Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The relationships between the ecological site quality and the site index and stem form of red alder in… Courtin, P. J. 1992

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

Item Metadata

Download

Media
831-ubc_1992_spring_courtin_paul_julian.pdf [ 3.97MB ]
Metadata
JSON: 831-1.0075164.json
JSON-LD: 831-1.0075164-ld.json
RDF/XML (Pretty): 831-1.0075164-rdf.xml
RDF/JSON: 831-1.0075164-rdf.json
Turtle: 831-1.0075164-turtle.txt
N-Triples: 831-1.0075164-rdf-ntriples.txt
Original Record: 831-1.0075164-source.json
Full Text
831-1.0075164-fulltext.txt
Citation
831-1.0075164.ris

Full Text

THE RELATIONSHIPS BETWEEN THE ECOLOGICAL SITE QUALITY AND THESITE INDEX AND STEM FORM OF RED ALDER IN SOUTHWESTERN B.C.byPAUL JULIAN COURTINB.S.F., University of British Columbia, 1978A THESIS SUBMITTED IN PARTIAL FULFILMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF FORESTRY(6 units)inTHE FACULTY OF GRADUATE STUDIES(DEPARTMENT OF FORESTRY)We accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIAAPRIL 1992© Paul Julian Courtin, 1992In presenting this thesis in partial fulfilment of therequirements for an advanced degree at the University of BritishColumbia, I agree that the Library shall make it freely availablefor reference and study. I further agree that permission forextensive copying of this thesis for scholarly purposes may begranted by the Head of my Department or by his representatives.It is understood that copying or publication of this thesis forfinancial gain shall not be allowed without my writtenpermission.Department of ForestryThe University of British Columbia2075 Wesbrook PlaceVancouver, CanadaV6T 1W5iiABSTRACTSite, soil, tree form, vegetation and foliar data werecollected from thirty-seven natural stands of red alder (Alnus rubra Bong.) from three biogeoclimatic subzones on VancouverIsland and the adjacent mainland of British Columbia. Based ontemperature and precipitation normals, three subzones of theCoastal Western Hemlock zone are suggested as being the mostsuitable for the growth of red alder. The results of quantitativeanalyses between different site variables indicate that a majorfactor accounting for the majority of variance in the data issoil reaction. When the stands were divided into two groups basedon pH, the majority of variation in site index was explained forboth soil and foliar nutrient variables.Other variables had weak correlations with site index. Soilmorphological and physical data were not sufficient to predictsite index, and the use of soil moisture and nutrient regimes ascategorical variables in site index regressions gaveinsignificant results.Diagnostic criteria applied to the understory vegetation ofthe stands could not differentiate site classes nor were standsdifferentiated on the basis of soil reaction group which seemedto dominate the separation of both soil and foliar variables.This is due to red alder's N-fixing ability, which tends toiiipromote species having nutrient rich indicative value regardlessof site. Tentative vegetation associations are proposed but theirrecognition needs further testing.The relationship between four measures of tree form of redalder with other variable domains did not indicate meaningfulresults. The hypothesis that form of alder differs with sitecannot be accepted based on these data.The synthesis of a qualitative site assessment forestimating the potential productivity of red alder was approachedfrom the recognition of soil parent materials while incorporatingsoil physical and morphological properties already in use infield guides.ivTABLE OF CONTENTSABSTRACT ^TABLE OF CONTENTS ^  ivLIST OF TABLES LIST OF FIGURES ^  viiLIST OF APPENDICES viiiLIST OF SYMBOLS ^  ixACKNOWLEDGEMENTS  xi1.0 INTRODUCTION  ^12.0 METHODS  ^52.1 Selection of Biogeoclimatic Units for Sampling  ^52.2 Stand Sampling  ^92.3 Soil Sampling and Chemical Analysis ^  142.4 Foliar Sampling and Chemical Analysis  152.5 Data Analysis ^  163.0 RESULTS AND DISCUSSION  243.1 Site Index and Soil Nutrients ^  243.2 Site Index and Foliar Nutrients  483.3 Vegetation Analysis ^  563.4 Relationships Between Variable Domains ^ 604.0 SITE QUALITY ASSESSMENT FOR RED ALDER ^  655.0 SUMMARY ^  706.0 LITERATURE CITED ^  737.0 APPENDICES ^  79vLIST OF TABLES1. Mean climatic data for southern coastal biogeoclimaticunits (precipitation in mm, temperature in °C) ^ 72. Minimum, maximum, mean and standard deviations (s.d.) forsoil nutrient concentration data ^  243. Correlations for site index and soil nutrient variables ... 254. Mahalanobis distances (D 2) calculated from soil nutrientvariables and associated significance levels for 37 stands.SALT and DEER are significantly different at thep=.05 level ^  305. Eigenvalues, cumulative % variance, component loadings, andcommunalities for four rotated components extracted from thecorrelation matrix of soil nutrient variables ^ 316. Regression analysis of site index and soil nutrients using acategorical variable based on cluster membership from a PCAanalysis ^  357. Regression analysis of site index and soil nutrients using acategorical variable based on cluster membership from a PCAanalysis ^  358. Minimums, maximums, means and standard deviations (s.d.) forsoil nutrients which statistically differentiate soil reactiongroups ^  379. Regression analysis of site index and soil nutrients usingstands with lower pH soils ^  4010. Regression analysis of site index and soil nutrients usingstands with higher pH soils  4011. Minimums, maximums, means and standard deviations (s.d.)for soil nutrient quantity data ^  4212. Correlations for site index and soil nutrient quantityvariables ^  4213. Eigenvalues, cumulative % variance, component loadings, andcommunalities for four rotated components extracted from thecorrelation matrix of soil quantity variables ^ 4414. Regression analysis of site index and soil nutrientsquantities using a categorical variable based on clustermembership from a PCA analysis of soil concentrations ^ 44viLIST OF TABLES (continued)15. Regression analysis of site index and soil nutrientquantities using a qualitative variable based on clustermembership from a PCA analysis of soil concentrations ^ 4516. Minimums, maximums, means and standard deviations (s.d.) offoliar nutrient concentration variables ^  4817. Correlations for site index and foliar nutrientvariables ^  4918. Mahalanobis distances (D2) calculated from foliar nutrientvariables and associated significance levels (p) for 37 stands.Three stands SALT, QUAD and GVWD are significantly differentat the .05 level^  5119. Eigenvalues, cumulative % variance and loadings for fiverotated components extracted from the correlation matrix offoliar nutrient variables ^  5220. Regression analysis based on all subsets of variables forfoliar nutrients versus site index^  5221. Regression analysis of site index and foliar nutrients usinga categorical variable based on cluster membership from a PCAanalysis of soil nutrient variables ^  5322. Regression analysis of site index and foliar nutrientsfor stands with higher pH soils  5423. Means, standard deviations (s.d.), and coefficients ofvariation (c.v.) for foliar nutrients sampled from 15 aldertrees from the Lund stand with number of samples required toestimate the mean within x% at y% confidence ^  5524. Diagnostic species determined from four site index classes 5825. Diagnostic species determined from soil reaction groups ^ 5926. Tentative classification of alder vegetation associations 5927. Correlations between foliar and soils nutrients ^ 6028. Results of canonical correlation analysis between the soilnutrient and foliar nutrient domains ^  6129. Results of canonical correlation analysis between the formand foliar nutrient domains ^  6330. Regression analysis of site index and soil physical andmorphological variables with elevation ^  64vi iLIST OF FIGURES1. Locations of 37 stands sampled in this study ^ 102. Numbers of stands by site index and site indexclasses ^  123. Sequence of steps in the analysis ^  184. The first two components of a PCA analysis on soilnutrient variables. Two stands, SALT and DEER need to betested as possible outliers. The other stands are plotted bytheir first letter ^  275. Andrews' plots for different subsets of the data:(a) all stands; (b)SALT and DEER stands omitted;(c) cluster 3 stands omitted; (d) pH high soils;(e) pH low soils ^  286. Normal probability plot of D 2 distances showing thatmultivariate normality for all stands can be assumed by theapproximate straight line. The points are plotted as the firstletter of each stand but this is not diagnostic ^ 297. Ordination of stands along the first two axes from aPCA analysis of soil nutrient variables and componentrotation. The axes are broken by lines determined fromminimum variance criteria. The first component reflects agradient in organic matter content, the second in soilreaction ^  348. The first two components of a PCA analysis on foliarnutrient variables. Stands are plotted as their firstletter except SALT as an outlier. Not all outliers appearfrom these axes ^  509. Site quality assessment for red alder ^  66viiiLIST OF APPENDICES1. Stand, site and soil properties for 37 stands of red alder. 792. Program to calculate Noy-Meir's partitioning ofcomponent axes ^  80Program to calculate Andrew's plots ^  823. List of plant species ^  84LIST OF SYMBOLS1. Tree form variablesDBH^diameter (cm) at breast height (1.3 m)DH ratio of diameter to tree height (m)DLC^diameter (cm) at the base of the live crownCHT height (m) to live crown2. Stand variablesAGE^total stand ageNS number of alder stems (hat )VOL^total whole stem alder volume (m3ha 1 )MAI mean annual volume increment (m3ha -lyr-1 )BA^total alder basal area (m2ha -1 )SI site index (m at 25 years)3. Physiographic variablesELE^elevation (m)ASP aspectSL^slopeS slope positionBG^biogeoclimatic subzoneP parent material4. Soil physical and morphological variablesBD^bulk density (g cm3 )CF coarse fragment content by weight (%)AH^Ah horizon depth (cm)G gleyed soil: soils with prominent mottling within thesurface 30 cmTEX^soil textureSAND (%)SILT^(%)CLAY (%)ixLIST OF SYMBOLS (continued)5. Soil nutrient variablesconcentrations^ quantities - kg ha -1PHH2O^pH in waterPHCACL pH in calcium chlorideTCC^total carbon (%)^ TCQTNC total nitrogen (%) TNQMINNC^mineralizable nitrogen (ppm)^MINNQPC^available phosphorus (ppm) PQTSC total sulfur (%)^ TSQASC^available SO, sulfur (ppm) ASQCAC exchangeable calcium (meq/100g)^CAQMGC^exchangeable magnesium (meq/100g) MGQKC exchangeable potassium (meq/100g)^KQCUC^available copper (ppm)^ CUQZNC available zinc (ppm) ZNQFEC^available iron (ppm) FEQMNC available manganese (ppm) MNQ6. Foliar nutrient concentrationsFN^nitrogen (%)FP phosphorus (%)FCA^calcium (%)FMG magnesium (%)FK^potassium (%)FS total sulfur (%)FCU^copper (ppm)FZN zinc (ppm)FFE^iron (ppm)FMN manganese (ppm)FB^boron (ppm)FAL aluminum (ppm)FAS^available S0,-sulfur (ppm)FAFE active iron (ppm)xiACKNOWLEDGEMENTSI am grateful to my thesis committee members, Drs. TimBallard, Peter Marshall, Karel Klinka and Hans Schreier for theirhelp and guidance in preparing this thesis.I thank Doug Hopwood, Chris Ferris and Harry Williams whovery ably carried out the majority of the stand descriptions andstem analysis in the field. Thanks also to Farida Bishay andCindy Durance who made the disc measurements. Pacific SoilAnalysis Inc. carried out the soil and foliar nutrient analyses.I also thank my colleagues at the Forest Service for theirassistance, particularly George Shishkov for his help duringstand reconnaissance and sampling, and Fred Nuszdorfer who asForest Sciences Officer was very supportive of this study.Many individuals from government and industry providedinformation about suitable alder stands for sampling. Iparticularly thank Bill Beese, MacMillan Bloedel Ltd., WoodlandsServices Division, and those divisional foresters who madeavailable the majority of stands for this study.My wife, Nuria, and sons Eric and Patrick were alwaysencouraging and understanding while Dad had to spend yet anothersunny weekend at UBC.This study is part of a project funded by the Canada-BritishColumbia Forest Resources Development Agreement, 1985-90 BacklogReforestation Program, Research and Development Sub-program,Project No. 2.54 (red alder component).11.0 INTRODUCTIONRed alder (Alnus rubra Bong.) is a fast growing hardwoodspecies primarily with a coastal distribution in the PacificRegion of Western North America. Its latitudinal range is from60°N in southeastern Alaska to 34 °N at Santa Barbara, California.It is a lowland to submontane species which is frost sensitiveand therefore generally occurs below 750m in B.C. (Krajina et al.1982). The most productive growth of alder in southwestern B.C.is within the Coastal Western Hemlock (CWH) biogeoclimatic zone,and within the very dry, dry, and very wet maritime subzones. Redalder occurs in other biogeoclimatic subzones either outside thestudy area or growing under climatic regimes where it cannotrealize its best growth.From a site standpoint, the best growth of red alder isoften seen on moist and rich soils characteristically, though notexclusively, of alluvial origin. In these riparian habitats,alder is often associated with brushy understory species such assalmonberry (Rubus spectabilis)(Franklin and Dyrness 1973). It isthese sites which provide the greatest challenge to foresters whowish to convert them to coniferous stands. The soil disturbancecreated by logging favours the establishment of alder, which is aprolific producer of light, widely dispersed seed. Many thousandsof stems per hectare become rapidly established and effectivelyshade out other species, either planted or natural. This is2particularly critical for Douglas-fir, a shade intolerantspecies, the best growth of which has been identified on thesesites in B.C.Past interest in red alder has come from varied and notalways compatible origins. Foresters, for the most part, havefocused on eliminating it from competing with coniferousplantations. It grows faster than any crop species in the PacificNorthwest (apart from black cottonwood) and thus is a strongcompetitor for light and moisture. Total eradication was theobjective but recently it has become practice to leave 50-100alder trees per hectare as part of a silvicultural prescription.This derives from alder's nitrogen fixing ability (50-200 kg Nha -lyr-i) (Hibbs and Cromack 1990), its resistance to rootpathogens, and its ability to increase soil organic matter(Tarrant and Trappe 1971). The current objective to try tomaintain a biologically diverse habitat in the forest as opposedto plantation monocultures also favours the retention of alder.Other interests have focused on alder's fast growth in shortrotation management for biomass or pulp (DeBell et al. 1978).Most recently, efforts have been directed towards developing agrowth and yield database for managing alder on sawlog rotations.This has been initiated by the Hardwood Silviculture Cooperativebased at Oregon State University and includes members fromgovernment and industry in the Pacific Northwest (HSC 1991). One3conspicuous deficit in our knowledge of alder is growth and yielddata, probably because alder inventory has always been greaterthan its utilization. Based on its growth and yield, asilvicultural regime can be developed for forest managers whoseobjective is to manage alder as a crop species.In order to meet these objectives, we also need to know howalder grows in relation to site (site being defined as thecomplex of factors which influence growth at a particularlocation). Studies of this type are commonly referred to as site-productivity studies and usually conclude with a statement aboutthe relationships between ecological site quality and an estimateof forest productivity. The most commonly used estimate of forestproductivity is site index which is the measure of the height ofdominant trees in a stand at a specified reference age.The major objective of this study is to investigate therelationships between red alder site index and ecological sitequality in a way compatible with what has already been done bysuch workers in B.C. as Green et al. (1989) and Klinka and Carter(1990). Their work focused on the three basic elements ofecological site quality, climate, soil moisture regime, and soilnutrient regime. Guides have been developed and used by fieldforesters in the Vancouver Forest Region to characterize sitesince 1977. They are based on assessments of ecological sitequality estimated from soil and physiographic variables and plant4indicator species within the limits of a biogeoclimatic subzone.They have evolved from the many studies carried out by V.J.Krajina and K. Klinka and their cooperators and provide anessential link between operational forestry and forest ecosystemdescription.In use, a qualitative assessment is an heuristic approachbecause the mechanism of how site factors (and theirinteractions) influence forest productivity is not addressed.Ideally, it has a quantitative basis which is theoreticallyderived from the measures between site factors. Both approachesare valid and complementary. This thesis proposes a qualitativeapproach to the site assessment for red alder based on aninvestigation of the quantitative relationships between sitefactors.Another objective was to test the hypothesis that tree formvaried with site. During stand reconaissance, tree form,especially crown length, appeared to change with site. Thisdimension, as well as other tree form variables were analyzedagainst other site factors.52.0 METHODS2.1 Selection of Biogeoclimatic Units for SamplingWithin the geographic distribution of red alder (includingAlaska, Idaho and California), mean annual precipitation rangesfrom 400 to 5600 mm and temperature extremes are from -30 ° to46 °C (Harrington 1990). These ranges are narrower in B.C. At aninitial stage of selecting alder for forest management, it wouldbe rational to suggest areas where alder might realize its bestproductivity on a climatic basis. This involved scrutinizing theclimate of southwestern B.C. with respect to the distribution ofred alder.The climate of B.C. can be conveniently partitioned usingbiogeoclimatic units (Krajina 1969). Although the boundaries ofthese units are not climatically derived and may not coincidewith climatic isolines, climatic change is reflected by thecharacteristic vegetation of the unit (Pojar et al. 1987). Redalder is a dominant component within two biogeoclimatic zones,the Coastal Douglas-fir (CDF) and Coastal Western Hemlock (CWH)zones.Ideally, in order to identify the best subzones for aldergrowth in B.C. a comparison of productivity and climatic factorsshould be made (Christie and Lines, 1979). Although yield tables6do exist for red alder there are no data on local yield whichmight be related to regional climate. Furthermore, the long termclimatic station network in B.C. is insufficient for thispurpose. Stations on the west coast of Vancouver Island andhigher elevation stations are particularly lacking.For the purposes of this study, climatic variables are usedas accessory characteristics to provide a background for thebiogeoclimatic units. Consequently, any evidence for selectingthe best subzones based on climate variables for the growth ofred alder is conjectural.Table 1 gives mean climatic data for those southern coastalbiogeoclimatic units where red alder grows. The climate summariesby biogeoclimatic subzone were derived from the published normals(Canadian Climate Program, 1980). The climate stations wereclassified into biogeoclimatic units according to the revision byKlinka et al. (1991).Table 1. Mean climatic data for southern coastal biogeoclimaticunits (precipitation in mm, temperature in °C).7index of continentalityCDFmm CWHxm CWHdm CWHmm CWHvm CWHvh20.1 38.6 52.8 45.4 74.5 95.8166.1 250.8 291.7 400.4 436.1 431.0198.7 363.0 498.0 470.0 752.4 899.6925.2 1505.3 1827.1 2349.3 2787.4 2951.53.2 1.8 1.9 -2.2 0.3 3.016.7 17.0 17.6 14.1 16.0 13.99.8 9.3 9.8 5.7 8.2 8.20.0 0.0 0.0 2.5 0.7 0.05.6 5.4 5.7 3.9 4.8 4.810.1 13.7 14.7 15.9 13.7 2.4precipitation of thedriest monthprecipitation of thewettest monthprecipitation April-Septembermean annualprecipitationtemperature of thecoldest monthtemperature of thewarmest monthmean annualtemperaturenumber of months withtemperature < 0 °number of months withtemperature > 10 °The CDFmm is distributed on the east side of VancouverIsland, the Gulf Islands and along the Sunshine coast. It rangesfrom sea level to about 500m. It is the driest subzone on thecoast with July and August being the driest months with a meanmonthly precipitation of 20mm, but this can be as low as 13mm.8This would limit the productivity of alder since it is nottolerant of moisture deficits. Although a moist site conditionmay alleviate climatic drought, it is still likely to becritical, especially at the initial phase when alder plantationsare likely to be established on scarified sites.The CWHmm is a montane subzone on the east side of VancouverIsland, that ranges in elevation from as low as 200m, but mostly400 to 800 m. Although growing season precipitation is probablyadequate for alder (470 mm), temperature is likely to be aconcern. Alder predominantly grows in a mesothermal climate whereaccording to Kiippen's climatic classification, the temperature ofthe coldest month is greater than -3 °C. This subzone approachesthis limit. In addition, it has a greater number of months <0 °Cand fewer months >10 °C. These are also criteria used by KOppenwhich relate vegetation to climate (see the description ofKOppen's classification in Trewartha (1982)). There are also somephysiographic limitations of this subzone for the use of alder.There are fewer alluvial sites in this montane subzone and thelimit of isostatic rebound was about 200 m, precluding any marinesoils here. These are probably the most productive parentmaterials for the growth of red alder.The CWHvh subzone is distributed along the west coast ofVancouver Island, where it extends inland a few km, and acrossits northern portion. There are no precipitation or temperature9limitations to red alder growth here. This is a very wet climatewith an equable temperature regime. The index of continentality,which is the difference of January and July mean temperaturesadjusted for latitude, is the lowest on the coast. However, theactual evapotranspiration (AE) here is likely to be low also.Major (1963) has used AE as an index of plant productivity,suggesting that higher AE's result in higher productivity. Thelow AE in this subzone results from cool temperatures and lowerradiation intensity from frequent cloud and fog.It is suggested that the three subzones discussed previouslymay have some climatic limitations for the best growth of redalder. The remaining three subzones in Table 1, CWHxm, CWHdm andCWHvm are less likely to have these limitations and were used insampling stands of red alder for this study.2.2 Stand SamplingThe reconnaissance of alder stands was carried out withinthe above selected subzones of the CWH zone. Stands were chosento represent a range of productivity. During reconnaissanceHarrington and Curtis' (1986) site index equation was used toestimate productivity. Stands had a minimum 80% alder componentbased on basal area with no large gaps in the canopy, age from 20years or older, and site homogeneity. Sites had to be homogeneousin terms of parent materials, slope, aspect, and understory10effectively limited elevation range. The geographic distributionof the stands was limited by access on Vancouver Island and theadjacent mainland (Figure 1). Appendix 1 lists the physiographicand biogeoclimatic data for the stands.Figure 1. Locations of 37 red alder stands sampled in this study.11Field sampling was conducted from July through September1989. Ecosystem description of the alder stands conformed to theprocedure described in Klinka et al. (1984) and Green et al.(1984). Plots were 400 m2 in area and either square orrectangular in shape. All vegetation growing on mineral soil orthe forest floor was recorded according to the significance scale(Klinka et al. 1984). Two soil pits were dug at subjectivelychosen locations representing modal conditions in microtopographyand surface features. Depressions and accumulations of decayingwood were avoided. Soils were pedogenically described andtaxonomically classified according to the Agriculture CanadaExpert Committee on Soil Survey (1987). Humus forms wereclassified after Klinka et al. (1981).All trees in the plot greater than 7.5 cm diameter outsidebark at breast height (1.3 m) were measured for diameter (DBH).Three dominant site trees, free of any visible damage, werechosen for stem analysis. Stem analysis involved sectioning thetree at ground level, 0.7 m, 1.3 m and thereafter at 1 mintervals to 15 m and thereafter at 2 m intervals to the top.These data were used in another study to develop height growthcurves for red alder. Total height, DBH, height to the base oflive crown (CHT), height to a fork (FHT), and diameter at thebase of the live crown (DLC) were determined from these data. Thebase of the live crown had living branches in at least threequadrants. Lean (LEAN) was measured as the angle (deg.) fromsite index classeslow^poor medium^good2I 112^3 41112vertical of the tree bole sighted at breast height along a linefrom ground level to 3m tree height. Site index (SI) for the plotwas determined from linear interpolation from the stem analysisdiscs and is the average of SI for the three site trees. Thereference age for site index was 25 years, total age. The rangeof site index was from 12.5 to 25.3 m and stands were groupedinto site index classes for comparison with other variabledomains (Figure 2).12 13 14 15 16 17 18 19 20 21 22 23 24 25 26site index (m/25yrs)Figure 2. Numbers of stands by site index and site indexclasses.FHT and LEAN were later deleted because major forks alongthe bole would have likely developed from past climatic damageand LEAN because lean in alder does not result in tension woodand consequent warping (Plank 1990). It was concluded that thesevariables would not be relevant in comparison with site13relationships. The variables diameter/height (DH), DLC, DBH andCHT were used as measures of tree form with stems ha -1 (NS)included to adjust for density.Tree volumes were calculated for the site trees from boltsresulting from stem analysis sections. The base bolt (groundlevel to 1.3 m) volume was calculated as a cylinder, the lastbolt (from the tip down to the last section) was calculated as acone and all other bolts were calculated according to Smalian'sformula (Avery and Burkhart 1983). Bolt volumes were summed fortotal tree volume. The correlation of these volumes compared tothe B.C. Forest Service whole stem volume equation for red alderas applied to site trees was 0.97 (B.C. Forest Service 1979).This was considered close enough to use the Forest Serviceequation for all plot trees. This equation is:volume (m3 ) = 10 (-4.431705+1.89057*LoglO(DBH)+1.09077*log10(height))Height for other than site trees was calculated for the aboveequation from a regression of height on dbh (Avery and Burkhart1983):height(m)=exp(3.6807-13.700+1/DBH).This equation had an R 2 of .60 and a standard error of .097. Thelowest DBH for site trees was 12.5 cm, so heights from trees with14DBH from 7.5 to 12.5 were extrapolated.2.3 Soil Sampling and Chemical AnalysisSoil chemical analysis was based on a composite sampling (15randomly located subsamples within the stand) from the surface 0-30 cm of the mineral soil. Soil pH was measured with a pH meterusing a 1:1 suspension in water and .01 M CaC1 2 . Total C wasmeasured using a Leco Induction Furnace (Bremner and Tabatabai1971). Total N was determined by semimicro-Kjeldahl digestionfollowed by colorimetric estimation of NH 4-N using a TechniconAutoanalyzer (Anonymous 1976). Mineralizable-N was determined byan anaerobic incubation procedure modified from Waring andBremner (1964). Released NH 4 was determined colorimetricallyusing a Technicon Autoanalyzer. Available phosphorus was measuredcolorimetrically using ascorbic acid reductant of the molybdatecomplex using the extractant of Mehlich (1978) on a 1:10soil:solution extract. Available SO 4-S was extracted withammonium acetate (Bardsley and Lancaster 1965); extracted sulfatewas reduced to sulfide by HI and then determined colorimetrically(Kowalenko and Lowe 1972). Exchangeable K, Mg, and Ca wereextracted with 1 M sodium acetate (pH 7) (Greweling and Peech1960). The displaced cations were analyzed by atomic absorptionspectrophotometry (Price 1978).Soil nutrient variables were expressed as concentrations on a15dry mass basis and converted to a kg ha -1 basis using bulkdensity (based on 10 samples per plot) corrected for coarsefragment content.2.4 Foliar Sampling and Chemical AnalysisFoliar samples were collected from the last week in Augustto the second week in September. Leaves were clipped from atleast three branches at the top third of the crown. Approximately40 g fresh weight of leaves was taken from each of 15 trees andcomposited for analysis. At one stand (Lund) the 15 samples wereanalyzed separately to provide a measure of sample sizerequirement for population estimates. The sample size wasiteratively determined from:n = t2 * cv2aezwhere t is the value from the Student's t Table, cv is thepercent coefficient of variation and ae is the percent allowableerror.The leaves were dried at 60 °C for 24 hours and ground with acoffee grinder. One gram samples were digested with H 2SO4-LiSO4-Se-H202 (Parkinson and Allen 1975). N and P were determinedcolorimetrically using a Technicon Auto-Analyzer andspectrophotometer, respectively. K, Ca, Mg, Mn, and Al weredetermined by atomic absorption spectrophotometry (AAS). Samples16were dry ashed for Cu, Zn, and Fe followed by AAS. B was dryashed and determined colorimetrically using the azomethine-Hmethod (Gaines and Mitchell 1979). Total S was determined using aLeco Sulfur Analyzer and available SO 4-S was extracted withboiling 0.01N HC1 and determined colorimetrically on a HI-Bismuthreducible distillate. Active Fe was extracted using 1N HC1(Oserkowsky 1933) and analyzed using AAS.2.5 Data AnalysisThe approach to analysis considered six data domains: a sitedomain containing site index as the only variable; a soil domainconsisting of morphological and physical properties; a soildomain consisting of chemical properties, and vegetation, foliarand form domains. A list of variables for all except thevegetation domain is given in the list of symbols (page ix). Alist of species is given in Appendix 3.The main body of the analysis was divided as being variable-dependent or stand- (i.e. or case-) dependent according to theobjective. That part of the study concerning site indexprediction using regression was considered to be variable-dependent. If the objective was to examine structure in the data,then the analysis was considered case-dependent. Data structurerefers to pattern in the data as opposed to noise. Patternoccurred if groups of stands showed similar responses within17variable domains. If meaningful relationships were not found withsite index versus a domain, then the data structure was examinedto determine if clusters of stands could reduce variance byapplying a categorical variable based on some variational breakin the data (Chatterjee and Price 1977). To this extent thedependence of the analysis on variables or stands wasinterchangeable. The approach is outlined in Figure 3.The all subsets approach was used in SI regression modelling(Chatterjee and Price 1977). Variable interactions were sometimesincluded if they improved the model.Most of the objectives of the analysis lent themselves tomultivariate rather than univariate analyses. In general,multivariate techniques consider the covariances betweenvariables which is advantageous for event prediction usingecological data characterized by multiple determinism (Kimmins1987). In other words, it is likely that several to many factorsare contributing to an outcome.or ENDyesCHARACTERISTICS OF CLUSTERS - PCA loading, MANOVAdescription and subjective input^• go to SYNTHESISARE CLUSTERS MEANINGFUL?nogo to SYNTHESISgo toSYNTHESISVARIABLE DEPENDENT STAND DEPENDENTORDINATION - PCA, TABULARgraphical depiction, stand distances, outliers?variable transformations and/or reductionunivariate and multivariate normalityDO STAND CLUSTERS RESULT?SITE INDEX PREDICTIONREGRESSION(categorical variables)REGRESSIONS MEANINGFUL?data toovariablefor domain,further samplesrequired...ENDyesTRY SITE INDEX PREDICTION AGAIN USING SUBSETSOF DATA BASED ON CLUSTERSyesyes—nSITEFOLIARnutrient^I'concentrations'SOILchemicalconcentrations& quantities1 FORMSITE QUALITY ASSESSMENSOILmorphological& physicalVEGETATIONData domains:SITE SOILmorphological& physicalSOILnutrientconcentrationsVEGETATION FOLIARnutrientconcentrationsFORM& quantitiesAnalysis:Synthesis:18Figure 3. Sequence of steps in the analysis.19Ordination is a technique that enables structure to bevisualized because it concentrates variance on fewer variables(components) and reduces noise (Pielou 1984). Stand ordinationwas carried out by various methods depending on the domain.Principal components analysis, PCA, (Gittins 1979) was most oftenused. It is the simplest of many ordination techniques and isadequate if the data show only linear trends and if Euclideandistance is appropriate as a distance measure (Gower 1986). Theconvention of accepting those components with eigenvalues greaterthan one from the correlation matrix was adopted (Legendre andLegendre 1983). It was also expected that the first fewcomponents would explain most of the variance (at least 75%);otherwise this technique might not have been useful forsummarizing the data (Kendall 1980). Non-linear ordination wasnot tried but might have provided useful results.As an aid in the interpretation of PCA, there are severalmethods of rotating the components to new positions whichconcentrate the loading of a variable on fewer components.Loadings are the correlations of the variables with components.The method used here was an orthogonal varimax rotation. Webster(1977) gives an example of this for soil chemical data.The detection of clusters in the data can be seen on thefirst two axes of the ordination, or the first three. Clusters of20plots may be visually obvious but there are objective ways ofpartitioning the axes. Noy-Meir's (1973) criteria of minimumvariance was adopted. A FORTRAN program to do this was writtenand is listed in Appendix 2.With more than three dimensions, Andrews' plots can be used(Andrews 1972, Everitt 1978). In these plots, variables arecoefficients in a wave function of the form:F (t) =x i/sqrt (2) +x2sin (t) +x3cos (t) +x4sin (2t) +x5cos (2t) . . . ,where x 1 — .xn are the n variables used and the function isplotted for values of t from -r to r. It was used here with theprincipal component scores as the values for the x variables. AFORTRAN program to graph Andrews' plots using the DISSPLA package(ISSCO 1985) is listed in Appendix 2.Plotting ordinations also served in the detection ofoutliers which can be thought of as single member clusters.Particular attention was paid to the detection of outliersbecause of their effect on regression (Chatterjee and Price 1977)and ordination (Pielou 1984). However, some testing of thesignificance of an outlier should be made before declaring itone. If multivariate normality can be assumed then thesignificance of an outlier is based on its Mahalanobis distance(D2) from the group centroid. Mahalanobis distances are21calculated as:D2 i = (X i -x) S -1 (X.-x)where S is the covariance matrix, x is the vector of meanvariable scores and x i is the vector of scores for case i. If thenormal probability plot of ordered D 2 distances is not a straightline then univariate normality was checked and adjustments madeas necessary. This involved log o and square root transformationsfor skewness and (1/2)log 1o ((x+1)/(x-1)) transformation forplatykurtosis. Although PCA does not assume multivariatenormality, the significance testing of correlations betweencomponents and variables does, as does the use of multivariateanalysis of variance (MANOVA).MANOVA has several advantages over univariate analysis ofvariance (Stevens 1969), one of which is that correlationsbetween variables are accounted for. A significant result fromMANOVA was succeeded by univariate t tests for each variable.The analysis of vegetation used two approaches. Indicatorspecies analysis was based on the indicator groups and the matrixof species x attributes defined by (Klinka et al. 1989). Mostspecies in the vegetation domain contributed to the analysis byhaving attributes based on one of five indicator groups: lifeform, climate, soil moisture and nutrient regime, and groundsurface materials. Each group has classes representing unique22attribute values. For example, the nutrient regime group hasthree classes which recognize species as being indicative of N-poor, N-medium or N-rich conditions. More information can befound in Emanuel (1989). For each stand a percentage contribution(after arcsine transformation) of species to a given class wereused as independent variables with SI as the dependent variable.Green et al. (1989) used this approach for estimating Douglas-fir site index.The second approach was based on the frequency of species interms of presence classes and their mean significance values.Diagnostic criteria l determined the usefulness of variousgrouping of stands.The relationships between data domains were investigatedusing canonical correlation analysis, CCA. N pairs of canonicalvariates are formed where N is the lesser of the number ofvariables from each domain. The first pair consists of onevariate from each domain which has the highest correlation, thenext pair have the next highest correlation between domains andare also orthogonal to the first, and so on. Gittins (1985)describes CCA using both soil and vegetation examples.The third stage of the analysis served to synthesize1Presence classes, significance values and diagnostic criteria are defined in Table 33 and based onPojar et al. 1987.23relationships between domains indicated at the analysis stage.The last section of Figure 3 details the network between domains.A connection between domains was made because it provided auseful and significant relationship. Of special interest was thedescription of relationships between three domains used in fieldsite quality assessment: site, soil physical and morphological,and vegetation.Most of the analysis was done using the BMDP (Dixon 1981)and MIDAS (Fox and Guire (1976) statistical packages. Vegetationtabling was based on the program written by Emanuel (1989)available on the UBC mainframe computer. Any major programswritten for this study are listed in Appendix 2.243.0 RESULTS AND DISCUSSION3.1 Site Index and Soil NutrientsDescriptive statistics for the soil nutrient concentrationdata are given in Table 2. PHCACL was used to represent pH, as itwas highly correlated with PHH2O (making one redundant), andbecause it provided a better cutpoint for the clusters that werelater recognized.Table 2. Minimums, maximums, means and standard deviations (s.d.)for soil nutrient concentration data.variable^ code minimum maximum mean s.d.pH (H2O) PHH2O^4.0^5.7^4.5^0.38pH (CaC1 2 )^ PHCACL^4.0 5.5^4.4^0.35total C (%) TCC^0.95^14.1^5.6^2.75total N (%) TNC 0.07^0.78^0.30^0.14mineralizable N (ppm)^MINNC^17.0^150.^64.3^35.82P (ppm)^ PC^3.0 30.^9.0^5.05Total S (%) TSC .007^0.13^.044^.025available SO4-S (ppm)^ASC^0.3 5.8^1.5^1.09exchangeable Ca (me/100g)^CAC 0.08^25.0^4.9^5.43exchangeable Mg (me/100g)^MGC^0.1 5.8^0.8^1.04exchangeable K (me/100g)^KC 0.05^0.51^0.12^.076available Cu (ppm)^CUC^0.3 6.2^2.2^1.69available Zn (ppm) ZNC 0.5^13.0^2.6^2.01available Fe (ppm) FEC^2.0^175.^42.4^33.98available Mn (ppm)^MNC 20.0^165.^56.4^30.25The correlations between site index and the soil nutrientvariables were low. PC had the highest correlation at .309 (Table3). Efforts to use these variables to form significant25relationships with site index were not successful. Insignificantcoefficients resulted from regression analysis of various modelswith standardized residual values of +or- 6.0. indicating highvariability. Thus, soil nutrient variables (either separately orin combination) from the 37 stands were not sufficient predictorsof site index.Table 3. Correlations for site index and soil nutrient variables.PHCACL .006TCC .036 .430TNC .125 .468 1.000MINNC .091 .421 .827 .776PC .309 -.050 .010 .096 .082TSC .159 .466 .718 .872 .641 .147ASC -.039 .201 .276 .385 .106 .185 .387CAC -.004 .823 .719 .746 .706 .100 .635 .382MGC -.156 .663 .369 .384 .522 -.043 .309 -.013 .683KC .168 -.026 .299 .259 .406 .616 .092 -.052 .222 .155CUC -.006 -.175 .271 -.293 -.208 -.027 -.292 -.163 -.247 -.238 .031ZNC .125 -.113 .073 .046 .110 .679 -.079 -.072 .043 .100 .796 .167FEC -.131 -.573 .549 -.542 -.484 .061 -.530 -.010 -.447 -.283 .161 .274 .058MNC .159 .046 .510 .319 .553 .028 .217 -.154 .222 .067 .420 .149 .294 -.228SI PHCACL TCC TNC MINNC PC TSC ASC CAC MGC KC CUC ZNC FECHigher correlations among the soils variables in Table 3indicated that the data might be summarized with fewer variablesand that clusters might result. Consequently, a PCA resulted in 4eigenvalues which were greater than 1.0 and explained 75% of thetotal variation.The plot of the first two axes (Figure 4) indicated twopossible outliers, DEER and SALT, that warranted furtherinvestigation. It was also shown by Andrews' plots that these26(and another stand - COOM) have different waves based on the fourPCA components compared to the rest of the stands. Figure 5demonstrates the visual aid of Andrews' plots in multidimensionalplotting. These were used with D2 distances to determineoutliers. Prior to testing whether the outliers weresignificantly different from the remaining data, transformationswere made for multivariate normality. The normal probability plotof D2 distances approximated a straight line (Figure 6). Thecalculated D2 distances reveal that the DEER and SALT plots aresignificantly different for these variables at p=.05. Thedistances and significance levels are given in Table 4. Outliersare sometimes reluctantly omitted from an analysis because it isnever certain whether they are representing extreme values in thepopulation. They often occur in small samples such as this. Inthe interests of trying to develop significant regression models,these stands were be omitted to reduce variability.27^8.1086+^ DEER5.9412+o 3.7739+^2 1.6065+^E^ CKS C G+G APMSMH KB^H^L CB W ZSM C L-0.5608+^RPP^L^B-2.7282+SALT-4.3421^0.1829^4.7079-2.0796^2.4454^6.9703component 1Figure 4. The first two components of a PCA analysis on soilnutrient variables. Two stands, SALT and DEER need to be testedas possible outliers. The other stands are plotted by their firstletter.4 .0 - 3.0^- 2.0^- 1.0^0.0^I r.()^2. 0- - pt i o 'pt.3.0 .n(b)9▪ - 4.0^- 3.0^- 2.▪ 0^-1.0^0.0^1.0^2.0^3.0^4.040 1 [7140 -3.0^-2.0^-1.0^0.0^i.o^2.0^3.0^4.01^p^ In^4 r ),(a)- 4.0 -3.0^-2.0^- 1.0^0.0^1.0^2.0^3.0^.0I^-^ ' r,,28 (C)oO- 4.0 - 3.0^- 2.0^-1.0^0.0^1.0^2.0^3.0^4.0I. -^ ' r((d)^ (e)Figure 5. Andrews' plots for different subsets of the data: (a)all stands; (b)SALT and DEER stands omitted; (c) cluster 3 standsomitted; (d) pH high soils; (e) pH low soils.1.0+0.8+MCCS0.6+GSKB0.4+EPMH0.2+^LBP+ AW+ GC+KZ0.0+ZBLCCHBDS290.0^0.2^0.4^0.6^0.8^1.0Figure 6. Normal probability plot of D 2 distances showing thatmultivariate normality for all stands can be assumed by theapproximate straight line. The points are plotted as the firstletter of each stand but this is not diagnostic.30Table 4. Mahalanobis distances (D2) calculated from soil nutrientvariables and associated significance levels for 37 stands. SALTand DEER are significantly different at the p=.05 level.plot D2 p plot^D2 p plot D2 pCCK2 7.018 .9340 MAGI^9.637 .7882 ZEB2 5.496 .9776CCK1 14.179 .4365 LOWR^8.364 .8695 AIRS 7.186 .9273BAIN 17.828 .2147 KENE^5.489 .9778 WOSS 7.245 .9249COR1 16.103 .3071 HQRG 20.306 .1208 SALT*33.581 .0024COR2 21.394 .0919 HQRP^9.564 .7933 DEER*26.944 .0196ELKM 10.188 .7483 BIG2 13.409 .4946 GVWD 13.170 .5132SARI 13.297 .5033 BIG1 20.196 .1241 LUND 11.998 .6065BTLK 8.591 .8563 SNOW 14.345 .4243 PENM 11.507 .6459KLA1 13.555 .4834 MENZ 15.618 .3373 PENR 10.298 .7401KLA2 15.458 .3476 QUAD 16.318 .2943 GOLD 6.698 .9457NITI 12.369 .5767 ROBL^7.746 .9021 SQUA 13.680 .4738LCOW 17.897 .2115 PREN^8.653 .8526COOM 20.421 .1174 ZEB1 18.251 .1956Therefore, a second PCA was run omitting these stands. Foureigenvalues were greater than 1.0 and accounted for 75% of thevariance. The components were varimax rotated to improve theirinterpretability. This resulted in the ordering of variables bycomponent loading in Table 5. The interpretation of thesecomponents is based on the magnitude of component loadings(correlations between the components and the variables). Stevens(1986) pointed out that studies suggest the critical values ofcorrelation coefficients should be increased 150-200% because ofthe orthogonal relationship of the components. The critical valueat p=.05 and df=33 is .334, times 175% is .584. Values greaterthan this probably indicate that the correlations aresignificantly different from zero. The loading on the firstcomponent is especially high for TNC, TCC, and MINNC and31secondarily for TSC, MNC and KC. FEC contrasts with thesevariables on this component. This can be interpreted as anorganic matter component, as the amount of C, N and S in soils isusually associated with soil organic matter content. The secondcomponent shows high correlations with PHCACL, CAC and MGC. Thiscomponent is interpreted to reflect soil reaction.Table 5. Eigenvalues, cumulative % variance, component loadings,and communalities for four rotated components extracted from thecorrelation matrix of soil nutrient variables.Eigenvaluecumulative % variancevariable^(1)^(2)^(3)4.9644^2.8735^1.293735.5 56.0^65.2component loadings(4)1.351674.9communalityTNC .1638 -.0889 -.0732 .9241.9402TCC .9251 .1954 -.0135 -.0084 .8942MINNC .8359 .3540 .0047 .1214 .8388TSC .8139 .1733 -.1125 .2301 .7582KC .7267 .0430 -.0854 .0790 .5435MNC .6832 .1306 .4544 -.1589 .7155FEC -.5893 -.4337 .3062 .2299 .6820PHCACL .1210 .9374 -.0293 -.1091 .9060CAC .4601 .8296 .1062 .0312 .9122MGC .4336 .8228 .1320 .1572 .9071CUC -.1764 -.0497 .8168 -.0174 .7011PC -.0549 -.1523 -.3501 .7535 .7165ZNC .2696 -.0236 .2966 .7375 .7051ASC .0132 -.4195 .2409 .2114 .2789The third component uniquely represents CUC. The importanceof Cu as a gradient in the data is not known. As a micronutrient,32Cu is involved with various enzyme and co-enzyme systemsconnected with oxidation reactions and in electron transport inphotosynthesis. Copper is thought to be essential for N-fixationin Alnus. Bond and Hewitt (1967) stated that Alnus cautinosa seedlings grew better when supplied with a concentration of .02ppm Cu compared to a control. They also mention that the Cutreated plants had higher FN concentrations. In this study therewas no evidence of correlation between CUC and FN. Copperdeficiencies are not known to occur in our forest soils but thereare few data available (Heilman 1979).The last component represents both PC and ZNC. P is oftencombined in organic form in soils and thus is associated withorganic matter but the available P analyzed in this study(Mehlich's method) is not a severe extraction and is associatedwith dissolved and weakly adsorbed P; it does not recover much ofthe organic P. This serves to differentiate this component fromcomponent one. ASC does not contribute significantly on any onecomponent, its communality2 is only 28%.The stands ordinated by the first two components are plottedin Figure 7. Noy-Meir's (1973) partitioning procedure was used asan aid because the stands were not obviously separated. It breakseach axis from * to *. Based on this plot three clusters were2The communality of a variable is calculated as the sum of squared correlations for the extractedcomponents. It indicates the percent of variance of a variable explained by the components. Similarly, thepercent of a variable's variance on a component is the square of its correlation with that component.33recognized: 1 - quadrant IV plots, 2 - quadrant III plots plusPREN, KLA1 and NITI, and 3 - quadrant I plots plus COR1. Thus,clusters are recognized largely on the basis of the first twocomponents which reflect gradients of soil organic matter andsoil reaction. The three clusters were differentiated on thebasis of pH (CaC1 2) and %total carbon content as follows:cluster(1)^(2)^(3)pH (CaC1 2) range^4.4-5.1^4.0-4.4^4.4-4.8total carbon (%) range^2.6-9.0^.95-7.1^9.6-14.1.By clustering the stands, the relationship between siteindex and soil nutrients improved. In the first analysis, cluster3 was omitted from regression analysis because only three standsrepresented higher organic matter and higher soil reaction.Clusters 1 and 2 were analyzed in the regression model using acategorical variable, GROUP, based on their membership. GROUP wasset to 1 if the stand belonged to cluster 1 and 0 if the standbelonged to cluster 2. The results of this analysis (Table 6)include an interaction variable PCxCAC.2.6709+ COOM+++++1.8421+c +o +m +p + SARI ZEB1o +n 1.0133+ MAGIe + BAINKENEn + HQRGt ++ AIRS2 + ZEB20.1845*^*BIG1 LCOWLOWRCOR2*CCK2LUND11BTLK QUADIVI I34+ PENM WOSS III II COR1+GVWD CCK1+ SNOW+ ROBL HQRP+ PENR0.6444+ BIG2 GOLD+ SQUA+ KLA2 ELKM PREN++ MENZ KLA1+ NITI-0.4732+-2.0756^-0.3076^1.4603-1.1916^0.5763^2.3443component 1Figure 7. Ordination of stands along the first two axes from aPCA analysis of soil nutrient variables and component rotation.The axes are broken by lines determined from minimum variancecriteria. The first component reflects a gradient in organicmatter content, the second in soil reaction.35Table 6. Regression analysis of site index and soil nutrientsusing a categorical variable based on cluster membership from aPCA analysis.variable coefficient s.e. t sig.constant 7.887 2.96 2.66 .031PC 68.03 17.12 3.98 .000PCxCAC -583.69 175.65 -3.32 .003MNC 207.94 69.79 2.98 .006GROUP -5.501 1.52 -3.62 .001n=31 F=6.51 R2=.500 s.e.=2.36In a second analysis GROUP equalled 1 for all stands abovethe break in component 2 and 0 for all stands below it. Thisdisregarded the separation of clusters based on organic matter(Table 7). It was concluded that the organic matter component isnot contributing much to the regression. A third regression usingtwo categorical variables representing all three clusters was notsignificant.Table 7. Regression analysis of site index and soil nutrientsusing a categorical variable based on cluster membership from aPCA analysis.variable coefficient s.e. t sig.constant 8.315 2.79 2.97 .006PC 63.46 15.49 4.10 .000PCxCAC -445.76 120.16 -3.71 .001MNC 185.47 58.80 3.15 .004GROUP 4.604 1.20 3.83 .001n=35 F=6.11 R2=.45 s.e.=2.34.36The implication of these results was that site indexprediction from soil nutrient variables improved if a categoricalvariable based on soil reaction is included in the model. Incontrast to this, an R 2 of .09 resulted using the same data froman all subsets regression of SI versus soil nutrient variableswhich identified only PC in its 'best' subset.A comparison of clusters 1 and 2 from the second analysis ofall soil nutrient variables using MANOVA revealed that theoverall multivariate test for group differences was significant(p=.0017) based on Hotelling's T 2 statistic. Subsequentunivariate t tests showed that only PHCACL, MINNC, CAC, MGC, andFEC were different between groups. The range of PHCACL betweenclusters provided a convenient cutpoint to separate them.Hereafter the clusters will be referred to as higher and lower pHsoils. The descriptive statistics for the clusters are given inTable 8.37Table 8. Minimums, maximums, means and standard deviations (s.d.)for soil nutrients which statistically differentiate soilreaction groups.cluster 1 - higher pH soilsvariable minimum maximum mean s.d.PHCACL 4.4 5.1 4.69 .200MINNC 30.0 150. 76.4 34.9CAC 2.75 17.5 7.12 4.40MGC .40 5.85 1.34 1.31FEC 8.0 70.0 26.8 16.0cluster 2 - lower pH soilsPHCACL 4.0 4.4 4.18 .131MINNC 17.0 143. 49.7 32.9CAC .080 7.50 1.56 1.69MGC .100 .950 .314 .220FEC 17.0 175. 59.3 39.5To summarize so far, the relationship between site index andsoil nutrient variables was not significant until two stands wereremoved and the remainder clustered on the basis of theirordination on the first and second components resulting from aPCA analysis. The second component was inferred to represent soilreaction because it was highly correlated with PHCACL, MGC andCAC. A cutpoint in PHCACL at 4.4 separated these clusters and theuse of a categorical variable based on cluster membership withPC, PCxCAC, and MNC resulted in a prediction of moderate siteindex.38Probably of more interest than the moderate prediction ofsite index was the fact that the greatest contribution to theregression from PC was from its correlation with lower pH soils,GROUP=2, which was .522 in contrast to higher pH soils, GROUP=1,at .120. CAC also contrasted between clusters with correlationsof -.592 for GROUP=1 with SI and -.190 for GROUP=2. Thisindicated that in lower pH soils SI is responding positively tohigher PC but is not significantly correlated with PC in higherpH ones. Also in higher pH soils, SI is responding negatively toCAC, but the relationship is insignificant in lower pH soils.There are some parallels between this result with regard toP and work done in alder stands in Washington (e.g., van Miegroetand Cole 1984, Compton 1990 and van Miegroet et al. 1990). Theyfound that soil P levels in a second rotation alder stand, thathad previously grown alder, were lower and that alder growth wassignificantly lower than in an alder stand that had previouslygrown Douglas-fir. They explain that this result is due to the N-fixing ability of alder. Nitrification (the conversion of NH 4 toNO3 - ) rates are particularly high under alder because of theproduction of NH 4 through fixation. Nitrification also resultsin the production of Fe which in combination with the mobility ofNO3 - , and the maintenance of electroneutrality in the soilsolution, causes cation leaching and acidification of soils.Under low pH the fate of P, which is already low in most forestsoils, is to be converted to unavailable forms. Alder is also39known to be P-demanding because of the energy demands of the N-fixing process. For example, 20 molecules of ATP are hydrolysedfor every 2 molecules of ammonia produced in fixation (Wild andJones 1988). Consequently, P nutrition is very important for thegrowth of alder and the paradox suggested by this research isthat alder promotes its unavailability.In this study the site index of alder seems to be respondingto P only in lower pH soils with lower base content. However bothsoils have similar levels of PC. Plants absorb P mostly asorthophosphate ions (H 2PO4 - and H2PO42- ) and their amount in thesoil is known to depend on pH. The adsorption of P by Fe and Aloxides increases with soil acidity and the solubility of thesecompounds is low (Tisdale et al. 1985). One important omission inthe soil chemical analysis was aluminum. At low pH, P tends to befixed by Al and other sesquioxides. In near-neutral soils, P isnot adsorbed so strongly.One conclusion might be that alder soils with a higher pHand content of exchangeable bases may be able to counteract theeffects of soil acidification, including the effect on P,resulting from nitrification under alder. In order to test this,regressions were performed for each pH group separately. When therelationships between SI and soil nutrient variables wereexamined for each group, the variables from the original model inTable 7 performed well only for lower pH soils (Table 9)40indicating that the relationships were not valid for subsets ofthe soils data. Running separate regressions yielded a verydifferent model for higher pH soils (Table 10).Table 9. Regression analysis of site index and soil nutrientsusing stands with lower pH soils.variable coefficient s.e. t sig.constant 5.37 3.53 1.52 .150PC 81.0 21.8 3.71 .002PCxCAC -705.3 280.7 -2.51 .025MNC 261.8 88.4 2.96 .010n=18 F=5.85 R2=.556 s.e.=2.32Table 10. Regression analysis of site index and soil nutrientsusing stands with higher pH soils.variable coefficient s.e. t sig.constant 22.98 0.72 31.89 .000MGC -8.412 1.50 -5.60 .000CUC -22.65 6.05 -3.74 .002n=17 F=15.8 R 2=.692 s.e.=1.54The results of this study are not conclusive, but tend toindicate that soils which are characterized by different levelsof base saturation and pH have very different nutrientrelationships with SI. Furthermore, soils in the higher pH classdo not seem to indicate that P availability is related to growth41as expressed by SI.In addition, there is some evidence that N-fixation isoptimum at certain ranges of pH. In an experiment by Wheeler etal. (1981), Alnus rubra grew best in the pH range from 4.5 to6.5. Below 4.5 the N-fixing nodules had lower specific gravityand contained less effective forms of Frankia (the bacteria thatfix N in alder root nodules).The relationships of soil nutrient variables and site indexclasses were also examined. The site index classes were <18 m,18-20, 20-22 and >22m (Figure 2, page 12). A MANOVA between thesefour groups did not yield significant results (p=.5201) accordingto Rao' F statistic, indicating that the null hypothesis of equalmean vectors cannot be rejected. Stratifying the soils by pHgroup left too few cases for class membership.The descriptive statistics for the soil nutrient quantity(kg ha -1 ) data are given in Table 11.42Table 11. Minimums, maximums, means and standard deviations(s.d.) for soil nutrient quantity data.variable code minimum maximum mean s.d.total C TCQ 20373. 212550. 100510. 39568.total N TNQ 1501. 9828. 5474. 2060.mineralizable N MINNQ 30.3 244.0 117.6 61.68P PQ 5.61 36.4 16.9 9.46total S TSQ 150.1 1948. 783.0 394.9avail SO4-S ASQ .650 11.2 2.70 1.95exchangeable Ca CAQ 42.9 6736. 1674. 1504.exchangeable Mg MGQ 13.6 1771. 195.0 291.9exchangeable K KQ 24.4 195.2 89.0 37.9available Cu CUQ .230 13.1 4.47 3.49available Zn ZNQ .620 12.7 4.83 2.86available Fe FEQ 1.56 375. 85.4 73.8available Mn MNQ 20.0 165. 56.4 30.2Multiple regressions of SI on soil nutrient quantitiesresulted in low R2 values similar to concentrations. Thecorrelation results are shown in Table 12.Table 12. Correlations for site index and soilvariables. nutrient quantityPHCACL .007TCQ -.024 .278TNQ .104 .083 .821MINNQ .076 .323 .733 .694PQ .233 -.146 -.079 .113 .141TSQ .114 .251 .674 .713 .589 .246ASQ -.166 -.055 -.074 .188 -.137 .195 .008CAQ -.053 .834 .452 .290 .648 -.111 .313 -.193MGQ -.076 .711 .502 .418 .707 .004 .488 -.185 .911KQ .152 -.076 .334 .491 .484 .423 .230 -.128 .164 .284CUQ .021 -.091 .030 .079 .110 .168 -.047 .115 -.037 -.014 .145ZNQ .048 -.166 .152 .207 .267 .458 .108 .055 .090 .230 .534 .245FEQ -.151 -.569 -.344 .131 -.199 .316 -.335 .292 -.384 -.315 .019 .261 .269MNQ .187 .056 .232 .006 .242 -.318 .040 -.344 .197 .209 .043 .036 .006 -.435SI PHCACL TCQ TNQ MINNQ PQ TSQ ASQ CAQ MGQ KQ CUQ ZNQ FEQ43The components extracted by PCA for quantities changedsomewhat (Table 13) from the components extracted forconcentrations. Here the second component is represented by MNQ-PQ-FEQ and the third component is represented by soil reaction.The same clustering of stands was seen along component 3 forquantities similar to concentrations, which was expected since itis dominated by PHCACL which does not transform. However, thecluster of higher organic matter soils was now represented byplots with either very little or no coarse fragment content andnot necessarily the highest TCQ. It was more important toemphasize an absolute organic matter content rather than relativeto coarse fragment content. This was one reason why thecomparison between domains was based on concentration data ratherthan quantities.Plots of all combinations of components were examined forclusters and although several ordinations seemed promising,regressions based on categorical variables yielded insignificantresults. On the basis of stand separation using the higher andlower pH soils from soil nutrient concentration analysis, and thesame variables expressed as quantities, a regression wasperformed (Table 14).44Table 13. Eigenvalues, cumulative % variance, component loadings,and communalities for four rotated components extracted from thecorrelation matrix of soil quantity variables.(1)^(2)^(3)^(4)eigenvalue^3.7791^2.4120^3.0607^1.1776cumulative % variance 27.0 44.2 66.1 74.5variable component loadings^ communalityTNQ .9376 .0316 .0540^.0221^.8835TCQ .8192 .2621 .2661^.0580^.8140MINNQ .7331 .0584 .5157^.1352^.8251TSQ .8132 .0039 .1784^-.1202^.7076KQ .7368 -.2785 .1274^.0558^.6398MNQ .0882 .8179 .1406^.2221^.7458PQ .2804 -.7410 -.0216^.0108^.6283FEQ -.2388 -.6894 -.2858 .2445^.6737PHCACL 0449 .1765 .9262 -.0430^.8928CAQ .2383 .1335 .9361 .0302^.9518MGQ .3779 .0548 .8807 .0078CUQ .0309 -.0918 .0070 .9119 .8410ZNQ .3572 -.5141 .2004 .3433 .5499ASQ -.1163 -.4933 -.1451^.2763^.3543Table 14. Regression analysis of site index and soil nutrientquantities using a categorical variable based on clustermembership from a PCA analysis of soil concentrations.variable coefficient s.e.^t^sig.constant^12.354^2.692^4.59^.000PQ^64.656^22.52^2.87^.007PQxCAQ^-261.07^119.0^-2.19^.036MNQ 131.64^65.53^2.01^.054GROUP^3.7738^1.619^2.33^.027n=35^F=2.99^R2=.285 s.e.=2.6645A better result using different variables is given in Table15.Table 15. Regression analysis of site index and soil nutrientquantities using a categorical variable based on clustermembership from a PCA analysis of soil concentrations.variable coefficient s.e. t sig.constant 12.98 1.936 6.70 .000MNQ 579.92 129.19 4.49 .000PQxMINNQ 259.33 129.22 2.01 .054MNQxMGQ -5048.5 1185.0 4.26 .000GROUP 4.750 1.31 3.63 .001n=35 F=6.39 R2=.460 s.e.=2.31The results of these analyses show that SI predictionbetween concentrations and quantities of soil nutrients is notconsistent in terms of the variables used. Since better resultswere achieved with concentrations, these were adopted torepresent soil nutrients for comparison with other domains.There are several aspects of alder nutrition, especially inrelation to N-fixation, that were not included in this study butmay have had a significant bearing on the results. Cobalt, as aconstituent of vitamin B12 , is essential for all N-fixingorganisms (Huss-Danell 1990). In the case of Rhizobium (N-fixingbacteria in leguminous plants), vitamin B 12 functions in theformation of leghaemoglobin which serves in 0 2 transport to46bacteria while maintaining an 0 2-free environment necessary forN-fixation within the nodule (Lynch and Wood 1988). The functionof haemoglobins in Frankia is less clear (Silvester et al. 1990).However, there were significant differences in the growth of redalder seedlings in a greenhouse experiment between 0 and 50 ppb(parts per billion) Co treatments (Russell et al. 1967). Cobaltmay be deficient in some soils formed by acid igneous rocks or insome weathered Podzols (Davies and Jones 1988) but these twoconditions were not sampled in the stands for this study. Lastly,Russel et al. (1967) suggested that other elements (Ni, Fe andMn) may interfere with Co in the synthesis of vitamin 13 12 .Another essential nutrient in N-fixation is molybdenumbecause it is a constituent of nitrogenase, the enzyme whichcatalyses the conversion of atmospheric N 2 to NH 4 in the rootnodules. Its requirement in this process is very low (Wild andJones 1988); furthermore Heilman (1979) states that Mo deficiencyhas not been reported for forest soils in the Pacific Northwest.Klinka and Carter (1990) applied categorical variables basedon moisture and/or nutrient regimes to predict SI for Douglas-fir with good success. Similar efforts here did not result insignificant SI estimates. There are several reasons for this. Thealder stands in this study represented a narrow range of actualmoisture regimes and nutrient regimes as determined from fieldguides (Green et al. 1984). Actual moisture regimes only47represented three of six classes: slightly dry, fresh, and moist;and nutrient regimes only two of five classes: rich and very rich(Appendix 1). In addition, stands with low site index occurred onboth moist sites with gleyed soils (e.g., HQRP) and on the driestof ecosystems sampled (e.g., ROBL). Some stands with Gleysols hadhigh SI if they had at least 30 cm of Ah horizon above a Bghorizon.Regression models using parent materials as categoricalvariables together with soil nutrients versus site index asdependent variable were not significant either. There was a rangeof SI within parent materials types as well. Some marine soilswere well drained with no prominent mottling and had high SIwhile others were poorly drained, and had low SI. Although aldertolerates waterlogging better than most species (Harrington1987), under conditions of low soil oxygen it will grow poorly.Alluvial soils also showed a range of SI. Skeletal alluvialsoils will dry out in summer especially on raised benches abovelow water. This likely occurs in two stands (WOSS and CCK1) withaccompanying lower SI. However, most alluvial soils showed highSI.483.2 Site Index and Foliar NutrientsThe minimums, maximums, means and standard deviations (s.d)for foliar nutrient concentrations are given in Table 16. Mostvalues appear to be similar to other studies, although Fe, Zn andMn are much lower here compared to Radwan (1987) and morevariable compared to the six stands sampled by DeBell and Radwan(1984).Table 16. Minimums, maximums, means and standard deviations(s.d.) of foliar nutrient concentration variables.variable code minimum maximum mean s.d.N^(%) FN 1.71 2.50 2.10 .165P^(%) FP 0.12 0.20 .161 .0206K^(%) FK 0.41 1.24 .849 .181Ca^(%) FCA 0.47 1.43 .876 .2046Mg^(%) FMG .108 .248 .167 .0372S^(%) FS .103 .171 .148 .0134Cu (ppm) FCU 5. 14. 8.3 1.64Zn (ppm) FZN 15. 45. 26.0 5.58Fe (ppm) FFE 42. 221. 68.8 39.6Mn (ppm) FMN 85. 464. 228.8 94.23B (ppm) FB 6. 52. 14.8 8.41Al^(ppm) FAL 16. 251. 71.5 54.5avail. SO4-S(ppm) FAS 13. 371. 81.0 66.2active iron (ppm) FAFE 38. 121. 54.3 18.2The highest correlation between foliar nutrients and siteindex was for FP at .54 (Table 17).49Table 17. Correlations for site index and foliar nutrientvariables.FN .334FP .545 .613FCA -.126 .015 -.071FMG -.444 .018 -.298 .378FK .269 .237 .271 .018 -.366FS .216 .558 .263 -.115 .161 -.008FCU -.332 .089 .088 -.019 .137 -.136 .168FZN -.072 .077 .090 .400 .256 .087 .032 .069FFE -.073 .226 .227 .162 .082 -.222 .139 .215 -.244FMN -.108 .261 .122 -.231 .092 .053 .028 -.027 .058 -.137FB .141 .022 -.049 .211 .027 .045 .337 -.096 -.012 .246 -.418FAL .008 .152 .144 .106 -.005 .002 .032 -.132 -.239 .808 .055 .229FAS .043 .074 .038 .136 .222 .139 .637 -.130 .120 .031 -.234 .527 .021FAFE .002 .260 .249 .040 .105 -.180 .218 .138 -.233 .944 -.075 .244 .799 .068SI25 FN FP FCA FMG FK FS FCU FZN FFE FMN FB FAL FASA stand ordination from PCA showed that SALT could again bea possible outlier (Figure 8). Not all potential outliers wereseen from this plot of the first two components and standdistance testing (D2 , Table 18) and Andrews' plots were alsouseful. Multivariate normality was first checked with a normalprobability plot of D2 distances similar to Figure 6. Thesignificance of these distances proved SALT to be an outlier(p=.0143) and with QUAD and GVWD borderline at the p=.05 level.It was decided to retain the latter two because all outlierexclusion would further reduce stand numbers when different datadomains were compared.7.2844+c^5.2384+e^3.1924+t^+^ Q2CL1.1464++ CC PAMKGP GCDCSB Z^BS• L R B+ P^H-0.8996+W^K^E-2.9456+SALTLB50-2.4401^0.9476^4.3353-0.7462^^2.6414^6.0291component 1Figure 8. The first two components of a PCA analysis on foliarnutrient variables. Stands are plotted as their first letterexcept SALT as an outlier. Not all outliers appear from theseaxes.51Table 18. Mahalanobis distances (D2) calculated from foliarnutrient variables and associated significance levels (p) for 37stands. Three stands SALT, QUAD and GVWD are significantlydifferent at the .05 level.stand D2 stand D2 stand^D2CCK2 7.780 .9005 MAGI 8.211 .8780 ZEB2 13.608 .4793CCK1 15.439 .3488 LOWR 10.163 .7502 AIRS^6.227 .9604BAIN 15.837 .3235 KENE 18.561 .1824 WOSS 18.763 .1742COR1 6.024 .9659 HQRG 8.846 .8408 SALT*27.976 .0143COR2 16.983 .2571 HQRP 14.799 .3920 DEER 17.766 .2176ELKM 5.276 .9816 BIG2 19.353 .1519 GVWD*23.983 .0460SARI 5.631 .9749 BIG1 10.613 .7161 LUND 14.566 .4084BTLK 10.978 .6877 SNOW 10.959 .6892 PENM 14.202 .4348KLA1 16.857 .2639 MENZ 15.225 .3630 PENR^7.542 .9118KLA2 10.210 .7467 QUAD*24.360 .0414 GOLD 15.358 .3542NITI 8.868 .8394 ROBL 16.755 .2695 SQUA 12.725 .5483LCOW 12.987 .5275 PREN 10.062 .7577COOM 16.747 .2699 ZEB1 13.759 .4678The results of a second ordination, excluding SALT, areshown in Table 19. Five components were extracted from thecorrelation matrix accounting for 76% of the variance andsubsequently varimax rotated. In contrast to the soils variables,all foliar correlations were significant with the componentsusing a critical value of .59 (similar to soil variables).52Table 19. Eigenvalues, cumulative % variance and loadings forfive rotated components extracted from the correlation matrix offoliar nutrient variables.eigenvaluecumulative% variancevariable(1)2.933821.0^(2)^(3)^(4)2.2413^2.1966^1.824937.0^52.7^65.7component loadings ^(5)1.427175.9communalityFFE .0448 .1168 .0199 -.2121 .9652.9509FAFE .9124 .1204 .1913 .0011 -.1728 .9134FAL .9064 .0444 .0082 .0116 .2027 .8648FN .1914 .8270 .2115 .0707 -.0213 .7707FP .2281 .7097 -.0682 -.3373 .0050 .6742FMN -.1060 .6292 -.2076 .1774 .0765 .4875FAS -.0874 -.0222 .8759 .0972 .1491 .8070FS -.0327 .4488 .8004 -.0538 -.2585 .9128F8 .2965 -.4082 .6321 -.1322 .1285 .6881FMG .0142 -.0167 .2864 .8387 -.2529 .8499FCA .2176 -.1807 -.1317 .7047 .1639 .6208FZN -.3697 .2521 -.3228 .6091 -.0748 .6810FCU .0312 .1088 -.1043 -.0732 -.8422 .7383FK -.1408 .3208 -.0592 -.2876 .6040 .5738Table 20 gives the results of an all subsets regression offoliar nutrients.Table 20. Regression analysis based on all subsets of variablesfor foliar nutrients versus site index.variable coefficient s.e. t sig.constant 12.24 4.87 2.51 .017FP 139.72 43.59 3.21 .003FMG -52.54 23.29 -2.26 .031FS 48.56 29.15 1.67 .106FCU -0.6466 0.223 -2.90 .007n=36 F=8.90 R 2=.535 s.e.=2.1353All component combinations were plotted in the search forstand clusters as a means of grouping the stands for regression.No grouping yielded results which improved on the results of theall subsets regression in Table 20. However, applying the GROUPvariable, which separated lower from higher pH soils used forsoil nutrients, was an improvement (Table 21).Table 21. Regression analysis of site index and foliar nutrientsusing a categorical variable based on cluster membership from aPCA analysis of soil nutrient variables.variable coefficient s.e. t sig.constant 17.78 4.28 4.15 .000FP 219.3 30.47 7.20 .000FK+FMG -79.83 17.15 -4.65 .000FS+FB 49.54 16.23 3.05 .005FAL 152.0 38.85 3.91 .000FFE -15.34 2.72 -5.63 .000GROUP 3.480 .574 6.06 .000n=36 F=17.34 R 2=.786, s.e.=1.51.This result indicated that soil reaction was a majorvariational gradient for foliar nutrients as well. The standswere divided as for concentrations based on pH group to rerunregressions. Although the results for the lower pH soils were notan improvement over Table 21, virtually all of the variation inSI was explained for the higher pH soils using foliar nutrients(Table 22). However, caution must be used in applying this resultgiven the small number of stands relative to the number of54variables.Table 22. Regression analysis of site index and foliar nutrientsfor stands with higher pH soils.variable coefficient s.e. t sig.constant 18.24 3.21 5.67 .001FN -445.7 46.85 -9.51 .000FP 415.2 28.91 14.36 .000FS 280.1 25.52 10.98 .000FCU -0.488 0.149 -3.28 .013FFE -15.10 2.01 -7.50 .000FMN -10.26 7.48 -1.37 .213FB 136.8 25.4 5.38 .001FAL 178.3 28.8 6.18 .000FAS -184.3 24.7 -7.46 .000n=17 F=43.37 R2=.982 adjusted R 2=.960 s.e.=0.523The sample size requirement for estimating the means offoliar nutrient concentrations within 10 and 20% and at 80, 90and 95% confidence is given in Table 23. Most estimates aresufficient for an allowable error of 20%.55Table 23. Means, standard deviations (s.d.), and coefficients ofvariation (c.v.) for foliar nutrients sampled from 15 alder treesfrom the Lund stand with number of samples required to estimatethe mean within x% at y% confidence.^Total^ Avail. Active-N^P ^Mg^K^S^Cu^Zn^Fe^Mn^B^Al SO4-S^Fe X     ppm mean 2.26 0.15 0.86 0.15 1.09 0.16 7.40 22.60 46.33 203.53 13.80 42.27 94.00 44.33s.d. 0.15 0.01 0.17 0.04 0.20 0.01 1.40 3.40 5.38 65.06 2.57 17.09 39.02 4.39c.v. 6.65 6.86 19.70 23.75 18.56 7.58 18.97 15.03 11.61 31.96 18.62 40.42 41.51 9.89number of samples required to be within x% of the mean at a y confidence levelx^y20% .80 1 1 3 3 2 1 2 2 1 5 2 8 8 120% .90 1 1 4 5 4 1 4 3 2 9 4 13 13 120% .95 1 1 6 7 5 1 5 4 3 12 5 18 19 210% .80 2 2 8 10 7 2 7 5 3 18 7 28 30 310% .90 2 3 12 17 11 3 11 8 5 29 11 46 48 410% .95 3 4 17 24 15 4 16 11 7 41 16 65 69 6563.3 Vegetation AnalysisThe regression of SI on indicator species groups did notprovide significant relationships. Only the life form groupprovided two significant but low correlations with SI, for fernsand deciduous shrubs. The latter was negatively correlatedindicating that SI decreases with increased frequency ofdeciduous shrubs in the stands. The abundance of salmonberry as adominant understory species in many high SI stands was confoundedwith the other shrubs listed in Appendix 3. Indicator speciesanalysis failed to give meaningful relationships for reasonsstated earlier: the stands were sampled from an area underrelatively uniform climatic conditions and within a narrow rangeof soil moisture and nutrient regimes.The site index classes (Figure 2) were used to group standsto determine if vegetation could indicate differences inproductivity (Table 24). There were three comparisons whichconsidered successive groupings: 1. <18 vs >18, 2. <20 vs >20 and3. <22 vs >22. On the basis of presence classes and significancevalues, the species are given diagnostic criteria which serve toseparate the groups. Of the 112 species recorded for the 37stands only 8 have diagnostic values for site index classes. Thisis indicative of the narrow range of nutrient regimes sampled andalso that alder stands tend to have rather homogenous understoryspecies composition. This homogeneity is due, in part, to alder's57N-fixing ability which tends to promote nitrophytic species whileexcluding other indicator values (e.g. oxylophytic species). Thisis illustrated in Table 25 where nitrophytic species arerepresented across the range of SI classes. Only Maianthemumdilatatum, Acer macrophyllum and Tsuga heterophylla are notconsidered nitrophytic (=preferring high N soils). It isinteresting to note that the only oxylophyte (=requiring acidicsoils), Tsuga heterophylla, occurs in abundance on the highest SIgroup. In the current field guide all the species in Table 24besides Tsuga heterophylla are indicative of moist and richecosystems. It was concluded that vegetation provides littledifferentiation of stands based on SI.Another separation of stands was based on the GROUP variableused to separate lower and higher pH soils. Three species havediagnostic value but do not give adequate separation of thesegroups (Table 25).Not enough alder stands were sampled to adequatelycharacterize vegetation associations. Especially lacking werestands on waterlogged soils associated with Carex spp. However,in the course of reconnaissance for this study, over a hundredstands were examined, so that a tentative classification ofassociations can be proposed based on cursory observation. Theclassification includes some information on physiographic andsoils conditions related to site index class (SIC) (Table 26).Diagnostic^5^12^11^9value Presence class% and mean species significance21 ^i1 IV 3.0IV 1.8I +.01^1.5^I +.0 II 2.0 IV 3.0III^4.1^II^3.14 1.1 III 2.9 IV 4.41 II 3.0I +.0 II 2.02 4.1^II^2.4 II^1.5Table 24. Diagnostic species determined from four site indexclasses.58Site index classes^ 1^2^3^4LT18^18-20^20-22^GT22Number of plotsVegetation unitsand speciesLT 18Galium triflorum^(d,c)^5 1.1 III 1.2 III +.0 III 1.2Veratrun viride (d) 3 1.2^I +.0^I +.0GT 20Claytonia sibiricaLT 20(d)^1 +.0 II^1.1^IV^1.0 III^1.5GT 20Acer macrophyllumRibes bracteosumLT 22Maianthemum dilatatumGT 22Acer macrophyllumTsuga heterophylla1Species diagnostic values: d - differential, dd - dominant differential, cd - constantdominant, c - constant, is - important companion (Pojar et al. 1987).2Presence classes as percent of frequency: I = 1-20, II = 21-40, III = 41-60, IV = 61-80,V = 81-100. If 5 plots or less, presence class is arabic value (1-5).Species significance class midpoint percent cover and range: + = 0.2 (0.1 -^0.3),^1^=0.7 (0.4 -^1.0),^2 =^1.6 (1.1^-^2.1),^3 = 3.6 (2.2 - 5.0), 4 = 7.5^(5.1 -^10.0),5 = 15.0 (10.1^-^20.0),^6 = 26.5 (20.1 - 33.0), 7 = 41.5^(33.1 -^50.0),8 = 60.0 (50.1^- 70.0), 9 = 85.0 (70.1 -^100).59Table 25. Diagnostic species determined from soil reactiongroups.pH group low^high<4.4^>4.4Number of plots Diagnostic^18^19Vegetation units value^Presence class andmean species significancepH < 4.4Polystichum munitum (cd) V 6.0 IV 5.7Rubus spectabilis (cd) V 7.7 IV 7.3Stachys cooleyae (d) III^2.6 I^+.4pH > 4.4Table 26. Tentative classification of alder vegetationassociations.Bottomland soils (alluvial and marine parent materials)Alder - salmonberry ^  SIC: H or M depending on CF content and soil textureAlder - swordfern  SIC: H or M depending on CF content and soil textureAlder - skunk cabbagel  SIC: H but needs at least 30 cm of well-aerated Ah abovegleyed horizon otherwise probably LAlder - sedge (Carex obnupta) ^ SIC: L (unsuitable)Upland soils (other than alluvial or marine parent materials e.g., colluvium, morainal, fluvial, eolian)Alder - salmonberry ^  SIC: MAlder - swordfern  SIC: M or P depending upon slope positionAlder - thimbleberry4  SIC: ? further sampling requiredAlder - grass ^  SIC: ? grass species dominated understory composition(Canada wildrye, weak false-manna and crinkle-awned fescue)1only one stand was sampled with this association (QUAD - Quadra Is.)2only observed in Chilliwack Valley and Veddar Mtn. associated with eolian or colluvial soils.3only one stand sampled (SALT - Saltspring Is.),common namesalmonberry^Rubus spectabilisswordfern Polystichum munitumskunk cabbage^Lysichitum americanunthimbleberry Rubus parviflorusCanada wildrye^Elymus canadensisweak false-manna^Torreyochloa paucifloracrinkle-awned fescue Festuca subuliflora603.4 Relationships Between Variable DomainsThe correlations between foliar and soil nutrients arelisted in Table 27.Table 27. Correlations between foliar and soils nutrients.FN^-.246 -.285 -.241 -.315^.152 -.113 -.200 -.288 -.273 -.147^.014 -.099^.340 -.156FP^-.422 -.292 -.232 -.266^.277 -.102^.049 -.454 -.416^.159^.140^.120^.251 -.194FCA^.478^.071^.004^.072 -.449^.004 -.174^.518^.388 -.037^.129 -.158 -.127^.166FMG^.234^.110^.166^.039 -.357^.167 -.141^.281^.421 -.028 -.209^.049 -.038 -.123FK^.052 -.174 -.206 -.028^.010 -.217^.048 -.037 -.099^.120 -.248 -.258^.134 -.095FS^-.069 -.064 -.026 -.125^.267^.115^.081 -.104 -.033 -.081 -.084^.026^.131 -.057FCU^-.035 -.439 -.425 -.297 -.117 -.333^.060 -.113 -.029 -.296^.306^.063^.343 -.156FZN^-.017 -.112 -.079 -.116 -.174 -.175 -.113^.013^.040 -.191 -.282^.272^.329 -.338FFE^.173 -.061 -.084 -.090^.078^.083 -.096^.095^.017^.060^.347 -.115 -.200 -.016FMN^-.637 -.298 -.280 -.376^.025 -.220^.212 -.549 -.572 -.348 -.205 -.167^.530 -.202FB^.350^.233^.189^.270 -.080^.230^.180^.298^.285^.093^.087 -.038 -.417^.352FAL^-.008^.002 -.021 -.064^.058^.114 -.096 -.042 -.134^.168^.227 -.250 -.170^.065FAS^.200^.290^.254^.217^.105^.239^.231^.178^.182^.291 -.455 -.020 -.302^.042FAF^.138 -.090 -.062 -.108^.200^.134 -.101^.031^.014^.036^.234 -.115 -.185 -.142PHCACL TCC^TNC MINNC^PC^TSC^ASC^CAC^MGC^KC^CUC^ZNC^FEC^MNCCanonical correlation analysis, CCA, investigated therelationships between soil and foliar nutrient variables (Table28). This was based on the components resulting from the PCA's ineach domain: four variables from the soils and five from thefoliar domain. Thus, 4 pairs of canonical variates resulted.61Table 28. Results of canonical correlation analysis between thesoil nutrient and foliar nutrient domains.canonical variate pair (1) (2) (3) (4)correlation .7209 .4723 .2590 .2294Soil nutrient domainsoil canonical variatesS1^S2^S3^S4^hw2foliar canonical variatesF1^F2^F3^F4^hb2organic matter 0.49 0.53 -0.61 -0.33 1.00 0.35 0.25 -0.16 -0.08 0.22soil reaction 0.81 -0.32 0.12 0.47 1.00 0.59 -0.15 0.03 0.11 0.38CUC 0.06 -0.77 -0.31 -0.56 1.00 0.05 -0.36 -0.08 -0.13 0.16PC-ZNC -0.31 -0.17 -0.72 0.60 1.00 -0.22 -0.08 -0.19 0.14 0.11var ext 0.25 0.25 0.25 0.25 1.00 0.13 0.06 0.02 0.01 0.22redundancy 0.13 0.06 0.02 0.01 0.22 0.13 0.06 0.02 0.01 0.22Foliar nutrient domainfoliar canonical variatesF1^F2^F3^F4 hw2soil canonical variatesS1^S2^S3^S4^hb2FFE-FAL 0.14 -0.19 0.31 -0.37 0.29 0.10 -0.09 0.08 -0.09 0.03FN-FP-FMN -0.77 0.25 0.28 0.43 0.92 -0.56 0.12 0.07 0.10 0.37FS-FB 0.39 0.40 -0.50 0.51 0.81 0.29 0.19 -0.13 0.12 0.14FCA-FMG-FZN 0.46 -0.15 0.68 0.54 0.99 0.33 -0.07 0.18 0.12 0.16FCU-FK 0.16 0.85 0.34 -0.35 0.99 0.12 0.40 0.09 -0.08 0.19var ext 0.20 0.20 0.20 0.20 0.80 0.10 0.04 0.01 0.01 0.17redundancy 0.10 0.04 0.01 0.01 0.17 0.10 0.04 0.01 0.01 0.17Bartlett's test statistic showed there was a significantcorrelation for the canonical variates (p=.0416). Theinterpretation of CCA relies on the correlation of variables(component scores) both within and between domains, the varianceextracted and redundancy of a domain, and the within (hw 2) andbetween (hb2) communalities (Gittins 1985). Since there were foursoils components and four pairs of canonical variates in theanalysis, the variance extracted is 100% and the communalitiesare all 1. The analysis is based on orthogonal components so thevariance extracted is equal for each canonical variate. The total62redundancy is low for the soils domain; only 22% of the variationin the soils domain is explainable by the foliar domain. This islargely due (13 of the 22%) to the first foliar variate (Fl)which is highly correlated with foliar FN-FP-FMN complex (-.77)relating to soil reaction on the first soil variate (S1) with acorrelation of .59. This indicates that as the soil reactioncomponent increases, foliar N-P-Mn decreases. The correlationbetween FMN and soil reaction was -.637 and probably reflects theimproved availability of soil Mn at low soil pH.The canonical correlations between form and soil nutrientdomains were insignificant (p=.3078) based on Bartlett's test.Table 29 gives the results of CCA between form and foliardomains. The analysis is directed in the sense that the effect offoliar variables on form is more meaningful than how form effectsfoliar, so part of the lower table is deleted. The only importantrelation seems to be with foliar variate 1 (F1) which iscorrelated with FCU-FK versus DBH and to a lesser extent DLC andDH on the first form variate (f1). Thus, as FCU-FK concentrationsin the alder foliage increase so does diameter.63Table 29. Results of canonical correlation analysis between theform and foliar nutrient domains.canonical variate pair (1) (2) (3) (4)correlation .7142 .4992 .4073 .1904Form domainform canonical variatesf1^f2^f3^f4 hw2foliar canonical variatesF1^F2^F3^F4 hb2CHT 0.43 0.39 0.82 -0.00 1.00 0.31 0.19 0.33 -0.00 0.24DBH 0.86 0.44 -0.01 -0.24 1.00 0.62 0.22 -0.00 -0.05 0.43DLC 0.62 -0.15 -0.47 -0.61 1.00 0.44 -0.07 -0.19 -0.12 0.25DH 0.55 0.64 -0.52 -0.13 1.00 0.39 0.32 -0.21 -0.02 0.30var ext 0.40 0.19 0.29 0.11 1.00 0.21 0.05 0.05 0.00 0.31redundancy 0.21 0.05 0.05 0.00 0.31 0.21 0.05 0.05 0.00 0.31Foliar nutrient domainfoliar canonical variatesF1^F2^F3^F4^hw2FFE-AL -0.41 0.10 0.89 0.03 0.97FN-FP-FMN 0.19 -0.55 0.08 0.78 0.95FS-FB 0.23 0.66 -0.12 0.23 0.55FCA-FMG-FZN -0.39 0.45 -0.14 0.58 0.71FCU-FK 0.77 0.22 0.41 0.06 0.82var ext 0.20 0.20 0.20 0.20 0.80redundancy 0.10 0.05 0.03 0.01 0.19This correlation does not seem to be immediatelyinterpretable. It was also noted that the correlation between FCUand NS (stems ha -1 ) was quite high at .70. The correlationbetween FCU and AGE was lower at -.435 which ruled out anyaccumulation effect, presumably it would not have been negativelycorrelated for accumulation, in any case. The highest correlationbetween FCU and any other foliar variable was FEQ at .431. Inaddition, this result was based on partialling out the effects ofstand density (using NS) in the correlations between the formvariables prior to running the CCA which made the results64independent of density effects. Apart from the relationshipbetween Cu and N-fixation (section 3.1), this relationshipremained uninterpretable at this time.The most significant relationship between SI and soilphysical and morphological varaiables was .28, by addingelevation (ELE) this increased somewhat (Table 30).Table 30. Regression analysis of site index and soil physical andmorphological variables with elevation.variable coefficient s.e. t sig.constant 26.35 3.43 7.69 .000CF 16.22 5.51 2.94 .006BD -134.25 53.7 -2.50 .018AH .150 .0577 2.60 .014CLAY -32.99 12.08 -2.73 .010SILT 16.55 7.30 2.27 .031ELE -.00756 .00347 -2.176 .038n=37 F=3.08 R2=.38 s.e.=2.51654.0 SITE QUALITY ASSESSMENT FOR RED ALDERA method of site quality evaluation for red alder inWashington and Oregon was proposed by Harrington (1986) whichscores sites based on 14 properties. The properties are arrangedin three factors based on geographic and topographic position,soil moisture and aeration during the growing season, and soilfertility and physical condition. For a given stand theproperties are evaluated and points tallied to be equivalent toWorthington et al. (1960) site index estimates (m/50yrs) for redalder. The stands for this study were scored using this systemand the ages and heights for the stands were used to calculateWorthington's site index. The correlation between the two valueswas 0.42. The correlation for Harrington's stands was 0.97 andusing another set of stands as validation, 0.96.Apart from the fact that the study areas are separatedgeographically, Harrington's system places greatest emphasis onfactor 1 - geographic and topographic position which includeselevation, physiographic position (slope position), slope andaspect, and growing season precipitation. Compared to the twoother factors this averaged 57% of the total score for sites. Inparticular, elevation can account for as much as 30%. The standssampled in this study were based on an a priori selection fromthree biogeoclimatic subzones which limited climatic variationand the range of elevation (30-650 m).gleyed soil: prominent mottling within the surface 30 cmcoarse textured soil: S or LS textures and >35% coarsefragments or SL,L or SCL textures and >70% coarse fragmentsalluvialor marineparentmaterials—yesM-G^ L-P^yes-no-1M-G^ L P^no-1P-Myes-yesgleyedsoilcoarse^no textured soilnP-Mmoist to very moist —norich to very richsites withoutcoarse texturedor gleyed soil^yes-lower slopeseepageotherparentmaterials66A simpler method for the qualitative assessment of site forred alder is proposed here and is largely based on propertiesalready in use for other species in the Vancouver Forest Region(Green et al. 1984). Greater emphasis is placed on certainproperties and some additional factors must be considered aswell. A flow chart outlining the assessment is in Figure 9. Usingthis chart, 68% of stands were correctly classified to site indexclass.Figure 9. Site quality assessment for red alder.The recognition of alluvial or marine parent materials isthe first step. Soils maps for most of the study area areavailable to assist identification (Jungen 1985). The majority ofthe stands in this study were found to have good site index on67these materials provided they were not coarse textured or gleyed(see definitions in Figure 9). As mentioned in Section 3.1,coarse textured alluvial soils are prone to water deficit in thedrier months. This was particularly true for skeletal soils (>75%CF content) where rooting appeared to be restricted to the firstfew centimetres of fine textured and thin Ah horizon.Gleying can also be indicative of poorer site for alder. Ifprominent mottling occurs within the surface 30 cm this isindicative of a high water table and combined with fine texturedsoils (silt loam and finer) this can restrict root growth and mayhave an effect on N-fixation. There are some compensatingfactors, however, for gleyed soil. A thick surface Ah horizon canmitigate the effects of poor drainage by providing a well-aeratedrooting zone. The role of soil fauna (especially earthworms) inmaintaining a well-aerated surface horizon may be crucial.For other parent materials, good growth of alder can beexpected on moist and rich ecosystems in lower slope positionswith seepage. Otherwise, medium growth can be expected. If thesoils are coarse textured and/or gleyed poor growth is likely.At this time it is not certain what role pH should play insite quality assessment. Based on the work in Washington (section3.1) there is a concern in planting "alder after alder" becauseof the effect of acidification on P, base leaching, Al toxicity68and perhaps even NO 3 - mobility. The first two concerns are fromdecreased P availability which alder seems to be sensitive to andthe removal of essential bases Ca 2+ and Me. Al toxicity in soilscan result in the reduced absorption of cations most notably Ca 2+because of the interference of the Al ions. Lastly, NO 3  leachingis a health concern in water supplies because of its toxiceffects. Although NO 3 - leachates under alder have been measuredin excess of EPA standards in Washington, there is no evidencethat local water supplies are affected by alder. For example, inthe Seymour watershed which partly supplies Greater Vancouver,measured NO3  levels are less than 0.01 of permissable amounts.Some of these concerns can be alleviated, at least intheory, by estimating some additional soil properties in thefield. A soils buffering capacity, which may be defined as itsability to adsorb and release W. , will reduce some acidificationeffects. Cation exchange capacities in soils are largelydetermined by their organic matter and clay contents and can bestrongly pH-dependent. P cycling in soils is dependent on organicmatter content. Most of the P in the soil solution derives fromthe mineralization of organic matter; very little results frommineral weathering, especially at the low pH of our forest soils.Al becomes toxic only at low pH; sesquioxides and organic matterin the soil also have some ability to adsorb NO 3 - at anionexchange sites, especially at low pH. The above discussionindicates that primarily the organic matter content of soils, and69secondarily, finer textures may compensate for some of theeffects of soil acidification under alder. The key will likelyindicate that a coarse textured soil low in organic matter is nota good site for alder.705.0 SUMMARYSeveral characteristics make red alder a potentially usefulspecies for forest management. Its fast growth combined with siteameliorative properties such as resistance to root pathogens, N-fixation, and ability to increase organic matter are alldesirable.Three biogeoclimatic subzones are suggested as having themost favourable climatic characteristics for the best growth ofred alder. Both temperature and climatic moisture regimes in theCWHxm, CWHdm and the CWHvm appear to be the most amenable for itsgrowth.Site quality assessment should rely on the quantitativerelationships between site factors which provide both therational and direction for making the assessment. In this studythe search for meaningful relationships centred around site indexas a measure of productivity but also included relationshipsbetween other variable domains.Initially, there seemed to be little relationship betweensite index and soil nutrient concentrations. The relationship wasbetter for foliar nutrients. The search for pattern usingprincipal components analysis and component partitioning revealedsoil reaction as a major gradient in the data. The effects of71soil acidification under red alder have been known for some time(Franklin et al. 1968) and recently there is evidence that thisacidification results in decreased P availability (Van Miegroet1990). This study demonstrates that the relationships betweensite index and soil and foliar nutrients is markedly effected bysoil reaction and it states what those relationships are. Abroader interpretation made here is that P availability, plus thedynamics of other soil and foliar nutrients, is dependent on soilreaction over the range sampled in this study. It is suggestedthat a pH of 4.4 be used to separate soils. The question of why asoil reaction gradient exists is fundamental. Obviously alder isnot acidifying all soils to the same extent. Whether this isdependent on the inherent chemical and/or physical properties ofthe soil was not determined.The analysis of vegetation and other site factors did notresult in useful relationships. The input of soil N from alderresults in the dominance of species indicating moist and richconditions. The form of alder showed only a weak relationshipwith foliar nutrients and no relationship with soil nutrients.The relationship between foliar Cu and diameter is notinterpretable.The relationships of alder site index versus soil moistureand nutrient regimes and parent materials were not significant.In order for good relationships to develop, sufficiently72contrasting regimes are necessary with a species that showsconsistent response. As an opportunistic species, alder hasproductivity-sites relationships which are difficult to classify.Adequate moisture is essential, but poorly drained soils willresult in poor growth.Parent materials alone do not provide the necessaryinformation on which to base a qualitative assessment. Theyencompass a wide range of conditions. But because the lowland,alluvial and marine soils in the study area often provide goodmoisture holding capacity, they provide the best conditions forgrowth. With additional soil properties, such as texture andcoarse fragment content, which may modify soil moisturecondition, soils can be keyed for the identification of siteindex class.736.0 LITERATURE CITEDAgriculture Canada Expert Committee on Soil Survey. 1987. TheCanadian system of soil classification. 2nd ed. Agric. Can.Publ. 1646.Andrews, D.F. 1972. Plots of high dimensional data. Biometrics28:125-136.Anonymous. 1976. Technicon Autoanalyzer. II. Methodology:individual/simultaneous determination of nitrogen and/orphosphorus in BD acid digests. Industrial Method No.329/74w/a.Avery, T.E. and H.E. Burkhart. 1983. Forest measurements. 3rd ed.McGraw-Hill Book Co.Bardsley, C.E. and J.D. Lancaster. 1965. Sulphur. pp. 1102-1116 .In Methods of Soil Analysis. Edited by C.A. Black, D.D.Evans, J.L. White, L.E. Ensminger, and F.E. Clark. AgronomyNo. 9 Am. Soc. Agron., Mad., WI. pp. 1102-1116.Bond, G. and E.J. Hewitt. 1967. The significance of copper fornitrogen fixation in nodulated Alnus and Casuarina plants.Plant and Soil. 27: 447-449.Bremner, J. and M.A. Tabatabai. 1971. Use of automated combustiontechniques for total carbon, total nitrogen, and totalsulfur analysis of soils. In Instrumental methods foranalysis of soils and plant tissue. edited by L.M. Walsh.Soil Sci. Soc. of Am., Mad., WI. pp. 1-16.British Columbia Forest Service. 1976. Whole stem cubic metrevolume equations and tables. B.C. Min. For., Victoria, B.C.Canadian Climate Program. 1980. Canadian climate normals.Temperature and precipitation, 1951-1980, British Columbia.Environment Canada. Atmospheric Environment Service.Chatterjee, S. and B. Price. 1977. Regression analysis byexample. John Wiley and Sons, New York.Christie J.M. and R. Lines. 1979. A comparison of forestproductivity in Britain and Europe in relation to climaticfactors. Forest Ecol. Manage. 2:75-102.Compton, J.E. 1990. Phosphorus distribution and cycling under redalder: Implications for second rotation alder growth andnutrition. M.Sc. thesis. University of Washington.Davies, B.E. and L.H.P Jones. 1988. Interactions between plant74roots and micro-organisms. In Russell's soil conditions andplant growth. Edited by A. Wild. Longman Scientific andTechnical, Harlow, England. pp. 780-814.DeBell, D.S., R.F. Strand and D.L. Reukema. 1978. Short-rotationproduction of red alder: some options for future forestmanagement. In Utilization and management of alder. Compiledby D.G. Briggs, D.S. DeBell and W.A.Atkinson. PacificNorthwest Forest and Range Experiment Station. Gen. Tech.Rep. PNW-70. USDA For. Serv. Portland, OR. pp. 231-244.DeBell, D.S. and M.A. Radwan. 1984. Foliar chemicalconcentrations in red alder stands of various ages . Plantand Soil. 77: 391-394.Dixon, W.J. (Editor). 1981. BMDP statistical software. Universityof California Press, Los Angeles, CA.Emanuel, J. 1984. F405:VTAB. Ecosystem analysis andclassification. University of British Columbia ComputingCentre.Everitt, B.S. 1978. Graphical techniques for multivariate data.Heinemann Educational Books Ltd., London.Fox, D.J. and K.E. Guire. 1976. Documentation for MIDAS. Stat.Res. Lab., Univ. of Michigan. Ann Arbor, MI.Franklin, J,F, C.T. Dyrness, D.G. Moore and R.F. Tarrant. 1968.Chemical soil properties under coastal Oregon stands ofalder and conifers. In Biology of alder. Edited by J.M.Trappe, J.F. Franklin, R.F. Tarrant and G.M. Hansen. PacificNorthwest Forest and Range Experiment Station. USDA For.Serv. Portland, OR. pp. 157-172.Franklin, J.F. and C.T. Dyrness. 1973. Natural vegetation ofWashington and Oregon. Pacific Northwest Research Station.Gen. Tech. Rep. PNW-8. USDA For. Serv. Portland OR.Gaines, T.P. and G.A. Mitchell. 1979. Boron determination inplant tissue by the azomethine H method. Comm. Soil. Sci.Plant Anal. 10: 1099-1108.Gittins, R. 1969. The application of ordination techniques. InEcological aspects of the mineral nutrition of plants.Edited by I.H. Rorison. Blackwell Scientific Publications,Oxford. pp. 37-66.Gittins, R. 1985. Canonical analysis: a review with applicationsin ecology. Springer Verlag. Berlin.Gower J.C. 1986. Introduction to ordination techniques. In75Developments in numerical ecology. Edited by P. Legendre andL. Legendre. Springer Verlag. New York. pp.Green R.N., P.J. Courtin and K. Klinka, R.J. Slaco and C.A. Ray.1984. Site diagnosis, tree species selection andslashburning guidelines for the Vancouver Forest Region.Land Manage. Handbook No. 8., Prov. of B.C., Min. of For.,Victoria, B.C.Green, R.N., P.L. Marshall and K. Klinka. 1989. Estimating siteindex of Douglas-fir (Pseudotsuga menziesii [Mirb.] Franco)from ecological variables in southwestern British Columbia.For. Sci. 35: 50-63.Greweling T. and M. Peech. 1960. Chemical soil tests. CornellAgric. Exp. Sta. Bull. 960.Harrington C.A. 1986. A method of site quality evaluation for redalder. Pacific Northwest Research Station. Gen. Tech. Rep.PNW-192. USDA For. Serv. Portland OR.Harrington, C.A. 1987. Responses of red alder and blackcottonwood seedlings to flooding. Physiol. Plantarum. 69:35-48.Harrington, C.A. 1990. Alnus rubra Bong.: red alder. In: Silvicsof North America: volume 2, hardwoods. R.M. Burns and H.Barbara, tech. coords. Agric. Handb. 654. USDA For. Serv.Washington D.C. pp. 116-123.Harrington C.A. and R.O. Curtis. 1986. Height growth and siteindex curves for red alder. Pacific Northwest ResearchStation. Res. Pap. PNW-358. USDA For. Serv. Portland OR.HSC. 1991.Hardwood Silviculture Cooperative. Annual Report.Forest Research Laboratory, College of Forestry, OregonState University. Corvallis, OR.Heilman P.E. 1979. Minerals, chemical properties and fertility offorest soils. In Forest Soils of the Douglas-fir Region.Edited by P.E. Heilman, H.W. Anderson and D.M. Baumgartner.Washington State University, Cooperative Extension Service.Pullman, WA. pp. 121-136.Hibbs, D.E and K. Cromack, Jr. 1990. Actinorhizal plants of thePacific Northwest. In: The Biology of Frankia andActinorhizal Plants. Edited by C.R. Schwintzer and J.D.Tjepkema. Academic Press Ltd. London. pp. 343-363.Huss-Danell, K. 1990. The physiology of actinorhizal nodules. InThe Biology of Frankia and Actinorhizal Plants. Edited byC.R. Schwintzer and J.D. Tjepkema. Academic Press Ltd.76London. pp. 129-156.ISSCO. 1985. DISSPLA. Display Integrated Software Systems andPlotting Language. User's Manual version 10.0. IntegratedSoftware Systems Corp. San Diego, CA.Jungen, J.R. 1985. Soils of southern Vancouver Island. MOETechnical Report 17. Report No. 44 British Columbia SoilSurvey. Surveys and Resource Mapping Branch. Min. of Env.Victoria, B.C.Kendall M.G. 1980. Multivariate analysis. Charles Griffin and Co.Ltd., London.Kimmins, J.P. 1987. Forest ecology. Macmillan Publishing Co., NewYork.Klinka, K., R.N. Green, R.L. Trowbridge and L.E. Lowe. 1981.Taxonomic classification of humus forms in ecosystems ofBritish Columbia. B.C. Min. For. Land Mgmt. Rep. 8,Victoria, B.C.Klinka, K., R.N. Green, P.J. Courtin, and F.C. Nuszdorfer. 1984.Site diagnosis, tree species selection, and slashburningguidelines for the Vancouver Forest region. B.C. Min. For.Land Mgmt. Rep. 25, Victoria, B.C.Klinka, K., V.J. Krajina, A. Ceska, and A.M. Scagel. 1989.Indicator plants of coastal British Columbia. Univ. of B.C.Press.Klinka K. and R.E. Carter. 1990. Relationships between site indexand synoptic environmental factors in immature Douglas-firstands. For. Sci. 36: 815-830.Klinka, K., J. Pojar and D.V. Meidinger. 1991. A revision ofbiogeoclimatic units of coastal British Columbia. NorthwestSci. 65: 32-47.Kowalenko, G.D. and L.E. Lowe. 1972. Observations on the bismuthsulphide colorimetric procedure for sulphate analysis insoil. Comm. Soil Sci. Plant Anal. 3: 79-86.Krajina, V.J. 1969. Ecology of forest trees in British Columbia.Ecol. Western N. Amer. 2(1):1-146Krajina, V.J., K. Klinka and J. Worrall. 1982. Distribution andecological characteristics of trees and shrubs of BritishColumbia. The University of British Columbia. Faculty ofForestry.Legendre, L. and P. Legendre. 1983. Numerical ecology. Elsevier77Scientific Publishing Co., Amsterdam.Lynch, J.M. and M. Wood. 1988. Interactions between plant rootsand micro-organisms. In Russell's soil conditions and plantgrowth. Edited by A. Wild. Longman Scientific and Technical.pp. 526-563.Major, J. 1963. A climatic index to vascular plant activity.Ecology 44: 485-498.Mehlich, A. 1978. New extractant for soil test evaluation ofphosphorous, potassium, magnesium, calcium, sodium,manganese, and zinc. Comm. Soil Sci. Plant Anal. 9: 477-492.Noy-Meir, I. 1973. Divisive polythetic classification ofvegetation data by optimized division on ordinationcomponents. J. Ecol. 61: 753-760.Oserkowsky, J. 1933. Quantitative relation between chlorophylland iron in green and chlorotic pear leaves. Plant Physiol.8: 449-468.Parkinson, J.A. and S.E. Allen. 1975. A wet oxidation procedurefor the determination of nitrogen and mineral nutrients inbiological material. Comm. Soil. Sci. Plant Anal. 6: 1-11.Pielou, E.C. 1984. The interpretation of ecological data. JohnWiley and Sons, New York.Plank, M.E., T.A. Snellgrove and S. Willits. 1990. Product valuesdispel "weed species" myth of red alder. Forest Prod. J. 40:23-28.Pojar, J., K. Klinka and D.V. Meidinger. 1987. Biogeoclimaticecosystem classification in British Columbia. For. Ecol.Manage. 22:119-154Price, W.J. 1978. Analytic atomic absorption spectrophotometry.Heydon and Son Ltd., London.Radwan, M.A. 1987. Effects of fertilization on growth and foliarnutrients of red alder seedlings. Res. Pap. PNW-RP-375.Portland, OR.Russell, S.A., H.J. Evans and P.Mayeux. 1968. Investigations onthe requirement and role of cobalt in Alnus rubra. InBiology of alder. Edited by J.M. Trappe, J.F. Franklin, R.F.Tarrant and G.M. Hansen. Pacific Northwest Forest and RangeExperiment Station. USDA For. Serv. Portland, OR. pp. 259-271.78Silvester, W.B., S.L. Harris and J.D. Tjepkema. 1990. Oxygenregulation and haemoglobin. In The Biology of Frankia andactinorhizal plants. Edited by C.R. Schwintzer and J.D.Tjepkema. Academic Press Ltd. London. pp. 157-176.Stevens, J. 1986. Applied multivariate statistics for the socialsciences. Lawrence Eribaum Associates, Publishers, NewJersey.Tarrant, R.F. and J.M. Trappe. 1971. The role of Alnus inimproving the forest environment. In Biological nitrogenfixation in natural and agricultural habitats. Edited byT.A. Lie and E.G. Mulder. Plant and Soil (Spec. Vol.). pp.355-348.Tisdale, S.L., W.L. Nelson, and J.D. Beaton. 1985. Soil fertilityand fertilizers. Macmillan Publishing Co., New York.Trewartha, G.T. 1980. An introduction to climate. McGraw-HillBook Co., New York.Van Miegroet, H. and D.W. Cole. 1984. The impact of nitrificationon soil acidification and cation leaching in a red alderstand. J. Environ. Qual. 13:586-590.Van Miegroet, H., D.W. Cole and P.S. Homann. 1990. The effect ofalder forest cover and alder forest conversion on sitefertility and productivity. In: Sustained productivity offorest soils. 7th North American Forest Soils Conference.Edited by S.P. Gessel, D.S. Lacate and R.F. Powers.University of British Columbia. Vancouver, B.C. pp. 333-354.Waring, S.A. and J.M. Bremner. 1964. Ammonium production in soilunder waterlogged conditions as an index of nitrogenavailability. Nature 201: 951-952.Webster, R. 1977. Quantitative and numerical methods in soilclassification and survey. Oxford University Press. Oxford.Wild, A. and L.H.P. Jones. 1988. Mineral nutrition of cropplants. In Russell's soil conditions and plant growth. 11thed. Edited by A. Wild. Longman Scientific and Technical,Harlow, England. pp. 69-112.Wheeler C.T., M.E. McLaughlin and P. Steele. 1981. A comparisonof symbiotic nitrogen fixation in Scotland in Alnus cautinosa  and Alnus rubra. Plant and Soil. 61: 169-188.Worthington, N.P., F.A. Johnson, G.R. Staebler and W.J. Lloyd.1960. Normal yield tables for red alder. Pacific NorthwestResearch Station. Res. Pap. 36. USDA For. Serv. Portland OR.79Appendix 1. Stand,alder1 2nssite and3-----4vol^maisoil properties5for 37 stands of red-12 --13--14 -15 --16--17--18-asp sl s bg p Ah tex g cf mr nr7----8----9-10--11stand age ba6SI ele1 CCK2 China Ck#2 58 900 486 8.4 46.0 22.5 400 f 0 f xm M 10^l n 15 m vr2 CCK1 China Ck#1 37 1325 299 8.1 34.2 18.9 300 f 0 f xm A 3^s n 85 f r3 BAIN Bainbridge Lk 26 2250 208 8.0 30.7 19.0 100 225 10 m xm A 4^sl n 60 f r4 COR1 Corrigan Ck#1 47 750 480 10.2 44.9 19.3 300 180 5 l xm F 4^sl n 40 m r5 COR2 Corrigan Ck#2 47 750 537 11.4 50.2 18.0 500 225 30 m xm C .5^sl n 40 sd r6 ELKM Elk Main 45 975 469 10.4 46.0 20.3 100 f 0 f vm A 5^s n 45 f r7 SARI Sarita Lk 38 525 470 12.4 41.2 24.9 100 f 0 f vm A 30^sl n 0 m vr8 BTLK Between Lakes 42 875 454 10.8 44.0 21.1 200 f 0 f vm A 6^sl n 40 m r9 KLA1 Klanawa #1 36 450 289 8.0 27.0 21.9 100 f 0 f vm A 6^sl n 20 m vr10 KLA2 Klanawa #2 38 1350 525 13.8 53.1 21.8 100 f 0 f vm A 4^sl n 30 m vr11^MITI^Nitinat Flats 47 675 382 8.1 35.6 23.0 30 f 0 f vm A 6 sil n 0 m vr12 LCOW Cowichan Lk 59 900 589 10.0 53.8 20.9 100 f 0 f vm A 14^cl y 0 m vr13 COOM Coombs 44 1200 309 7.0 34.6 16.5 50 f 0 f xm W 16^L y 0 m vr14 MAGI Maggie Lk 29 900 313 10.8 33.2 22.6 50 f 0 f vm A 15^Is n 30 f vr15 LOUR Lowry Lk 53 975 399 7.5 40.5 18.2 200 f 0 f xm A 40^l n 0 m vr16 KENE Kennedy Lk East 42 675 542 12.9 49.6 20.4 50 f 0 f vm W 4 sil n 0 m vr17 HQRG HQ Ck#1 46 575 235 5.1 24.0 18.7 50 f 0 f xm W 15^l n 0 m vr18 HQRP HQ Ck#2 45 1350 234 5.2 29.2 12.5 50 f 0 f xm W 6^sl y 0 m vr19 BIG2 Bigtree Ck #2 45 850 298 6.6 31.3 16.4 300 f 0 f xm A 9^s n 0 m vr20 BIG1 Bigtree Ck #1 43 1325 378 8.8 41.2 18.0 250 f 0 f xm F 4^sl n 10 m vr21 SNOW Snowdon Rd 40 625 283 7.1 28.1 21.0 100 320 25 m xm W 2^Ls n 40 f vr22 MENZ Menzies Bay 42 1100 345 8.2 36.4 18.6 100 f 0 f xm W 4^sl n 0 f vr23 QUAD Quadra Is 62 825 443 7.1 42.6 23.3 50 f 0 f xm W 30 sil y 0 m vr24 ROBL Roberts Lk 51 1700 228 4.5 30.6 14.3 650 225 20 u xm M -^sl n 25 sd r25 PREN Prenticeville 39 825 299 7.7 31.0 21.3 100 f 0 f xm W 10 sil n 0 f vr26 ZEB1 Zeballos #1 36 1050 280 7.8 31.2 18.3 250 f 0 f vm F 5^sl n 60 f vr27 ZEB2 Zeballos #2 33 900 340 10.3 34.3 25.3 75 f 0 f vm A 16^l n 5 m vr28 AIRS Airstrip Rd 47 700 422 9.0 39.9 21.0 300 f 0 f xm F 6^is n 0 f vr29 WOSS Woss 65 775 394 6.1 38.7 13.8 250 f 0 f vm A <1^is n 40 m r30 SALT Saltspring Is 49 575 307 6.3 29.2 21.6 150 f 0 f xm M 7 sil n 0 m vr31 DEER Deer Ck 28 1150 235 8.4 27.6 22.2 500 360 70 m dm C 10^sl n 65 sd r32 GVWD GVWD 28 2575 357 12.8 42.3 18.4 200 80 20 L dm F <1^sit n 10 f r33 LUND Lund 51 675 456 8.9 41.7 22.9 150 120 5 f dm F 6^sl n 30 f vr34 PENM Pender upper 62 450 508 8.2 43.3 19.4 150 270 20 m dm F 1^sl n 40 sd r35 PENR Pender lower 69 350 532 7.7 44.1 21.1 150 250 5 1 dm W 4^is n 20 f vr36 GOLD Gold River 29 1925 329 11.4 40.8 19.8 300 f 0 f vm A 5^Ls n 40 f r37 SQUA Squamish 28 875 204 7.3 23.2 23.8 400 f 0 f dm F -^sl n 50 sd r1 total stand age2 number of alder stems ha -1 >7 5 cm dbh3 whole stem alder volume m3ha-^-4 mean annual increment m3halyr 45 basal area mha-16 site index (m/25yrs)7 elevation (m)8 aspect: azimuth, f=flat9 slope (%)10 slope position: u=upper, m=middle, 1=lower, f=flats11 biogeoclimatic subzone: xm=very dry maritime CWH subzone, dn=dry maritime CWH subzone, vm=very wetmaritime CWH subzone12 parent material: A=alluvial, C=colluvial, F=fluvial, M=morainal W=marine13 Ah horizon depth (cm)14 texture: s=sand, ls=loamy sand, sl=sandy loam, l=loam, sil=silt loam15 gleying (y=yes, n=no): soils with prominent mottling within the surface 30 cm of mineral soil16 coarse fragment content (%weight)17 actual moisture regime: sd=slightly dry, f=fresh, m=moist18 nutrient regime: r=rich, vr=very rich80Appendix 2. Computer programs.C Program: PART - partitions components of a PCA analysis basedC^ on minimizing variance criteria.CC ISORT is a FORTRAN callable internal sort routine.C F is a matrix of component scores where the first element ranksC the stand position on the component and the second is theC score.CREAL ID(35),F(2,35),MG,MG1,MG2,MINSP2DO 20 1=1,35READ(2,10)ID(I),F(2,I)10^FORMAT(A4,36X,F12.5)F(1,I)=I20 CONTINUECALL ISORT(F,2,35,1,35,2,3,0,&30,&50)GO TO 7030 WRITE(6,40)40 FORMAT(' ERROR IN SORT PARAMETERS')GO TO 18050 WRITE(6,60)60 FORMAT(' ERROR IN SORT MEMORY')GO TO 18070 SG=0.GN=35.DO 80 J=1,35SG=SG+F(2,J)80 CONTINUEMG=SG/GNS2=0.DO 90 J=1,35S2=S2+(F(2,J)-MG)**290 CONTINUECC PARTITION BETWEEN EACH ADJACENT PAIR OF CASESCMINSP2=1000.DO 160 J=1,34JJ=J+1CC CALCULATE MEANS FOR THE GROUPSCSG1=0.G1N=0.DO 100 K=1,JG1N=G1N+1.SG1=SG1+F(2,K)CCC100 CONTINUEMG1=SG1/G1NSG2=0.G2N=0.DO 110 K=JJ,35G2N=G2N+1.SG2=SG2+F(2,K)110 CONTINUEMG2=SG2/G2NCC CALCULATE SQUARED DEVIATIONS FOR EACH GROUPCD21=0.DO 120 K=1,JD21=D21+(F(2,K)-MG1)**2120 CONTINUED22=0.DO 130 K=JJ,35D22=D22+(F(2,K)-MG2)**2130 CONTINUECIF(G1N.EQ.1.)G1N=2.IF(G2N.EQ.1.)G2N=2.S12=D21/(G1N-1.)S22=D22/(G2N-1.)SP2=S12+S22DS2=S2-(S12+S22)IF(SP2.GT.MINSP2)GO TO 140MINSP2=SP2J1=JJ2=JJ140^WRITE(6,150)ID(F(1,J)),ID(F(1,JJ)),SP2,DS2150^FORMAT(1X,A4,1X,A4,2F10.5)160 CONTINUEWRITE(6,170)ID(F(1,J1)),ID(F(1,J2)),MINSP2170 FORMAT(/1X,A4,1X,A4,F10.5)180 STOPEND8182C Program: AND (Andrew's plots)C $R *FTN SCARDS=ANDC $R-LOAD+*DISSPLA 2=file containing component scoresCREAL*8 XPOINT,XSTEP,XRIGHT,P(5),PTS,TREAL LABEL(100),X(750),Y(750)NF=4NCASE=17NPTS=10NPTS1=NPTS+1PTS=NPTSCALL DSPDEV('PLOT')CALL NOBRDRLENGTH = 80CALL TITLE (LABEL, 100,1^' t = -pi to +pi $', 100, ' F(t)^$', 100, 4.6, 3.6)XMIN=-3.14XMAX=3.14YMIN =-10.YMAX =10.CXSTEP=(XMAX-XMIN)/99.0D0CALL GRAF (XMIN, 'SCALE', XMAX, YMIN, 'SCALE', YMAX)CALL FRAMECDO 70 M=1,NCASEREAD(2,10)ID,(P(N),N=1,NF)10^FORMAT(A4,5F12.5)T=-3.14D0-6.28D0/PTSDO 30 I=1,NPTS1T=T+6.28D0/PTSX(I)=TY(I)=P(1)/DSQRT(2.0D0)+P(2)*DSIN(T)+P(3)*DCOS(T)+P(4)*DSIN(2.0D0*T)WRITE(7,20)ID,X(I),Y(I)20^FORMAT(A4,1X,2F5.1)30 CONTINUEWRITE(7,40)40 FORMAT(/)CC --- Plot the dataCCALL CURVE (X, Y, NPTS1, -1)CC^Plot the fitted curveCXPOINT = XMINXRIGHT = XMAXCC83DO 50 I = 1, 750T = XPOINTX(I)=XPOINTY(I)=P(1)/DSQRT(2.0D0)+P(2)*DSIN(T)+P(3)*DCOS(T)+P(4)*DSIN(2.0D0*T)NCPNTS=IXPOINT = XPOINT + XSTEPIF (XPOINT.GT .XRIGHT) GO TO 6050 CONTINUE60 CONTINUECALL CURVE (X,Y,NCPNTS,O)70 CONTINUECALL ENDPL (0)CALL DONEPLSTOPENDAppendix 3. List of plant species.84Coniferous treesAbies amabilisAbies grandisPicea sitchensisPinus monticolaPseudotsuga menziesiiThuja plicateTsuga heterophyllaBroad-leaved treesAcer macrophyllumAlnus rubraCornus nuttalliiNalus fuscaEvergreen shrubsGaultheria shallonMahonia nervosaDeciduous shrubs(Dougl. ex Loud.) Forbes(Dougl. ex D. Don) Lindl(Bong.) Carr.Dougl. ex D. Don in Lamb(Mirb.) FrancoDonn ex D. Don in Lamb.(Raf.) Sarg.PurshBong.Audub. ex Torr. & Gray(Ref.) Schneid.Pursh(Pursh) Nutt.Acer circinatumAcer glabrumCornus stoloniferaHolodiscus discolorLonicera involucrataMenziesia ferrugineaOplopanax horridusPhysocarpus capitatusPrunus emarginataRhamnus purshianaRibes bracteosumRibes lacustreRubus parviflorusRubus spectabilisRubus ursinusSalix lucidaSambucus racemosaSymphoricarpos albusVaccinium ovalifoliumVaccinium parvifoliumViburnum eduleFernsAdiantum pedatumAthyrium filix-feminaBlechnum spicantDryopteris expansaEquisetum hyemaleEquisetum telmateiaGymnocarpium dryopterisPolystichum munitumPteridium aquilinumGraminoidsBromus sitchensisCarex deweyanaPurshTorr.Michx.(Pursh) Maxim.(Richards.) Banks ex Spreng.Sm.(Sm.) Miq.(Pursh) Ktze.(Dougl. ex Hook.) Walp.DC.Dougl. ex Hook.(Pers.) Poir.Nutt.PurshCham. & Schlecht.Muhl.L.(L.) BlakeSm. in ReesSm. in Rees(Michx.) Raf.L.(L.) Roth(L.) Roth(Presl) Fraser-Jenkins & JermyL.Ehrh.(L.) Newm.(Kaulf.) Presl(L.) Kuhn in DeckenTrin.Schwein.85Carex hendersonii^BaileyCarex obnupta BaileyCarex rossii Boott in Hook.Cinna latifolia^ (Trey. ex Goeppert.) Griseb. in Ledeb.Elymus canadensisElymus glaucus Buckl.Festuca subulata^Trin. in Bong.Festuca subuliflora Scribn. in MacounJuncus ensifolius Wikstr.Scirpus microcarpus^PreslTorreyochloa pauciflora^(Presl) ChurchTrisetum canescens Buckl.Trisetun cernuum^Trin.HerbsAchlys triphylla^(Sm.) DC.Actaea rubra (Ait.) Willd.Adenocaulon bicolor^Hook.Anaphalis margaritacea (L.) Benth. & Hook. f. ex C.B. ClarkeAruncus dioicus (Walt.) Fern.Asarum caudatum^ Lindl.Boykinia elata (Nutt.) GreeneCicuta douglasii (DC.) Coult. & RoseCircaea alpina^ L.Circaea pacifica Aschers. & MagnusCirsium arvense (L.) Scop.Claytonia sibirica^L.Cornus canadensis L.Dicentra formosa (Haw.) Walp.Disporum hookeri^(Torr.) NicholsonFragaria vesca L.Galium triflorum Michx.Geum macrophyllum^Willd.Heuchera micrantha Dougl. ex Lindl.Lactuca muralis (L.) Fresn.Lilium columbianum^Hanson ex BakerLysichitum americanum Hult. & St. JohnMaianthemum dilatatum^(Wood) Nelson & MacBrideOsmorhiza chilensis Hook. & Arn.Petasites palmatus (Ait.) GrayPrunella vulgaris^L.Smilacina stellata (L.) Desf.Spiranthes romanzoffiana^Cham.Stachys cooleyae^HellerStreptopus amplexifolius^(L.) DC. in Lam. & DC.Streptopus roseus Michx.Tellima grandiflora^(Pursh) Dougl. ex Lindl.Tiarella laciniata Hook.Tiarella trifoliata L.Tolmiea menziesii^(Pursh) Torr. & GrayTrautvetteria caroliniensis^(Walt.) VailTrientalis latifolia^Hook.Trillium ovatum PurshUrtica dioica^ L.Veratrum viride Ait.Viola glabella R. Br. in Richard.MossesHylocomium splendensKindbergia oreganaKindbergia praelongaLeucolepis menziesiiPlagiomnium insigne(Hedw.) B.S.G.(Sull.) Ochyra(Hedw.) Ochyra(Hook.) Steere ex L. Koch(Mitt.) Kop.86Plagiothecium undulatum^(Hedw.) B.S.G.Polytrichum alpinum Hedw.Rhizomnium glabrescens^(Kindb.) Kop.Rhizomnium magnifolium (Horik.) Kop.Rhytidiadelphus loreus^(Hedw.) Warnst.Rhytidiadephus triquetrus^(Hedw.) Warnst.LiverwortsConocephalum conicum^(L.) Lindb.Plagiochila porelloides (Torr. ex Nees) K. Muell.

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.831.1-0075164/manifest

Comment

Related Items