UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Vegetation-environment relationships in a subalpine wet meadow and a brackish tidal marsh Drewa, Paul B. 1992

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

Item Metadata


831-ubc_1992_spring_drewa_paul.pdf [ 7.18MB ]
JSON: 831-1.0086237.json
JSON-LD: 831-1.0086237-ld.json
RDF/XML (Pretty): 831-1.0086237-rdf.xml
RDF/JSON: 831-1.0086237-rdf.json
Turtle: 831-1.0086237-turtle.txt
N-Triples: 831-1.0086237-rdf-ntriples.txt
Original Record: 831-1.0086237-source.json
Full Text

Full Text

VEGETATION-ENVIRONMENT RELATIONSHIPS IN A SUBALPINE WET MEADOWAND A BRACKISH TIDAL MARSHByPAUL BOGUSLAW DREWAB.Sc., The University of British Columbia, 1989A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF SCIENCEinTHE FACULTY OF GRADUATE STUDIES(Department of Botany)We accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIAApril 1992© Paul Boguslaw Drewa, 1992In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature) Department of  BotanyThe University of British ColumbiaVancouver, CanadaDate April 27, 1992.DE-6 (2/88)ABSTRACTCommunity structure, community-environment relationships, therole of land surface elevation as a determinant of plant speciescomposition, and aspects of scale were examined and comparedbetween a brackish tidal marsh in the Squamish estuary and asubalpine wet meadow in Garibaldi Park in southwestern BritishColumbia. Vegetation abundance, soil variables, and groundelevation data were collected from 225, 0.5 X 0.5 m quadratssystematically located at five metre intervals in a 40 X 120 msampling grid in both sites. The effect of changing the samplingscale was examined through simulation rather than by resampling inthe field using different quadrat sizes. Larger quadrat sizes weresimulated by aggregating adjacent quadrats in the grid andcalculating average values for species and environmental variables.Five aggregation scales (referred to as 'agg' levels) were formed:aggl (0.5 X 0.5 m), agg4 (5 X 5 m), agg6a (5 X 10 m), agg6b (10 X5 m), and agg9 (10 X 10 m). At each scale, minimum variancecluster analysis and canonical correlation analysis were used todescribe community structure by selecting a dendrogram level tosegregate the vegetation data into subcommunities. Vegetation datawere correlated with environmental data using canonicalcorrespondence analysis. To evaluate which scale provided theclearest picture of community structure (yielded the largestbetween and smallest within-cluster variability estimate), ananalysis of variance was performed on canonical correspondence- iii -analysis first and second axis scores using the selected dendrogramlevel for stratification at each scale. This helped to provide anoverall between and within-cluster variability estimate for eachscale. The role of elevation as a determinant of vegetationpattern was investigated by regressing a canonical axisrepresenting species variables against a canonical axisrepresenting elevation. Residuals, representing that proportion ofthe variation in vegetation unexplained by elevation, were savedand correlated with the environmental variables to examine if othervariables unrelated to elevation shared strong relationships.At most scales the marsh study site is composed of twosubcommunities: upper and lower. The upper zone, characterized bysoils of greater sand and organic content but less clay than thelower, is resident to many species common to Pacific coastalmarshes. The lower zone is a monospecific stand of Carex lvnqbvei that is exposed to high-low tide alternation which may removeorganic content, sand, and deposit clay. Salinity did not sharestrong correlative relationships with the vertical distribution ofplant species. However, soils were most saline and acidic in a lowarea near the upper marsh that was apparently not exposed to tidalflushing. Strong correlations'between residuals and carbon, sand,and clay content suggest sources other than elevation such as tidesand the species themselves may influence edaphic factors which inturn share relationships with vegetation pattern.- iv -Generally, the subalpine meadow is composed of threesubcommunities: forb meadow (upper), heath (middle), and sedgemeadow (lower). Greater sand and electrical conductivity in uppermeadow soils suggest a well-exposed and well-drained area. Thelower subcommunity characterized by mostly Carex niqricans, possesssoils of greater clay and organic content. Soils generally tend tobe less acidic than the upper zone suggesting that leaching may beoccurring as water drains from the upper meadow into the lower.Simulation sampling with a rectangular quadrat positionedperpendicular to vegetational banding (agg6b), defined eightsubcommunities in both sites. In addition, overall within-assemblage variability was least and between-assemblage variabilitywas greatest suggesting that observation clarity is maximized atagg6b. Correlations among environmental variables and species axesgenerally become stronger at progressively coarser scales. Inparticular, subcommunity-pH relations were unnoticeable at aggl butstrengthened at agg4 in both sites. However, a strong agg4correlation weakened at agg9 in the tidal marsh, recognizingexceptions. An hierarchical approach reminds one to be cautiouswhen assessing the 'importance' of environmental variables.Results of this study suggest the importance of environmentalfactors, estimated by their correlations with vegetation pattern,may depend on the scale at which the data are analyzed.- v -TABLE OF CONTENTSPageABSTRACT^ iiTABLE OF CONTENTS^LIST OF TABLES viiiLIST OF FIGURES^LIST OF PHOTOGRAPHS xiiACKNOWLEDGEMENTS^ xiiiCHAPTER 1: INTRODUCTION^ 11.1 Prelude 11.2 Marsh Literature Review^ 21.3 Alpine/Subalpine Literature Review^ 71.4 Hierarchy Theory^ 101.5 Objectives^ 13CHAPTER 2: METHODS 152.1 Study Sites 152.1.1 Tidal Marsh^ 152.1.2 Subalpine Wet Meadow^ 152.2 Field Data Collection 202.3 Laboratory Data Collection 232.4 Data Analysis^ 252.4.1 Altering the Scale of Observation^ 252.4.2 Classification of Subcommunities 272.4.3 Interpolation of Soil Variables 292.4.4 Vegetation-Environment Relationships^ 302.4.5 Overall Between-Within ClusterVariability Assessment^ 312.4.6 Closer Examining the Effect of EL^ 32CHAPTER 3: RESULTS^ 343.1 Tidal Marsh and Wet Meadow:A Brief Comparison^ 343.2 Brackish Tidal Marsh 343.2.1 Aggl Scale (0.5 X 0.5 Metre ObservationUnit) ^ 343.2.1.1 Community Structure^ 343.2.1.2 Environmental VariableRelationships^ 373.2.1.3 Community-EnvironmentRelationships 443.2.2 Agg4 Scale (5 X 5 Metre ObservationUnit) ^ 543.2.2.1 Community Structure^ 54- vi - Environmental VariableRelationships^ 543.2.2.3 Community-EnvironmentRelationships 553.2.3 Agg9 Scale (10 X 10 Metre ObservationUnit) ^ 573.2.3.1 Community Structure^ 573.2.3.2 Environmental VariableRelationships^ 593.2.3.3 Community-EnvironmentRelationships 593.2.4 Agg6a and Agg6b Scales (5 X 10 Metre and10 X 5 Metre Observation Units) ^ 613.2.4.1 Community Structure and Environ-mental Variable Relationships^ 613.2.4.2 Comutunity-EnvironmentRelationships^ 623.2.5 EL Influence Verification 633.2.5.1 Agg1, 4, 6a, 6b, 9 Scales^ 633.2.6 Tidal Marsh Discussion 633.3 Subalpine Wet Meadow^ 673.3.1 Aggl Scale (0.5 X 0.5 Metre ObservationUnit) ^ 673.3.1.1 Community Structure^ 673.3.1.2 Environmental VariableRelationships^ 783.3.1.3 Community-EnvironmentRelationships 783.3.2 Agg4 Scale (5 X 5 Metre ObservationUnit) ^ 913.3.2.1 Community Structure^ 913.3.2.2 Environmental VariableRelationships^ 923.3.2.3 Community-EnvironmentRelationships 933.3.3 Agg9 Scale (10 X 10 Metre ObservationUnit) ^ 943.3.3.1 Community Structure^ 943.3.3.2 Environmental VariableRelationships^ 953.3.3.3 Community-EnvironmentRelationships 953.3.4 Agg6a and Agg6b Scales (5 X 10 Metre and10 X 5 Metre Observation Units) ^ 963.3.4.1 Community Structure and Environ-mental Variable Relationships^ 963.3.4.2 Community-EnvironmentRelationships^ 993.3.5 EL Influence Verification 1003.3.5.1 Aggl, 4, 6a, 6b, 9 Scales^ 1003.3.6 Subalpine Wet Meadow Discussion 102CHAPTER 4: SYNTHESIS^ 1074.1 Quadrat Shape and Orientation^ 1074.2 Noisy Data and Redundancy Estimates^ 1094.3 Hierarchical Perspective: An Assessment^ 110BIBLIOGRAPHY^ 113APPENDIX A 120APPENDIX B^ 122LIST OF TABLESPageI. Mean precipitation and temperature data,1951-1980, for Squamish (tidal marsh) and Alta Lake(subalpine wet meadow) (Environment Canada 1980) ^ 17II. Redundancy and R, estimates for dendrogramlevels C2-C10 at different scales in the SquamishMarsh. Highest redundancy and R, estimates at eachscale are marked with an '*' ^ 36III. Species mean aerial coverage class datafor subcommunities at different scales in theSquamish Marsh. Those species with a mean aerialcoverage class estimate >1 were deemed to berepresentative of a particular subcommunity. Speciesnames corresponding to the codes used below may befound in Appendix A. Integers directly above meanand standard deviation estimates representsubcommunities at each scale^ 39IV. Pearson correlations between environmentalvariables at different scales in the Squamish Marsh.EL, relative ground level elevation; C, carboncontent; EC, electrical conductivity; SA, sandcontent; CY, clay content; pH, soil acidity^ 43V. Pearson correlations between environmentalvariables and species axes I and II at differentscales in the Squamish Marsh. EL, relative groundlevel elevation; C, carbon content; EC, electricalconductivity; SA, sand content; CY, clay content;pH, soil acidity^ 49VI. Summarized environmental data for the mainsubcommunities recognized at the different scalesof analysis (agg levels) in the tidal marsh. Integersdirectly above mean and standard deviation estimatesrepresent subcommunities at each scale. EL, relativeground level elevation; C, carbon content; EC,electrical conductivity; SA, sand content; CY, claycontent; pH, soil acidity^ 50VII. Pearson correlations at different scalesbetween environmental variables, residuals, and acanonical axis representing species variables in theSquamish Marsh. AXIS, canonical correlation axis;RESD, residuals; EL, relative ground level elevation;C, carbon content; EC, electrical conductivity;SA, sand content; CY, clay content; pH, soil acidity^ 64VIII. Redundancy and R, estimates for dendrogramlevels C2-C10 at different scales in the subalpinewet meadow. Highest redundancy and Rc estimates ateach scale are marked with an '*' ^ 68IX. Species mean aerial coverage class datafor subcommunities at different scales in the sub-alpine wet meadow. Those species with a mean aerialcoverage class estimate >1 were deemed to berepresentative of a particular subcommunity. Speciesnames corresponding to the codes used below maybe found in Appendix A. Integers directly above meanand standard deviation estimates represent subcommunitiesat each scale^ 70X. Pearson correlations between environmentalvariables at different scales in the subalpine wetmeadow. EL, relative ground level elevation; C,carbon content; EC, electrical conductivity; SA,sand content; CY, clay content; pH, soil acidity;FI, soil sample particles < 2 mm^ 79XI. Pearson correlations between environmentalvariables and species axes I and II at differentscales in the subalpine wet meadow. EL, relativeground level elevation; C, carbon content; EC,electrical conductivity; SA, sand content; CY,clay content; pH, soil acidity; FI, soil sampleparticles < 2 mm^ 85XII. Summarized environmental data for themain subcommunities recognized at the differentscales of analysis (agg levels) in the subalpinewet meadow. Integers directly above mean andstandard deviation estimates represent subcommunitiesat each scale. EL, relative ground level elevation;C, carbon content; EC, electrical conductivity;SA, sand content; CY, clay content; pH, soil acidity;FI, soil sample particles < 2 mm^ 89XIII. Pearson correlations at different scalesbetween environmental variable, residuals,and a canonical axis representing species variablesin the subalpine wet meadow. AXIS, canonicalcorrelation axis; RESD, residuals; EL, relative groundlevel elevation; C, carbon content; EC, electricalconductivity; SA, sand content; CY, clay content; pH,soil acidity; FI, soil sample particles < 2 mm^ 101- x -LIST OF FIGURESPage1. Maps showing the location of the tidalmarsh and the subalpine wet meadow study sites^ 162. "Systematic grid" sampling designsuperimposed on topographic profiles of marshand subalpine meadow study sites. Numbers alongthe X and Y axes refer to transects; the Z axisdenotes elevation in metres^ 243. Simulation sampling with larger quadratsizes by aggregating neighbouring 0.5 X 0.5 mquadrats (aggl level): agg4 (a); agg9 (b);agg6a (c); agg6b (d). '99' denotes omitted quadratsbecause of an unavailability of neighbouringquadrats to group with^ 264. A dendrogram (a) and grouping variables (b)offering alternative means of presenting MVCAresults. C2-C10 refer to different dendrogramlevels in (a) and respective grouping variablesin (b) ^ 285. Whittaker diversity curves for the marshand subalpine meadow study areas^ 356. Grid maps showing subcommunity layout atdifferent scales in the tidal marsh study site:aggl (a), agg4 (b), agg9 (c), agg6a (d), and agg6b (e) ^ 387. Subcommunity-environment biplots at differentscales for the marsh study site: aggl (a), agg4 (b),agg9 (c), agg6a (d), and agg6b (e). Subcommunitiesare represented by 50% confidence ellipses. Ellipseswere unable to be produced where those subcommunitieswere represented by three or fewer sampling units.Each environmental variable is represented by a vector.EL, relative ground level elevation; C, carbon content;EC, electrical conductivity; SA, sand content;CY, clay content; pH, soil acidity^ 458. Two-dimensional isopleths displayingEL and soil variable zonation patterns at the agglscale in the marsh study site: CY, clay content (a);SA, sand content (b); C, carbon content (c); EL,relative ground level elevation (d); EC, electricalconductivity (e); and pH, soil acidity (f) ^ 529. Overall between and within-clustervariability estimates for all scales in the marshstudy site. Unstandardized estimates shown alonginner isoclines; standardized estimates shown alongouter isocline^ 5810. Grid maps showing subcommunity layout atdifferent scales in the subalpine wet meadow studysite: aggl (a), agg4 (b), agg9 (c), agg6a (d), andagg6b (e) ^ 6911. Subcommunity-environment biplots atdifferent scales for the subalpine meadow studysite: aggl (a), agg4 (b), agg9 (c), agg6a (d),and agg6b (e). Subcommunities are representedby 50% confidence ellipses. Ellipses were unableto be produced where those subcommunities wererepresented by three or fewer sampling units. Eachenvironmental variable is represented by a vector.EL, relative ground level elevation; C, carbon content;EC, electrical conductivity; SA, sand content; CY,clay content; pH, soil acidity; FI, soil sampleparticles < 2 mm^ 8212. Two-dimensional isopleths displaying ELand soil variable zonation patterns at the aggl scalein the subalpine meadow study site: EL, relativeground level elevation (a); EC, electrical conductivity(b); SA, sand content (c); CY, clay content (d); C, carboncontent (e); FI, soil sample particles < 2 mm (f);and pH, soil acidity (g) ^ 8613. Overall between and within-clustervariability estimates for all scales in the sub-alpine meadow. Unstandardized estimates shownalong inner isoclines; standardized estimates shownalong outer isocline^ 97LIST OF PHOTOGRAPHSPageI: Photograph of the Squamish Marsh^ 18II: Photograph of the subalpine wet meadowadjacent to Mimulus Lake in Garibaldi ProvincialPark, British Columbia^ 21ACKNOWLEDGEMENTS The author expresses sincerest thanks to committee members Dr.G.E. Bradfield, Department of Botany, University of BritishColumbia, for his assistance, advice, and encouragement; Dr. W.B.Schofield, Department of Botany, University of British Columbia,for his encouragement and help with bryophyte identification; andDr. R.E. Foreman, Department of Botany, University of BritishColumbia, for his valuable suggestions and encouragement.Special thanks to Dr. L.M. Lavkulich, Department of SoilScience, University of British Columbia, for permission to use hislaboratory facilities as well as Mr. B.W. von Spindler, Departmentof Soil Science, University of British Columbia, for help withlaboratory soil analyses.The author is indebted to Drs. G.B. Straley, Department ofBotany, University of British Columbia, and A. Ceska, Royal BritishColumbia Museum, for their help in identifying vascular plantspecies.Sincerest thanks also goes to the British Columbia Ministry ofParks for granting permission to work in Garibaldi Provincial Parkand to Miss J. Lawrence for her help in collecting field data aswell as my wife, Mrs. A.S. Drewa, for her help in collecting fielddata and encouragement.CHAPTER 1: INTRODUCTION1.1 Prelude Plant community ecology is concerned largely with thedescription and explanation of plant distribution patterns in thefield. Moreover, plant ecology attempts to interpret therelationship of plants to their environments (Billings 1964). Anenvironment is a complex of many factors that interact not onlywith an organism but also among themselves, changing continuouslythrough time (Billings 1964; Greig-Smith 1983). The difficultiesin studying communities are twofold: 1) the number and uniquenessof communities and community components far exceed the number ofindividual items that an ecologist considers for investigation, and2) communities and their components are integrated, yet our mindsapproach communities by a succession of individual thoughts (Simon1962; Daubenmire 1968; Elsasser 1969; Gauch 1982).Recognizing that plant ecology is a "product of interactionbetween communities and ecologists through observations andanalysis" (Gauch 1982), a study was undertaken to examine andbriefly compare vegetation-environment relationships in the contextof a brackish tidal marsh and a subalpine wet meadow.Alpine/subalpine habitats have provided excellent opportunities tostudy species and community pattern. Vegetation pattern is usuallysharply accentuated in subalpine/alpine regions because oftopographic diversity and often changes abruptly related to rapid- 2 -shifts in environmental gradients (Douglas and Bliss 1977). Thoughmarsh systems tend to have fewer environmental gradients and,therefore, fewer species, vegetation pattern tends to be moredistinctive. Marsh habitats also tend to be highly productiveproviding critical habitats for large numbers of bird, fish, andmammalian species. Their easy accessibility also provide excellentopportunities to study vegetation pattern. The benefits ofstudying these habitats are threefold. First, issues concerningvegetation pattern and vegetation-environment relationships can beexamined in both systems within the confines of a relatively smallarea and within the limited time frame allotted for a MSc. thesis.Second, valuable contributions already exist in the scientificliterature, providing helpful insights and suggestions for futureresearch. Third, results from this study may provide new insightsinto marsh as well as subalpine/alpine park management. Despitethe differences between the two habitat types, a brief comparisonbetween a brackish tidal marsh and a subalpine wet meadow may be inorder since they are both influenced by fresh water and may, infact, represent opposite ends of a fresh water gradient. Differentspecies of the same genus such as Carex, Potentilla, Actrostis,Deschampsia, and Juncus may be found in most subalpine meadow andbrackish tidal marsh systems in the Pacific Northwest.1.2 Marsh Literature ReviewTopography directly controls the submergence/emergence ratio- 3 -of a marsh through its interaction with tides. Dawe and White(1982) emphasized elevation as playing a major role in determiningvegetation zonation in the Little Qualicum River estuary, BritishColumbia. While soil texture, type, and salinity were mentioned asplaying relatively important roles, all were dependent on elevationand its interaction with tides. Vegetation pattern was alsoconcluded to be controlled primarily by elevation in the Nanoose-Bonell estuary on Vancouver Island, British Columbia. Once again,Dawe and White (1986) stressed the interaction between elevationand tides and how elevation is responsible for controlling thesubmergence/emergence ratio of the marsh. Elevation as well assoil texture controlled species distribution along the verticalgradient of the marsh while inundating water salinities determinedthe species' horizontal distribution. Three major plantcommunities were identified along a vertical gradient of a tideinfluenced meadow on Chichagof Island, Alaska (Stephens andBillings 1967). Communities were clearly dependent on elevationand its interaction with tides. Soil characteristics such as pH,cation exchange capacity, and exchangeable sodium, calcium,magnesium, and potassium were also deemed important. Campbell andBradfield (1989) recognized vegetation-elevation relationships intwo estuarine marshes of northern British Columbia. The Yakounmarsh showed a closer connection between vegetation pattern andelevation and a clearer zonation of communities than the Dalamarsh. The Yakoun marsh experiences steeper gradients insubmergence time and flooding frequency, offering an explanation- 4 -for the difference in vegetation-elevation relationships betweenthe two.Despite other researchers' emphasis on topography/elevation asprimarily determining vegetation distribution, Disraeli and Fonda(1979) have stated that elevation shows not a direct but instead,an indirect effect on vegetation distribution. Through its effecton other environmental factors like tidal inundation, sand and siltcontent of the soil, and soil moisture, elevation indirectly causedthe Nooksack delta marsh species to be broadly zoned into two marshtypes at Bellingham Bay, Washington. Mahall and Park (1976)attempted to investigate why two distinct zones of Salicornia virqinica and Spartina foliosa were prominent in salt marshes ofnorthern San Francisco Bay. Experiments in which Salicornia andSpartina plants were exposed to artificial tides indicated thatinhibition of growth through reduced daylight, inhibition of re-rooting and the production of new branches prevented Salicornia seedlings from advancing seaward to the Spartina zone. Majorportions of two monotypic stands were found to have different tidalrelationships, frequency, and duration of flooding in a salt marshin Davis Bay, Mississippi (Eleuterius and Eleuterius 1979). Tidalphenomena per se could not be shown to account for salt marshzonation and the clear demarcation between zones. Eleuterius andEleuterius (1979) urged that other environmental factors,especially edaphic conditions and possibly biotic interactions,need to be investigated.- 5 -Vince and Snow (1984) described patterns of plant speciesdistribution, plant abundance, and environmental factors on SusitnaFlats, an Alaskan subarctic marsh. Though there was littletopographic relief and soil texture was similar throughout theFlats, vegetation zones differed with respect to floodingfrequency, rate of siltation, soil organic content, moisturecontent, redox potential, and salinity. Waterlogging and soilsalinity were found to segregate most of the vegetation zones.Vince and Snow (1984) emphasized that it is unlikely that tidalinundation per se determines plant distribution on Susitna Flats.Dawson and Bliss (1987) recognized a salinity and soil extractable-ion gradient on a high arctic brackish marsh. They concluded thatplant zonation exists at least in part as a function of plantresponse to gradients in edaphic characteristics. A study wasconducted at the brackish tidal marsh along the shore of Skagit Baynear Mt. Vernon, Washington by Ewing (1983). Principal componentsanalysis indicated that salinity and soil texture were stronglycorrelated with the first generated factor while elevation and soilredox potential with the second and third respectively. Ewing(1983) concluded that brackish intertidal marshes of the PacificNorthwest are most profoundly affected by water salinity.Community composition is also affected by soil texture, soil redoxpotential, and elevation. Conversely, Disraeli and Fonda (1979)clearly stated that "salinity played no significant role" on abrackish marsh at Bellingham Bay, Washington.- 6 -An experimental study of the role of edaphic conditions wasconducted by Snow and Vince (1984). Salt tolerance of each speciescorresponded with the soil salinity in the zone of occurrence.Snow and Vince (1984) also emphasized that the results from a twoyear reciprocal transplant experiment demonstrated the influence ofbetween-species competition on zonation. Bertness and Ellison(1987) concluded physical disturbance and interspecific competitionto be major determinants of spatial pattern in a salt marshcommunity in New England. Interspecific competition was cited asa major determinant of pattern by 1) transplant studies, 2)distribution and rapid recruitment of rare high marsh plants inareas where dominant high marsh plants are numerically lacking, and3) the rapid closure of disturbance-generated bare patches andrelative rarity of early colonists in undisturbed vegetation.Patterning of New England salt marsh plant communities is aninteractive product of physical constraints on plant success,predation pressure, physical disturbance, and interspecificcompetition (Bertness and Ellison 1987). Resource competition forsoil nitrogen and light was examined in a brackish tidal marshlocated at Brunswick Point, British Columbia by Pidwirny (1990).Two vegetation zones of Scirpus americanus and Carex lvngbvei wereidentified. Pidwirny suggested that Scirpus is dominant in the lowmarsh because it is a better competitor for nitrogen. Conversely,Carex may be dominant in the high marsh because of its greaterbiomass and height, thus making it a superior competitor for light.- 7 -1.3 Alpine/Subalpine Literature ReviewSnow distribution in alpine/subalpine systems is nonuniformbecause of the interaction between wind and topography (Billingsand Bliss 1959). Topographic diversity in conjunction with windnot only affect snow cover depth but the rate of snowmelt which inturn is often responsible for steep and abrupt environmentalgradients (Douglas and Bliss 1977; Oberbauer and Billings 1981;Olyphant 1984; Evans and Fonda 1990). Abrupt changes in speciescomposition occur resulting in a mosaic of plant communities. Inthe North Cascade Range of both British Columbia and Washington,phenological patterns of vegetation shared strong correlativerelationships with the snowmelt date and early season temperatureregimes (Douglas and Bliss 1977). At Rocky Mountain National Park,vegetational and phenological differences were also correlated withthe melting back of late season snowbanks (Holway and Ward 1963).In their study, persistent snow cover delayed normal plantdevelopment, replaced certain species by different ones and mayhave contributed to the failure to complete certain life cyclephases. Length of the growing season differed by two to threeweeks between successive communities along a snow depth gradient,a function of topography, in the Olympic Mountains, Washington(Kuramoto and Bliss 1970). In an area of New Zealand alpineherbfield, Weir and Wilson (1987) examined micro-zonation pattern.Discriminant functions analysis suggested snow cover, slope, andexposure as important correlates of vegetation pattern. Bray-- 8 -Curtis ordination was used to analyze vegetation pattern on SignalMountain in the Canadian Rockies (Hrapko and La Roi 1978). Again,snow depth and duration, correlated with topography, were concludedas being primarily responsible for species composition. Suchmicrotopographic factors as aspect, slope, and drainage patternswere critical determinants of nineteen plant communities in thesouthern Chilcotin Mountains (Selby and Pitt 1984). In the alpinetundra of the Colorado Front Range, topographic changes wereresponsible for a mosaic of snow cover conditions, resulting insharp contrasts in moisture conditions. Moisture conditions, inturn, were found to be strongly related to bryophyte distributionpatterns (Flock 1978).Snow depth and duration may influence soil conditions, such assoil type (Knapik et al. 1973), moisture, and temperature (Douglasand Bliss 1977). Many biological processes are dependent on soiltemperature. Processes include the decomposition of organicmatter, release of nutrients as well as their uptake by plantroots. Growth, development, and life cycle characteristics of soilorganisms can all be related to soil temperature (Brown et al.1980). Soil temperatures are also of importance because of theirimpact upon carbon and nitrogen cycles (Nimlos et al. 1965).Though soil temperature was believed to have little effect onvegetation pattern, soil moisture was emphasized as being the mostcritical factor affecting vegetational differences above and belowa snow bank in the Snowy Range of the Medicine Bow Mountains,- 9 -Wyoming (Billings and Bliss 1959). Within an alpine ecosystem onthe Beartooth Plateau, approximately thirty miles southwest of RedLodge, Montana, soil moisture was the dominant environmental factorin determining the distribution of three stand types (Nimlos et al.1965).Interactions between topography, precipitation, wind,insolation, soil texture and other factors ultimately govern waterrelations (Isard 1986). The distribution of water in space andtime has been emphasized as largely controlling the spatial patternof vegetation (Isard 1986; de Molenaar 1987). De Molenaar (1987)emphasized that water has a direct causal effect on vegetation asa "nutrient" and acts as a solvent for all other plant nutrients.Water has an influence on aeration, redox conditions, and acidity,nitrogen and phosphorous supply as well as metal ion availabilityin the rhizosphere. Humus formation and breakdown, nutrientcycling, leaching, and other aspects of soil formation depend onthe interaction of the water regime and relayed phenomena withplant cover (de Molenaar 1987).Holway and Ward (1963) demonstrated the importance of surfacemeltwater as a thermoregulatOr on the microclimate in a studyinvolving the artificial application of meltwater to many species.Results showed a delay in flowering for the majority of speciesranging from a week to as much as one month. Water availabilitydoes not simply differ horizontally along topographic gradients.- 10 -It also varies temporally and vertically (Oberbauer and Billings1981). Groundwater flow was investigated in fen meadows in Dutchstream valleys where they are influenced by both deep (calciumrich) and shallow (calcium poor) groundwater flows (Grootjans etal. 1988). They found that distribution of groundwater patternsinfluenced plant species and showed distinct relationships with thedistribution of different groundwater types. Trophic gradients invirgin mires are mainly responses to differences in the ioniccontent of the groundwater as well as the flow rate.1.4 Hierarchy TheoryEcologists have generally recognized ecosystems as possessingspecific levels of organization or exhibiting hierarchic structure(Simon 1962; Allen and Starr 1982; O'Neill et al. 1986). Ecosystemcomplexity is attributed to vegetation-environment relationshipsoperating at varying spatial and temporal scales (Allen 1987). Ourunderstanding of ecological complexity depends critically onmethods used to describe it (Simon 1962). In general, the samplingstrategy, quadrat size, data transformations, and preferredstatistical tools influence not only the collection of ecologicaldata but also its subsequent interpretation. In particular,quadrat size is of critical importance when examining ecologicalattributes of a system. Choosing a quadrat of certain size anddimensions to sample a system is equivalent to choosing a specificscale to observe relationships. By selecting only one scale (ie.quadrat^size)^to observe ecological^relationships,^ananthropocentric bias is introduced. "An ecosystem's attributes,perhaps the most critical in terms of its proper functioning, maybe virtually unobservable because the chosen scale may havesuppressed them" (Allen and Starr 1982).Hierarchy theory leads to observational approaches thatattempt to circumvent the anthropocentric bias by recognizing theneed to examine ecosystem relationships at different scales.Hierarchy theory uses the concept of level as one of its organizingprinciples (Allen 1987) and relies on the use of multivariatestatistical methods for examining different scales within avegetation community. In conjunction with multivariate statisticalmethods, various kinds of data transformation are used to changethe scale at which dominant processes that are reflected in thedata structure may be observed. As the scale of perception ischanged through the use of these tools, the ecologist's eye moveseither up or down through the hierarchy (Allen and Starr 1982).Despite the intuitive appeal of hierarchy theory in relationto ecosystem organization, few published studies exist. Allen andWyleto (1983) used an hierarchical approach to investigate the roleof fire from 1951 to 1972 on vegetation of the Curtis Prairie ofthe Arboretum of the University of Wisconsin. Two levels oforganization were displayed in two separate principal componentanalyses. At a fine-grained scale, analyses of species cover data- 12 -revealed fire to act as a disturbance with the potential to altergreatly the individual patterns of species cover within thevegetation. At a coarse-grained scale, derived by transforming thecover data to presence/absence data, the analysis indicated thatfire was acting as a stabilizing factor, maintaining speciesdiversity within the vegetation. Allen and Wyleto (1983) concludedthat the removal of fire would, in fact, represent a coarse-scaledisturbance in the grassland system just as the presence of firewas acting as a disturbance at a fine scale. Allen et al. (1984)investigated the effects of data transformation on phytoplanktondaily abundance data over two years, for thirty species from atemperate lake in Llyn Maelog, North Wales. Using this approachled to an improved understanding of the dynamics of the lakeecosystem in terms of species turnover, seasonal areas ofattraction, and uniqueness of individual sample dates. Maintainingthe scale but altering the reference or observation point isanother strategy which can be used to improve understanding on howecosystems are organized. For example, Bradfield and Orloci (1975)used cluster analysis to generate a classification of some openbeach vegetation in southwestern Ontario, and then useddiscriminant analysis to assign a separate set of quadrats,obtained at the same sampling scale, to the pre-established groups.Such repositioning within a level of an hierarchy allows greaterinsight into the effects of inferred processes operating at thatscale.- 13 -An hierarchical approach does not completely eliminate ananthropocentric perspective since the different levels oforganization within a community are defined by observer-chosencriteria (Allen et al. 1984). Based on experience, however, itseems logical to examine various processes at different scales thanat only one. Examination of different levels can only enhance ourunderstanding of vegetation pattern. The hierarchical conceptresults in a sharpening of the concepts and issues involved(Bossort et al. 1977).1.5 Objectives Before the ecological investigation was initiated, a number ofobjectives were established. The first objective was to describeand compare community pattern/structure in a subalpine wet meadowand a brackish tidal marsh. The second objective was to examinerelationships between measured environmental variables andcommunity pattern/structure in both marsh and subalpine meadowstudy sites. Given the emphasis in the marsh and subalpine/alpinescientific literature on topography and its influence on speciesdistribution patterns, a closer examination of the role of landsurface elevation as a determiAant of plant species composition inboth systems constituted a third objective. The use of quadrats ofspecific size and dimensions during field sampling introduces ananthropocentric bias when observing ecological relationships.Thus, a fourth objective was an attempt to alleviate some of this- 14 -bias by investigating what effect different quadrat sizes (scales)have on the perception of community structure and community-environment relationships. Here, the underlying hypothesis wasthat environmental factor 'importance', estimated by theircorrelations with vegetation pattern, as well as perception ofcommunity structure, are dependent on the scale at which the dataare analyzed.- 15 -CHAPTER 2: METHODS 2.1 Study Sites 2.1.1 Tidal Marsh The marsh study was conducted in a brackish tidal marsh at theSquamish estuary. Adjacent to the town of Squamish, locatedapproximately 45 km north of Vancouver, British Columbia, theSquamish Marsh is influenced by glacially fed water from theSquamish River and by Howe Sound, a saltwater source (Figure 1).In reference to Table I, a weather station adjacent to the marsh inSquamish, B.C., recorded 2247 mm as an average of the total annualprecipitation from 1951-1980. Most of the precipitation occursduring the fall and winter months in the form of rain (2109.5 mm).Temperatures tend to be mild during the spring-summer growingseason (approximately 16°C) and colder during the fall and winter(October to April) (4.6°C) (Environment Canada 1980). While thelower marsh is characterized by vast stands of Carex lvnqbvei, theupper is a mixture of marsh species common to other coastal marshesof the Pacific Northwest such as Potentilla Pacifica, Deschampsia cespitosa, and Triqlochin maritimum (Hutchinson et al. 1989)(Photograph I).2.1.2 Subalpine Wet MeadowApproximately 37 km north of Squamish, is the Black Tuskrecreation area of Garibaldi Provincial Park. A wet meadow areaStudyP7SitePanoramaidGaribaldiLake4 5Km- 16 -Figure 1: Maps showing the location of the tidal marsh and thesubalpine wet meadow study sites.- 17 -Table I: Mean precipitation and temperature data, 1951-1980, forSquamish (tidal marsh) and Alta Lake (subalpine wet meadow)(Environment Canada 1980).SquamishMean^Std. Dev.Alta LakeMean^Std. Dev.Precipitation (mm)May-September 79.6 50.1 53.04 30.32October-April 264.1 99.7 164.3 69.2Total Annual 2247.0 244.3 1415.4 226.1Total Annual Rainfall 2109.5 800.6Total Annual Snowfall 177.1 657.4Temperature (°C)May 11.5 1.0 9.0 1.1June 14.3 1.3 12.5 1.8July 16.6 0.9 15.3 1.5August 16.3 1.3 14.9 1.5September 13.7 1.2 11.9 1.4October-April 4.6 1.3 0.6 1.9- 18 -Photograph I: Photograph of the Squamish Marsh.- 20 -adjacent to Mimulus Lake in the Black Tusk meadows was deemedsuitable for study. The wet meadow site has an aspect of 135°(southeast) and is located well within the Coast Mountain Range atan elevation of approximately 1700 m (Figure 1). A nearby weatherstation at Alta Lake, adjacent to the town of Whistler, B.C., hasrecorded total mean annual precipitation as being approximatelyhalf (1415.4 mm) of that in Squamish. However, Alta Lake, morerepresentative of the subalpine study site, receives almost sixtimes the amount of snow per year (657.4 mm as opposed to 177.1 mmat Squamish) (Table I). Though temperatures are comparably mild,compared to Squamish during the spring and summer months, freezingtemperatures, or at least temperatures close to freezing aretypical of the Black Tusk area during the winter (Table I)(Environment Canada 1980). While the lower portion of the site ischaracterized by a virtually monospecific stand of Carex nigricans other species such as Phyllodoce empetriformis, Cassiope mertensiana, Lupinus latifolius, and Luetkea pectinata become moreabundant in the middle and upper portions of the study area(Photograph II). All of these species are common throughoutGaribaldi Provincial Park (Brink 1959; Archer 1963).2.2 Field Data CollectionDuring July and August of 1990, field data were collectedusing 0.5 X 0.5 m quadrats systematically located every 5 m alongtransects arranged in a 40 X 120 m "systematic grid" at both meadow- 21 -Photograph II: Photograph of the subalpine wet meadow adjacent toMimulus Lake in Garibaldi Park, British Columbia.- 2 2 -- 23 -and marsh sites (Figure 2). Within the 225 quadrats of eachsampling grid, vascular plant species were identified and assignedvalues of 1, 2, 3, 4, or 5 designating their occurrence in one offive aerial coverage classes (<5%, 6-25%, 26-50%, 51-75%, 76-100%).Nomenclature followed Hitchcock and Cronquist (1973). If a largelog, rock or some other obstacle prevented data collection at aspecific location, then the quadrat was positioned to allowappropriate data collection adjacent to the obstacle. Ground levelelevation, relative to the lowest quadrat location (EL), was alsorecorded for each quadrat by using a survey level. Bryophytes andlichens were recorded in the meadow area but, except for Sphagnumwarnstorfii and Cetraria subalpina, were not included in dataanalyses. Bryophyte identification followed the nomenclature ofStotler and Stotler (1977) and Ireland et al. (1987). Bryophytesand lichens were not encountered during sampling in the marsh.In addition to the vegetation data, soil samples werecollected from every other quadrat along alternate transects withinthe sampling grid. In this fashion, sixty-five soil samples ofapproximate dimensions, 15 x 15 x 15 cm, were collected from bothmarsh and meadow sites.2.3 Laboratory Data CollectionSoil samples were air dried (Jackson 1958; Davidescu andDavidescu 1982) in individual plastic trays at approximately 20°C.- 24 -Figure 2:^"Systematic grid" sampling design superimposed ontopographic profiles of marsh and subalpine meadow study sites.Numbers along the X and Y axes refer to transects; the Z axisdenotes elevation in metres.Meadow- 25 -After drying, the samples were gently crushed with a rolling pin tobreak up aggregate soil particles and were subsequently passedthrough a 2 mm sieve. That fraction of soil samples failing topass through the sieve (meadow soils only) was weighed and used todetermine a separate variable (FI) for data analysis (FI thepercentage of "fines" (particles <2 mm) in the total). All 130soil samples from meadow and marsh sites were analyzed for pH,electrical conductivity (EC), carbon content (C), and percent sand(SA) and clay (CY) using procedures outlined by Lavkulich (1981).Soil pH was measured using a 1:2, soil:distilled water ratio,whereas EC measurements were recorded from a 1:2 volume extractfollowing the recommendation of Rhoades (1982).2.4 Data Analysis 2.4.1 Altering the Scale of ObservationInterpretation of community structure as well as community-environment relationships may be a function of scale. The scale ofobservation is dependent on the quadrat size used during fieldsampling. Raw vegetation and environmental data matrices representthe most detailed scale of observation (aggl). Neighbouring casesin both matrices were aggregated in groups of four (agg4), six(width-wise) (agg6a), six (length-wise) (agg6b), and nine (agg9) tosimulate sampling with larger quadrat sizes (5 X 5 metres, 5 X 10metres, 10 X 5 metres, and 10 X 10 metres respectively) (Figure 3).This was equivalent to examining a system .at progressively coarser- 26 -Figure 3:^Simulation sampling with larger quadrat sizes byaggregating neighbouring 0.5 X 0.5 m quadrats (aggl level): agg4(a); agg9 (b); agg6a (c); agg6b (d). '99' denotes omitted quadratsbecausewith.aof an unavailability of neighbouring quadrats to group1 . 1, 2, 2, 3, 3, 4, 4, 99, 1, 1, 1, 2, 2, 2, 3, 3, 3,1, 1, 2, 2, 3, 3, 4, 4, 99, 1, 1, 1, 2, 2. 2, 3, 3, 3,5 , 5, 6, 6, 7, 7, 8, 8, 99, 1, 1, 1, 2, 2, 2, 3, 3, 3,5, 5, 6, 6, 7, 7, 8, 8, 99, 4, 4, 4, 5, 5, 5, 6, 6, 6,9, 9, 10, 10, 11, 11, 12, 12. 99, 4, 4, 4, 5, 5, 5, 6, 6, 6,9, 9, 10, 10, 11, 11, 12, 12, 99, 4, 4, 4, 5, 5, 5. 6, .6, 6,13, 13, 14, 14, 15, 15, 16, 16, 99, 7, 7, 7, 8, 8, 8, 9, 9, 9,0, 13, 14, 14, 15, 15, 16, 16, 99, 7, 7, 7, 8. 8, 8, 9, 9. 9,17, 17, 18, 18, 19, 19, 20, 20, 99, 7, 7, 7, 8, 8, 8, 9, 9, 9,17, 17, 18, 18, 19, 19, 20, 20, 99, 10, 10, 10, 11, 11, 11, 12, 12, 12,21, 21, 22, 22, 23, 23, 24, 24, 99, 10, 10, 10, 11, 11, 11, 12, 12, 12,21, 21, 22, 22, 23, 23, 24, 24, 99, 10, 10, 10, 11, 11, 11, 12, 12, 12,25, 25, 26, 26, 27, 27, 28, 28, 99, 13, 13, 13, 14, 14, 14, 15, 15, 15,25, 25, 26, 26, 27, 27, 28, 28, 99, 13, 13, 13, 14, 14, 14, 15, 15, 15,29, 29, 30, 30, 31, 31, 32, 32, 99, 13, 13, 13, 14, 14, 14, 15, 15, 15,29, 29, 30, 30, 31, 31, 32, 32, 99, 16, 16, 16, 17, 17, 17, 18, 18, 18,33, 33, 34, 34, 35, 35, 36, 36, 99, 16., 16, 16, 17, 17, 17, 18, 18, 18,33, 33, 34, 34, 35, 35, 36, 36, 99, 16, 16, 16, 17, 17, 17, t8, 18, 18,37, 37, 38, 38, 39, 39, 40, 40, 99, 19, 19, 19, 20, 20, 20,'21, 21, 21,37, 37, 38, 38, 39, 39, 40, 40, 99, 19, 19, 19, 20, 20, 20, 21, 21, 21,41, 41, 42, 42, 43, 43, 44, 44, 99, 19, 19, 19, 20, 20, 20, 21, 21, 21,41, 41, 42, 42, 43, 43, 44, 44, 99, 22, 22, 22, 23, 23, 23, 24, 24, 24,45, 45, 46, 46, 47, 47, 48, 48, 99, 22, 22, 22, 23, 23, 23, 24, 24, 24,45, 45, 46, 46, 47, 47, 48, 48, 99, 22, 22, 22, 23, 23, 23, 24, 24, 24,99, 99, 99, 99, 99, 99, 99, 99, 99 99, 99, 99, 99, 99, 99, 99, 99, 99c d1, 1, 1, 2, 2, 2, 3, 3, 3, 1, 1, 2, 2, 3, 3, 4, 4, 99,1, 1, 1, 2, 2, 2, 3, 3, 3, 1, 1, 2, 2, 3, 3, 4, 4, 99,4, 4, 4, 5, 5, 5, 6, 6, 6, 1, 1, 2, 2, 3, 3, 4, 4, 99,4, 4, 4, 5, 5, 5, 6, 6, 5, 5, 5, 6, 6, 7, 7, 8, 8, 99,7, 7, 7, 8, 8, 8, 9, 9, 9, 5, 5, 6, 6, 7, 7, 8, 8, 99,7, 7, 7, 8, 8, 8, 9, 9, 9, 5, 5, 6, 6, 7, 7, 8, 8, 99,10, 10, 10, 11, 11, 11, 12, 12, 12,- 9, 9, 10, 10, 11, 11, 12, 12, 99,10, 10, 10, 11, 11, 11, 12, 12, 12, 9, 9, 10, 10, 11, 11, 12, 12, 99,13, 13, 13, 14, 14, 14, 15, 15, 15, 9, 9, 10, 10, 11, 11, 12, 12, 99,13, 13, 13, 14, 14, 14, 15, 15, 15, 13, 13, 14, 14, 15, 15, 16, 16, 99,16, 16, 16, 17, 17, 17, 18, 18, 18, 13, 13, 14, 14, 15, 15, 16, 16, 99,16, 16, 16, 17, 17, 17, 18, 18, 18, 13, 13, 14, 14, 15, 15, 16, 16, 99,19, 19, 19, 20, 20, 20, 21, 21, 21, 17, 17, 18, 18, 19, 19, 20, 20, 99,19, 19, 19, 20, 20, 20, 21, 21, 21, 17, 17, 18, 18, 19, 19, 20, 20, 99,22, 22, 22, 23, 23, 23, 24, 24, 24, 17, 17, 18, 18, 19, 19, 20, 20, 99,22, 22, 22, 23, 23, 23, 24, 24, 24, 21, 21, 22, 22, 23, 23, 24, 24, 99,25, 25, 25, 26, 26, 26, 27, 27, 27, 21, 21, 22, 22, 23, 23, 24, 24, 99,25, 25, 25, 26, 26, 26, 27, 27, 27, 21, 21, 22, 22, 23, 23, 24, 24, 99,28, 28, 28, 29, 29, 29, 30, 30, 30, 25, 25, 26, 26, 27, 27, 28, 28, 99,28, 28, 28, 29, 29, 29, 30, 30, 30, 25, 25, 26, 26, 27, 27, 28, 28, 99,31, 31, 31, 32, 32, 32, 33, 33, 33, 25, 25, 26, 26, 27, 27, 28; 28, 99,31, 31, 31, 32, 32, 32, 33, 33, 33, 29, 29, 30, 30, 31, 31, 32, 32, 99,34, 34, 34, 35, 35, 35, 36, 36, 36, 29, 29, 30, 30, 31, 31, 32, 32, 99,34, 34, 34, 35, 35, 35, 36, 36, 36, 29, 29, 30, 30, 31, 31, 32, 32, 99,99, 99, 99, 99, 99, 99, 99, 99. 99 99, 99, 99, 99, 99, 99, 99, 99. 99- 27 -scales. Regrettably, it was impossible to aggregate all quadratsfor each aggregation scheme because of an unavailability ofneighbouring quadrats to group with. Where quadrats wereaggregated in groups of four and six (length-wise), the lastquadrat of each transect as well as the last transect itself wereomitted. The last transect was omitted from data analyses wherecases were grouped by nine and six (width-wise). At eachaggregation level or observation scale, mean values of speciesaerial coverage and environmental variables were used to assesscommunity structure and its relationships with environmentalvariables.2.4.2 Classification of SubcommunitiesIn order to examine community structure and how the perceptionof community structure may change at different scales, minimumvariance cluster analysis (MVCA) was performed on both marsh andmeadow vegetation matrices using the MIDAS statistical program (Foxand Guire 1976) on the Amdahl 470 V6-II mainframe computer at theUniversity of British Columbia. Using Euclidean distance as adissimilarity measure, MVCA provided a dendrogram and nine groupingvariables for each scale (Pigure 4). Categorical groupingvariables offer an alternative way of presenting MVCA results.Organized in a matrix, each quadrat was assigned a number dependingon which subcommunity it belonged to at a particular dendrogramlevel. For this purpose, canonical correlation analysis (CCorA)ren ul r co 0)N CO CD m Y: r, c0 •-• CO '7 Ul CD (D 01 ry rncy CU^CV CVCV CV CV CV CV CV CU CV CU CV CV CV CV CV CV^  CV^CV CV CI CV CV VI CI CV CV rn m CV VI rn rlCU^ rn mm vmmvvmrn^rn ,rrq ^ NT CV NI NT LI, Nt NT LC) V1 N: N7 (.0 LC) NT U-) Ul V")w- Q^CV CV CU CV V1 CI Un ul co Ul Ul UD UD In U1 co co Ul co co UDw- V •- CV CU CU CV U) m^u3 r, up up r- r, uD up r- N. up r-^N.at^CV CV CV CV un rn Lc) UD r- UD UP CO CO UD UD a) CD up CO a) N.Un CU rn rn en m UD Y: VD N. CD r■^cn 0) N. r- CP CD N. cn cn co-- up CU PI cr vl Cs VI N. op cn CO OD CD CD 07 OD CD CD W (D C) cno'cl) b•co zUrU VI yr U) UD r- OD cn CD CV rn cr uD N. OD CD CD N rn yr^  CV CV CV CU CV-1 0 t17)•H^• H)-4LI-I 4.4 04-1O 0.10C.) !rr.1c`I UU,Q N(1)al 4-)rC1• a)t:51•H1(O UrCi t7) H1(15-1-) (r)Q)W(ti Cal0)• 44 gO 0 aiTi^t3)0 •• 1(1)r0 • rL,• (19 —^(1)'ZS01)^• H•• • H'`J4 4-)^Q.)05-QaiQ.) 4-1 • H01 4--) 4-19-1H • r-I (Ci[1.4 cc$ "0U0co z- 29 -(Gittins 1985) was used to provide quantitative guidelines forselecting among the nine possible partitionings of the vegetationin the dendrogram for each aggregation level. Application ofCCorA, using MIDAS, involved correlating the covariancerelationships of species with grouping variables corresponding toeach of the nine dendrogram levels. The useful statisticsgenerated from these analyses included canonical correlationcoefficients (R e ), which provided an overall measure of therelationships between species and the grouping variables, andredundancy, defined as the proportion of variation in the speciesvariables explained by a particular grouping variable. Because theRe-values are sensitive to distributional peculiarities in a dataset (Kowalski 1972; Thissen et al. 1981), redundancy estimates wereused to help describe subcommunity zonation.2.4.3 Intervolation of Soil VariablesGiven logistic constraints during the field work, it waspossible to obtain soil samples for only 65 of the 225 quadrats inthe marsh and meadow sampling grids. In order to estimate thevalues of soil variables (pH, EC, C, SA, CY, and FI (meadow only))at quadrat locations where soil_ samples had not been obtained, aninterpolation technique known as kriging (Krige 1966) was used.Kriging, based on regionalized variable theory, is one of the mostreliable interpolation techniques available because it provides notonly an optimal interpolation estimate but also a complementary- 30 -variance estimate (Webster 1985; Robertson 1987). Semivariance andbest fit anisotropic model analyses, performed on the untransformedsoil data by program GS+ (Gamma Design Software 1991) providedpunctual kriged soil variable estimates, and two dimensionalisopleths displaying the zonation pattern of each soil variable.Before proceeding with further multivariate analyses of the soildata, the soil variables were tested for normality with both aKolmogorov-Smirnov one-sample test (Kolmogorov 1933; Smirnov 1939)and a more robust Lilliefors test (Lilliefors 1967).Transformations of soil variables that did not conform to a normaldistribution (log, square root, winsorizing and trimmed means) wereattempted but provided isopleths that were less meaningful andlacked the desirable map details provided by those obtained withthe untransformed data. Two dimensional isopleths, showing theextent of topographic variation within the sampling grids, werealso generated.2.4.4 Vegetation-Environment RelationshipsVegetation-environment relationships were examined withcanonical correspondence analysis (CCA) (Ter Braak1986,1987a,1987b). CCA determines the major axes of compositionalvariation that are also constrained to be linear combinations ofthe environmental variables. For each of the marsh and meadow datasets, a Monte Carlo permutation test was used to evaluate thesignificance of the first canonical axis and the trace statistic- 31 -(summation of eigenvalues of the first four canonical axes). TheCCA's and Monte Carlo permutation tests were performed usingprogram CANOCO (Ter Braak 1987b). Ordination diagrams wereproduced to illustrate graphically the extent of vegetation-environment relationships. The locations of the previouslydetermined subcommunities were shown by superimposing 50%confidence ellipses on the ordination diagrams. CANOCO alsocalculated Pearson correlation coefficients among the environmentalvariables and the CCA axes which greatly assisted interpretation ofthe ordination axes. Correlative relationships were deemedrelevant for those variables with r> Overall Between-Within Cluster Variability - Assessment A subcommunity-environment biplot with subcommunitiesrepresented as 50% confidence ellipses provides a qualitativeassessment of between and within-cluster (subcommunity)variability. In order to evaluate which observation scale providedthe clearest picture of community structure, a quantitativeassessment was deemed necessary. An analysis of variance wasperformed on CCA first and second axis scores using the selectedgrouping variable for stratification. This provided a between,within, and total sum of squares estimate for each axis. For eachaxis, the between sum of squares estimate was divided by the totalsum of squares. The quotient was subsequently multiplied by theaxis eigenvalue with the understanding that each axis did not- 32 -account for 100% of the variation in the data sets. Between sum ofsquares, expressed as a percentage for each axis, were subsequentlyadded together providing an overall between sum of squares value.This value was subtracted from the sum of CCA axes I and IIeigenvalues, ultimately providing an overall percentage of withinsum of squares. For each scale overall percentage estimates ofbetween and within-cluster variability were standardized forpurposes of comparing community structure clarity across scales.Scales that possessed relatively greater between and less within-cluster variability values segregated quadrats into relativelytighter and more independent groups. This was judged to give aclearer picture of community structure as opposed to scales ofrelatively less between and greater within-group estimates sincesubcommunities are not as compact and tend to overlap more.2.4.6 Closer Examining the Effect of ELEL has been recognized as a major determinant of speciescomposition through its interaction with the tidal regime in manymarsh systems. Through its interaction with wind, and their jointinfluence on snow distribution and soil variables, EL has also beenrecognized as a major determin.Lt of species composition in meadowsystems. In view of past findings, it was deemed necessary tocloser examine whether other environmental variables unrelated toEL shared strong correlative relationships with speciescomposition. Using a method outlined by Bradfield and Campbell- 33 -(1986), principal components analysis (PCA) was performed on bothmarsh and meadow vegetation data sets. CCorA was then used toexamine relationships among the first three PCA axes and EL.Canonical axis scores, summarizing the PCA axes, were subsequentlyregressed against the quadrat scores along the canonical axis ofEL. This not only examined the relationship between EL and thevegetation data summarized by the canonical axis but providedresiduals as well. Residuals represent that proportion of thevariation in vegetation unexplained by EL. Residuals were savedand subsequently correlated with the environmental variables toexamine if other variables unrelated to EL shared strongrelationships. This set of correlations was compared to a setbetween the canonical axis scores representing the vegetation dataand environmental variables. Similar correlations between the twosets would suggest a strong lack of EL influence on vegetationpattern.- 34 -CHAPTER 3: RESULTS 3.1 Tidal Marsh and Wet Meadow: A Brief ComparisonIn reference to Figure 5, Whittaker diversity curves indicatethe greater diversity of species in the wet meadow (36 species)compared to the tidal marsh (19 species). The diversity curverepresenting the tidal marsh possesses a steeper slope than themeadow because of the overwhelming dominance of Carex lynqbyei,Potentilla Pacifica, and Triqlochin maritimum. As in the tidalmarsh, the genus Carex is a predominant constituent of the wetmeadow vegetation. Specifically, Carex niqricans and Carexspectabilis are the main species. The more gradually slopingdiversity curve and greater number of species in the subalpinemeadow may suggest the presence of more microhabitats than in themarsh. The names of marsh and meadow species shown in order ofdecreasing abundance of each species, as well as encounteredbryophytes in the wet meadow may be found in Appendix A.3.2 Brackish Tidal Marsh 3.2.1 Aggl Scale (0.5 X 0.5 Metre Observation Unit) Community StructureTable II summarizes the results of applying CCorA todendrogram levels C2-C10 obtained with cluster analysis of the agglscale marsh vegetation data. Dendrogram level C2 provided thelargest redundancy (15.60%) and R c (0.9309) estimates; thus, two10080■Marsh^Meadow604020- 35 -Figure 5: Whittaker diversity curves for the marsh and subalpinemeadow study areas.SPECIES- 36 -Table II: Redundancy and R, estimates for dendrogram levels C2-C10at different scales in the Squamish Marsh. Highest redundancy andRc estimates at each scale are marked with an '*'.Aggl ScaleRedundancy(%) R,Agg4 ScaleRedundancy(%) R„C 2^*15.60 *0.9309 C 2^*29.21 *0.9943C 3 8.71 0.8684 C 3^12.45 0.9646C 4^6.75 0.862 C 4 8.19 0.9615C 5 5.51 0.8424 C 5^6.45 0.9576C 6^5.56 0.8569 C^6 6.09 0.9553C 7 6.9 0.8778 C 7^4.88 0.9609C 8^7.98 0.8826 C 8 6.03 0.9760C 9 7.43 0.8644 C 9^6.66 0.9740C10^6.76 0.8551 C10 6.13 0.9765Agg6a Scale Agg6b ScaleRedundancy(%) R. Redundancy(%)C 2^*44.38 *0.9797 C 2^45.92 0.9694C 3^37.07 0.9382 C 3^42.5 0.9607C 4^29.86 0.9307 C 4^45.34 0.9851C 5^34.63 0.9625 C 5^44.68 0.9799C 6^37.8 0.9599 C 6^45.15 0.9897C 7^40.84 0.9743 C 7^46.22 0.9912C 8^41.6 0.9744 C 8^*47.22 0.9921C 9^41.6 0.9736 C 9^46.97 *0.9932C10^38.81 0.9661 C10^46.43 0.9925Agg9 ScaleRedundancy RC 2 *39.55 0.9686C 3 34.0 0.9479C 4 33.11 0.9493C 5 33.05 0.9606C 6 34.76 0.9521C 7 36.96 *0.9692C 8 36.48 0.9677C 9 37.58 0.9640C10 37.5 0.9688- 37 -subcommunities may be recognized (Figure 6a).^Subcommunity 1represents the lower marsh and is virtually a monospecific stand ofCarex lvnqbvei. Potentilla pacifica also characterizessubcommunity 1 but tends to occur mainly in close proximity to theboundary with subcommunity 2. Subcommunity 2, representative ofthe upper marsh, is characterized by C. lvnqbyei, P. pacifica, aswell as Triqlochin maritimum, Deschampsia cespitosa, and Aqrostis alba (Table III). Because P. pacifica, as a member of subcommunity1, is found only in the vicinity of subcommunity 2, the transitionbetween the two subcommunities tends to be gradual. Environmental Variable RelationshipsEL interacts with many of the measured environmental variablesin the marsh study area. In reference to Table IV, EL ispositively correlated with C (0.58) and SA (0.46) and negativelycorrelated with CY (-0.44). C and SA share a positive correlativerelationship of 0.78 and share negative correlative relationshipswith CY (-0.54 and -0.68 respectively). Because the study site waslocated on a slight incline, more SA and C are prevalent in theupper marsh area compared to the lower marsh where soils containmore CY. Though protected by 'a dyke, the lower marsh is adjacentto a water channel that overflows into the lower marsh zone duringhigh tide. High-low tide alternation may act as a flushingmechanism that may remove much organic content (C) and contributeto CY accumulation in the lower marsh. Because it is relativelyNNNNNN     W7 ^CO ^CO ^CO ^h^h^CO5VC^V^V7 ^W^N^N^CO ^CONV V h N CO W W, r V N W W W W          rNNNNrrrrNNN^NN^NNNN^^NNNNrcirv.h^NNNN^ImNNNr,..NN^NNNNv.,N^r..-NNNN^.-^r^..^N^N^N^N^Nr r N N N N N Nr•^r.^,^N^N^N^N^NN N NN N N N N N N NN N N N N N N N N NN N N N N N N N N N13- 39 -Table III:^Species mean aerial coverage class data forsubcommunities at different scales in the Squamish Marsh. Thosespecies with a mean aerial coverage class estimate > 1 were deemedto be representative of a particular subcommunity. Species namescorresponding to the codes used below may be found in Appendix A.Integers^directly^above mean^and^standard deviation^estimatesrepresent subcommunities at each scale.Aggl Scale1^ 2Mean^Std Dev^Mean^Std DevCARLYN 4.36 1.01 1.22 0.85POTPAC 1.12 1.15 1.73 0.61TRIMAR 0.36 0.79 1.97 1.14DESCES 0.04 0.24 1.19 1.45STEHUM 0.10 0.37 0.76 1.12AGRALB 0.09 0.40 2.65 1.34ASTEAT 0.01 0.15 0.22 0.53TRIWOR 0 0.11 0.39LATPAL 0 0.16 0.44JUNBAL 0.02 0.13 0.78 1.21HORBRA 0.01 0.10 0.56 0.99SONARV 0.01 0.07 0.05 0.23RANCYM 0.12 0.46 0.03 0.16SIUSUA 0.19 0.45 0.03 0.16ELYMSP 0.01 0.10 0.22 0.75CONPAC 0 0.03 0.16ATRPAT 0.06 0.30 0.03 0.16PLANAR 0.02 0.16 0SCIMAR 0.07 0.54 0- 40 -Table III:Agg4 Scale(Continued)1Mean^Std Dev2Mean Std DevCARLYN 4.21 0.88 1.17 0.47POTPAC 1.26 1.01 1.75 0.55TRIMAR 0.40 0.57 1.96 0.62DESCES 0.06 0.25 1.54 0.73STEHUM 0.12 0.27 0.92 0.82AGRALB 0.27 0.59 2.46 0.49ASTEAT 0.01 0.08 0.33 0.49TRIWOR 0 0.17 0.20LATPAL 0 0.25 0.27JUNBAL 0.04 0.13 1.00 0.74HORBRA 0.02 0.12 0.83 0.66SONARV 0.01 0.04 0.08 0.20RANCYM 0.13 0.30 0.04 0.10SIUSUA 0.18 0.32 0.04 0.10ELYMSP 0.01 0.05 0.33 0.61CONPAC 0 0.04 0.10ATRPAT 0.07 0.16 0.04 0.10PLAMAR 0.02 0.09 0SCIMAR 0.01 0.04 0Agg6a Scale1Mean Std Dev2Mean Std DevCARLYN 1.57 0.91 4.16 0.93POTPAC 1.67 0.57 1.20 0.95TRIMAR 1.93 0.63 0.45 0.54DESCES 1.50 0.72 0.03 0.09STEHUM 0.90 0.69 0.10 0.15AGRALB 2.20 0.52 0.26 0.60ASTEAT 0.27 0.43 0.01 0.06TRIWOR 0.13 0.22 0LATPAL 0.20 0.28 0JUNBAL 0.97 0.57 0.02 0.05HORBRA 0.77 0.45 0SONARV 0.07 0.15 0.01 0.03RANCYM 0.03 0.08 0.12 0.24SIUSUA 0.07 0.09 '0.18 0.28ELYMSP 0.27 0.43 0.01 0.04CONPAC 0.03 0.08 0ATRPAT 0.03 0.08 0.06 0.13PLAMAR 0 0.02 0.07SCIMAR 0 0.07 0.30- 41 -Table III: (Continued)Agg6b Scale1^ 2^ 3Mean Std Dev^Mean Std Dev^Mean Std DevCARLYN 2.33 0.17 1.00 0.50POTPAC 1.72 0.19 2.17 1.67TRIMAR 1.94 0.54 2.33 1.50DESCES 0.94 0.54 2.33 1.50STEHUM 1.00 0.58 0.83 0.33AGRALB 2.17 0.44 2.33 2.00ASTEAT 0 0.17 1.00TRIWOR 0.06 0.10 0.33 0.17LATPAL 0.06 0.10 0.17 0.50JUNBAL 0.50 0.44 1.33 1.50HORBRA 0.44 0.42 1.00 1.50SONARV 0.11 0.19 0 0RANCYM 0.11 0.10 0 0SIUSUA 0.11 0.10 0 0ELYMSP 0.06 0.10 0 1.33CONPAC 0.06 0.10 0 0ATRPAT 0.06 0.10 0 0PLAMAR 0 0 0SCIMAR 0 0 04 5 6Mean Std Dev Mean Std Dev Mean Std DevCARLYN 3.00 0.43 3.17 0.23 2.67POTPAC 2.37 0.09 1.58 0.35 0TRIMAR 0.92 0.29 0.33 1.67DESCES 0.25 0.32 0.09 0.12 0STEHUM 0.38 0.31 0.25 0.11 0.17AGRALB 1.25 0.35 0.58 0.35 0ASTEAT 0.04 0.09 0.17 0.23 0TRIWOR 0 0 0LATPAL 0.04 0.09 0 0JUNBAL 0.04 0.09 0 0HORBRA 0 0 0SONARV 0 0.09 0.12 0RANCYM 0.13 0.09 0.75 0.35 1.00SIUSUA 0.04 0.09 0.09 0.12 0.17ELYMSP 0.04 0.09 0 0CONPAC 0 '0 0ATRPAT 0.13 0.16 0.50 0.24 0PLAMAR 0 0.25 0.11 0SCIMAR 0.04 0.09 0 0- 42 -Table III: (Continued)Agg6b Scale (Continued)7^ 8Mean Std Dev^Mean Std DevCARLYN 4.31 0.46 4.83 0.27POTPAC 2.19 0.42 0.40 0.43TRIMAR 0.33 0.33 0.08 0.20DESCES 0 0STEHUM 0.02 0.06 0.04 0.10AGRALB 0.05 0.12 0.01 0.05ASTEAT 0 0TRIWOR 0 0LATPAL 0 0JUNBAL 0.05 0.08 0.01 0.05HORBRA 0 0SONARV 0 0RANCYM 0 0.03 0.06SIUSUA 0.38 0.21 0.14 0.20ELYMSP 0 0CONPAC 0 0ATRPAT 0 0.03 0.06PLAMAR 0 0SCIMAR 0 0Agg9 Scale1 2Mean Std Dev Mean Std DevCARLYN 1.56 0.84 3.56 0.84POTPAC 1.75 0.46 1.72 0.99TRIMAR 1.97 0.54 0.61 0.53DESCES 1.38 0.77 0.05 0.14STEHUM 0.88 0.74 0.16 0.27AGRALB 2.25 0.60 0.48 0.73ASTEAT 0.25 0.44 0.03 0.13TRIWOR 0.13 0.19 0LATPAL 0.19 0.26 0JUNBAL 0.84 0.72 0HORBRA 0.72 0.63 0SONARV 0.06 0.18 0.02 0.06RANCYM 0.06 0.12 0.28 0.45SIUSUA 0.06 0.12 t.09 0.15ELYMSP 0.25 0.53 0.03 0.09CONPAC 0.03 0.09 0ATRPAT 0.03 0.09 0.16 0.22PLAMAR 0 0.05 0.14SCIMAR 0 0.02 0.06- 43 -Table IV: Pearson correlations between environmental variables atdifferent scales in the Squamish Marsh. EL, relative ground levelelevation; C, carbon content; EC, electrical conductivity; SA, sandAgg1Ccontent; CY,ScaleELclay content; pH,C^pHsoil acidity.EC^SA0.5766pH 0.1421 0.0104EC -0.0993 0.1655 -0.2036SA 0.4565 0.7788 0.0310 0.2550CY -0.4437 -0.5377 -0.0040 -0.2670 -0.6802Agg4 ScaleEL C pH EC SAC 0.6541pH 0.4130 0.1470EC 0.0869 0.3973 -0.3510SA 0.6523 0.9100 0.2358 0.2913CY -0.6811 -0.8241 -0.1916 -0.4066 -0.8904Agg6a ScaleCpHECSACYEL C pH EC SA0.69320.3426-0.01680.6093-0.66690.04590.36520.8955-0.8268-0.42310.0404-0.01500.3940-0.4781 -0.9051Agg6b ScaleCEL C pH EC SA0.7170pH 0.4882 0.0839EC 0.1186 0.4446 -0.3634SA 0.6982 0.9292 0.1161 0.3499CY -0.7211 -0.8585 -0.1014 -0.4440 -0.9245Agg9 ScaleEL C pH EC SAC 0.7457pH 0.4646 0.0285EC 0.0405 0.4092 -0.5524SA 0.6545 0.9147 -0.0999 0.4489CY -0.6933 -0.8437 0.0781 -0.5219 -0.9321- 44 -higher and farther away from a main water channel, the upper zonemay be less disturbed, allowing organic content accumulation andreceiving relatively less CY. Community-Environment RelationshipsThe results from CCA are summarized as a subcommunity-environment biplot in Figure 7a. The arrangement of subcommunitiesindicates their similarities in relation to the main axessummarizing variation in vegetation and environmental conditions.Biplot interpretation also involves examination and comparison ofthe environmental vectors whose lengths indicate the relativeimportance of the different variables. The greater the vectorlength, the stronger the correlative relationship between thatenvironmental variable and subcommunity(ies), relative to otherenvironmental variables. Each environmental vector points in thedirection of maximum change of that environmental variable (TerBraak 1987a, 1987b).The first and second ordination axes of the CCA haveeigenvalues of 0.38 and 0.13 respectively. The first canonicalordination axis eigenvalue and trace statistic are both significant(p<0.05). Figure 7a shows two very distinct subcommunities.Vector CY is the only environmental variable that points in thedirection of subcommunity 1. Its direction as well as its lengthemphasize the relationship between subcommunity 1 and CY. CY is a1.5a 2.5CYCo•z(-0.50.5EC- 45 -Figure 7: Subcommunity-environment biplots at different scales forthe marsh study site: aggl (a), agg4 (b), agg9 (c), agg6a (d), andagg6b (e).^Subcommunities are represented by 50% confidenceellipses.^Ellipses were unable to be produced where thosesubcommunities were represented by three or fewer sampling units.Each environmental variable is represented by a vector. EL,relative ground level elevation; C, carbon content; EC, electricalconductivity; SA, sand content; CY, clay content; pH, soil acidity.-1.5-2.5 ^-2.5^-1.5^-0.5^0.5^1.5^2.5AXIS 11.40.7-0.7AXIS 1Figure 7: (Continued)b 2.1- 46 -EC-1.4CY-2.1c 2.21.1N(0Xa0.0CY-1 . 1-2.2-2.1^I^I^1^I^I^I-1.4 -0.7^0.0^0.7^1.4^2.1CI^I^I^ I-2.2^-1.1^0.0^1.1^2.2AXIS 156ECpHFigure 7: (Continued)d 2.2 -- 47 -1.1ECpHt 1^1^10.0^1.1^2.2AXIS 1-2.2e 2.2-2.21.1NwXQ0.0CY-1 .1 -2.2-2.2^-1.1^0.0^1.1^2.2AXIS 1- 48 -positive correlate of the first species axis (0.64) (Table V).Referring to Table VI, average CY content found in subcommunity 1is 29.7% as opposed to 21.2% in subcommunity 2 (upper marsh).However, subcommunity 2 soils possess relatively more SA (24.0%)and C (11.6%) compared to the lower marsh (15.3% and 8.5%respectively). In Figure 7a, vectors representing EL, C, and SAare also relatively long and point in the direction of subcommunity2; moreover, they share negative correlative relationships with thefirst species axis (-0.63, -0.67, and -0.70 respectively) (TableV). Figures 8a, b, and c display CY, SA, and C zonation patternsrespectively, reaffirming upper marsh soils as possessing greaterSA and C content and lower marsh soils possessing more CY.While EL is an important determinant of species axis 1, it mayalso be responsible for the positive correlative relationshipbetween EC and species axis 2 (0.34) (Table V). The EL gradient inthe marsh lacks consistent step-wise elevation change. Figure 8ddisplays a low area near the upper marsh where EL is approximatelyequivalent to that of the lower marsh (0.42-0.61 m). As a result,the low area near the upper marsh receives more salt water per serelative to most of the study site, but is not exposed to theflushing process unique to the lower marsh. Hence, this area hasa preponderance of salt deposits (Figure 8e).- 49 -Table V: Pearson correlations between environmental variables andspecies axes I and II at different scales in the Squamish Marsh.EL, relative ground level elevation; C, carbon content; EC,electrical conductivity; SA, sand content; CY, clay content; pH,soil acidity.Aggl Scale^ Agg4 ScaleSPP AXIS 1 SPP AXIS 2 SPP AXIS 1 SPP AXIS 2EL -0.6274 -0.1067 0.7479 -0.2881C -0.6667 -0.1310 0.8004 0.1881pH -0.1806 -0.1491 0.4354 -0.5772EC -0.1643 0.3390 0.2395 0.4055SA -0.7003 0.0602 0.8735 0.2110CY 0.6444 -0.1420 -0.8813 -0.2023Agg6a ScaleSPP AXIS 1 SPP AXIS 2Agg6b ScaleSPP AXIS 1 SPP AXIS 2EL 0.7337 -0.1934 0.8305 -0.1697C 0.7820 -0.0403 0.8228 0.2254pH 0.3822 -0.4472 0.3434 -0.6612EC 0.2728 0.5489 0.2778 0.4595SA 0.8536 0.1206 0.9098 0.1656CY -0.8655 -0.1992 -0.9071 -0.2612Agg9 ScaleSPP AXIS 1 SPP AXIS 2EL 0.8405 -0.1648C 0.8175 -0.0640pH 0.2596 -0.5401EC 0.2489 0.6638SA 0.8797 0.0940CY -0.8898 -0.2421- 50 -Table VI:^Summarized environmental data for the mainsubcommunities recognized at the different scales of analysis (agglevels) in the tidal marsh. Integers directly above mean andstandard deviation estimates represent subcommunities at eachscale. EL, relative ground level elevation; C, carbon content; EC,electrical conductivity; SA, sand content; CY, clay content; pH,soil acidity.Agg1 Scale1Mean Std Dev2Mean Std DevEL (m) 0.57 0.17 0.84 0.23C (%) 8.48 1.96 11.61 2.35pH 5.31 1.11 5.80 1.02EC (mmhos/cm) 4.25 1.40 4.74 1.09SA (%) 15.25 4.73 24.00 4.24CY (%) 29.67 6.11 21.21 3.13Agg4 Scale1Mean Std Dev2Mean Std DevEL (m) 0.60 0.15 0.88 0.16C (%) 8.72 1.74 12.04 1.46pH 5.37 0.59 6.07 0.14EC (mmhos/cm) 4.27 0.87 4.65 0.24SA (%) 15.26 3.89 25.00 2.66CY (%) 29.82 4.17 20.11 1.72Agg6a Scale1Mean Std Dev2Mean Std DevEL (m) 0.84 0.18 0.59 0.14C (%) 11.81 1.07 8.66 1.70pH 5.98 0.24 5.29 0.58EC (mmhos/cm 4.70 0.50 4.36 1.02SA (%) 25.16 2.91 15.46 3.75CY (%) 20.43 1.39 29.52 4.14- 51 -Table VI: (Continued)Agg6b Scale1^ 2^ 3Mean Std Dev^Mean Std Dev^Mean Std DevEL (m)^0.76^0.06 0.92 1.01C^(%) 11.50^2.36^12.40^12.07pH 5.57^0.60 6.09 6.23^EC (mmhos/cm) 4.13^0.28^4.72 4.82SA (%)^23.87^5.63 25.01^25.27CY (%) 21.36^3.23^20.83 18.374 5 6Mean Std Dev^Mean Std Dev^Mean Std DevEL (m)^0.76^0.03 0.65^0.01 0.51C^(%) 10.49^0.98^11.22^1.12^8.61pH 5.69^0.51 5.03^0.74 3.95EC (mmhos/cm) 4.58^0.63^5.21^0.36^6.29SA (%)^20.00^1.90 19.66^0.43 16.46CY (%) 25.54^2.65^24.28^1.67^24.457 8Mean Std Dev^Mean Std DevEL (m)^0.70^0.02 0.49^0.13C^(%) 8.67^0.88^7.67^1.26pH 5.65^0.08 5.34^0.37EC (mmhos/cm) 3.83^0.54^4.18^0.73SA (%)^14.11^1.48 13.15^1.84CY (%) 30.70^1.90^32.46^2.50Agg9 Scale1 2Mean Std Dev^Mean Std DevEL (m)^0.75^0.15 0.57^0.13C^(%) 10.89^1.80^8.21^1.27pH 5.29^0.86 5.43^0.36EC (mmhos/cm) 4.93^0.99^4.15^0.78SA (%)^22.17^4.10 14.13^2.24CY (%) 22.55^3.03^31.11^2.8458^120Metres12041.020.6(L)0.30 '58Metres- 52 -Figure 8:^Two-dimensional isopleths displaying EL and soilvariable zonation patterns at the aggl scale in the marsh studysite: CY, clay content (a); SA, sand content (b); C, carboncontent(c), EL, relative ground level elevation (d); EC, electricalconductivity (e); and pH, soil acidity (f).a^ CY41.0$4+a^20.6C0.30<21.90^<25.51bI^I<29.13^<32.74^<36.36 v.SAC4C1<14.76 <18.73 <22.70^<26.67^<30.64 xI^1<7.33^<8.90^<10.48^<12.06^<13.64 x12058MetresI^1 Pam111<0.81^<1.01 Metres<0.41 <0.61<0.20d^ EL43.6• 22.0I)58^120Metres12041.0• 20.4-)0.30 58Metres- 53 -Figure 8: (Continued)e^ ECI^I Emmil^<3.83^<5.29^<6.75^<8.21^<9.67 Mmhos/cmf^ pH41.0^$.^20.60.30 1 himmi<4.80 <5.21 <5.62 <6.03^<6.44- 54 -3.2.2 Agg4 Scale (5 X 5 Metre Observation Unit Community StructureAt agg4, level C2 in the dendrogram yields the highestredundancy (29.21%) and R c (0.9943) values (Table II). Again, twosubcommunities are recognized in the marsh (Figure 6b). WhileCarex lynqbyei and Potentilla Pacifica continue to berepresentative species of subcommunity 1 (lower marsh), the samespecies mentioned at agg1 as well as Juncus balticus characterizethe upper marsh (Table III). Environmental Variable RelationshipsIn reference to Table IV, EL shares stronger correlativerelationships with C, SA, and CY (0.65, 0.65, and -0.68respectively) compared to agg1. At this scale, EL clearly sharesa positive correlative relationship with pH (0.41) as well. WhileSA and CY maintain stronger relationships with C (0.91 and -0.82respectively), EC is well correlated not only with C (0.40) butwith pH (-0.35) as well. In addition, EC as well as SA sharenegative correlative relationships with CY (-0.41 and -0.89respectively).An EL gradient is also recognized at agg4 where greater SA, Cand less acidic soils are found in the upper marsh. Conversely,the lower end of the gradient contains relatively more CY.- 55 - Community-Environment RelationshipsThe first and second ordination axes of the CCA haveeigenvalues of 0.45 and 0.08 respectively. The first canonicalordination axis eigenvalue and trace statistic are significant(p<0.05).Figure 7b tells an almost identical story to that previouslydescribed at agg1. Stronger correlations between CY, SA, EL, C,and the first species axis are apparent at this scale (-0.88, 0.87,0.75, and 0.80 respectively) (Table V). All three environmentalvectors representing SA, EL, and C pass through subcommunity 2(Figure 7b). Moreover, subcommunity 2 contains more C and SA andless CY than subcommunity 1 (Table VI).A positive correlative relationship between pH and speciesaxis 1 is stronger at this scale (0.44) (Table V). Figure 7b -suggests that the relative importance of pH is better observed atagg4. The length of the pH vector is not as short in relation toSA, C, and EL vectors as opposed to Figure 7a. Differences in pHbetween subcommunities are not as clear at a finer scale (aggi)(Table VI) mostly because pH mdasurements within each subcommunityare quite variable (high standard deviation estimates). At agg4 amore noticeable difference between subcommunities is evident(relatively lower standard deviation estimates). Whilesubcommunity 2 is located in relatively less acidic soils (6.1),- 56 -lower marsh soils are slightly lower (5.4) (Table VI).Though EC is a weak correlate of species axis 1 (0.24) (TableV), a slight salinity difference between subcommunity 1 and 2 isapparent (4.3 mmhos/cm and 4.7 mmhos/cm respectively) (Table VI).As in agg1, EC shares a stronger correlative relationship, however,with species axis 2 (0.41) (Table V), emphasizing the effect of thelow area near the upper marsh on EC. At agg4, pH is also clearlyrecognized as a correlate of species axis 2 (-0.58) (Table V).Though Figure 8f displays how pH is clearly affected by the lowarea near the upper marsh at agg1, pH and the second species axisshared a rather weak correlative relationship (0.15) (Table V).Changing observation graininess from aggl to agg4 has brought outan important feature whose view may have been obstructed among muchnoise and clutter at aggl. CY is not well correlated with thesecond species axis (-0.20) perhaps suggesting that itsdistribution remains for the most part unaffected by a lack ofconsistency in EL change (Figure 8a). Nevertheless, the unique lowarea near the upper marsh possesses soils of relatively less CYpossibly suggesting the absence of a flushing mechanism.At agg4 ellipses 1 (lower' marsh) and 2 (upper marsh) (Figure7b) are respectively smaller and larger than the correspondingellipses at aggl (Figure 7a). More variability is included in thelower marsh at aggl whereas within-assemblage variability forellipse 2 appears to be less. Because of simulation sampling at a- 57 -coarser scale (agg4), some of the variability has been transferredfrom the low marsh at aggi, to what is now defined as the uppermarsh at agg4 possibly better defining a gradual transition betweenupper and lower subcommunities (Figure 7b). In reference to Figure9, standardized estimates of (overall) between and within-clustervariability suggest that agg4 defines community structuremarginally better than aggi. Between-assemblage variability isslightly greater (57.9%) and within-assemblage variability is less(42.1%) at agg4 as opposed to aggl (5 .4.4% and 45.6% respectively).Noteworthy is a close association between C and SA vectors inFigure 7b and a close association between EL and C vectors inFigure 7a. To suggest that C and SA variables have a closerelationship at agg4 but less so at aggi should be supported byPearson correlations showing the same trend. Such evidence islacking. At both scales, SA always has a stronger correlativerelationship with C than C and EL (Table IV).3.2.3 Acm9 Scale (10 X 10 Metre Observation Unit) Community StructureAt agg9, C2 (39.55%) and C7 (0.9692) provide the highestredundancy and Rc estimates respectively (Table II). Similar toaggl and agg4, the marsh study site is divided into twosubcommunities (Figure 6c). Both Carex lvngbvei and Potentilla Pacifica characterize the lower marsh (subcommunity 2) and the same- 58 -Figure 9: Overall between and within-cluster variability estimatesfor all scales in the marsh study site. Unstandardized estimatesshown along inner isoclines; standardized estimates shown alongouter isocline.10080Agg9AgglAgg6a^Agg4gg9Aggl^gg6aAgg4Agg6b.Agg620 40 60 80 100BETWEEN (Z)604020- 59 -species excluding Juncus balticus described at aggi and agg4characterize the upper (subcommunity 1) (Table III). Environmental Variable RelationshipsEnvironmental interactions at this scale reinforce what hasbeen revealed at agg4. Generally, correlations are stronger to the' extent of emphasizing an additional interaction. SA and EC sharea positive correlative relationship (0.45) perhaps due to the lowelevated area located close to the upper marsh where there isrelatively greater SA, EC, and less CY (Table IV). Community-Environment RelationshipsCCA reported eigenvalues of 0.45 and 0.08 for the first andsecond ordination axes. The first canonical ordination and tracestatistic are significant (p<0.05) as well.Figure 7c reveals an almost identical story to that describedin agg1 and agg4. Supported by stronger correlations with speciesaxis 1 (Table V), SA, EL, and C vectors are associated withsubcommunity 1, and subcommuriity 2 (lower) is associated withvector CY. Relatively more SA and C are found in the upper marshand more CY is characteristic of the lower (Table VI).At agg4 pH was noticeably well correlated with species axis 1- 60 -(0.44).^At agg9, however, the correlative relationship hasweakened (0.26) (Table V). The upper marsh possessed slightly lessacidic soils at agg4 compared to the lower (6.1 and 5.4respectively). At agg9, however, the difference between upper andlower subcommunities is more subtle (5.3 and 5.4 respectively)(Table VI). Ellipses representing the upper marsh haveconsistently increased, and ellipses representing the lower marshhave consistently decreased at progressively coarser scales (agg1,4, to agg9) (compare a, b, and c of Figure 7). At agg9 (Figure7c), ellipse 1 (upper marsh) has ballooned in size compared toellipse 2 (upper marsh) (Figure 7b). In contrast is ellipse 2 atagg9 (lower marsh) (Figure 7c) which is better defined as amonospecific stand of Carex lvnqbvei. Its within-assemblagevariability is considerably less compared to ellipse 1,representative of the lower marsh at agg4 (Figure 7b). Becausevariability is disproportionately distributed to a greater degreeat agg9 as opposed to agg1 and agg4, overall, standardized within-assemblage variability is greater (49.40%) than aggi and agg4estimates, which in turn has decreased variability between groups(50.60%) (Figure 9). Sampling with a larger quadrat (agg9) hasincorporated more variability into the upper zone. Variability maybe attributed mostly to veget4tion structure affected by the lowelevated area described at finer scales as being located near theupper marsh. This may also disrupt the correlative relationshipbetween pH and species axis 1 (Table V) and explain a subtle pHdifference between upper and lower zone soils (Table VI).- 61 -3.2.4 Agg6a and Agg6b Scales (5 X 10 Metre and 10 X 5 Metre) Observation Units Community Structure and Environmental VariableRelationshipsWhen a 5 X 10 meter quadrat (agg6a) is imposed on the studysite, dendrogram level C2 provides the highest redundancy (44.38%)and Rc (0.9797) values (Table II). Thus, the marsh is partitionedin two subcommunities (Figure 6d). Similar to agg1, agg4, and agg9scales, the two subcommunities represent upper and lower marshzones, and are well represented by the same species at agg9 (TableIII).Simulation sampling with a quadrat of same dimensions butpositioned length-wise (agg6b) provides the highest redundancy and13., values at C8 (47.22%) and C9 (0.9932) respectively (Table II).At this scale, the marsh is composed of eight distinctive groups(Figure 6e). In reference to Table III, subcommunities 1 to 3 arecharacterized by Potentilla pacifica, Triqlochin maritimum, andAqrostis alba. While Carex lvnqbvei and Stellaria humifusa arefound in 1, C. lvnqbvei is also found in 2 along with Deschampsia cespitosa, Juncus balticus', and Hordeum brachvantherum.Subcommunity 3 is also well represented by J. balticus, H.brachvantherum, D. cespitosa, Elvmus sp., and Aster eatonii.Subcommunities 4, 5, and 6 are characterized by C. lvnqbvei. Inaddition, subcommunity 4 is also represented by Potentilla - 62 -pacifica, Triqlochin maritimum, and Aqrostis alba. P. pacifica isalso found in subcommunity 5 and T. maritimum, and Ranunculus cvmbalaria are found in subcommunity 6 as well. Subcommunities 7and 8 represent the lower marsh: 7 characterized almostexclusively by C. lvnqbyei and P. pacifica, and 8 characterized byC. lvnqbyei.Simulation sampling with both a 5 X 10 meter quadrat and a 10X 5 meter quadrat reveal very similar environmental variablerelationships to aggl, agg4, and agg9 scales (Table IV). Community-Environment RelationshipsFirst and second axes eigenvalues of 0.44, 0.13 for agg6a and0.46, 0.08 for agg6b as well as a significant (p<0.05) firstcanonical axis eigenvalue and trace statistic were reported fromCCA. Environmental variable first and second species axesrelationships are the same not only between agg6a and agg6b butalso the same as that previously described for agg4 (Table V)(Figures 7d and e).Simulation sampling at ag.46a recognized the same subcommunitynumber as aggl, agg4, and agg9. However, at agg6b perception ofonly two subcommunities appears to be lost. Interestingly, agg6bprovides the clearest overall perception of community structure.In reference to Figure 9 between-assemblage variability is much- 63 -greater (88.5%) and overall within-assemblage variability is muchless (11.5%) than other scales.3.2.5 EL Influence Verification3.2.5.1 Aggl, 4, 6a, 6b, 9 ScalesCorrelations between environmental variables and a canonicalcorrelation axis representing species variables (summarized bythree PCA axes) as well as residuals, confirm EL influence on soilcharacteristics (Table VII). Environmental variable-canonical axiscorrelations reveal very similar trends previously discussed at allscales. Environmental variable-residual correlations are not asstrong as environmental variable-canonical axis correlations.However, several environmental variable-residual correlations arequite respectable perhaps suggesting the existence of an ELgradient with less influence on edaphic factors. Independent ofEL, SA, CY (soil texture) and C (organic content) share noticeablecorrelative relationships with residuals for all scales (TableVII).3.2.6 Tidal Marsh DiscussionThe Squamish marsh study site is generally composed of twosubcommunities: upper and lower. Generally, this corresponds withHutchinson et al.'s (1989) general description of the Squamishestuary where the lower zone is virtually a monospecific stand of- 64 -Table VII: Pearson correlations at different scales betweenenvironmental variables, residuals, and a canonical axisrepresenting species variables in the Squamish Marsh. AXIS,canonical correlation axis; RESD, residuals; EL, relative groundlevel elevation; C, carbon content; EC, electrical conductivity;SA, sand content; CY, clay content; pH, soil acidity.Agg1 ScaleEL^C^pH^EC^SA^CY AXIS 0.7211^0.6363^0.1349^0.0377^0.5599 -0.4444RESD 0^0.2879^0.0583^0.1667^0.3454 -0.2126Agg4 ScaleEL^C^pH^EC^SA^CY AXIS 0.8466^0.7516^0.3923^0.0917^0.7340 -0.7275RESD 0^0.3278^0.1580^0.1400^0.3210 -0.2928Agg6a ScaleEL^C^pH^EC^SA^CY AXIS 0.8586^0.7561^0.3850^0.0161^0.6824 -0.6830RESD 0^0.3051^0.2316^0.1387^0.3041 -0.2531Agg6b ScaleEL^C^pH^EC^SA^CY AXIS 0.9103^0.7810^0.4536^0.0897^0.7791 -0.7332RESD 0^0.2711^0.1914^0.0610^0.3448 -0.2178Agg9 ScaleEL^C^pH^EC^SA^CY AXIS 0.9075^0.7844^0.4080^0.0295^0.7250 -0.7063RESD 0^0.2542^0.1228^0.0640^0.3323 -0.2513- 65 -Carex lvnqbvei and the upper zone is a mixture of wetland speciessuch as Potentilla pacifica, C. lvnqbvei, Triqlochin maritimum,Juncus balticus, and Deschampsia cespitosa. Communities of similarspecies composition have been documented also in a brackish marshon Lulu Island in Richmond, British Columbia (Hutchinson 1982) andin a fjord head marsh in northern coastal British Columbia(Campbell and Bradfield 1989). Furthermore, virtually monospecificstands of Carex lvnqbvei have been extensively described alongtidal marshes in the Pacific Northwest (Disraeli and Fonda 1979;Dawe and White 1982; Ewing 1983; Vince and Snow 1984). At allscales, upper marsh soils are associated with greater SA and Cwhile lower marsh soils possess more CY. The predominance of C.lvnqbvei on clayey soils has also been documented for a brackishintertidal marsh in the Puget Sound area of Washington (Ewing1983). Because the lower zone is closest to a water channel, high-low tide alternation may be responsible for CY accumulation and Cremoval. In general, the vertical distribution of vegetation didnot share strong correlative relationships with EC. This may be indisagreement with Hutchinson et al. (1989) since they foundDeschampsia in more saline locations. However, this result is inagreement with similar marsh research done in a brackish marsh inBellingham Bay, Washington (Didraeli and Fonda 1979) and the LittleQualicum River estuary, Vancouver Island, British Columbia (Daweand White 1982) where salinity was found to play a very minor rolein the vertical distribution of communities. Though communitystructure for the most part corresponds well with EL, a presumed EL- 66 -gradient lacked consistent step-wise elevation change. One sucharea was located near the upper marsh that shared approximately thesame EL as the lower zone, yet lacked a tidal flushing mechanism.As a result, salt deposition and lower pH characterized this area.Generally, upper marsh soils had slightly greater pH than lower atmost scales but can only be perceived as having a minor role indetermining species composition. A somewhat "flawed" EL gradientmay explain strong correlations between residuals data and edaphicfactors. Specifically, tides and the decomposition of the organicmaterial provided by the species themselves may contribute todifferences in soil texture (SA and CY) and organic content (C)between upper and lower subcommunities which in turn sharerelationships with plant distribution patterns.Employment of different sampling units as well as CCorA toselect among nine possible subcommunity schemes per MVCA revealdifferent aspects of community structure. At aggl the upper marshis represented by ellipse 2 in Figure 7a which is smaller thanellipse 1. That is, within-assemblage variability is greater inthe lower as opposed to the upper marsh. At agg4, (Figure 7b)ellipse 2, representing the upper, and ellipse 1, representing thelower, have increased and 'decreased in size respectively.Employing a square quadrat of larger area to sample the marsh sitehas included some of the variability inherent in the lower at agglinto the upper at agg4. Perhaps at this scale, a smooth transitionbetween upper and lower is best observed. Overall within-- 67 -assemblage variability is shown to be less and between-assemblagevariability is greater at agg4 than aggl (Figure 9), emphasizingthe redistribution of variation from the lower to the upper ellipsebetween aggi and agg4 scales. A monospecific stand of Carexlvngbvei is best represented at agg9 (Figure 7c). Here, within-assemblage variability is considerably less for the lowersubcommunity (ellipse 2) as opposed to ellipse 1 where upper marshboundaries have shifted seaward incorporating the aforementionedarea of EL gradient inconsistency.3.3 Subaltine Wet Meadow3.3.1 Aggl Scale (0.5 X 0.5 Metre Observation Unit) Community StructureAt the agg1 observation scale, the two group level yields thehighest IR, value(0.9426) and the five group level yields thehighest redundancy value (8.86%) (Table VIII). Five reasonablydistinct subcommunities may be recognized (Figure 10a).Subcommunities 1 and 2 are located in the lower meadow area besideMimulus Lake. Whereas subcommunity 2 is predominately amonospecific stand of Carex nicrricans, subcommunity 1 contains thisspecies in mixture with Leptatrhena pyrolifolia, Caltha biflora,Epilobium anagallidifolium, and Agrostis thurbergiana (Table IX).Between high and low meadow areas is subcommunity 3 (Figure 10a)characterized by Carex nicrricans, Caltha biflora, Luetkea pectinata, Ericreron peregrinus, Cassiope mertensiana, and- 68 -Table VIII:^Redundancy and R, estimates for dendrogram levels C2-C10^at^different^scales^in the subalpine wet meadow.^Highestredundancy and R, estimates at each scale are marked with an 'Aggl Scale Agg4 ScaleRedundancy(%)^R, Redundancy(%)^R,C 2^8.34 *0.9426 C 2^20.39 0.9723C 3 8.82 0.9317 C 3^*22.06 0.9725C 4^8.53 0.9407 C 4^20.40 0.9762C 5^*8.86 0.9404 C 5^21.72 0.9745C 6 7.89 0.9264 C 6^21.28 0.9730C 7^8.03 0.9247 C 7^20.48 0.9685C 8 8.43 0.9297 C 8^21.34 0.9724C^9^8.76 0.9329 C 9^21.38 *0.9789C10 8.42 0.9265 010^21.04 0.9752Agg6a Scale Agg6b ScaleRedundancy(%) R,C 2^23.14 0.9826 C 2^21.40 0.9907C 3^*26.08 0.9889 C 3^24.10 0.9893C 4^23.37 0.9899 C 4^23.88 0.9853C 5^25.40 0.9896 C 5^22.10 0.9908C 6^24.68 0.9896 C^6^23.36 0.9929C 7^25.26 0.9916 C 7^24.38 0.9937C 8^25.22 0.9924 C 8^*24.69 0.9934C 9^24.79 0.9936 C 9^24.29 0.9937C10^25.47 *0.9948 010^24.41 *0.9941Agg9 ScaleRedundancy(%)C 2^26.06 0.9883C 3^*28.71 0.9963C 4^25.81 0.9987C 5^26.87 *0.9994C 6^28.36 0.9989C 7^28.22 0.9989C 8^28.14 0.9988C 9^27.94 0.9984C10^27.22 0.99781^ 11^ 11^ 1^ 22^ 1^ 22^ 2^ 23 3 33 3 33 3- 69 -Figure 10: Grid maps showing subcommunitY layout at differentscales in the subalpine wet meadow study site: aggl (a), agg4 (b),agg9 (c), agg6a (d), agg6b (e).a b1 3 4 4 5 5 5 5 44 4 5 4 4 4 4 5 44 4 4 4 4 4 4 5 44 4 4 4 4 4 5 4 54 4 4 4 4 4 4 4 54 5 4 4 5 4 4 4 44 4 4 5 4 4 3 3 34 4 4 4 4 4 3 4 34 4 3 5 5 4 3 3 34 4 3 5 5 4 3 3 34 3 3 5 5 5 3 3 33 4 3 4 5 3 3 3 33 4 3 5 5 3 3 3 33 3 4 3 3 3 3 3 11 4 3 3 3 3 1 1 11 3 4 2 2 4 1 1 12 1 2 2 2 1 4 1 12 2 2 2 2 2 2 1 11 2 2 2 2 2 2 1 22 2 2 1 2 2 1 1 12 2 2 2 2 2 2 1 22 2 2 2 2 2 2 2 22 2 2 .. 2 2 2 1 2 12 2 2 2 2 2 2 2 21 2 2 2 2 2 2 2 21 1 1 11 1 1 11 1 1 11 1 1 21 1 1 22 1 1 23 1 1 23 3 3 33 3 3 33 3 3 33 3 3 33 3 3 3MIMULUS LAKEMIMULUS LAKECd eMIMULUS LAKE1 11 11 1 11 121 122 1 22 2 23 3 33 3 33 3 33 3 33 3 31 1 2 31 1 2 21 3 2 45 3 3 56 5 6 67 7 7 78 8 8 78 8 8 7MIMULUS LAKEMIMULUS LAKE- 70 -Table IX:^Species mean aerial coverage class data forsubcommunities at different scales in the subalpine wet meadow.Those species with a mean aerial coverage class estimate > 1 weredeemed to be representative of a particular subcommunity. Speciesnames corresponding to the codes used below may be found inAppendix A. Integers directly above mean and standard deviationestimates represent subcommunities at each scale.Aggl Scale1Mean Std Dev2Mean Std Dev3Mean Std DevSENTRI 0.04 0.20 0.02 0.12 0.09 0.29CARNIG 3.08 0.56 4.12 1.07 1.21 1.03LEPPYR 2.62 0.85 0.26 0.64 0.67 0.97LUEPEC 0.54 0.81 0.14 0.46 1.98 1.10ERIPER 0.69 0.79 0.10 0.29 1.21 0.97HIEGRA 0 0 0.28 0.63EPIANA 0.96 0.72 0.58 0.79 0.49 0.67CALBIF 2.19 1.10 0.67 0.90 1.72 1.14CASMER 0.27 0.53 0.02 0.12 2.37 1.09POACUS 0.38 0.57 0.17 0.41 0.19 0.39JUNCSP 0.31 0.47 0.10 0.34 0.79 0.86PHYEMP 0.35 0.75 0.02 0.12 1.74 1.36VERWOR 0.04 0.20 0.03 0.17 0.26 0.49POTFLA 0.04 0.20 0.02 0.12 0CARSPE 0.04 0.20 0.26 0.59 0.53 0.70PETFRI 0 0 0.02 0.15LUPLAT 0 0 0.74 1.18VALSIT 0 0.02 0.12 0.14 0.47LUZPAR 0 0 0.07 0.34PHLALP 0.04 0.20 0 0.02 0.15ANEOCC 0 0 0.07 0.26ABILAS 0 0 0ERYGRA 0 0 0CASPAR 0 0 0.09 0.29POALEP 0 0 0RANESC 0 0 0.02 0.15ANTALP 0.27 0.60 0.08 0.36 0.14 0.41TRISPI 0 0 0.02 0.15VACDEL 0 0 0.07 0.34LYCSEL 0.04 0.20 0 0.12 0.32KATMIC 0 0.03 0.17 0.19 0.45AGRTHU 0.96 0.66 '0.21 0.51 0.23 0.48EQUARV 0.23 0.43 0.02 0.12 0CETSUB 0 0 0.37 0.79SPHWAR 0.12 0.59 0.02 0.12 0PEDBRA 0 0 0- 71 -Table IX: (Continued)Agg1 Scale (Continued)4^ 5Mean Std Dev^Mean Std DevSENTRI 0.49 0.96 0.30 0.56CARNIG 2.21 1.05 1.30 0.97LEPPYR 0.04 0.27 0LUEPEC 0.66 1.23 1.22 1.13ERIPER 1.22 1.04 0.70 0.82HIEGRA 0.70 0.87 0.65 0.71EPIANA 0.48 0.61 0.26 0.54CALBIF 0.33 0.77 0.35 0.78CASMER 0.15 0.47 0.13 0.46POACUS 0.34 0.54 0.17 0.39JUNCSP 0.93 0.88 0.91 0.90PHYEMP 0.01 0.12 0.09 0.29VERWOR 0.31 0.50 0.35 0.57POTFLA 0.49 0.96 0.09 0.42CARSPE 1.72 0.88 1.87 1.01PETFRI 0.01 0.12 0LUPLAT 0.36 0.79 3.48 0.90VALSIT 1.30 1.27 1.04 1.11LUZPAR 0.19 0.47 0.13 0.34PHLALP 0.01 0.12 0.04 0.21ANEOCC 0.43 0.87 0.13 0.46ABILAS 0.03 0.17 0ERYGRA 0.01 0.12 0CASPAR 0.33 0.64 0.17 0.49POALEP 0.37 0.60 0.17 0.58RANESC 0.28 0.49 0.26 0.54ANTALP 0.06 0.24 0TRISPI 0.06 0.30 0.13 0.34VACDEL 0 0.13 0.63LYCSEL 0 0KATMIC 0 0AGRTHU 0.01 0.12 0.04 0.21EQUARV 0 0CETSUB 0.12 0.41 0SPHWAR 0 0PEDBRA 0 O.04 0.21- 72 -Table IX: (Continued)Agg4 Scale1^ 2^ 3Mean Std Dev^Mean Std Dev^Mean Std DevSENTRI 0.48 0.59 0 0.04 0.12CARNIG 1.90 0.85 0.85 0.42 3.63 0.78LEPPYR 0.11 0.30 0.60 0.84 0.76 0.74LUEPEC 0.70 0.78 2.45 0.84 0.44 0.51ERIPER 1.07 0.66 1.70 0.62 0.43 0.59HIEGRA 0.64 0.48 0.20 0.33 0.01 0.05EPIANA 0.42 0.38 0.30 0.21 0.65 0.55CALBIF 0.51 0.70 1.25 0.64 1.17 0.99CASMER 0.40 0.58 2.25 0.94 0.27 0.51POACUS 0.28 0.32 0.30 0.21 0.13 0.15JUNCSP 0.82 0.51 0.90 0.76 0.33 0.53PHYEMP 0.22 0.48 1.40 0.58 0.29 0.48VERWOR 0.38 0.30 0.15 0.14 0.08 0.14POTFLA 0.40 0.55 0 0CARSPE 1.65 0.58 0.60 0.38 0.30 0.39PETFRI 0.01 0.05 0 0.01 0.05LUPLAT 1.33 1.18 0.20 0.27 0.04 0.16VALSIT 1.25 0.66 0.20 0.21 0.01 0.05LUZPAR 0.16 0.36 0.05 0.11 0PHLALP 0.02 0.07 0.05 0.11 0ANEOCC 0.40 0.54 0 0ABILAS 0.02 0.07 0 0ERYGRA 0.01 0.05 0 0CASPAR 0.26 0.36 0.25 0.31 0.02 0.08POALEP 0.32 0.39 0 0RANESC 0.30 0.38 0 0ANTALP 0.02 0.07 0.20 0.21 0.18 0.30TRISPI 0.08 0.18 0.05 0.11 0VACDEL 0.03 0.16 0 0LYCSEL 0 0.10 0.14 0.05 0.13KATMIC 0.01 0.05 0.15 0.22 0.06 0.13AGRTHU 0.03 0.16 0.20 0.45 0.39 0.41EQUARV 0 0 0.05 0.10CETSUB 0.17 0.33 0.20 0.21 0.01 0.05SPHWAR 0 0 0PEDBRA 0.01 0.05 0 0- 73 -Table IX: (Continued)Agg6a Scale1^ 2^ 3Mean Std Dev^Mean Std Dev^Mean Std DevSENTRI 0.49 0.53 0.05 0.08 0.02 0.09CARNIG 2.01 0.67 1.10 0.59 3.68 0.69LEPPYR 0.06 0.22 0.57 0.57 0.94 0.86LUEPEC 0.64 0.64 2.14 0.49 0.40 0.48ERIPER 0.99 0.61 1.36 0.42 0.38 0.50HIEGRA 0.70 0.44 0.33 0.32 0.01 0.04EPIANA 0.37 0.28 0.48 0.40 0.70 0.51CALBIF 0.32 0.54 1.43 0.71 1.23 1.00CASMER 0.24 0.33 2.00 0.63 0.21 0.44POACUS 0.32 0.32 0.17 0.17 0.19 0.18JUNCSP 0.89 0.36 0.74 0.49 0.28 0.42PHYEMP 0.04 0.10 1.60 0.53 0.20 0.28VERWOR 0.33 0.32 0.21 0.21 0.07 0.11POTFLA 0.42 0.51 0 0.02 0.09CARSPE 1.76 0.28 0.64 0.35 0.26 0.36PETFRI 0.01 0.04 0.02 0.06 0LUPLAT 1.19 0.88 0.86 0.82 0VALSIT 1.32 0.56 0.14 0.15 0LUZPAR 0.19 0.28 0.07 0.19 0PHLALP 0.02 0.09 0.02 0.06 0ANEOCC 0.39 0.48 0.05 0.08 0ABILAS 0.02 0.06 0 0ERYGRA 0.01 0.04 0 0CASPAR 0.24 0.36 0.21 0.19 0.01 0.04POALEP 0.35 0.39 0 0RANESC 0.31 0.35 0 0ANTALP 0.01 0.04 0.14 0.18 0.17 0.21TRISPI 0.07 0.14 0.05 0.08 0VACDEL 0.04 0.13 0.07 0.13 0LYCSEL 0 0.05 0.08 0.04 0.13KATMIC 0 0.12 0.19 0.06 0.14AGRTHU 0 0.24 0.38 0.41 0.40EQUARV 0 0.02 0.06 0.07 0.11CETSUB 0.11 0.17 0.33 0.40 0.01 0.04SPHWAR 0 0 0.01 0.04PEDBRA 0.01 0.04 0 0- 74 -Table IX: (Continued)Agg6b Scale1^ 2^ 3Mean Std Dev^Mean Std Dev^Mean Std DevSENTRI 1.13 0.38 0.04 0.08 0.25 0.22CARNIG 1.97 0.62 2.58 0.55 1.42 0.78LEPPYR 0.17 0.37 0 0.08 0.17LUEPEC 0.17 0.24 0.71 0.72 0.75 0.59ERIPER 1.40 0.55 0.46 0.53 1.08 0.63HIEGRA 0.33 0.42 1.00 0.54 0.08EPIANA 0.43 0.30 0.38 0.42 0.17 0.14CALBIF 0.27 0.28 0.17 0.33 0.67 0.56CASMER 0.17 0.37 0.17 0.19 0.54 0.42POACUS 0.40 0.35 0.29 0.16 0.21 0.32JUNCSP 0.53 0.25 1.38 0.39 0.58 0.29PHYEMP 0.07 0.15 0 0.29 0.48VERWOR 0.50 0.24 0.17 0.24 0.38 0.28POTFLA 1.03 0.27 0.08 0.17 0.08 0.17CARSPE 1.93 0.63 1.42 0.40 1.71 0.16PETFRI 0.03 0.07 0 0LUPLAT 0.47 0.45 0.71 0.25 2.83 0.30VALSIT 1.47 0.43 1.58 0.52 1.04 0.67LUZPAR 0.03 0.07 0.50 0.43 0.04 0.08PHLALP 0.07 0.09 0 0ANEOCC 0.57 0.58 0.21 0.25 0.50 0.64ABILAS 0.03 0.07 0 0ERYGRA 0 0.04 0.08 0CASPAR 0.23 0.25 0.17 0.33 0.21 0.32POALEP 0.60 0.42 0.29 0.34 0.13 0.25RANESC 0.53 0.46 0.08 0.10 0.29 0.25ANTALP 0 0.04 0.08 0.04 0.08TRISPI 0.03 0.07 0 0.17 0.24VACDEL 0 0.13 0.25 0LYCSEL 0 0 0KATMIC 0 0 0AGRTHU 0 0 0EQUARV 0 0 0CETSUB 0 0.17 0.33 0.25 0.50SPHWAR 0 0 0PEDBRA 0 0 0.04 0.08- 75 -Table IX: (Continued)Agg6b Scale (Continued)4^ 5^ 6Mean Std Dev^Mean Std Dev^Mean Std DevSENTRI 0 0.06 0.10 0.11 0.19CARNIG 0.50 1.11 0.51 2.06 0.38LEPPYR 0 0.56 0.59 1.28 0.51LUEPEC 2.50 2.39 0.35 1.50 0.44ERIPER 1.17 1.83 0.50 1.39 0.25HIEGRA 0.17 0.56 0.69 0.11 0.19EPIANA 0.33 0.56 0.35 0.94 0.42CALBIF 0.83 1.28 0.35 2.50 0.17CASMER 3.33 1.22 0.63 1.67 0.44POACUS 0.50 0.22 0.10 0.22 0.10JUNCSP 0.17 0.89 0.54 1.28 0.10PHYEMP 0.33 1.67 0.76 1.17 0.33VERWOR 0.17 0.22 0.10 0.39 0.25POTFLA 0 0 0CARSPE 1.00 0.83 0.33 0.50 0.44PETFRI 0 0 0.06 0.10LUPLAT 0.33 0.67 0.88 0.61 0.67VALSIT 0.33 0.28 0.19 0.06 0.10LUZPAR 0.17 0 0PHLALP 0 0 0.06 0.10ANEOCC 0 0.06 0.10 0ABILAS 0 0.06 0.10 0ERYGRA 0 0 0CASPAR 0.17 0.56 0.59 0.11 0.10POALEP 0 0 0RANESC 0 0.06 0.10 0ANTALP 0 0.28 0.10 0.17 0.29TRISPI 0 0.17 0VACDEL 0 0 0LYCSEL 0 0.11 0.10 0.22 0.19KATMIC 0 0.17 0.17 0.17 0.17AGRTHU 0 0.06 0.10 0.83 0.33EQUARV 0 0 0CETSUB 0.17 0.44 0.25 0.06 0.10SPHWAR 0 0 0PEDBRA 0 0 0- 76 -Table IX: (Continued)Agg6b Scale (Continued)7^ 8Mean Std Dev^Mean Std DevSENTRI 0.03 0.07 0CARNIG 3.44 0.66 4.17 0.48LEPPYR 1.08 0.58 0.19 0.27LUEPEC 0.50 0.21 0.14 0.22ERIPER 0.56 0.46 0HIEGRA 0.03 0.07 0EPIANA 0.92 0.53 0.28 0.20CALBIF 1.64 0.39 0.22 0.17CASMER 0.25 0.31 0POACUS 0.06 0.09 0.14 0.16JUNCSP 0.39 0.39 0.03 0.07PHYEMP 0.17 0.15 0.08 0.20VERWOR 0.08 0.09 0POTFLA 0 0CARSPE 0.14 0.13 0.39 0.46PETFRI 0 0LUPLAT 0 0VALSIT 0 0LUZPAR 0 0PHLALP 0 0ANEOCC 0 0ABILAS 0 0ERYGRA 0 0CASPAR 0.03 0.07 0POALEP 0 0RANESC 0 0ANTALP 0.25 0.31 0.06 0.14TRISPI 0 0VACDEL 0 0LYCSEL 0 0KATMIC 0.08 0.14 0AGRTHU 0.61 0.23 0.06 0.14EQUARV 0.11 0.14 0CETSUB 0 0SPHWAR 0 0PEDBRA 0 0- 77 -Table IX: (Continued)Agg9 Scale1^ 2^ 3Mean Std Dev^Mean Std Dev^Mean Std DevSENTRI 0.49 0.53 0.07 0.09 0.01 0.04CARNIG 2.05 0.61 1.35 0.72 3.79 0.62LEPPYR 0.06 0.19 0.78 0.64 0.83 0.81LUEPEC 0.62 0.67 1.96 0.44 0.30 0.22ERIPER 0.94 0.59 1.39 0.38 0.28 0.42HIEGRA 0.67 0.36 0.35 0.40 0.01 0.04EPIANA 0.35 0.26 0.61 0.40 0.65 0.46CALBIF 0.28 0.36 1.63 0.77 1.07 0.89CASMER 0.22 0.20 1.78 0.67 0.11 0.18POACUS 0.32 0.29 0.26 0.13 0.14 0.11JUNCSP 0.86 0.27 0.80 0.45 0.22 0.29PHYEMP 0.04 0.08 1.39 0.71 0.12 0.14VERWOR 0.33 0.30 0.24 0.16 0.04 0.06POTFLA 0.43 0.48 0 0.02 0.07CARSPE 1.75 0.13 0.67 0.44 0.25 0.29PETFRI 0.01 0.04 0.02 0.05 0LUPLAT 1.21 0.83 0.70 0.58 0VALSIT 1.35 0.48 0.15 0.13 0LUZPAR 0.20 0.27 0.06 0.14 0PHLALP 0.02 0.07 0.02 0.05 0ANEOCC 0.40 0.44 0.06 0.09 0ABILAS 0.01 0.04 0.02 0.05 0ERYGRA 0.01 0.04 0 0CASPAR 0.19 0.27 0.26 0.32 0.01 0.04POALEP 0.36 0.37 0 0RANESC 0.30 0.32 0.04 0.09 0ANTALP 0.01 0.04 0.19 0.15 0.14 0.20TRISPI 0.06 0.11 0.06 0.06 0VACDEL 0.04 0.11 0.06 0.09 0LYCSEL 0 0.11 0.14 0KATMIC 0 0.13 0.16 0.04 0.08AGRTHU 0 0.35 0.45 0.35 0.34EQUARV 0 0.02 0.05 0.07 0.12CETSUB 0.09 0.15 0.31 0.35 0SPHWAR 0 0 0.01 0.04PEDBRA 0.01 0.04 0 0- 78 -Phyllodoce empetriformis (Table IX). Subcommunities 4 and 5 arefound mostly in the high meadow. Both are represented by Juncus sp., Carex spectabilis, Valeriana sitchensis, and Carex nigricans.However, 4 is also well represented by Erigeron perecirinus, and 5is represented by Luetkea pectinata and Lupinus latifolius (TableIX). Environmental Variable RelationshipsSimilar to the marsh, EL interacts with many of the measuredenvironmental variables in the subalpine meadow study area. Inreference to Table X, EL is positively correlated with EC and SA(0.69 and 0.34 respectively), and is negatively correlated with CY(-0.71). Relatively higher EC values are present where soilcontains relatively more SA (0.32) and relatively less CY (-0.49).Considering that the study site was located on an incline, areas ofgreater CY are found near the lower end, adjacent to Mimulus Lake,possibly resulting from CY being carried by water flow during snowmelt. Higher areas within the community are well-drained (greaterSA content) and receive more exposure. Better drainage as well asmore evaporation on the higher areas may explain greater EC and SApresence. Community-Environment RelationshipsThe first and second ordination axes of the CCA have- 79 -Table X: Pearson correlations between environmental variables atdifferent scales in the subalpine wet meadow. EL, relative groundlevel elevation; C, carbon content; EC, electrical conductivity;SA, sand content; CY, clay content; pH, soil acidity; FI, soilsample particles < 2 mm.Aggi ScaleCpHECSACYFIEL C pH EC SA CY-0.1481-0.27650.69310.3419-0.70940.16850.12690.2452-0.0024-0.04340.0893-0.12230.24530.14360.03970.3238-0.49180.1597-0.2607-0.0682 -0.0791Agg4 ScaleCpHECSACYFIEL C pH EC SA CY-0.1668-0.53360.90080.5681-0.83770.33690.1414-0.1653-0.0990-0.00720.2708-0.4870-0.19870.41670.05540.5741-0.68360.2518-0.52090.0568 -0.3757Agg6a ScaleCpHECSACYFIEL C pH EC SA CY-0.1461-0.66560.90930.6148-0.81980.29830.1947-0.1959-0.1487-0.04160.2107-0.6221-0.33590.47540.01040.5792-0.66710.2169-0.6440-0.1149 -0.2675Agg6b ScaleCpHECSACYFIEL C pH EC SA CY-0.2040-0.58190.93480.5825-0.85730.32980.3326-0.2631-0.10270.01790.3760-0.5103-0.10760.42830.0653'^0.5565-0.73970.2567-0.59490.0708 -0.4128Table X:^(Continued)Agg9 ScaleEL^C pH- 80^-EC SA CYC^-0.1809pH^-0.7255EC^0.9429SA^0.6004CY^-0.8405FI^0.29680.3193-0.3186-0.1257-0.02580.3104-0.6888-0.24870.4650-0.02490.5746-0.73150.1920-0.6939-0.0620 -0.3054- 81 -eigenvalues of 0.33 and 0.14 respectively. The first canonicalaxis eigenvalue and trace statistic are significant (p<0.05).Figure lla shows five subcommunities, arranged in an arch orhorseshoe shape. This suggests an environmental variable has astrong effect on the ordination (Ter Braak 1987b). EL is not onlystrongly correlated with species axis one (-0.84) (Table XI), butis represented in Figure lla by a vector of considerable length,suggesting that EL is a major determinant of the first speciesaxis. Moreover, axis 2 is simply a quadratic function of thefirst. Figure 12a confirms the presence of what appears to be adefinite EL gradient. EC and SA are also negatively correlatedwith the first species axis (-0.66 and -0.33 respectively) (TableXI). All three vectors pass directly through subcommunities 4 and5 located in the high meadow area (Figure 11a). Table XII showssubcommunities 4 and 5 are located in soils of greater SA and EC asopposed to subcommunities 1 and 2, located at the low end of themeadow. Subcommunities 1 and 2, however, are associated with soilsthat have slightly more CY (12.5% and 13.4% respectively) thansubcommunities 3, 4, and 5 (11.9%, 7.8%, and 7.8% respectively)offering an interpretation of a CY-species axis 1 correlativerelationship of 0.57 (Table XI). Figures 12b, c, and d reaffirmthe presence of greater EC and"SA as well as less CY respectivelyin upper as opposed to lower meadow soils. C, the highestcorrelate with species axis 2 (0.37) (Table XI), is mostlyassociated with predominantly monospecific stands of Carex nicfricans (subcommunity 2) (11.4%). Other subcommunities contain- 82 -Figure 11: Subcommunity-environment biplots at different scalesfor the subalpine meadow study site: aggl (a), agg4 (b), agg9 (c),agg6a (d), and agg6b (e). Subcommunities are represented by 50%confidence ellipses. Ellipses were unable to be produced wherethose subcommunities were represented by three or fewer samplingunits. Each environmental variable is represented by a vector.EL, relative ground level elevation; C, carbon content; EC,electrical conductivity; SA, sand content; CY, clay content; pH,soil acidity; FI, soil sample particles < 2 mm.a^ -1.5-2.5 ^-2.5 -1.5^-0.5^0.5^1.5^2.5AXIS 10.8-0.8- 1.6 ^-1.6 -0.8^0.0 0.8^1.6- 83 -Figure 11: (Continued)b 1.6 AXIS 1C 1.60.8CP 0.0X-0.8-1.6 ^-1.6 -0.8^0.0^0.8^1.6AXIS 1Figure 11: (Continued)d 1.6- 84 -0.80.0Q-0.8 -1.6 ^-1.6 -0.8 0.0^0.8^ -0.8 0.0 0.8 1.6AXIS 1e 1.6AXIS 1- 85 -Table XI: Pearson correlations between environmental variables andspecies axes I and II at different scales in the subalpine wetmeadow. EL, relative ground level elevation; C, carbon content;EC, electrical conductivity; SA, sand content; CY, clay content;pH, soil acidity; FI, soil sample particles < 2 mm.Aggl ScaleSPP AXIS 1 SPP AXIS 2Agg4 ScaleSPP AXIS 1 SPP AXIS 2EL -0.8366 0.0896 -0.8671 0.1324C 0.2080 0.3689 0.2803 0.5604pH 0.2810 -0.0230 0.5312 -0.0099EC -0.6616 0.0407 -0.9048 0.0031SA -0.3266 0.1222 -0.6027 0.2015CY 0.5716 -0.3166 0.7069 -0.3512FI -0.0954 0.3520 -0.1935 0.5162Agg6a ScaleSPP AXIS 1 SPP AXIS 2Agg6b ScaleSPP AXIS 1 SPP AXIS 2EL -0.8756 -0.0422 -0.8847 0.1286C 0.1753 0.6065 0.3412 0.5514pH 0.7002 0.0723 0.5863 0.1446EC -0.8828 -0.1747 -0.9316 0.0094SA -0.6257 0.1435 -0.5968 0.3038CY 0.7361 -0.2712 0.7233 -0.3286FI -0.2448 0.4200 -0.1763 0.5981Agg9 ScaleSPP AXIS 1 SPP AXIS 2EL -0.8763 -0.1211C 0.2295 0.6260pH 0.6871 0.2662EC -0.8927 -0.2365SA -0.6644 0.1781CY 0.7545 -0.1807FI -0.2696 0.489512039.820.0a)4-)0.20I^I58Metres- 86 -Figure 12:^Two-dimensional isopleths displaying EL and soilvariable zonation patterns at the aggl scale in the subalpinemeadow study site: EL, relative ground level elevation (a); EC,electrical conductivity (b); SA, sand content (c); CY, clay content(d); carbon content C (e); FI, soil sample particles < 2 mm (f);and pH, soil acidity (g).a^ EL58^120MetresI^1^<1.39^<2.50^<3.60^<4.71^<5.82 Metresb^ EC39.820.00.20<0.31^<0.44^<0.56^<0.69^<0.82 Mmhos/cmI^I• 20.0VSA39.80.20 58Metres120- 87 -Figure 12: (Continued)<40.37^<45.21^<50.05^<54.89^<59.74 zd^ CY39.8V• 20.0V0.258^120Metres[I<6.70^<9.77^<12.83^<15.90^<18.96 ze C39.8• 20.0V0.20^58^120MetresI^I0<8.83^<11.22^<13.62^<16.01^<18.40 z12039.85.^20.00.20 58MetresU 20.0V0.20 58Metres120pH39.8- 88 -Figure 12: (Continued)f^ FII^I [nag<42.54^<53.22^<63.89^<74.57^<85.24 v.I^1<3.99^<4.19^<4.40^<4.60^<4.80- 89 -Table XII:^Summarized environmental data for the mainsubcommunities recognized at the different scales of analysis (agglevels) in the subalpine wet meadow. Integers directly above meanand standard deviation estimates represent subcommunities at eachscale. EL, relative ground level elevation; C, carbon content; EC,electrical conductivity; SA, sand content; CY, clay content; pH,soil acidity; FI, soil sample particles < 2 mm.Aggl Scale1^ 2^ 3Mean Std Dev^Mean Std Dev^Mean Std DevEL (m)^1.54^0.98 0.77^0.37 2.71^0.74C^(%) 9.66^3.63^11.39^4.89^8.31^1.72pH 4.39^0.91 4.55^0.58 4.01^0.93EC (mmhos/cm) 0.42^0.28^0.31^0.16^0.55^0.14SA (%)^50.12^5.17 46.57^7.15 47.03^13.41CY (%) 12.49^2.63^13.36^2.60^11.86^4.10FI (%)^51.43^19.77 62.13^22.85 50.32^11.774 5Mean Std Dev^Mean Std DevEL (m)^4.09^1.05 4.09^1.14C^(%) 8.99^1.06^9.41^1.86pH 3.80^1.10 3.87^0.85EC (mmhos/cm) 0.69^0.13^0.71^0.12SA (%)^54.37^7.66 54.72^3.08CY (%) 7.83^2.95^7.75^3.76FI (%)^57.63^13.89 58.84^12.49Agg4 Scale1 2 3Mean Std Dev^Mean Std Dev^Mean Std DevEL (m)^4.09^1.06 2.86^0.56 1.10^0.52C^(%) 8.86^1.10^8.55^1.07^10.39^2.73pH 3.78^0.50 4.12^0.25 4.51^0.37EC (mmhos/cm) 0.70^0.08^0.56^0.05^0.33^0.10SA (%)^54.18^3.47 48.61^5.85 47.35^4.95CY (%) 7.88^2.73^12.78^2.96^12.81^1.91FI (%)^59.15^9.92 48.24^7.35 57.00^12.75Agg6a Scale1 2 3Mean Std Dev Y Mean Std Dev^Mean Std DevEL (m)^4.33^0.90^2.69^0.54 1.07^0.49C^(%) 9.05^1.09 8.69^0.72^10.32^2.47pH 3.76^0.34^4.08^0.27 4.53^0.29EC (mmhos/cm) 0.71^0.06 0.59^0.04^0.33^0.08SA (%)^54.71^3.17^49.14^3.88 47.67^4.98CY (%) 7.29^2.57 12.59^2.59^12.70^1.90FI (%)^59.18^9.92^50.17^6.68 56.81^13.02- 90 -Table XII: (Continued)Agg6b Scale1^ 2^ 3Mean Std Dev^Mean Std Dev^Mean Std DevEL (m)^4.63^0.77 4.46^0.76 3.71^1.16C^(%) 8.63^0.48^9.79^0.98^8.46^1.28pH 4.01^0.31 3.45^0.26 3.76^0.37EC (mmhos/cm) 0.77^0.04^0.70^0.05^0.66^0.03SA (%)^54.38^4.91 54.56^2.10 54.44^2.82CY (%) 7.13^1.65^6.37^2.08^9.06^3.69FI (%)^68.25^10.55 56.01^4.11 53.46^3.554 5 6Mean Std Dev^Mean Std Dev^Mean Std DevEL (m)^3.52^2.57^0.37 2.08^0.09C^(%) 9.54 7.66^0.89^8.76^0.53pH 3.84 3.79^0.68 4.22^0.47EC (mmhos/cm) 0.56 0.54^0.10^0.52^0.04SA (%)^50.03^49.55^5.00 44.32^5.10CY (%) 9.75 12.52^1.95^12.42^3.11FI (%)^53.67 44.77^4.93 48.22^9.137^ 8Mean Std Dev^Mean Std DevEL (m)^1.22^0.38 0.70^0.23C^(%) 8.95^1.02^12.55^1.97pH 4.62^0.09 4.63^0.09EC (mmhos/cm) 0.35^0.08^0.27^0.04SA (%)^50.54^3.96 46.15^4.02CY (%) 12.35^1.53^13.59^1.69FI (%)^54.52^9.07 64.56^12.25Agg9 Scale1 2 3Mean Std Dev^Mean Std Dev^Mean Std DevEL (m)^4.37^0.87 2.57^0.57 0.98^0.42C^(%) 9.11^1.05^8.40^0.58^10.65^2.01pH 3.74^0.20 4.01^0.35 4.63^0.07EC (mmhos/cm) 0.71^0.06^0.54^0.05^0.33^0.09SA (%)^54.73^3.37 47.83^3.54 48.57^4.99CY (%) 7.16^2.48^12.36^2.36^12.80^1.75FI (%)^59.54^9.39^' 48.96^5.60 58.33^12.73- 91 -slightly less (Table XII) (also see Figure 12e).Subcommunity 3, though well represented by species unique toit, may serve as a transition belt between high and low meadowareas. SA content in subcommunity 3 (47.0%), for example, is verysimilar to that in subcommunities 1 and 2 (50.1% and 46.6%respectively) (Table XII). Also, subcommunity 3 shares fourrepresentative species found in high and low meadow areas: Caltha biflora, Carex nigricans, Erigeron peregrinus, and Luetkea pectinata (Table IX).FI shares a strong correlative relationship with species axis2 (Table XI), yet Figure 12f confirms the difficulty in providinga possible explanation.3.3.2 Agg4 Scale (5 X 5 Metre Observation Unit) Community StructureAt agg4, the highest redundancy and Rc values are found atlevels C3 (22.06%) and C9 (0.9789) respectively (Table VIII).Thus, the study site is divided into fewer zones than aggl: highmeadow, mid-meadow, and low meadow (Figure 10b). Subcommunity 1,representative of the high meadow, is characterized in Table IX byCarex nigricans, Erigeron peregrinus, Carex spectabilis, Lupinus latifolius, and Valeriana sitchensis.- 92 -Ericieron peregrinus is also found in subcommunity 2 (Figure10b), but is largely represented by Cassiope mertensiana andPhyllodoce empetriformis as well as Luetkea pectinata, and Caltha biflora (Table IX). Its representation as a transition beltbetween upper and lower meadow subcommunities is not as defined atthis scale as opposed to subcommunity 3 at aggl (Figure 10a).Subcommunity 3, shown in Table IX is predominately a lowmeadow stand of Carex nicrricans interspersed with Caltha biflora. Environmental Variable RelationshipsIn reference to Table X, EL shares stronger correlativerelationships with EC, SA, and CY (0.90, 0.57, and -0.84respectively) compared to aggl. EC also shares stronger positiveand negative relationships with SA and CY (0.57 and -0.68respectively) as opposed to aggl.Again referring to Table X, relationships between pH and otherenvironmental variables are more pronounced at this scale. WhileEL and EC are negatively correlated with pH (-0.53 and -0.49respectively), CY shares a pdsitive correlative relationship of0.42. Of interest is a negative correlative relationship betweenpH and SA (-0.20). Though this relationship is rather weak, it is,nevertheless, negative--unlike a positive relationship at aggl(0.25) (Table X). Soils are slightly less acidic in the low meadow- 93 -area. Since lower meadow soils contain relatively more CY andrelatively less SA, lower meadow soils have presumably greaterwater holding capacity and are less susceptible to leachingcompared to upper meadow soils. In addition, the low meadow areais located at the bottom of a slope: an ideal location for adecrease in water flow rate and subsequent water settlement.EL is also positively correlated with FI (0.34). Conversely,FI is negatively correlated with CY (-0.38) (Table X). Community-Environment RelationshipsCCA reported first and second axes eigenvalues of 0.33 and0.16 respectively, and a significant (p<0.05) trace statistic andfirst canonical ordination axis eigenvalue.The subcommunity-environment biplot shown in Figure lib tellsan almost identical story to that previously described at a finerscale (aggl). Mostly monospecific stands of Carex niciricans (subcommunity 3) in Table XII are found in soils of slightlygreater CY and C but contain less SA and EC as in subcommunity 1.Subcommunity 2 represents a' community between two differentextremes, is distinctive in species composition, and containsenvironmental variable values that are, again, usually similar toor between those values of subcommunities 1 and 3.- 94 -Differences in vector length are not as variable at this scale(Figure 11b). Generally, environmental variables have strongercorrelative relationships with species axes 1 and 2. Referring toTable XI, SA and pH, for example, have stronger correlativerelationships with species axis one (-0.60 and 0.53 respectively).In addition, pH vector length is not as short in relation to vectorCY in Figure llb as opposed to Figure 11a, possibly betterrecognizing pH-subcommunity relationships at this scale. Despitewhat appears to be clear pH zonation pattern at aggl (Figure 12g),relative pH differences between subcommunities are not very clearat a finer scale (aggl) (Table XII) mostly because pH measurementswithin each subcommunity are quite variable (rather high standarddeviation estimates). At a coarser scale, however, a morenoticeable trend is evident (lower standard deviation estimates).Subcommunity 3 is located in slightly less acidic soils (4.5) asopposed to subcommunities 1 and 2 (3.8 and 4.1 respectively) (TableXII).Interestingly, FI shares a strong correlative relationshipwith species axis 2 (Table XI). A definitive trend, however,appears to be lacking along a height gradient (Table XII).3.3.3 Aqq9 Scale (10 X 10 Metre Observation Unit) Community StructureAt agg9, redundancy and R, estimates are highest at levels C3- 95 -(28.71%) and C5 (0.9994) respectively (Table VIII).^Here, themeadow is divided into three subcommunities (Figure 10c)characterized by the aforementioned species at agg4. One subtledifference, however, is the improved representation (aerialcoverage class greater or equal to one) of Carex niciricans insubcommilnity 2 (Table IX). Environmental Variable RelationshipsEnvironmental variable interactions at this observation scalereinforce what was revealed at agg4. Generally, correlationsbetween environmental variables are stronger, to the extent ofemphasizing additional interactions. The most noticeableinteraction is between C and pH (0.32) as well as C and EC (-0.32)(Table X). Greater C is found in the low meadow areas where soilstend to be less acidic. Organic content may also be contributingto relatively better water holding capacity in soils resulting inless leaching. Community-Environment RelationshipsFirst and second axes' eigenvalues of 0.33 and 0.20respectively and a first canonical ordination axis eigenvalue andtrace statistic significant at 0.05 were reported from CCA. Asreported at the agg4 scale, same subcommunity-environmentrelationships are relevant at this scale except that correlations- 96 -between relevant environmental variables and the first two speciesaxes are generally stronger (Table XI). Of interest is theincreased ellipse size of subcommunity 2 and decreased ellipse sizeof 1 and 3 in Figure 11c compared to Figure lib. This may suggestthat the distinctiveness of all three communities can be moreclearly defined at agg9 as opposed to agg4. Within-assemblagevariability of subcommunities 1 and 3 appears to be less at agg9(Figure 11c) as opposed to agg4 (Figure lib), clearly redefiningupper and lower meadow zones. Conversely, within-assemblagevariability in ellipse 2 has increased from agg4 (Figure 11b) toagg9 (Figure 11c). Perhaps this more clearly defines subcommunity2 as a transition belt between upper and lower zones (Figure 10c),compared to agg4 (Figure 10b) and aggl (Figure 10a). Referring toFigure 13, agg4 has a larger overall estimate of within-clustervariability (31.10%) and smaller estimate of between-clustervariability (68.90%) compared to aggl (22.87% and 77.13%respectively). At agg9 community structure is more clearly definedthan both aggl and agg4 since overall between-assemblagevariability is 82.88% and within-assemblage variability is 17.12%.3.3.4 Agq6a and Aqq6b Scales (5 X 10 Metre and 10 X 5 Metre Observation Units) Community Structure and Environmental VariableRelationshipsAt agg6a, levels C3 (26.08%) and C10 (0.9948) yield theAgg4AgglAgg6aAgg9Agg6bIIIIII IIIIIIIiiiiiiIIIAgg6b- 97 -Figure 13: Overall between and within-cluster variabilityestimates for all scales in the subalpine meadow. Unstandardizedestimates shown along inner isoclines; standardized estimates shownalong outer isocline.1008060z402020^40^60^80^100BETWEEN (°M- 98 -highest redundancy and R, values respectively (Table VIII). Whenfield sampling is simulated with a rectangular quadrat positionedwidth-wise (agg6a), the wet meadow is, therefore, partitioned intothree zones: upper, middle, and lower subcommunities (Figure 10d).The same representative species at agg9 also characterize the threesubcommunities at agg6a (Table IX). At agg6b, levels C8 (24.69%)and C10 (0.9941) yield the highest redundancy and R, (Table VIII).Hence, simulation sampling with a quadrat of the same area butpositioned length-wise, divides the meadow into eightsubcommunities (Figure 10e). In reference to Table IX, upper zonesubcommunities 1, 2, and 3 are characterized by Carex nigricans,Carex spectabilis, and Valeriana sitchensis. Subcommunity 1 isalso represented by Senecio trianqularis, Erigeron peregrinus, andPotentilla flabellifolia. Hieracium gracile and Juncus sp. alsocharacterize subcommunity 2. Subcommunity 3 also has apreponderance of E. peregrinus and Lupinus latifolius. The middlezone subcommunities 4, 5, and 6 are well represented by E.peregrinus, Luetkea pectinata, and Cassiope mertensiana. While C.spectabilis is another representative of subcommunity 4, Caltha biflora, Phvllodoce empetriformis, and C. nigricans alsocharacterize subcommunity 5, and C. biflora, C. nigricans, P.empetriformis, Juncus sp., and Leptarrhena pvrolifolia alsorepresent subcommunity 6. The lower zone comprises twosubcommunities: 7 characterized by C. nigricans, C. biflora, and L.pvrolifolia, and 8, composed mostly of C. nigricans.- 99 -When scale is maintained but rectangular quadrat placement isaltered during simulation field sampling, environmental variableinteractions at both agg6a and agg6b levels reveal similar trendsto those at the agg4 and agg9 scales (Table X). Community-Environment RelationshipsFirst and second axes eigenvalues of 0.34, 0.20 for agg6a,0.33, 0.17 for agg6b, as well as a significant (p<0.05) firstcanonical axis eigenvalue and trace statistic significant werereported from a CCA. Environmental variable-first and secondspecies axes relationships are the same not only between agg6a andagg6b, but also the same as that previously described for agg4 andagg9 scales (Table XI).Ellipses 1 and 3 in Figure 11d are smaller than ellipses 1 and3 at agg4 (Figure 11b) yet larger than corresponding ellipses atagg9 (Figure 11c). Conversely, ellipse 2 in Figure 11d is largerthan ellipse 2 at agg4 (Figure lib) but smaller than agg9 (Figure11c). Thus, within-assemblage variability consistently decreasesin subcommunities 1 and 3 and increases in subcommunity 2 asprogressively larger sampling' units (agg4, agg6a, to agg9) areused. The distinctiveness of upper and lower zones and theinterpretation of subcommunity 2 as a transition belt may becomeincreasingly clearer at progressively coarser scales (agg4, agg6a,and agg9). Standardized between and within-variability estimates- 100 -at agg4, 6a, and 9 support this notion where between-assemblagevariability gradually increases and within-assemblage variabilitygradually decreases. Similar to the marsh, largest between (92.7%)and smallest within (7.4%) estimates are provided at agg6b (Figure13).3.3.5 EL Influence Verification3.3.5.1 Aggl, 4, 6a, 6b, 9 ScalesCorrelations between environmental variables and a canonicalcorrelation axis representing species variables (summarized bythree PCA axes) as well as residuals confirm EL influence onvegetation pattern (Table XIII). Environmental variable-canonicalaxis correlations reveal very similar trends previously discussedat all scales. Independent of EL, environmental variable-residualcorrelations are generally not as strong as environmental variable-canonical axis correlations perhaps confirming the presence of awell-defined EL gradient, influencing species variables in the wetmeadow study area. However, at agg6b and agg9, correlativerelationships between C and residuals are quite strong (-0.27 and -0.26 respectively). This may suggest that relationships betweenorganic content and vegetatiOn are independent of EL to somedegree. At these scales, vegetation (specifically Carex niqricans)may be clearly observed to contribute to the (C) organic content inlower meadow soils.- 101 -Table XIII:^Pearson correlations at different scales betweenenvironmental variables, residuals, and a canonical axisrepresenting species variables in the subalpine wet meadow. AXIS,canonical correlation axis; RESD, residuals; EL, relative groundlevel elevation; C, carbon content; EC, electrical conductivity;SA, sand content; CY, clay content; pH, soil acidity; FI, soilsampleAgglparticles < 2 mm.ScaleEL^C^pH EC SA CY FIAXIS 0.7563 -0.2322 -0.2969 0.6606 0.3375 -0.5103 0.0062RESD 0 -0.0837 -0.0555 0.1511 0.0496 0.0776 -0.0180Agg4 ScaleEL C pH EC SA CY FIAXIS 0.8504 -0.3974 -0.5475 0.8924 0.5990 -0.6684 0.0756RESD 0 -0.2344 -0.0320 0.2041 0.1457 0.1136 -0.0815Agg6a ScaleEL^C^pH^EC^SA^CY^FI ^AXIS 0.8693 -0.3660 -0.7307^0.9001^0.6162 -0.6812^0.0309RESD 0^-0.1913 -0.1782^0.1833^0.1198^0.0900 -0.1244Agg6b ScaleEL C pH EC SA CY FIAXIS 0.8715 -0.4469 -0.6392 0.9110 0.5672 -0.6886 0.0528RESD 0 -0.2719 -0.0628 0.1746 0.0648 0.1554 -0.1175Agg9 ScaleEL C pH EC SA CY FIAXIS 0.8865 -0.4342 -0.7968 0.9136 0.6055 -0.7067 0.0154RESD 0 -0.2632 -0.1790 0.1465 0.1047 0.1095 -0.1514- 102 -3.3.6 Subalpine Wet Meadow DiscussionThe wet meadow study site is generally composed of threezones: upper, middle, and lower. These three zones may correspondto major vegetation types in Garibaldi Provincial Park: forbmeadow (upper), heath (middle), and sedge meadow (lower) (Brink1959; Archer 1963; Brooke et al. 1970). Similar vegetation typeshave been documented in other alpine/subalpine regions of thePacific Northwest. The upper subcommunity represented in themeadow site by such species as Valeriana sitchensis, Carex spectabilis, Lupinus latifolius, and Ericreron perecirinus have beenextensively described by Kuramoto and Bliss (1970) in the OlympicMountains, Douglas and Bliss (1977) on steep well-drained slopes inthe North Cascade Range, and Evans and Fonda (1990) on windwardslopes of Excelsior Ridge in the North Cascades. The heathsubcommunity dominated by Phyllodoce empetriformis, Cassiope mertensiana, and Luetkea pectinata has also been identified as acommon community type by Kuramoto and Bliss (1970) in the OlympicMountains and Douglas and Bliss (1977) and Evans and Fonda (1990)in the North Cascades. Of all the major vegetation zones in thewet meadow study area, the Carex nigricans dominated subcommunity(lower meadow) is the most widespread. It has been found in theOlympic Mountains (Kuramoto and Bliss, 1970), east to the CanadianRockies (Knapik et al. 1973); Hrapko and La Roi 1978),and NorthCascades (Douglas and Bliss, 1977;Evans and Fonda 1990). C.nigricans snowbed communities have also been found by del Moral- 103 -(1979) in the Enchantment Lakes Basin, Washington.At all scales, subcommunities found in the upper zone sharestrong correlative relationships with SA and EC while lower zonesubcommunities share strong correlative relationships with CY andC. Lower meadow-pH relations are also apparent at most scales.The upper meadow may receive more exposure to climatic factors suchas insolation than the middle and lower meadow. Also, Phyllodoce empetriformis and Carex nigricans, dominant species in middle andlower subcommunities respectively, were found to better insulatesoil from solar radiation and higher temperatures on ExcelsiorRidge in the North Cascades than Valeriana sitchensis,representative of the upper meadow (Evans and Fonda 1990). Thoughsoil temperature was not recorded in this study, Evans and Fondas'(1990) findings may offer an hypothesis as to why EC values werefound to be higher in the upper meadow (possibly a result of highertemperatures and thus higher evaporation rates leading to greaterdeposition of salts in upper soil horizons), than in the middle andlower subcommunities. The upper meadow is also well-drainedbecause of greater sand content allowing water to flow easily intoa catch basin (lower meadow) where soils possess relatively greaterCY and C. Rapid water percolation through slightly more acidicsoils in the upper meadow may be promoting a leaching process;superior water holding capacity a result of greater CY and (C)organic accumulation from Carex nigricans in the lower may help toslow this process. Relatively less acidic soils may help retain- 104 -the distinctiveness of the lower sedge subcommunity by deterringheath species such as Phyllodoce empetriformis and Cassiope mertensiana from invading since they are accustomed to soils thatpossess sufficient organic accumulation but are of a slightly moreacidic nature (Brooke et al. 1970). Though water does not standover the sedge area, the water table is generally high (Brink1959,1964) possibly because of longer snow persistence (Brooke etal. 1970), ground and above ground water flow during snow melt, andpossible ground water influence from Mimulus Lake. A high watertable may also deter P. empetriformis and C. mertensiana frominvading the sedge area (Brooke et al. 1970).Brink (1959) reported forb meadow soils as having higher pH asopposed to those soils found in a sedge-Caltha meadow. Thisconflicts with the findings in the Mimulus Lake wet meadow area andperhaps suggests a need to study more intensively the communitystructure of smaller area. This may be not only more informativebut better recognize that broad descriptions of extremely variablehabitats may obscure important issues of ecological complexity.Soil properties are a function of regional climate, topography,biota, and parent materials (Jenny 1941). These factors maycontribute to pH variability in Garibaldi Park.Community pattern has long been recognized as a function ofsnow distribution and duration (Billings and Bliss 1959; Holway andWard 1963; Bell and Bliss 1979; Isard 1986; Evans and Fonda 1990).- 105 -Snowmelt as influenced by EL in the wet meadow may have a profoundinfluence on the three encountered subcommunity types. The mostsnow usually accumulates in Carex niqricans dominated basins andremains there until late July to early August (Evans and Fonda1990;Kuramoto and Bliss 1970; Hrapko and La Roi 1978; Selby andPitt 1984). On better drained areas, heath communities typical ofthe heath subcommunity in the Mimulus Lake area are usuallyreleased earlier from snow in early June to early July. Atsteeper, more exposed and well drained forb meadow type areas, snowhas disappeared between late May to early June (Douglas and Bliss1977) Since EL has a profound effect in the wet meadow community,a snow melt gradient as influenced by EL may ultimately be a majordeterminant of community pattern by influencing a plant's growingseason (Kuramoto and Bliss 1970).Employment of different quadrat sizes as well as CCorA toselect among nine possible subcommunity schemes per MVCA revealdifferent aspects of vegetation structure. At aggl the meadowstudy site is generally divided into the aforementioned threezones. Much variability is evident, however, with the subdivisionof the upper meadow into two subcommunities (Figures 10a and 11a).A transition belt can barely be observed suggesting gradualvegetation change along a pronounced EL gradient. The lower meadowis predominately a monospecific stand of Carex niqricans. Theinterspersion of less abundant species such as Caltha biflora isclearly evident with the division of the lower into two- 106 -subcommunities. Not only are upper, middle, and lower meadow areasmore clearly defined as simply three subcommunities at agg4, agg6a,and agg9, but the distinctiveness of all three subcommunities isalso evident at progressively coarser scales. Ellipsesrepresenting subcommunities 1 (upper) and 3 (lower) becomeprogressively smaller from agg4, agg6a, to agg9 (Figures lib, c,and d). Thus, within-assemblage variability consistently decreasesin subcommunities 1 and 3 from agg4, 6a, to agg9 which may meanthat their distinctiveness becomes progressively more defined.Conversely, an ellipse representative of subcommunity 2 appears togrow larger. That is, within-assemblage variability consistentlyincreases in subcommunity 2 from agg4, 6a, to agg9 emphasizing itspresence as a transition belt. This is clearly illustrated inFigure 10b where subcommunity 2 is distinct but is mostly confinedto one side of the grid. In Figures 10c and d, a transition belthas phased into view providing a clearer picture than agg1 and agg4(Figures 10a and b). Figure 13 reaffirms improved perception ofcommunity structure as one moves from agg4, 6a, to agg9. Overallwithin-assemblage variability progressively decreases andvariability between groups increases.- 107 -CHAPTER 4: SYNTHESIS 4.1 Quadrat Shape and OrientationThough it has long been the custom to employ square quadratsduring field sampling, there exists strong support for the use ofrectangular sampling units. Variance per unit area has been foundto be lower in rectangular plots than in square plots of the samearea (Clapham 1932; Kalamkar 1932; Justesen 1932). The results ofthis thesis may conform with those who have found this to be truewhen the rectangular sampling unit is positioned at right angles tothe observed vegetational or soil banding (Clapham 1932; Bormann1953). During field sampling, a rectangular quadrat oriented inthis fashion is more likely to include more (species) variability,ultimately reducing heterogeneity between sampling units (Kalamkar1932; Greig-Smith 1983). In both the marsh and meadow systems, theuse of rectangular quadrats oriented at right angles to theobserved bands of vegetation (agg6b) facilitated the recognition ofsubcommunities. That is, overall (standardized) within-assemblagevariability was least and between-assemblage variability wasgreatest at agg6b (10 X 5 m) (Figures 9 and 13). Interestingly, atagg6b in both marsh and meadow systems, redundancy estimates werehighest at the dendrogram level yielding eight groups (C8) (TablesII and VIII). Since vegetation heterogeneity between 10 X 5 mquadrats (agg6b) is presumed to be less than at other scales,relatively fewer differences between sampling units exist at agg6b.Fewer, yet better defined, differences between quadrats at agg6b- 108 -may be unique to only a few cases. Since the clustering algorithmmay have used these distinctive differences as criteria tosegregate quadrats into groups and only a few quadrats share thesedifferences per group, more groups are formed. An example of thepreceeding explanation is illustrated in Appendix B. Threesubcommunities can be easily observed in the vegetation data matrixat agg6a in the subalpine meadow. However, at agg6b it isconsiderably more difficult to observe three or eightsubcommunities. This may illustrate less heterogeneity betweensampling units at agg6b as opposed to agg6a. Differences inspecies composition and abundance at agg6b, appear to be fewer thanat agg6a, but those that do exist are much more noticeable. Inparticular, Cassiope mertensiana is most abundant withinsubcommunity 2 at agg6a; however, at agg6b, C. mertensiana tends tovary more erratically. Since quadrat 12 in agg6b is the onlysampling unit where C. mertensiana was given an aerial coverageclass of 4, this could have influenced the clustering algorithm indefining case 12 as a unique subcommunity. Another example isLupinus latifolius where it tends to be most abundant insubcommunity 1, at agg6a. At agg6b, abundance of L. latifolius tends to be more variable between subcommunities; specifically, itis most abundant in subcommdnity 3. These and other subtledifferences may explain why eight subcommunities were defined inboth the marsh and meadow at agg6b. Fewer, yet more distinctivedifferences between quadrats ultimately contribute to maximizingbetween and minimizing within-subcommunity variability estimates at- 109 -agg6b compared to other scales.4.2 Noisy Data and Redundancy Estimates The restrictions of a finite number of samples and the use ofvarious measurement scales to estimate species abundances have beenrecognized by Gauch (1982) as sources of noise in vegetation data.Furthermore, the chance distribution and establishment ofindividual plants, faunal activity, disturbance (Gauch 1982), andmixed and largely unpredictable species' responses to manyenvironmental gradients (Austin 1980) are other possible causes ofnoisy data. In this study, dendrogram levels were used at eachscale to describe community structure. Redundancy estimates foreach scale in both meadow and marsh systems have been shown to bequite low (6-47%). Much variation in the vegetation data sets wasleft unexplained and may be attributed to noise. There are twomajor setbacks that warrant concern. First, the clusteringalgorithm attempts to agglomerate samples that contain a mixture ofinterpretable variation and noise. Some of the noise may have beeninterpretable had either a different quadrat size or shape orsampling strategy been used to capture more information. Thevariation required to explain' some of the noise may operate atdifferent scales to those imposed by the observer. Noise, may bean indicator of a conflict between ecological complexity andimposed anthropocentric scales used to (insufficiently) assessvariation. Second, the clustering algorithm is forcing variation- 110 -to be segregated into somewhat artificial groupings. Natural zonesmay correspond to those dictated by the method or they may not.Given these considerations, it is not surprising that eachdendrogram level was only able to explain a small percentage of thetotal variation.4.3 Hierarchical Perspective: An Assessment Employment of different observation scales (quadrat sizes) inboth marsh and subalpine meadow, suggests that the scale at whichobservations are made will undoubtedly affect our perception ofvegetation-environment relationships as well as communitystructure. In this study, correlations among environmentalvariables and species axes generally tended to become stronger atprogressively coarser scales. Correlations that were regarded asweak (unimportant) at a fine scale usually became more noticeableas larger quadrat sizes were used. Subcommunity-pH relations inboth marsh and subalpine meadow may serve as an example where aweak aggl correlation became stronger at agg4. However, a strongagg4 correlation weakened at agg9 in the tidal marsh, recognizingexceptions to this generalization. In the marsh andsubalpine/alpine literature, 'researchers have recognized theinteraction of many factors in determining pattern, and many haveassessed the importance of measured environmental variables.Moreover, different factors have been attributed as having a majorinfluence on vegetation pattern in marsh and meadow systems. One- 111 -of the reasons for this may be that each study site is unique.However, most of these studies have used only one quadrat size andshape. Changing the scale and reference point in this studydemonstrated that the strength of vegetation-environmentcorrelations is a function of scale. The 'importance' ofenvironmental factors, estimated by their correlation withvegetation pattern, may be dependent on the scale at which the datawere analyzed. An hierarchical approach warns one to be cautiouswhen 'ranking' the importance of environmental variables.Community structure has been demonstrated also to be a function ofscale. An hierarchical perspective reinforces the notion that thecriteria used for defining a community or subcommunity are entirelyman-selected and may or may not correspond to undefinable naturalzones in the field.In summary, multivariate statistical techniques and othertools have been used to filter out noise in the hope of isolatinginterpretable variation. Benefits of comparing differently scaledanalyses are fourfold: 1) certain vegetation-environmentrelationships are clarified by appearing progressively stronger atcoarser scales, 2) at coarser scales, new relationships may beuncovered and established relationships at finer scales may phaseout of view, 3) interpretation of community structure may change asa function of scale, and 4) confidence is added to the observersoverall interpretation of ecological complexity. Given thatecological reality has been described as "loosely ordered,- 112 -complexly patterned, multiply determined" (Whittaker 1952), anhierarchical perspective can contribute significantly to anecologist's existing arsenal of statistical techniques andobservation skills.- 113 -BIBLIOGRAPHYAllen, T.F.H.^1987.^Hierarchical complexity in ecology:^anoneuclidean conception of the data space. Vegetatio 69:17-25.Allen, T.F.H., and T.B. Starr. 1982. Hierarchy: Perspectives for Ecological Complexity. The University of Chicago Press,Chicago.Allen, T.F.H., and E.P. Wyleto. 1983. A hierarchical model forthe complexity of plant communities. Journal of Theoretical Biology 101: 529-540.Allen, T.F.H., D.A. Sadowsky, and N. Woodhead.^1984.^Datatransformation as a scaling operation in ordination ofplankton. Vegetatio 56: 147-160.Archer, A.C. 1963. Some synecological problems in the alpine zonein Garibaldi Park. MSc. thesis, University of BritishColumbia, Vancouver, 129 pp.Austin, M.P. 1980. Searching for a model for use in vegetationanalysis. Vegetatio 42: 11-21.Bell, K.L., and L.C. Bliss.^1979.^Autecology of Kobresia bellardii:^Why winter snow accumulation limits localdistribution. Ecological Monographs 49: 377-402.Bertness, M.D., and A.M. Ellison. 1987. Determinants of patternin a New England salt marsh plant community. Ecological Monographs 57(2): 129-147.Billings, W.D.^1964.^Plants and the Ecosystem.^WadsworthPublishing Company, Inc., Belmont, California.Billings, W.D., and L.C. Bliss.^1959.^An alpine snowbankenvironment and its effects on vegetation, plant development,and productivity. Ecology 40: 388-397.Bormann, F.H. 1953. The statistical efficiency of sample plotsize and shape in forest ecology. Journal of Ecology 34:474-487.Bossort, A.K., M.A. Jasieniuk, and E.A. Johnson. 1977. Levels oforganization. Bioscience, 27: 82.Bradfield, G.E., and A. Campbell.^1986.^Vegetation-elevationcorrelation in two dyked marshes of northeastern VancouverIsland: a multivariate analysis. Canadian Journal of Botany64: 2487-2494.- 114 -Bradfield, G.E., and L. Orloci.^1975.^Classification ofvegetation data from an open beach environment in southwesternOntario: cluster analysis followed by generalized distanceassignment. Canadian Journal of Botany 53(5): 495-502.Brink, V.C. 1959. A directional change in the subalpine forest-heath ecotone in Garibaldi Park, British Columbia. Ecology 40: 10-16.Brink, V.C. 1964. Plant establishment in the high snowfall alpineand subalpine regions of British Columbia. Ecology 45(3):431-438.Brooke, R.C., E.B. Peterson, and V.J. Krajina.^1970.^Thesubalpine mountain hemlock zone. Ecology of Western NorthAmerica 2: 147-349.Brown, J., P.C. Miller, L.L. Tieszen, and F.L. Bunnell. 1980. AnArctic Ecosystem: The Coastal Tundra at Barrow, Alaska.Dowden, Hutchinson, and Ross, Stroudsburg, Pennsylvania.Campbell, A.C., and G.E. Bradfield. 1989. Comparison of plantcommunity-environment relations in two estuarine marshes ofnorthern British Columbia. Canadian Journal of Botany 67:146-155.Clapham, A.R.^1932.^The form of the observational unit inquantitative ecology. Journal of Ecology 20: 192-197.Daubenmire, R. 1968. Plant Communities. Harper, New York.Davidescu, D., and V. Davidescu. 1982. Evaluation of Fertility by Plant and Soil Analysis. Abacus Press, Kent, England.Dawe, N.K., and E.R. White. 1982. Some aspects of the vegetationecology of the Little Qualicum River estuary, BritishColumbia. Canadian Journal of Botany 60: 1447-1460.Dawe, N.K., and E.R. White. 1986. Some aspects of the vegetationecology of the Nanoose-Bonell estuary, Vancouver Island,British Columbia. Canadian Journal of Botany 64: 27-34.Dawson, T.E., and L.C. Bliss. 1987. Species patterns, edaphiccharacteristics, and plant water potential in a high-arcticbrackish marsh. Canadian Journal of Botany 65: 863-868.de Molenaar, J.G. 1987. An ecohydrological approach to floral andvegetational patterns in arctic landscape ecology. Arctic andAlpine Research 19(4): 414-424.- 115 -del Moral, R. 1979. High elevation vegetation of the EnchantmentLakes Basin, Washington. Canadian Journal of Botany 57:1111-1130.Disraeli, D.J., and R.W. Fonda. 1979. Gradient analysis of thevegetation in a brackish marsh in Bellingham Bay, Washington.Canadian Journal of Botany 57(5): 465-475.Douglas, G.W., and L.C. Bliss. 1977. Alpine and high subalpineplant communities of the North Cascades Range, Washington andBritish Columbia. Ecological Monographs 47: 113-150.Eleuterius, L.N., and C.K. Eleuterius. 1979. Tide levels and saltmarsh zonation. Bulletin of Marine Science 29(3): 394-400.Elsasser, W.M. 1969. Acausal phenomena in physics and biology:A case for reconstruction. American Scientist 57: 502-516.Environment Canada. 1980. Temperature and Precipitation, 1951-1980, British Columbia. Atmospheric Environment Service,Downsview, Ontario.Evans, R.D., and R.W. Fonda.^1990.^The influence of snow onsubalpine meadow community pattern, North Cascades,Washington. Canadian Journal of Botany 68: 212-220.Ewing, K.^1983.^Environmental controls in Pacific Northwestintertidal marsh plant communities.^Canadian Journal of Botany 61: 1105-1116.Flock, J.W. 1978. Lichen-bryophyte distribution along a snow-cover soil-moisture gradient, Niwot Ridge, Colorado. Arctic and Alpine Research 10(1): 31-47.Fox, D.J. and K.E. Guire. 1976. Documentation for MIDAS. ThirdEdition. Statistical Research Laboratory, University ofMichigan.Gamma Design Software. 1991. GS+-Geostatistics for the Agronomicand Biological Sciences (version 1.1). Gamma Design Inc.,Plainwell, Michigan.Gauch, H.G. 1982. Multivariate Analysis in Community Ecology.Cambridge University Press, New York.Gittins, R. 1985. Canonical Analysis. A review with application in ecology. Springer Verlag, Berlin.Greig-Smith, P. 1983. Quantitative Plant Ecology. Third Edition.University of California Press, Berkeley and Los Angeles.- 116 -Grootjans, A.P., R. Van Diggelen, M.J. Wassen, and W.A. Wiersinga.. 1988. The effects of drainage on groundwater quality andplant species distribution in stream valley meadows.Vegetatio 75: 37-48.Hitchcock, C.L., and A. Cronquist. 1973. Flora of the Pacific Northwest. University of Washington Press, Seattle.Holway, J.G., and R.T. Ward. 1963. Snow and meltwater effects inan area of Colorado alpine. The American Midland Naturalist 69(1): 189-197.Hrapko, J.0., and G.H. La Roi. 1978. The alpine tundra vegetationof Signal Mountain, Jasper National Park. Canadian Journal of Botany 56: 309-332.Hutchinson, I.^1982.^Vegetation-environment relations in abrackish marsh, Lulu Island, Richmond, B.C. Canadian Journal of Botany 60: 452-462.Hutchinson, I., A. Campbell-Prentice, and G.E. Bradfield. 1989.Aquatic plant resources of the Strait of Georgia. Pages 50-60in K. Vermeer and R.W. Butler, editors. The ecology andstatus of marine and shoreline birds in the Strait of Georgia, British Columbia. Special Publication of the CanadianWildlife Service, Ottawa.Ireland, R.R., G.R. Brassard, W.B. Schofield, and D.H. Vitt. 1987.Checklist of the mosses of Canada II. Lindbergia 13: 1-62.Isard, S.A. 1986. Factors influencing soil moisture and plantcommunity distribution on Niwot Ridge, Front Range, Colorado,U.S.A. Arctic and Alpine Research 18(1): 83-96.Jackson, M.L. 1958. Soil Chemical Analysis. Prentice Hall Inc.,Englewood Cliffs, New Jersey.Jenny, H. 1941. Factors of Soil Formation. McGraw-Hill, NewYork.Justesen, H.S. 1932. Influence of size and shape of plots on theprecision of field experiments with potatoes. Journal of Agricultural Science 22: '365-372.Kalamkar, R.J.^1932.^Experimental error and the field plottechnique with potatoes. Journal of Agricultural Science 22:373-383.Knapik, L.J., G.W. Scotter, and W.W. Pettapiece.^1973. Alpinesoil and plant community relationships of the sunshine area,Banff National Park. Arctic and Alpine Research 5(3): A161-A170.- 117 -Kolmogorov, A. 1933. Sulla determinazione empirica di una leggedi distribuzione. Giornalle dell Instituto Italiano degli Attuari 4: 1-11.Kowalski, C.J.^1972.^On the effects of non-normality on thedistribution of the sample product-moment correlationcoefficient. Applied Statistics 21: 1-12.Krige, D.G. 1966. Two dimensional weighted moving average trendsurfaces for ore-evaluation. Journal of the South African Institute of Mining and Metallurgy 66: 13-38.Kuramoto, R.T., and L.C. Bliss.^1970.^Ecology of subalpinemeadows^in the Olympic Mountains, Washington. Ecological Monographs^40(3): 317-347.Lilliefors, H.W.^1967.^On the Kolmogorov-Smirnov test fornormality with mean and variance unknown. Journal of the American Statistical Association 64: 399-402.Mahall, B.E., and R.B. Park. 1976. The Ecotone Between Spartina foliosa and Salicornia virginica in salt marshes of NorthernSan Francisco Bay. III. Soil aeration and Tidal Immersion.Journal of Ecology 64: 811-819.Nimlos, T.J., R.C. McConnell, and D.L. Pattie.^1965.^Soiltemperature and moisture regimes in Montana Alpine Soils.Northwest Science 39(4): 129-138.Oberbauer, S.F., and W.D. Billings. 1981. Drought tolerance andwater use by plants along an alpine topographic gradient.Oecologia 50: 325-331.Olyphant, G.A.^1984.^Insolation topoclimates and potentialablation in alpine snow accumulation basins: Front Range,Colorado. Water Resources Research 20: 491-498.O'Neill, R.V., D.L. DeAngelis, J.B. Waide, and T.F.H. Allen. 1986.A Hierarchical Concept of Ecosystems. Princeton UniversityPress, Princeton, New Jersey.Pidwirny, M.J. 1990. Plant zonation in a brackish tidal marsh:descriptive verification'of resource-based competition andcommunity structure. Canadian Journal of Botany 68: 1689-1697.Rhoades, J.D. 1982. Soluble salts. Pages 167-179 in A.L. Page,R.H.miller, and D.R. Keeney, editors. Methods of Soil Analysis, Part 2. Chemical and Microbiological Properties.American Society of Agronomy-Soil Science Society ofAmerica, Madison, Wisconsin.- 118 -Robertson, G.P. 1987. Geostatistics in ecology: interpolatingwith known variance. Ecology 68: 744-748.Selby, C.J., and M.D. Pitt. 1984. Classification and distributionof alpine and subalpine vegetation in the Chilcotin Mountainsof southern British Columbia. Syesis 17: 13-37.Simon, H.A. 1962. The architecture of complexity. Proceedings of the American Philosophical Society 106: 467-482.Smirnov, N.V. 1939. Sur les ecarts de la courbe de distributionempirique. Recueil Mathematique  N.S. 6: 3-26.Snow, A.A., and S.W. Vince. 1984. Plant zonation in an Alaskansalt marsh. II. An experimental study of the role of edaphicconditions. Journal of Ecology 72: 669-684.Stephens, F.R., and R.F. Billings. 1967. Plant Communities of atide-influenced meadow on Chichagof Island, Alaska. Northwest Science 41(4): 178-183.Stotler, R., and B.C. Stotler.^1977.^A checklist of theliverworts and hornworts of North America. The Bryologist 80:407-428.Ter Braak, C.J.F. 1986. Canonical correspondence analysis: a neweigenvector technique for multivariate direct gradientanalysis. Ecology 67: 1167-1179.Ter Braak, C.J.F. 1987a. The analysis of vegetation-environmentrelationships by canonical correspondence analysis. Vegetatio 69: 69-77.Ter Braak, C.J.F. 1987b. CANOCO-a FORTRAN program for canonicalcommunity ordination by partial detrended canonicalcorrespondence analysis, principal components analysis andredundancy analysis (version 2.1). TNO Institute of AppliedComputer Science, Wageningen.Thissen, D., L. Baker, and H. Wainer. 1981. Influenced-enhancedscatterplots. Psychological Bulletin 90: 179-184.Vince, S.W., and A.A. Snow. 1984. Plant zonation in an Alaskansalt marsh. I. Distribution, abundance and environmentalfactors. Journal of Ecology 72: 651-667.Webster, R. 1985. Quantitative spatial analysis of soil in thefield. Pages 1-70 in B.A. Stewart, editor. Advances in Soil Science Volume 3. Springer-Verlag, New York.Weir, D.A., and J.B. Wilson. 1987. Micro-pattern in an area ofNew Zealand alpine vegetation. Vegetatio 73: 81-88.- 119 -Whittaker, R.H.^1952.^A study of summer foliage insectcommunities in the Great Smoky Mountains.^Ecological Monographs 22: 1-44.- 120 -APPENDIX AThe following is a list of encountered plant species and theirrespective codes in the tidal marsh and wet meadow. A list ofencountered bryophytes in the meadow is also included.Nomenclature follows Hitchcock and Cronquist (1973) for vascularplants and Ireland et al. (1987) and Stotler and Stotler (1977) forbryophytes.Tidal MarshEncountered Plant Species^ CodeCarex lvngbvei Hornem. CARLYNPotentilla pacifica Howell POTPACTriglochin maritimum L.^ TRIMARAgrostis alba L.^ AGRALBSium suave Walt. SIUSUAStellaria humifusa Rottb. STEHUMDeschampsia cespitosa (L.) Beauv.^ DESCESRanunculus cvmbalaria Pursh^ RANCYMJuncus balticus Willd. JUNBALHordeum brachvantherum Nevski HORBRAAtriplex patula L.^ ATRPATAster eatonii (Gray) Howell^ ASTEATLathyrus palustris L. LATPALElvmus sp.^ ELYMSPScirpus maritimus L.^ SCIMARTrifolium wormskjoldii Lehm.^ TRIWORSonchus arvensis L. SONARVPlantago maritima L. PLAMARConioselinum pacificum (Wats.) Coult and Rose^CONPACWet MeadowEncountered Plant SpeciesCarex nigricans Retz.^ CARNIGCarex spectabilis Dewey CARSPEErigeron peregrinus (Pursh) Greene^ ERIPERCaltha biflora D.C. CALBIFEpilobium anagallidifolium L. EPIANAJuncus sp.^ JUNCSPLuetkea pectinata (Pursh) Kuntte.^ LUEPECValeriana sitchensis Bong.^ VALSITLeptarrhena pvrolifolia (D. Don) R.Br. LEPPYRCassiope mertensiana (Bong.) G. Don CASMERPoa cusickii Vasey^ POACUSHieracium gracile Hook. HIEGRALupinus latifolius Agardh^ LUPLATAgrostis thurbergiana Hitchc AGRTHUPhyllodoce empetriformis (Sw.) D. Don^ PHYEMPVeronica wormskjoldii Roem. and Schult. VERWOR- 121 -Senecio trianqularis Hook.^ SENTRIRanunculus eschscholtzii Schlecht.^ RANESCCastille -ja parviflora Bong. CASPARPoa leptocoma Trin.^ POALEPAnemone occidentalis Wats.^ ANEOCCPotentilla flabellifolia Hook. POTFLAAntennaria alpina Gaertn. ANTALPCetraria subalpina Imsh. CETSUBLuzula parviflora (Ehrh.) Desv.^ LUZPARKalmia microphylla (Hook.) Heller KALMICEquisetum arvense L.^ EQUARVTrisetum spicatum (L.) Richter TRISPILycopodium selaqo L. LYCSELPhleum alpinum L. PHLALPVaccinium deliciosum Piper^ VACDELPetasites friqidus (L.) Fries PETFRISphagnum warnstorfii Russ. SPHWARAbies lasiocarpa (Hook.) Nutt. ABILASErythronium grandiflorum Pursh^ ERYGRAPedicularis bracteosa Benth. PEDBRAEncountered BryophytesLescuraea radicosa (Mitt.) Moenk.Polytrichum piliferum Hedw.Polytrichum sexanqulare Brid.Kiaeria blyttii (Schimp.) Broth.Dichodontium olympicum Ren. and Card.Rhacomitrium sudeticum (Funk) B.S.G.Cladonia chlorophaea (Floerke ex Somm.) Spreng.Aulacomnium palustre (Hedw.) Schwaegr.Drepanocladus uncinatus (Hedw.) WarnstDrepanocladus aduncus (Hedw.) WarnstDesmatodon latifolius (Hedw.) Brid.Pohlia sp.Pohlia nutans (Hedw.) Lindb.Lophozia floerkei (Web. and Mohr) Schiffn.Nardia qeoscvphus (De Not.) Lindb.Bryum sp.Philonotis fontana (Hedw.) Brid.Cephalozia bicuspidata (L.) Dum.Dicranum scoparium Hedw.Brachvthecium reflexum (Starke'ex Web. and Mohr) B.S.G.- 122 -APPENDIX BVegetation (species X quadrats) data matrices at agg6a and 6b inthe subalpine meadow.^Data matrices are divided intosubcommunities 1-3 at agg6a and 1-8 at agg6b.^Species namescorresponding to the codes used below may be found in Appendix A.Cover scale values:^-3^(26-50%),^4^(51-75%),Agg6a(absent),^1^(<5%5^(76-100%).1 2aerial cover),^2^(6-25%),3111111234567890134711111222568901222222223333333234567890123456CARSPE 22223222222232 11111 1 ^ 11111-11LUPLAT 1231-211112133111113VALSIT 12232132222121-1111 ^SENTRI 21-21-11-1-111-- --11-1 HIEGRA -12-1211111111111 -11-1 POACUS 111111--1111--111---11111111-11--11-JUNCSP 11212112111211 -2121212-1111 1 ^VERWOR 1--1--11-11111 11111-111 1 ^ 1 ---POTFLA 2--11-11-11 ^ 1 ^LUZPAR -11111-11 ^ 1 ^ANEOCC ---1--11-2111- --1--1ABILAS ---1 ^ 1 ^ERYGRA ----1 CASPAR ---1---1-1121-1 -1-111 -1 ^POALEP ---11-1112111RANESC ---11-11-1111TRISPI ^ 1^11^1 1-1PEDBRA 1CARNIG 222333332332211111222343453443544554ERIPER 21121-111223221222212122111 ^ 1- --LUEPEC 112 -- 21 - 2 - 11113332232221111111 ^ 1---CASMER 1 ---- 1 ---- 111133222232111 ----1 PETFRI 1^ 1 ^PHLALP 1 1VACDEL 1 -1^1LYCSEL ^ 1^1-11 ^KATMIC 1^1111 1 CETSUB -1--11--^1-21111-1 PHYEMP 1^ 112232221111111 -1 ^CALBIF 1 21112212333231231121-2-12LEPPYR 1 12112213212112--2--2EPIANA 1111 - 1111111111-111212121121111-2111ANTALP ^ 1 ^ 11--11111111 ^AGRTHU 1111121111-1--1--1EQUARV 1 1-^1^1-^1^1SPHWAR ^ 1- 123 -Agg6bSENTRIVERWORPOTFLACARSPEVALSITPHLALPANEOCCABILASPOALEPRANESCHIEGRAJUNCSPLUZPARERYGRALUPLATTRISPIVACDEL -PEDBRA -CASMER 1POACUS 1LUEPEC 1ERIPERCALBIF 1PHYEMP 1CASPAR -ANTALP -KATMIC -CETSUB -LEPPYR 1EPIANA 1PETFRI -LYCSEL -AGRTHUEQUARV -CARNIG 2SPHWAR -1^2^3 4 5^6^7^81^1111^1111222222322223312569378140452368790 2348236790122221^1---111---111111^1-1^1111111^11^1^1 ^21111-1---1 ^2323221222222121111-11--111-2-11122222321221111111 ^11  ^1 ^-1121-1-1-21----1 ^1  ^1 ^-1111-1-1-1 ^1-111-1-1-111-1 ^--1111111111112-1-1-1 ^11111221211111211222^111^1 ^---1-111-1---1-  ^1 ^111-111143331-12121^1-1-111 ^1 ^1 ^11-11141222221-111111111111-1-111111111----1--1-1-1--1-212112332312211111111----22112-1-112222322222111211--1---1-11212123332223221111-11111322211-111-1 ^-111---1-11-12-11-1--1 ^1^1^111^11111^1 ^11-1111 ^- 1  ^1^1111^1 ^1^-21212211221111---11-11111-11111111211112121111-11  ^1 ^11-1111121111111 ^11113233422311111232^54334544545


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items