handbook on Reference Methods for Soil Testing. The Council for SoilTesting and Plant Analysis. Athens, GA. 101 pp.Arnborg, T. 1960. Can we find a common platform for the different schools of forest typeclassification? In Symposium on Forest Types and Forest Ecosystems, Silva Fennica105:13-15.Attiwill, P.M. 1986. Interactions between carbon and nutrients in the forest ecosystem. TreePhysiol. 2: 389-399.Auchmoody, L.R. 1985. Evaluating growth responses to fertilization. Can. J. For. Res. 15: 877-880.Auchmoody, L.R., and S.M. Filip. 1973. Forest fertilization in the eastern United States:Hardwoods. Gen. Tech. Rep. NE-3, U.S. Dept. of Agric., Forest Service, NortheasternForest Experiment Station, Broomall, PA. pp 211-222.Baker, J.B., and W. K. Randall. 1975. Foliar nitrogen and potassium variation in cottonwood asaffected by genetic and site factors. Proc. Ninth Cent. States For. Tree Improv. Conf.1974, pp. 106-111.Baker, J.B. and W.M. Broadfoot. 1976. Soil requirements for site selection for Aigeiros poplarplantations. Pp. 225-238 in, B. A. Thielges and S.B. Lambs (eds.), Proceedings of aConference on Eastern Cottonwood and Related Species. Louisiana State University,Baton Rouge, LA.181Baker, J.B. and W.M. Broadfoot. 1979. A practical field method of site evaluation forcommercially-important southern hardwoods. Gen. Tech. Rep. SO-26. U.S.D.A.For. Serv., South. For. Res. Stn., New Orleans, LA. 51 pp.Ball, D.F., and W.M. Williams. 1968. Variability of soil chemical properties in twouncultivated Brown Earths. J. Soil. Sci. 15 :84-92.Ball, D.F., and W.M. Williams. 1971. Further studies on variability of soil chemical properties:Efficiency of sampling programmes on an uncultivated brown earth. J. Soil. Sci.22:60-68.Ballard, R. 1978. Effect of first rotation phosphorus applications on fertilizer requirements ofsecond rotation radiata pine. N.Z. J. For.Sci. 8:135-145.Ballard, R. 1980. The Phosphorus Nutrition and Fertilization of Trees. Pp. 763-804, in R.C.Dinauer (ed.). The Role of Phosphorus in Agriculture. Proc. of a Symposium, June 1-3,1976, Muscle Shoals, AL. Soil Sci. Soc. Amer., Madison. WI.Ballard, T.M. 1981. Foliar Analysis Research. Fac. Forestry, Univ. of B.C. 132 pp.Ballard, T.M. 1985. Foliar analysis for deficiency diagnosis and fertilization prescription inforestry: Variability considerations. Proceedings of the 9th British Columbia SoilScience Workshop, Vancouver, British Columbia. B.C. Ministy of EnvironmentTech. Rep. 16:142-153.Ballard, T.M., and R. E. Carter. 1986. Evaluating forest stand nutrient status. Land Manage.Rep. No. 20, B.C. Min. For. and Lands, Prov. of British Columbia. Victoria, B.C. 60pp.Ballard, T.M. and N. Majid. 1985. Use of pretreatment increment data in evaluating tree growthresponse to fertilization. Can. J. For. Res. 15: 18-22.Banner, A., R.N. Green, K. Klinka, D.S. McLennan, D.V. Meidinger, F.C. Nuszdorfer, and J.Pojar. 1990. Site classification for coastal British Columbia: a first approximation.B.C. Min. For., Victoria, B.C. 2 pp. (coloured pamphlet)Barber, S.A. 1977. Nutrients in the soil and their flow to plant roots. Pp. 161-169 in J.K.Marshall (ed.). The belowground ecosystem: A synthesis of plant-associatedprocesses. Range Science Dept., Science Series No. 26. Colorado State University,Fort Collins, Colorado.Bardsley, C.E., and J.D. Lancaster. 1965. Sulfur. Pgs. 1102-1116 in C.A. Black, D.D. Evans, J.L. White, L.E. Ensminger, and F.E. Clark (eds.). Methods of Soil Analysis.Agronomy No. 9, Amer. Soc. Agron., Madison, WI.Barnes, R.L. 1962. Glutamine synthesis and translocation in pine. Plant Physiol. 37:232-326.Barnes, R.L. 1963. Organic nitrogen compounds in tree xylem sap. For. Sci. 9:98-102.182Beaufils, E.R. 1973. Diagnosis and recommendation integrated system (DRIS). Soil Sci. Bull.No. 1, Dept. of Soil Sci. and Agro-Meterology. Univ. of Natal, Pietermaritzburg,S.Afr. 132 pp.Beckett, P.H.T., and R. Webster. 1971. Soil variability: a review. Soils Fertilizers 34:1-15.Bengston, G.W. 1968. Placement Influences the Effectiveness of Phosphates for Pine Seedlings.Pp. 51-62, in C.T. Youngberg and C.B. Davey (eds.). Tree Growth and Forest Soils.Proc., Third North American Soils Conference, Raleigh, NC. Oregon State UniversityPress, Corvallis, OR.Beverly, R.B., J.C. Stark, J.C. Ojala and T.W. Embleton. 1984. Nutrient diagnosis of 'Valencia'oranges by DRIS. J. Amer. Hort. Soc. 109 (5): 649-654.Beverly, R.B., M.E. Sumner, W.S. Letsch and C. 0. Plank. 1986. Foliar diagnosis of soybean byDRIS. Comm. in Soil Sci. Plant Anal. 17 (3): 237-256.Bickelhaupt, D.H., R. Lea, D.T. Tarbet, and A.L. Leaf. 1979. Seasonal weather regimesinfluence interpretation of Pinus resinosa foliar analysis. Soil Sci. Soc. Amer. J.43: 417-420.Binkley, D., and S.C. Hart. 1989. The components of nitrogen availability assessments in forestsoils. Advances in Soils Science 10: 57-112.Blackmon, B.G., and E.H. White. 1972. Nitrogen fertilization increases cottonwood growth onold-field soil. U.S. Dep. Agric. For. Serv. Res. Note SO-143. 5 pp.Blackmon, B.G., J.B. Baker, and D.T. Cooper. 1979. Nutrient use by three geographic sourcesof eastern cottonwood. Can. J. For. Res. 9:532-534.Blyth, J.E. and D.C. MacLeod. 1978. The significance of soil variability for forest soil studiesin northwest Scotland. J. Soil Sci. 29:419-430.Bobcock, K.L., and O.J.W. Gilbert. 1957. The disappearance of leaf litter under differentwoodland conditions. Plant and Soil IX (2):179-185.Boishot, P., M. Coppenet„ and J. Hebert. 1950. The fixation of phosphoric acid on calciumcarbonate in soils. Plant and Soil 2:311-322.Bonner, F.T., and W.M. Broadfoot. 1967. Growth response of eastern cottonwood to nutrientsin sand culture. Res. Note SO-65, U.S. Department of Agriculture, Forest Service,South. For. Res. Stn. New Orleans, LA. 4 p.Bowen, G.D. 1964. Tree roots and the use of soil nutrients. Pp 147-197 in G.D. Bowen andE.K.S. Nambiar (eds.). Nutrition of Plantation Forests. Academin Press, Orlando,FA.Bowersox, T.W., and W.W. Ward. 1976a. Growth and yield of close-spaced young hybridpoplars. Forest Sci. 22:449-454.183Bowersox, T.W., and W.W. Ward. 1976b. Economic analysis of short rotation fibre productionsystem for hybrid poplar. J. For. 74:750-753.Bremner, J., and M.A. Tabatabai. 1971. Use of automated combustion techniques for totalcarbon, total nitrogen, and total sulfur analysis of soils. Pgs. 1-16 in L.M. Walsh(ed.). Instrumental methods for analysis of soils and plant tissue. Soil Sci. Soc.Amer., Madison, WI.Bremner, J., and C.S. Mulvaney. 1982. Nitrogen - Total. Pgs. 595-624 in A. L. Page. (ed.).Methods of soil analysis. Agronomy No. 9 , Part 2. A. Soc. Agron. and Soil Sci.Soc. Amer., Madison, WI.Brendemuenl, R.H. 1968. The phosphorus placement problem in forest fertilization. Pp. 43-50.in C.T. Youngberg and C.B. Davey (eds.). Tree Growth and Forest Soils. Proc., ThirdNorth American Soils Conference, Raleigh, NC. Oregon State University Press, Corvallis,OR.Brink, V.C. 1956. Survival of plants under flood in the lower Fraser River Valley. Ecology35:94-95.Broadfoot, W.M. 1960. Field guide for evaluating cottonwood sites. Occas. Pap. No. 178, U.S.Department of Agriculture, Forest Service, South. For. Res. Stn. New Orleans, LA.6 p.Broadfoot, W.M. 1967. Shallow water impoundment increases soil moisture and growth ofhardwoods. Proc. Soil Sci. Soc. Amer. 31:562-564.Broadfoot, W.M. 1969. Problems in relating soil to site index for southern hardwoods. For. Sci.15:354-364.Broadfoot, W.M., and R.E. Farmer Jr. 1969. Genotype and moisture supply influence nutrientcontent of eastern cottonwood foliage. For. Sci. 15(1):46-48.Broadfoot, W.M., and H.L. Williston. 1973. Flooding effects on southern forests. J.Forestry:584-587.Cannell, M.G.R. 1980. Productivity of closely-spaced young poplar on agricultural soils inBritain. Forestry 53:1-21.Cannell, M.G.R., and R.I. Smith. 1980. Yields of mini-rotation, closely-spaced hardwoods intemperate regions: Review and Appraisal. Forest Sci. 26(3):415-428.Carmean, W.H. 1954. Site quality for Douglas-fir in southwestern Washington and its relation toprecipitation, elevation, and physical soil properties. Soil Sci. Soc. Amer. Proc.18:330-334.Carmean, W.H. 1965. Black oak site quality in relation to soil and topography in southeasternOhio. Soil Sci. Soc. Amer. Proc. 29:308-312.184Carmean, W.H. 1970. Tree height growth patterns in relation to site and soil. Pp. 499-512 inC.T. Youngberg and C.C. Daveys (eds.) Tree Growth and Forest Soils. NorthAmerican Forest Soils Conference, Proceedings.Carmean, W.H. 1972. Site index curves for upland oaks in the Central States. Forest Sci.18:102-120.Carmean, W. H. 1975. Forest site quality evaluation in the United States. Adv. in Agronomy (9)1:209-269.Carter, R.E., and L.E. Lowe. 1986. Lateral variability of forest floor properties under second-growth Douglas-fir and the usefulness of compositing techniques. Can. J. For. Res.16:1128-1132.Carter, R.E., and K. Klinka. 1990. Relationships between growing-season soil water deficit,mineralizable soil nitrogen, and site index of coastal Douglas-fir. For. Ecol.Manage. 30:301-311.Carter, R.E., and R.P. Brockley. 1990. Boron deficiencies in British Columbia: Diagnosis andtreatment evaluation. For. Ecol. and Manage. 37:83-94.Chapin, F.S. III, K. van Cleve, and P. Vitousek. 1986. The nature of nutrient limitation in plantcommunities. Am. Nat. 27:148-158.Chatterjee, S., and B. Price. 1977. Regression Analysis by Example. John Wiley and Sons, NewYork. 228 pp.Cohen, J. 1968. Multiple regression as a general data-analytical system. Psychological Bulletin70 (6): 426-443.Coile, T.S. 1952. Soil and the growth of forests. Advances in Agronomy 16:239-398.Cole, C.V., S.R. Olsen, and C.O. Scott. 1953. The nature of phosphate sorption by calciumcarbonate. Soil Sci. Soc. Amer. Proc. 17:352-356.Cole, D.W., E.D. Ford, and J. Turner. 1990. Nutrients, moisture, and productivity of establishedforests. For. Ecol. Manage. 30:283-299.Courtin, P.J. 1992. The relationship between ecological site quality and site index and stem formof red alder in south-western Briitish Columbia. M.S.For. Thesis, Dept. of For. Sci.,Fac. of Forestry, Univ. of British Columbia. Vancouver, B.C. 87 pp.Courtin, P.J., M.C. Feller, and K. Klinka. 1983. Lateral variability in some properties ofundisturbed forest soils in southwestern British Columbia. Can. J. Soil Sci. 63:529-539.Courtin, P.J., M.C. Feller, and J.P. Demaerchalk. 1988. An approach to the quantitativeclassification of nutrient regimes of forest soils. Can. J. Bot. 66:2640-2653.Covell, R.R., and D.C. McClurkin. 1967. Site index of loblolly pine on Ruston soils in thesouthern Coastal Plain. J. Forest. 65:263-264.185Cox, G.S., R.C. McConnell, and L.M. Matthew. 1960. Ponderosa pine productivity in relation tosoil and landform in western Montana. Soil Sci. Soc. Amer. Proc 24:139-142.Crist, J.B., and D.H. Dawson. 1975. Anatomy and dry weight yields of two Populus clonesgrown under intensive culture. Res. Pap NC-113, U.S. Dept. of Agric., Forest Service,North Central Forest Experiment Station,Critchfield, W.B. 1960. Leaf dimorphism in Populus trichocarpa. Amer. J. Bot. 47:699-711.Curran, P.M. 1984. Soil testing for phosphorus availability to some conifers in BritishColumbia. M. Sc. Thesis, Department of Soil Science, Faculty of Forestry,University of British Columbia, Vancouver, B.C. 130 pp.Czapowskyj, M.M., and L.O. Safford. 1979. Growth response to fertilizer in a young aspen-birch stand. Res. Note NE-274, U.S. Dept. of Agric., Northeastern Forest ExperimentStation, Broomall, PA.Daubenmire, C.R. 1976. The use of vegetation in assessing the productivity of forest lands. Bot.Rev. 42:115-143.Dawson, D.H., and J.G. Isebrands, and J.C. Gordon. 1976. Growth, dry weight yields, andspecific gravity of 3-year old Populus grown under intensive culture. Res. Pap NC-122,U.S. Dept. of Agric., Forest Service, North Central Forest Experiment Station,Day, F.P. Jr., and C.D. Monk. 1977. Seasonal nutrient dynamics in the vegetation of a southernAppalachian watershed. Amer. J. Bot. 64(9):1126-1139.DeBell, D. S. 1990. Populus trichocarpa Torr. and Gray - Black Cottonwood. Pp. 570-576 inBurns, R.M., and B.H. Honkala. (eds). Silvics of North America. USDA For. Serv.Agric. Handbook. No. 654. 877 p.DeBell, D.S., and M.A. Radwan. 1979. Growth and nitrogen relations of coppiced blackcottonwood and red alder in pure and mixed stands. Bot. Gaz. 140 (Suppl.). S97-S101.Demeritt, M.E. Jr. 1990. Poplar Hybrids. Pp. 577-582 in Burns, R.M., and B.H. Honkala. (eds).Silvics of North America. USDA For. Serv. Agric. Handbook. No. 654. 877 p.Devlin, R.M. 1966. Plant Physiology. Van Nostrand Rheinhold Co., New York. 446 pp.Dijkshoorn, W., and A.L. van Wijk. 1967. The sulphur requirements of plants as evidenced bythe sulphur-nitrogen raio in the organic matter - A review of the published data.Plant and Soil 26:129-157.Drees, L.R., and L.P. Wilding. 1973. Element variability within a sampling unit. Soil Sci.Soc. Amer. Proc. 37:82-87.Dyer, M.E., and R. L. Bailey. 1987. A test of six methods for estimating true heights from stemanalysis data. Forest Sci. 33:3-13.186Eis, S. 1962. Statistical analysis of several methods for estimation of forest habitats and treegrowth near Vancouver, B.C. Faculty of Forestry, Forestry Bulletin No. 4. Univ. ofBritish Columbia, Vancouver, B.C.Ek, A.R., and D.H. Dawson. 1976. Yields of intensively grown Populus: actual and projected.Pp. 5-9 in, Intensive plantation culture: five years' research. Gen. Tech. Rep. NC-21, U.S.Dept. of Agric., Forest Service, North Central Forest Experiment Station,Ellis, R.C. 1975. Sampling deciduous broadleaved trees for the determination of leaf weight andfoliar nutrient concentrations. Can. J. For. Res. 5:310-317.Ellis, R.C., and F.W. von Althen. 1973. Fertilization of adult and juvenile hardwoods in Ontario.Gen. Tech. Rep. NE-3, U.S. Dept. of Agric., Forest Service, Northeastern ForestExperiment Station, Broomall, PA. pp 174-182.Epstein, E. 1972. Mineral Nutrition of Plants: Principles and Perspectives. Wiley and Sons,New York.Everard, J. 1973. Foliar analysis. Sampling methods, interpretations, and application of theresults. Quart. Jour. Forestry 67:51-66.Fife, D.N., and E.K.S. Nambiar. 1982 Accumulation and retranslocation of nutrients indeveloping needles in relation to seasonal growth of young radiata pine trees. Ann.Bot. 50:817-829.Fife, D.N., and E.K.S. Nambiar. 1984. Movement of nutrients in radiata pine needles in relationto the growth of shoots. Ann. Bot. 54:303-314.Flanagan, P.W., and K. van Cleve. 1983. Nutrient cycling in relation to decomposition andorganic matter quality in taiga ecosystems. Can. J. For. Res. 13:795-817.Food and Agricultural Organization of the United Nations. 1958. Poplars in forestry and landuse. FAO Forestry and Forest Products Studies, Number 12, Rome, Italy.Freese, F. 1962. Elementary Forest Sampling. USDA Forest Service, Agricultural HandbookNo. 232. 91pp.Gadgil, P.D. 1972. Effects of waterlogging on mycorrhizas of radiata pine and Douglas-fir. N.Z. J. For. Sci. 2:222-226.Gagnon, J.D. 1975. A simple and quick method to assess yearly diameter response tofertilization in natural forest stands. Can. For. Serv. Bimon. Res. Notes. 31(4): 25-26.Gaines, T.P., and G.A. Mitchell. 1979. Boron determination in plant tissues by azomethine Hmethod. Comm. Soil Sci. Plant Anal. 10:1099-1108.Gauch, H. 1984. Multivariate analysis in community ecology. Cambridge Univ. Press. London.298 pp.187Giles, D.G., T.A. Black, and D. Spittlehouse. 1985. Determination of growing season soil waterdeficits on a forested slope using water balance analysis. Can. J. For. Res. 15:107-114.Green, R. N., P.J. Courtin, K. Klinka, R.J. Slaco, and C.A. Ray. 1984. Site diagnosis, treespecies selection, and slashburning guidelines for the Vancouver Forest Region.Land Manage. Handbook No. 8. B.C. Ministry of Forests, Victoria, B.C. 54 pp.Green, R. N., P. L. Marshall, and K. Klinka. 1989. Estimating site index of Douglas-fir(Pseudotsuga menziesii [Mirb.] Franco) from ecological variables in southwesternBritish Columbia. For. Sci. 35:50-63.Greenwood, D.J. 1969. Effect of oxygen distribution in the soil on plant growth. Pp. 202-223in W.J. Whittington (ed.), Root Growth. Butterworths, London.Grewelling, T., and M. Peech. 1960. Chemical soil tests. Cornell Agric. Exp. Sta. Bull. 960.Grier, C.C., and J.G. McColl. 1971. Forest floor characteristicswithin a small plot in Douglas-fir in western Washington. Soil Sci. Soc. Amer. Proc. 35:988-991.Griffin, R.A., and J. J. Jurinak. 1973. The interaction of phosphate with calcite. Soil Sci. Soc.Amer. Proc. 37:847-850.Guha, M.M., and R.L. Mitchell. 1965a. The trace and major element composition of the leavesof some deciduous trees. I. Sampling techniques. Plant Soil 23:323-338.Guha, M.M., and R.L. Mitchell. 1965b. The trace and major element composition of the leavesof some deciduous trees. II. Seasonal changes. Plant Soil 26:90-112.Guthrie, T.F., and L. E. Lowe. 1984. A comparison of methods for total sulphur analysis of treefoliage. Can. J. For. Res. 14:470-473.Hagglund, B. 1981. Evaluation of forest site productivity. Comm. For. Bureau, For. Abstracts,42(11):515-527.Haines, S.G., and G. Cleveland. 1981. Seasonal variation in soil properties of five forest soils insouthwest Georgia. Soil Sci. Soc. Amer. J. 45:139-143.Harrington, C. 1986. A method of site evaluation for red alder. Gen. Tech. Rep. PNW-192.U.S. Department of Agriculture, Forest Service, Pacific Northwest Research Station,Portland OR: 22 p.Harrington, C. A. 1987. Responses of red alder and black cottonwood seedlings to flooding.Physiol. Plant. 69:35-48.Harrington, C.A., and D. S. DeBell. 1984. Effects of irrigation, pulp mill sludge, and repeatedcoppicing on growth and yield of black cottonwood and red alder. Can. J. For. Res.14:844-849.Harris, M., D.G. Horvitz, and A.M. Mood. 1948. On the determination of sample sizes indesigning experiments. Am. Stat. Assoc. J.:391-402.188Harrison, A.F. 1979. Variation of four phosphorus properties in woodland soils. Soil Biol.Biochem. Vol. 11:393-403.Hart, J.B. Jr., A.L. Leaf, and S.J. Stutzbach. 1969. Variation in potassium available to treeswithin an outwash soil. Soil Sci. Soc. Amer. Proc. 33:950-954.Heilman, P.E. 1985. Sampling and genetic variation of foliar nitrogen in black cottonwood andits hybrids in short rotation. Can. J. For. Res. 15:1137-1141.Heilman, P.E., D.V. Peabody Jr., D.S. DeBell„ and R.F. Strand. 1972. A test of close-spaced,short rotation culture of black cottonwood. Can. J. For, Res. 2:456-459.Heilman, P.E., and D.V. Peabody Jr. 1981. Effect of harvest cycle and spacing on productivityof black cottonwood in intensive culture. Can. J. For. Res. 11:118-123.Heilman, P., and R.F. Stettler. 1983. Phytomass production in young mixed plantations of Alnusrubra (Bong.) and cottonwood in western Washington. Can. J. Microbiol. 29:1007-1013.Heilman, P., and R.F. Stettler. 1985a. Mixed, short rotation culture of red alder and blackcottonwood: Growth, coppicing, nitrogen fixation, and allelopathy. Forest Sci. 31(3):607-616.Heilman, P., and R.F. Stettler. 1985b. Genetic variation and productivity of Populus trichocarpaand its hybrids. II. Biomass production in a 4 year plantation. Can. J. For. Res. 15:384-388.Heinsdorf, D. 1968. Enrahrung und Wachstum junger Kiefern auf einem an Magnesium undKalium armen Talsandboden nach mehrmaliger mineralischer Duengueng. Arch.Forstwes. 17:813-838.Hicken, E. J., and H.M. Sichingabula. 1988. The geomorphic impact of the catastrophicOctober 1984 flood on the planform of the Squamish River, southwestern BritishColumbia. Can. J. Earth Sci., 25:1078-1087.Hoyle, M.C. 1965. Variation in foliage composition and diameter growth of yellow birch withseason, soil, and tree size. Soil Sci. Soc. Amer. Proc. 29:475-480.Ike, A.F., and J.L. Clutter. 1968. The variability of forest soils in the Georgia Blue RidgeMountains. Soil Sci. Soc. Amer. Proc. 32:284-288.Ingestad, T. 1962. Macro element nutrition of pine, spruce, and birch seedlings in nutrientsolutions. Medd. statens Skogsforskningsinst 51(7):1-150.Isebrands, J.G., N.D. Nelson, D.I. Dickmann, and D.A. Micheal. 1983. Yield physiology ofshort rotation, intensively cultured poplars. U.S. Dept. of Agric., For. Ser., GTR NC-91.pp. 77-93Johnson, C.M. and H. Nishita. 1952. Microestimation of sulphur in plant materials, soils, andirrigation waters. Anal. Chem. 24:736-742.189Jolicoeur, P., and J. E. Mosimann. 1960. Size and shape variation in the painted turtle. Aprincipal component analysis. Growth. 24:339-354.Jones, J.R. 1969. Review and comparison of site evaluation methods. Res. Pap. RM-51, U.S.Department of Agriculture, Forest Service, Rocky Mountain Forest and RangeExperimental Station, Fort Collins, CO. 22 p.Kabzems, R. D., and K. Klinka. 1987a. Initial quantitative characterization of soil nutrientregimes. I. Soil properties. Can. J. For. Res. 17:1557-1564.Kabzems, R. D., and K. Klinka. 1987b. Initial quantitative characterization of soil nutrientregimes. H. Relationships among soils, vegetation, and site index. Can. J. For. Res.17:1565-1571.Kayahara, G.J. 1991. An investigation of relationships between western hemlock ecosystems,ecological site quality, and productivity in the Coastal Westerm Hemlock Zone ofBritish Columbia. M. Sc. Thesis, Dept. of Forest Sci., Faculty of Forestry, Univ. ofBritish Columbia, Vancouver, B.C. 157 p.Kelly, J., and M.J. Lambert. 1972. The relationship between sulphur and nitrogen in the foliageof P. radiata. Plant and Soil 37:395-408.Kennedy, H.E. Jr., and R.M. Krinard. 1974. 1973 Mississippi River flood's impact on naturalhardwood forests and plantations. Res. Note SO-177. U.S. Dept. Agric., For. Ser.,Southern Forest Exp. Sta., New Orleans, LA. 6 pp.Kimmins, J.P., and B.C. Hawkes. 1978. Distribution and chemistry of fine roots in a whitespruce-subalpine fir stand in British Columbia: Implications for management. Can.J. For. Res. 8(3):265-279.Klinka, K., M.C. Feller, L.M. Lavkulich, and A. Kozak. 1980. Evaluation of methods ofextracting soil cations for forest productivity studies in southwestern BritishColumbia. Can. J. Soil Sci. 60:697-705.Klinka, K., R.N. Green, R.L. Trowbridge, and L.E. Lowe. 1981. Taxonomic classification ofhumus forms in ecosystems of British Columbia. Land Manage. Rep. No. 8, B.C.Ministry of Forests, Victoria, B.C. 54 pp.Klinka,K., P.J. Courtin, and R.N. Green. 1984. An approach to quantitative classification ofhygrotopes and trophotopes of forest soils. IUFRO Workshop on QuantitativeAssessment of Forest Sites, with Special Reference to Soils. Swiss Federal Insituteof Forestry Research, CH-8903 Birmensdorf, Switzerland. September 10-15, 1984.40 pp.Klinka, K., and V.J. Krajina. 1987. Ecosystems of the University of British Columbia ResearchForest, Haney, B.C. Faculty of Forestry, University of British Columbia,Vancouver, B.C. 123 pp.Klinka, K., R.E. Carter, M.C. Feller, and Q. Wang. 1989a. Relations between site index, salal,plant communities, and sites in coastal British Columbia. For. Sci. 36:815-830.190Klinka, K., V.J. Krajina, A. Ceska, and A.M. Scagel. 1989b. Indicator Plants of Coastal BritishColumbia. University of B.C. Press, Vancouver, B.C. 288 pp.Klinka,. K, and R.E. Carter. 1990. Relationships between site index and synoptic environmentalvariables in immature coastal Douglas-fir stands. For. Sci. 36:815-830.Klinka, K., Q. Wang, and R. E. Carter. 1990. Relationships among humus forms, forest floornutrient properties, and understory vegetation. For. Sci. 36:564-581.Klinka, K., J. Pojar, and D.S. McLennan. 1990. Classification of forested landscapes in coastalBritish Columbia. In, Proceedings: Landscape Approaches to Wildlife andEcosystem Management. Canadian Society for Landscape Ecology andManagement, University of British Columbia, Vancouver, B.C.Knapp, T.R. 1978. Canonical correlation analysis: A general parametric significance-tesingsystem. Psychological Bulletin 85 (2): 410-416.Knight, P.J. 1978. Foliar concentrations of ten mineral nutrients in nine Pinus radiata clonesduring a 15-month period. N.Z. J. For. Sci. 8:351-368.Kolster, H.W., and H.A. van der Meiden. 1979. Wood production at very short rotations.Populier 16:3-7.Kowalenko, G.F., and L. E. Lowe. 1972. Observations on the bismuth sulfide colorimetricprocedure for sulfate analysis in soil. Commun. Soil Sci. Plant Anal. 3:79-86.Kozlowski, T.T. 1982. Water Supply and Tree Growth. Part II: Flooding. Forestry Abstracts43(3):145-158.Krajina, V, K. Klinka, and J. Worrall. 1982. Distribution and Ecological Characteristics ofTrees and Shrubs of British Columbia. Faculty of Forestry, University of BritishColumbia, Vancouver, B.C. 131 pp.Kramer,P.J. 1969. Plant and Soil Water Relationships. A Modern Synthesis. McGraw-Hill, NewYork.Kumata, A., A. Kawana, and H. Kataoka. 1988. Trends in foliar elements in various phases ofgrowth of deciduous trees. pp. 22-23. In Forest site evaluation and long-termproductivity. Cole, D. and S. Gessel (eds.). Contribution No. 63. University ofWashington, Inst. of Forest Resources. Seattle, Wa.Larson, P.R. and J.G. Isebrands. 1972. The relation between leaf production and wood weight infirst-year sprouts of two Populus clones. Can J. For. Res. 2: 98-104.Lavender, D.P. 1970. Foliar analysis and how it is used - A review. Research Note No. 52,School of Forestry, Forest Research Laboratory, Oregon State University, Corvallis,Oregon. Res. Note 52. 8 pp.Lea, R., W.C. Tierson, D.H. Bickelhaupt, and A.L. Leaf. 1979a. Stand treatments and samplingtime of hardwood foliage. I. Macro-element analysis. Plant Soil 51:515-533.191Lea, R., W.C. Tierson, D.H. Bickelhaupt, and A.L. Leaf. 1979b. Stand treatments and samplingtime of hardwood foliage. II. Micro-element analysis. Plant Soil 51:535-550.Leaf, A.L., J.E. Berglund, and R.E. Leonard. 1970. Annual variation in foliage of fertilizedand/or irrigated red pine plantations. Soil Sci. Soc. Amer. Proc. 34:667-682.Leaf, A.L. 1973. Plant analysis as an aid in fertilizing forests. Pp. 427-454 in Walsh, L.M., andJ. D. Beaton (eds.) Soil Testing and Plant Analysis. Rev. ed. Soil Sci. Soc. Amer.,Madison, WI.Leech, R.H., and Y.T. Kim. 1979a. Assessment of nutrient requirements of poplars. pp. 15-1 to15-15 In, D.C.F. Fayle, L. Zsuffa„ and H.W. Henderson (eds.), Poplar research,Management, and Utilization in Canada. Ont. Min. Nat. Resour. Forest Res. Info.Paper No. 102., Toronto, Ontario.Leech, W.S. and Y.T. Kim. 1979b. Assessment of nutrient requirements of poplars. Proc. NorthAmerican Poplar Council Meeting, Ont. Min. of Nat. Res., Info. Paper No.102. 16 pp.Leech, W.S. and Y.T. Kim. 1981. Foliar diagnosis and DRIS as a guide to fertilizer amendmentsin poplar plantations. For. Chron. 57(1): 17-21.Letzsch, W.S. and M. E. Sumner. 1983. Computer program for calculating DRIS indices.Comm. in Soil Sci. Plant Anal. 14 (9): 811-815.Lewis, T. 1976. The till-derived podzols of Vancouver Island. Ph. D. thesis. University ofBritish Columbia. Vancouver, B.C. 158 pp.Leyton, L. 1948. Mineral nutrition relationships of forest trees. Forestry Abstracts 9(4):399-408.Leyton, L. 1958. The relationship between the growth and mineral nutrition of conifers. Pp.323-345 in K.V. Thimann (ed.), The Physiology of Trees. Ronald Pres, NY.Lipas, E. 1979. Alternative methods for calculating growth response to fertilization. Seloste:Vaihtoehtoisia menetelmia lannoitusreaktion laskemiseksi. Comm. Inst. For. Fenn.97(7):1-56.Lowe, L. E. 1972. Aspects of chemical variability in forest humus layers under a maturewestern hemlock-western redcedar stand. Can. J. For. Res. 2:487-489.Luttmerding, H.A., D.A. Demarchi, E.C. Lea, D. V. Meidinger, and T. Vold (eds.). 1990.Describing ecosystems in the field. 2nd ed. B.C. Ministry of Forests, Victoria, B.C.213 pp.Mader, D.L. 1963. Soil variability - a serious problem in soil-site problems in the North-east.Soil Sci. Soc. Amer. Proc. 27:707-709.Mader, D.L., and H.W. Lull. 1968. Depth, weight, and water storage of the forest floor in whitepine stands in Massachusetts. U.S. For. Serv., Res. Paper NE-109, N.E. ForestExperimental Sta., Upper Darby. PA.192Marshall, P.L. 1987. A microcomputer program for assisting in the design of sample randomsamples. For. Chron. 64:422-425.Marshall, P.L., and K. Jahraus. 1987. Sample size for foliar analysis of coastal Douglas-fir. Can.J. For. Res. 17:1240-1245.Marshall, P.L., V. M. LeMay, and A. Nussbaum. 1992. Sample size adjustment to reduce theprobability of exceeding a specified confidence interval. For. Chron. 68(6):747-751.McFee, W.W., and E. L. Stone. 1965. Quantity, distribution, and variability of organic matterand nutrients in a forest Podzol in New York. Soil Sci. Soc. Amer. Proc. 29:432-436.McHargue, J.S., and W.R. Roy. 1933. Mineral and nutrient content of the leaves of some foresttrees at different times in the growing season. Bot. Gaz. 94:381-393.McLain, D.H. 1974. Drawing contours from arbitrary data points. The Computer Journal17:318-324.McLennan, D.S. 1990. Spatial variation in black cottonwood (Populus trichocarpa) foliarnutrient concentrations at seven alluvial sites in coastal British Columbia. Can. J.For. Res. 20 (7):1089-1097.Mehlich, A. 1978. New extractant for soil test evaluation of phosphorus, magnesium, calcium,sodium, manganese, and zinc. Comm. Soil Sci. Plant Anal. 9:477-492.Mellilo, J.M., and J.D. Aber. 1980. Nitrogen and lignin control of hardwood leaf litterdecomposition and dynamics. Ecology 63(3):621-626.Miller, W.F. 1966. Annual changes in foliar nitrogen, phosphorus, and potassium levels ofloblolly pine (Pinus taeda L.) with site and weather factors. Plant and Soil24(3):369-378.Miller, R.E., and R.F. Tarrant. 1983. Long-term response of Douglas-fir to ammonium nitratefertilizer. For. Sci. 29:127-137.Minore, D. 1968. Effects of artificial flooding on seedling survival and growth of sixnorthwestern tree species. Res. Note PNW-92, U.S. Dept. Agric., For. Ser., PacificNorth West Forest and Range Exp. Sta., Olympia, WA. 12 pp.Mitchell, H.L. 1936. Trends in nitrogen, phosphorus, potassium, and calcium content of leavesof some forest trees during the growing season. Black Rock Forest Paper 1:29-44.Mitchell, H.L., and R.F. Chandler. 1939. The nitrogen nutrition and growth of certain deciduoustrees of northeastern United States. Black Rock Forest Bulletin 11. 94 pp.Mize, C.W., and R.C. Schultz. 1985. Comparing treatment means correctly and appropriately.Can. J. For. Res. 15:1142-1148.Mollitor, A.V., A.L. Leaf, and L.A. Morris. 1980. Forest soil variability on northeasternfloodplains. Soil Sci. Soc. Amer. J. 44:617-620.193Monserud, R.A. 1984. Height growth and site index curves for inland Douglas-fir based onstem analysis data and forest habitat type. For. Sci. 30(4):943-965.Morrison, I.K. 1972. Variation with crown position and leaf age in content of seven elements inleaves of Pinus banksiana Lamb. Can. J. For. Res. 2:89-94.Morrison, I.K. 1974a. Within-tree variation in mineral content of leaves of young balsam fir.For. Sci. 20:276-278.Morrison, I.K. 1974b. Mineral nutrition of conifers with special reference to nutrient stautusinterpretation. A review of literature. Can. For. Ser. Pub. No. 1343.Morrison, I.K. 1985. Effect of crown position of foliar concentrations of 11 elements in Acersaccharum and Betula alleghaniensis trees on a till soil. Can. J. For. Res. 15:179-183.Mueller-Dombois, D., and H. Ellenberg. 1974. Aims and methods of vegetation ecology. JohnWiley and Sons, New York. 547 pp.Myers, C.A., and J.L. van Deusen. 1960. Site index of ponderosa pine in the Black Hills fromsoil and topography. J. For. 58:548-555.Nuszdorfer, F.N., K.L. Kassay, and A.M. Scagel. 1985. Biogeoclimatic units of coastal BritishColumbia. 1:500,000 Colour Map. Research Branch, B.C. Ministry of Forests,Victoria, B.C.Oserkowsky, J. 1933. Quantitative relation between chlorophyll and iron in green and chloroticpear leaves. Plant Physiol. 8:449-468.Ostman, N.L., and G.T. Weaver. 1982. Autumnal transfer by retranslocation, leaching, andlitter fall in a chestnut oak forest in southern Illinois. Can. J. For. Res. 12:40-51.Palmer, C.L. 1991. Short rotation culture of Populus and Larix: A literature review. For. Can.,Ont. Region, Sault Ste. Marie, Ont. COFRDA Rep. 3306. 66p.Parkinson, J.A., and S.E. Allen. 1975. A wet oxidation procedure for the determination ofnitrogen and mineral nutrients in biological material. Comm. Soil Sci. Plant Anal.6:1-11.Peech, M. 1965. Hydrogen-ion activity. Pgs. 914-926 in, C.A. Black, D.D. Evans, J. L. White,L.E. Ensminger, and F.E. Clark (eds.). Methods of Soil Analysis. Agronomy No. 9,Amer. Soc. Agron., Madison, WI.Peterson,D.L., and G.L. Rolfe. 1982. Seasonal variation in nutrients of floodplain and uplandforest soils of central Illinois. Soil Sci. Soc. Amer. J. 46:1310-1315.Peterson, D.L., and G.L. Rolfe. 1985. Temporal variation in nutrient status of a floodplain forestsoil. Forest Ecology and Management 12:73-82.Peterson, D.L., and R.D. Hammer. 1986. Soil nutrient flux: A component of nutrient cycling intemperate forest ecosystems. Forest Sci. 32(2):318-324.194Pfister, R.D., B.L. Kovalchik, S.F. Arno, and R.C. Presby. 1977. Forest habitat types ofMontana. Gen. Tech. Rep. INT-34. U.S. Dept. Agric., Forest Serv., IntermountainForest Range and Experiment Station, Ogden, UT.Pharis, R.P., and P.J. Kramer. 1964. The effects of nitrogen and drought on loblolly pineseedlings. I. Growth and composition. For. Sci. 10(2):143-150.Pielou, E.C. 1975. The Interpretation of Ecological Data. John Wiley and Sons, New York. 263pp.Pojar, J., K.Klinka, and D.V. Meidinger. 1987. Biogeoclimatic ecosystem classification inBritish Columbia. For. Ecol. Manage. 22:119-154.Ponnamperuma, F.N. 1984. Effects of flooding on soils. Pp. 9-45 in, T.T. Koslowski (ed.).Flooding and Plant Growth. Academic Press Inc., Orlando, FA.Powers, R.F. 1980. Mineralizable soil nitrogen as an index of nitrogen availability to foresttrees. Soil Sci. Soc. Amer. J. 44:1314-1320.Powers, R.F. 1984. Sources of variation in mineralizable soil nitrogen. Pp. 25-31 in NitrogenAssessment Workshop held May 19-20. RFNRP Report No. 2. College of ForestResources, Univ. of Washington, Seattle, WA..Quesnel, H.J., and L.M. Lavkulich. 1980. Nutrient variability of forest floors near Port Hardy,British Columbia, Canada. Can. J. Soil Sci., 60:565-573.Radwan, M.A., and D.S. DeBell. 1988. Nutrient relations in coppiced black cottonwood and redalder. Plant and Soil 106:171-177.Regehr, D.L., F.A. Bazzaz, W.R. Bogess. 1975. Photosynthesis, transpiration, and leafconductance of Populus deltoides in relation to flooding and drought.Photosynthetica 9:52-61.Richards, B.N., and D.I. Bevege. 1972. Principles and practice of foliar analysis as a basis forcrop logging in pine plantations. I. Basic considerations. Plant and Soil 36:109-119.Robertson, G.P. 1987. Geostatistics in ecology : Interpolating with known variance. Ecology68:744-748.Robertson, G.P., M.A. Houston, F.C. Evans, and J.M. Tiedje. 1988. Spatial variability in asuccessional plant community: Patterns of nitrogen availability. Ecology 69: 1517-1524.Roe, A.L. 1958. Silvics of Black Cottonwood. Misc. Pub. 17. U.S. For. Serv., IntermountainFor. and Range Exp. Sta., Moscow, ID.Russell, E.W. 1974. Soil Conditions and Plant Growth. Longman, London.Safford, L.O., and S.M. Filip. 1974. Biomass and nutrient content of 4 year old fertilized andunfertilized northern hardwood stands. Can. J. For. Res. 4:549-554.195Salonius, P.O., R.A. Fisher, and M.K. Mahendrappa. 1982. An alternative method of measuringfertilizer effects in forest stands. Can. J. For. Res. 12: 146-150.Sampson, A.W., and R. Samisch. 1935. Growth and seasonal changes in composition of oakleaves. Plant Physiol. 10:739-751.Schmidt, M.G., and W.H. Carmean. 1988. Jack pine site quality in relation to soil andtopography in north central Ontario. Can. J. For. Res. 18:297-305.Schutz, C.J., and J.M. de Villiers. 1986. Foliar diagnosis and fertilizer prescription in forestry -The DRIS system and its potential. Proceedings: 18th IUFRO World Congress,Ljubljana. 16 pp.Scott-Russell, R.R. 1977. Plant Root Systems. Their Function ans Interaction with the Soil.McGraw-Hill, N.Y.Shear, C.B., H.L. Crane, and A.T. Myers. 1946. Nutrient element balance: a fundamentalconcept in plant nutrition. Proc. Amer. Soc. Hort. Sci. 51:319-326.Shear, C.B., H.L. Crane, and A.T. Myers. 1948. Nutrient element balance: application of theconcept to the interpretation of foliar analysis. Proc. Amer. Soc. Hort. Sci. 51:319-326.Sheriff, D.W., E.K.S. Nambiar, and D.N. Fife. 1986. Relationships between nutrient status,carbon assimilation, and water use efficiency in Pinus radiata D. Don needles. TreePhysiol. 2:73-88.Slavinsky, H.C. 1977. Soil variability along a topographic sequence, University of BritishColumbia Endowment Lands. M. Sc. thesis, University of British Columbia.Vancouver, B.C. 203 pp.Smit, B.A. 1988. Selection of flood-resistant and susceptible seedlings of Populus trichocarpaTorr. and Gray. Can. J. For. Res. 18:271-275.Smith, J.H.G. 1957. Some factors indicative of site quality for black cottonwood. J. For.55:578-580.Smith, J.H.G. 1980. Growth and yield of poplar in British Columbia. Paper presented at 1980Meeting, Poplar Council of Canada. Faculty of Forestry, University of British Columbia,Vancouver, B.C.Soil Survey Staff. 1975. Soil Taxonomy: A basic system of soil classification for making andinterpreting soil surveys. U.S. Dept. Agric. Soil Conservation Service, Agric.Handbook No. 436.Spittlehouse, D.L., and T.A. Black. 1981. A growing season water balance model applied to twoDouglas-fir stands. Water Resour. Res. 17:1651-1656.Spurr, J.E., and B.V. Barnes. 1980. Forest Ecology. John Wiley and Sons. New York. 687pp.196Stettler, R.F., R.C. Fenn, P.E. Heilman, and B. Stanton. 1988. Populus trichocarpa x Populusdeltoides hybrids for short rotation culture: variation patterns and four-year fieldperformance.Sumner, M.E. 1977a. Use of the DRIS system in foliar diagnosis of crops at high yield levels.Comm. in Soil Sci. Plant Anal. 8(3): 251-268.Sumner, M.E. 1977b. Preliminary NPK foliar diagnostic norms for wheat. Comm. in Soil Sci.Plant Anal. 8(2): 149-167.Sumner, M.E. 1978. Interpretation of nutrient ratios in plant tissue. Comm. in Soil Sci. PlantAnal. 9(4):335-345.Sumner, M.E. 1979. Interpretation of foliar analyses for diagnostic purposes. Amer. Soc. Agron.J. 71: 343-348.Switzer, G.L., and L.E. Nelson. 1972. Nutrient accumulation and cycling in loblolly pine (Pinustaeda L.) plantation ecosystems: the first twenty years. Soil Sci. Soc. Amer. Proc.36:143-147.Switzer, G.L., L.E. Nelson, and J.B. Baker. 1976. Accumulation and distribution of dry matterand nutrients in Aigeiros poplar plantations. Pp. 359-369 in, B. A. Thielges and S.B.Lambs (eds.), Proceedings of a Conference on Eastern Cottonwood and Related Species.Louisiana State University, Baton Rouge, LA.Tamm, C.O. 1951. Seasonal variation in the composition of birch leaves. Physiol. Plant. 4:461-469.Timmer, V.R. 1985. Response of a hybrid poplar clone to soil acidification and liming. Can. J.Soil Sci. 65: 727-735.Timmer, V.R. and L.D. Morrow. 1984. Predicting fertilizer growth response and nutrient statusof jack pine by foliar diagnosis. Pages 335-351 in, E.L. Stone (ed.), Forest Soils andTreatment Impacts. Proc. 6th North Am. For. Soil Conf., Knoxville, Tenn..Timmer, V.R. and P.N. Ray. 1988. Evaluating soil nutrient regime for black spruce in theOntario Claybelt by fertilization. For. Chron. 45:40-46.Timmer, V.R. and E. L. Stone. 1978. Comparative foliar analysis of young balsam fir fertilizedwith nitrogen, phosphorus, potassium, and lime. Soil Sci. Am. J. 42:125-130.Titus, S.J. 1987. A graphic approach to summarizing forestry data. For. Chron. 38:347-350.Troedsson, T., and C.O. Tamm. 1969. Small-scale spatial variation in forest soil properties andits implications for sampling procedures. Studia Forestalia Suecica, Nr. 74. RoyalCollege of Forestry, Stockholm. 18 pp.Tukey, J.W. 1977. Exploratory Data Analysis. Reading, MA. Addison-Wesley.Turner, J., M.J. Lambert, and S.P. Gessel. 1977. Use of foliage sulphate concentrations topredict response to urea application by Douglas-fir. Can. J. For. Res. 7:476-480.197van Cleve, K. 1973. Short-term growth response to fertilization in young quaking aspen. J. For.758-759.van den Driessche, R. 1974. Prediction of mineral nutrient status of trees by foliar analysis. Bot.Rev. 40(3):347-394.van der Kamp, B. 1979. Decay-resistance owing to near-anaerobic conditions in cottonwoodwetwood. Can. J. For. Res. 9 (1):39-44.van Slycken, J. 1984. Results of an investigation of groudwater behaviour in alluvial soils for aperiod of two years. IUFRO Workshop on qualitative and quantitative assessment offorest sites, with special reference to soil. Swiss Federal Inst. of Forest Research,Birmensdorf, Switzerland.Velleman, P.C., and D.C. Hoaglin. 1981. Applications, Basics, and Computing of ExploratoryData Analysis. Boston:Duxbury Press.Verry, E.S., and D.R. Timmons. 1976. Elements in leaves of trembling aspen clone by crownposition and season. Can. J. For. Res. 6:436-440.von Athen, F.W. 1973. Fertilization of juvenile hardwoods. Gen. Tech. Rep. NE-3, U.S. Dept.of Agric., Forest Service, Northeastern Forest Experiment Station, Broomall, PA. pp 183-187.Walker, L.C. 1962. The effects of water and fertilizer on loblolly and slash pine seedlings. SoilSci. Soc. Amer. Proc. 26:197-200.Wallihan, E.F. 1944. Chemical composition of leaves in different parts of sugar maple trees. J.For. 42:684Wang, Q. 1992. Ecological and height growth analysis of some sub-boreal, immature lodgepolepine stands in central British Columbia. Ph.D. Thesis, Faculty of Forestry, Univ. ofBritish Columbia, Vancouver, B.C. 183 p.Waring, S.A., and J.M. Bremner. 1964. Ammonium production in soil under waterloggedconditions as an index of soil nitrogen availability. Nature, Lond. 201:951-952.Weaver, T., and F. Forcella. 1979. Seasonal variation in soil nutrients under six Rocky Mountainvegetation types. Soil Sci. Soc. Amer. J. 43:589-593.Weetman, G. F., and C. G. Wells. 1990. Plant analyses as an aid in fertilizing forests. Chapter 25in, Soil Sci. Soc. Amer., Soil Testing and Plant Analysis, 3rd ed. Soil Sci. Soc.Amer. Book Series No. 3., Madison, Wisconsin.Weetman, G.F. 1981. Predicting fertilizer needs and responses by soil and tissue tests. Invitedpaper, XVII IUFRO World Congress, Kyoto. 16 pp.Weetman, G.F., and C.G. Wells. 1990. Plant analyses as an aid in fertilizing forests. Chapter25, in Soil Testing and Plant Analysis, 3rd ed. Soil Sci. Soc. Amer., SSSA Book Series,no. 3. Madison, Wisconsin.198Wells, C. G., and L.J. Metz. 1963. Variation in nutrient content of loblolly pine needles withseason, age, soil, and position on the crown. Soil Sci. Soc. Amer. Proc. 27:90-93.White, D.P. 1954. Variation in the nitrogen, phosphorus, and potassium contents of pineneedles with season, crown position and sample treatment. Soil Sci. Soc. Amer.Proc. 18:326-330.White, D.P., and R.S. Wood. 1958. Growth variations in a red pine plantation influenced by adeep-lying soil layer. Soil Sci. Soc. Amer. Proc. 22:174-177.White, E.H., and M.C. Carter. 1970a. Relationships between foliage, nutrient levels, and growthof young natural stands of Populus deltoides Bartr. in C.T. Youngberg and C.B.Davey (eds.) Tree growth and forest soils. Proc. 3rd North Am. For. Soils Conf. pp283-294.White, E.H., and M.C. Carter. 1970b. Properties of alluvial soils supporting young stands ofeastern cottonwood in Alabama. Res. Note SO-111, U.S. Department ofAgriculture, Forest Service, Southern Forest Research Station. New Orleans, LA. 4PP.White, J.D., L.T. Alexander, and E.W. Clark. 1972. Fluctutations in the inorganic constituentsof inner bark of lobolly pine with season and soil series. Can. J. Bot. 50:1287-1293.Whyte, A.G.D., and D.J. Mead. 1976. Quantifying responses to fertilizer in growth of radiatapine. N.Z. J. For. Sci. 6:432-444.Wilde, S.A. 1970. Soils and forest growth: their relationship in terms of regression analysis.BioScience 20:101-102.Wilkinson, L. 1990. SYSTAT: The System for Statistics. Evanston, IL. 677 pp.Will, G.M. 1965. Increased phosphorus uptake by radiata pine in Riverhead Forest followingsuperphosphate applications. N.Z. J. For. 10:33-42.Woodwell, G.M. 1974. Variation in the nutrient content of leaves of Quercus alba, Quercuscoccinea, and Pinus rigida in the Brookhaven forest from bud burst to abscission.Amer. J. Bot. 61:749-753.Woollons, R.C. and A.G.D. Whyte. 1988. Analysis of forest fertilizer experiments: obtainingbetter precision and extracting more information. For. Sci. 34(3):769-780.Youngberg, C.T. 1966. Forest floors in Douglas-fir forests.: I. Dry weight and chemicalproperties. Soil Sci. Soc. Amer. Proc. 30:406-440.

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0075152/manifest

Comment

Related Items