Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Relationships between site index of Sitka spruce (Picea sitchensis(Bong.) Carr.) and measures of ecological… Pearson, Audrey F. 1992

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

Item Metadata

Download

Media
831-ubc_1992_fall_pearson_audrey_frances.pdf [ 3.31MB ]
Metadata
JSON: 831-1.0075211.json
JSON-LD: 831-1.0075211-ld.json
RDF/XML (Pretty): 831-1.0075211-rdf.xml
RDF/JSON: 831-1.0075211-rdf.json
Turtle: 831-1.0075211-turtle.txt
N-Triples: 831-1.0075211-rdf-ntriples.txt
Original Record: 831-1.0075211-source.json
Full Text
831-1.0075211-fulltext.txt
Citation
831-1.0075211.ris

Full Text

RELATIONSHIPS B E T W E E N SITE INDEX O F SITKA S P R U C E (Picea sitchensis (Bong.) Carr.) A N D M E A S U R E S O F E C O L O G I C A L SITE QUALITY IN T H E E A S T E R N Q U E E N CHARLOTTE ISLANDS by AUDREY FRANCES PEARSON B.Sc.(Honours) University of British Columbia, 1983  A THESIS SUBMITTED IN PARTIAL F U L F I L L M E N T OF THE REQUIREMENTS FOR THE D E G R E E OF MASTER OF SCIENCE in T H E FACULTY O F G R A D U A T E STUDIES (DEPARTMENT O F FORESTRY)  We accept this thesis as conforming to the required standard  T H E UNIVERSITY OF'BFÛTISH C O L U M B I A OCTOBER  1992  © Audrey Frances Pearson, 1992  In  presenting  degree  this  thesis  in  at the University of  partial fulfilment  of  department  this thesis for scholarly or  by  his  or  the  requirements  for  her  I further agree that permission for  purposes  representatives.  may be granted  permission.  Department The University of British Columbia Vancouver, Canada  DE-6 (2/88)  advanced  It  is  extensive  by the head of  understood  that  publication of this thesis for financial gain shall not be allowed without  Date  an  British Columbia, I agree that the Library shall make it  freely available for reference and study. copying  of  copying  my or  my written  ABSTRACT  Relationships between measures of ecological site quality and height growth of Sitka spruce [Picea sitchensis (Bong.) Carr), measured as site index (metres at 50 years at breast height), were investigated for the eastern Queen Charlotte Islands using multiple linear regression techniques. Temperature and light were assumed to be similar i n a region of similar climate, expressed as biogeoclimatic subzone, and the investigation focused on measures of moisture and nutrients. This study tested the hypothesis that the classification units of biogeoclimatic ecosystem classification system are good indirect estimates of ecological site quality so are able to predict forest growth.  Fifty-five plots were chosen for study to represent a wide range i n ecological site quality and geographic variation. Individual measures of moisture and nutrients were represented as categorical variables; soil moisture and nutrient regime; and continuous variables; soil nutrients and predicted moisture deficit i n combination with soil physical measures. Soil moisture and nutrient regimes were represented as spectra of indicator species groups. The synoptic effect of individual measures of moisture and nutrients were expressed as site associations and assumed to be represented by plant associations. The relationship between the categorical and continuous variables for soil nutrients was also examined.  The most successful models used the B E C classification units; soil moisture regime, soil nutrient regime and site association. Multiple linear regression models using either soil moisture plus soil nutrient regimes or site associations were equally successful i n explaining variation i n (ln) site index (adjusted R2 = 0.80; 12 = 0.79). However, the former should be considered the  Ill  superior model for predictive purposes since the variance of the latter model was not homogeneous.  The soil nutrient regime model was as successful i n  explaining variation i n site index as the best continuous model (mineral soil mineralizable N plus forest floor mineralizable N, extractable calcium and potassium) (adjusted  = 0.40 versus 0.42). The soil moisture regime model  was more successful than those using continuous variables (adjusted  =  0.45 versus no relationship). The lack of relationship was attributed to inaccuracy of the water balance models due to their lack of calibration for this location and forest type. There was a moderate relationship with site index and plant associations (adjusted 2^ = 0.49); however vegetation is subject to change with succession over time so would be difficult to apply for predictive purposes. Indicator species groups showed no relationship with site index, likely due to the paucity of understorey species from canopy closure and intense browsing by introduced deer.  The soil nutrient regime classification was well supported by a discriminant analysis using soil chemical nutrients which correctly classified the plots i n 8 3 % of the cases, on average. The soil moisture regime classification could not be tested since the models used to estimate moisture deficit i n this study were considered too inaccurate. Continuous synoptic versus categorical models could also not be compared for the same reason.  It was concluded that the units of the biogeoclimatic ecosystem classification were good indirect measures of ecological site quality and good predictors of forest growth. The equations generated i n this study need to be tested with independent data and further investigation is necessary to determine moisture relationships for Sitka spruce.  TABLE OF CONTENTS Abstract Table of Contents List of Tables List of Figures List of Appendices List of Symbols Acknowledgements C H A P T E R 1 INTRODUCTION 1.1 Objectives 1.2. Literature Review 1.3 The Approach  ii iv v vi vi .vli viii 1 1 3 9  C H A P T E R 2. T H E STUDY A R E A 2.1 Physiography 2.2 Geology and Surficial Materials 2.3 Climate 2.4 Forest Vegetation  11 11 11 12 15  C H A P T E R 3 MATERIALS A N D M E T H O D S 3.1 Field Methods 3.2 Laboratory Analysis 3.3 Water Balance 3.4 Ecological Analysis and Classification 3.4.1 Vegetation Analysis and Classification 3.4.2 Site Analysis and Classification 3.5 Relationships Between Site Index and Ecological Variables  18 18 20 21 23 23 24  CHAPTER 4 RESULTS 4.1 Vegetation Classification 4.2 Soil Moisture Analysis 4.3 Soil Nutrient Analysis 4.3.1 Cluster Analysis for Soil Nutrients and Soil Nutrient Regimes 4.3.2 Discriminant Analysis for Soil Nutrients and Soil Nutrient Regimes . . . 4.4 Site Classification 4.5 Relationships Between Sitka Spruce Site Index and Ecological Site Quality  27 27 32 34 37 39 42 .47  C H A P T E R 5 DISCUSSION  55  C H A P T E R 6 CONCLUSIONS  63  LITERATURE CITED  .64  APPENDICES  72  25  LIST OF TABLES 1) Selected climatic characteristics for the C W H w h l variant  13  2) Diagnostic combination of species for plant associations  28  3) Means and standard deviations for selected environmental characteristics for plant associations  30  4) Means and standard deviations for selected characteristics by soil moisture regimes  33  5) Means and standard deviations for selected characteristics by soil nutrient regimes  35  6) Mean and standard deviations for Sitka spruce site index by soil moisture and soil nutrient regime  36  7) Group membership predicted by discriminant analysis  41  8) Constants and coefficients of classification functions for soil nutrient regimes determined by discriminant analysis 9) Means and standard deviations for selected characteristics for site associations  41 43  10) Models for regression of Sitka spruce site index and selected categorical and continuous variables  48  11) Stability of linear multiple regression nutrient equations  .49  LIST OF FIGURES 1) Location of the study plots i n the Queen Charlotte Islands 2)  19  8 0 % confidence ellipses for principal component scores for plant alliances along first two principal component axes  29  3)  B o x plots of Sitka spruce site index by plant association  31  4)  Cluster analysis for soil nutrients and soil nutrient regimes  38  5)  Discriminant analysis for soil nutrients and soil nutrient regimes  40  6)  Mean Sitka spruce site index for site associations  44  7)  Mean Sitka spruce site index for beach side, ocean-spray affected site associations  8)  45  Box plots of Sitka spruce site index by site association  46  9)  Predicted versus actual site index and residual analysis for soil moisture plus soil nutrient regime regression model 10) Predicted versus actual site index and residual analysis for site association regression model  51  11) Edatopic grid for Sitka spruce site index  54  52  LIST OF APPENDICES Appendix A Alphabetical list of plant species Appendix B Reciprocal averaging iterations for vegetation classification . . . . Appendix C 8 0 % confidence ellipses for principal component analysis of plant associations Appendix D Pearson correlation matrix for soil nutrients  72 73 79 82  LIST OF SYMBOLS Classification BEC  biogeoclimatic ecosystem classification  SMR SD F M VM W  soil moisture regime slightly dry fresh moist very moist wet  SNR P M R VR  soil nutrient regime poor medium rich very rich  Soil Variables FF MS  forest floor mineral soil  TOTC TOTN MINN P S Ca Mg K SUMCAT  total carbon total nitrogen mineralizable nitrogen phosphorus sulphur extractable calcium extractable magnesium extractable potassium sum of extractable calcium, magnesium and potassium carbon/nitrogen ratio pH  CN PH  Vlll  ACKNOWLEDGEMENTS I am very grateful to my committee members, Gary Bradfield and Peter Marshall, for being my guides to the world of statistics, an area 1 have developed a dangerous fondness for. Thanks also go to Rob Pollock formerly of the B . C . Forest Service, Del Williams of the B . C . Forest Service, Ron Bronstein formerly of Western Forest Products, Sue Craven of Fletcher Challenge and J o h n D u n c a n of MacMillan Bloedel for their kind assistance i n my field work on the Islands. J o h n Barker of Western Forest Products was also a valuable information source. 1 am grateful for the funding provided by the B . C . Forest Service. Special thanks go to Diane Hanson (nee Tamis), my trusty field assistant who measured the dozens of trees, dug the meters of soil pits and weighed the kilos of rocks cheerfully and competently. M y office mate, Gordon Kayahara, has been my invaluable companion. Reid Carter was always ready with helpful advice. Qingli Wang and Gaofeng Wang helped keep the laughter flowing. XieXie Peng-you! A n d to everyone who makes the Islands a very special place for me, How-a, Thank you. A n d to my Mother, who cheerfully paid whatever bills my education presented to her - 1 couldn't have done it without you! Following my heart has never been constrained for the mere lack of money. For that, 1 am very deeply grateful.  CHAPTER 1 INTRODUCTION  1.1 Objectives  One of foresters' great preoccupations is quantifying forest productivity and therefore the amount of wood that the landbase is potentially capable of producing. Forest productivity, as an ecosystem process, usually refers to the accumulation of photosynthate by the canopy, its allocation into tissue, a n d losses through respiration and consumption by herbivores (Waring and Schlesinger 1985). Foresters are virtually always only concerned with wood production, i.e. the amount of biomass or volume accumulated i n the stem and its rate of increase (Spurr and Barnes 1980).  Direct assessments of productivity are very difficult since insufficient long-term studies exist for reliable data sets (Daubermire 1976; Spurr and Barnes 1980). Height growth, however, can be measured indirectly as site index, the height of dominant trees at a reference age, and inferences made to volume (Monserud 1984). Therefore, site index serves an indirect measure of volume. Site quality has two similar but not identical meanings i n the literature. Carmean (1975) used it to mean the "productive capacity" of the land for growing trees, i.e., an index of performance which is usually measured as site index. Spurr and Barnes (1980) defined site quality as "the s u m total of all of the factors affecting the capacity to produce forests", such as climate and edaphic factors. Since the terms site index and site quality are used interchangeably i n the literature and it is not always obvious whether the investigator is referring to the productive capacity of the site, the factors which determine the latter, or height of trees at a reference age. To alleviate the confusion. Carter and K l i n k a (1991) chose to define site quality following Spurr  and Barnes (1980) as "ecological site quality" which is equivalent to the earlier concept of an "operational environment", the phenomena that affect a plant during its life cycle (Mason and Langeheim 1957).  The phenomena that effect the plant during its life cycle and, therefore its groAvth, are ultimately a function of five factors that cannot be replaced temperature, light, moisture, nutrients and mechanical forces (Livingston and Shreve 1921; Hills 1952; Major 1963; Waring and Major 1964; Bakuzis 1969; Krajina 1969). Therefore areas of similar temperature, light, moisture, nutrients and mechanical forces should have similar potentials for growth. These factors are also difficult to measure directly, so they too estimated indirectly.  The biogeoclimatic ecosystem classification system (BEC) is based on the concept of an operational environment, although mechanical forces are generally not considered. Temperature and light are estimated through delineating regions of similar climate (subzones) (Pojar et al. 1987). A site association consist of all sites that have similar or equivalent physical properties and the same vegetation potential (op cit). Therefore, the B E C variables are potentially good indicators of ecological site quality and good predictors of forest productivity. The objective of this study was to test the ability of the biogeoclimatic classification system's measures of ecological site quality to predict growth of Sitka spruce [Picea sitchensis (Bong.) Carr.) measured as site index (m @ 50yr at breast height age).  1.2 Literature Review  Bole productivity is virtually always represented by site index, the height of dominant trees at a reference age (Spurr 1952; Hagglund 1981; Monserud 1984). While there are certainly caveats with respect to the use of site index (Monserud 1984), dominant trees are generally a reliable index of stand volume since height growth is closely related to volume growth and is unaffected by stand density (Monserud 1984; Thrower 1989).  There have been hundreds of studies that have attempted to find relationships between site index and measures of ecological site quality. Comprehensive reviews have been published by Carmean (1975) and Hagglund (1981). Measures can be characterized as biotic and abiotic factors and are often used i n combination, such as i n a classification system (Spurr and Barnes 1980). Biotic factors include units of vegetation classification systems and frequency of indicator species. Abiotic factors include measures of climate, soil properties and topography. Techniques generally involve using the chosen measures of ecological site quality as independent variables i n regression analyses and/or testing for differences between proposed classes of variables of interest (Hagglund 1981). No one variable or approach has been consistently successful, and results vary widely with tree species and ecosystem.  Indicator plant analysis is based on the concept that plants serve as "phytometers", which integrate the s u m of environmental factors experienced by vegetation on the site (Spurr and Barnes 1980). Plants that occur only on a narrow range of environmental conditions can indicate that condition e.g., high available nitrogen, low soil moisture. Daubermire (1976) reviewed the use of vegetation to assess productivity and concluded that indicator plants are  effective measures of ecological site quality (e.g., Cajander 1926; Rowe 1956; McLean and Bolsinger 1973). Other authors caution that understorey plants in the forest do not necessarily only reflect site since they are not always independent of other factors, such as stand density, disturbance and past history (Spurr and Barnes 1980; Spies and Barnes 1985). Forest trees may also be more deeply rooted than understorey species; therefore, they can be influenced by different factors (Carmean 1975). Indicator plants may also only be applicable i n areas that have not been harvested or otherwise severely altered by humans (Spurr and Barnes 1980; Schonau 1988).  Green et al. (1989) found a good relationship between site index of second-growth Douglas-fir (Pseudotsuga menziesii (Mirb.) Franco) and indicator species groups for climate, soil moisture and soil nitrogen. However, Green (1989) found a poor relationship between western redcedar site index [Thiy'a plicata (Donn exD. Don i n Lamb.) and indicator species groups i n the Queen Charlotte Islands, which he attributed to poor understorey development due to canopy closure and browsing pressure from introduced deer.  Climate parameters used to represent ecological site quality have included latitude, longitude, growing degree days, temperature, wind, precipitation and évapotranspiration (Hagglund 1981). The first four variables are often used as means of stratifying by geographic area or similar climate (e.g., Farr and Harris 1979; 1983).  Climate on a local scale is often expressed  by topographic variables such as aspect, elevation and slope (e.g., Worrell and Malcolm 1990a).  The majority of studies concerned with quantifying and predicting forest growth have focused on soil parameters, both chemical and physical properties  (Carmean 1975; Hagglund 1981). Measures such as soil taxonomic unit, drainage class, and depth of rooting, are often used (op cit.). Usually these alone are insufficient to explain a large amount of variation i n site index and analjrtical variables such as soil nutrients or soil moisture content are needed. Carter and K l i n k a (1990) found promising results with a model of mineralizable N and moisture deficit as predicted by the Energy/Soil Limited Water Balance Model (Spittlehouse and Black 1981) and site index of Douglas-fir. Green (1989) found a good relationship between western redcedar site index and total N and volumetric moisture content.  However, relationships involving nutrients are unsuccessful when applied over large areas and require laboratory analysis (Com and Pluth 1984). Monserud et al. (1990) found that soil factors, both chemical and physical properties, accounted for far less variation i n Douglas-fir site index than the combination of elevation and either habitat type, longitude or precipitation. They attributed the poor result to site factor interactions over the very large study area or their failure to measure the true causes of productivity.  No generally accepted measure of soil moisture has been determined since it is not very straightforward to measure what the tree actually experiences (Broadfoot 1969; Stone 1978). Models that use climate data to determine growing season évapotranspiration and therefore moisture deficit such as Thornthwaite et al. (1957) and Spittlehouse and Black (1981) are potentially a means for overcoming the difficulty of field measurements.  Previous investigations into relationships of ecological site quality and productivity of Sitka spruce have occurred i n southeast Alaska, the Pacific coast, and northern Great Britain where spruce from the Queen Charlotte  Islands were planted as a timber tree after World War I. There has been a previous descriptive study i n the Queen Charlotte Islands (Day 1957) and a study investigating the productivity of forests of the Masset Lowlands that indirectly examined Sitka spruce (Green 1989).  The first attempt to relate Sitka spruce site index to ecological measures occurred i n Southeast Alaska. Stephens et al (1968) concluded that site index of Sitka spruce could be estimated from soil depth and drainage. They identified seven soil groups and estimated site index for each. Their mean site index values were then applied to the soil ecosystem classification system developed i n the Tongass National Forest and used as a means of assessing available timber volume (Rogers 1988). Use of the classification system was discontinued i n 1980 (Ford et al 1988)  Ford et aL(1988) examined the relationship of Sitka spruce site index i n Southeast A l a s k a to four soil variables commonly used i n classification: depth of mineral soil, soil drainage class, organic carbon content and coarse fragment content. They found only soil drainage and coarse fragment content to be significantly related to site index. However, there were no significant differences i n site index between individual drainage or coarse fragment classes. They concluded that soil taxa were not sufficient to predict site index of Sitka spruce, i n contrast to Stephens et al (1968).  Farr and Harris (1979; 1983) looked at the range of Sitka spruce along the Pacific coast. They tried to integrate the variables potentially important to growth into an expression of overall dominant climatic factors. They found that site index of Sitka spruce was strongly correlated with latitude and mean annual growing degree days. Site index decreased northward at the rate of approximately 0.8 m per degree of latitude. They considered that temperature  considered that temperature was more important than moisture deficit, since the latter is not a concern i n a climate where summer drought is rare.  In Great Britain, there have been several studies that have investigated relationships between climate and soil factors and productivity of Sitka spruce, measured as yield class (m^/ha/year). Worrell (1987) found there was a very smcdl difference i n yield class attributable to soil types. Plots i n w h i c h there was restricted rooting on wet, excessively clayey or shallow soils resulted i n a lower potential yield class. Yield class was significantly correlated with year of planting so the effect of soil was confounded by s i M c u l t u r a l input which increased with time (i.e. sites planted later received more site preparation and silvicultural treatment). Therefore, results were inconclusive.  Blyth and Macleod (1981a) found that total N i n the "humus" was strongly related to jàeld class. In a companion paper, Blyth and MacLeod (1981b) found that a small number of variables, including depth to mottling and total P of the h u m u s layer were able to account for 60-86% of the variation i n local yield class, but that the regression equations developed varied too greatly over short distances to be useful. They found physiographic and soil physical properties to be more useful for predictive purposes.  Worrell and Malcolm (1990a; 1990b) found that indices of temperature and windiness were strongly related to spruce 5àeld class. The amount of variation explained was only marginally improved by adding soil taxonomic class. Estimates of rainfall or potential water deficit showed no relationship.  Day (1957), i n a descriptive study of ecological requirements of Sitka spruce i n the Queen Charlotte Islands, concluded that dominance of spruce is  largely associated with the site characters related to soil water supply a n d drainage. He concluded that spruce is dominant on sites where there is a large enough supply of aerated water. The better drained the site, the more spruce is predominant. As this condition changes, western hemlock [Tst^a heterophylla (Raf. Sarg.) and western redcedar become the dominants. Spruce is generally the better competitor on better sites, but on less productive sites, it becomes even with hemlock and cedar. Green (1989) found similar results for the Masset Lowlands.  Day also considered that spruce is favoured by relatively high nutrient status. Its presence along the sea shore is related to nutrient enrichment from sea spray, especially phosphorus. Spruce is associated with sites where a high degree of nitrification occurs. O n poorly drained sites, nitrification is poor as is spruce growth.  Green (1989) examined variation i n tree growth i n imperfectly and poorly drained sites i n the Masset Lowlands. He used site index of western recedar as his measure of productivity and concluded that the site index of the other tree species, including Sitka spruce, was correlated with it. He found a good relationship between cedar site index, total N and volumetric moisture content. Sitka spruce showed a different pattern than did western hemlock and lodgepole pine {Pinus contortavar. contorta Dougl ex. Loud). Spruce height growth was comparably slower on poor sites and m u c h higher on better sites.  In terms of its autecology, Sitka spruce requires large amounts of phosphorous and balanced amounts of available calcium with magnesium (Krajina et al 1982). Its growth is greater when nitrates are more prevalent than ammonium compounds, but it will tolerate the latter. It has a high flood  resistance, and can survive under prolonged but not permanent flooding. Sitka spruce is an indicator of nitrogen rich soils, and does not occur on moisture or nutrient deficient sites (Klinka et al 1989).  1.3 The Approach  The B E C system is potentially well suited to characterize ecological site quality (Green et al. 1989). The classification deals primarily with three ecosystem elements: climate, vegetation, and soil (including topography and parent materials) (Pojar et al. 1987). The recognized units thus result from a synthesis of vegetation, climate and soil data (op cit.). A site association consists of all sites that have similar or equivalent physical properties and the same vegetation potential. Therefore, site classification organizes ecosystems on the basis of more or less stable environmental attributes and according to the concept of ecological equivalence.  The objective of this study was to test the ability of B E C measures of ecological site quality to predict growth of Sitka spruce, measured as site index (m @ SCtyrs at breast height age). Since the study occurred within a region of similar climate represented as a biogeoclimatic subzone, temperature and light were considered constant and the investigation focused on measures of moisture and nutrients.  Individual measures of moisture and nutrients were represented as categorical variables; soil moisture and nutrient regime; and continuous variables; soil nutrients and moisture deficit (Thornthwaite et al. 1957; Spittlehouse and Black 1981) i n combination with soil physical measures. Soil moisture and nutrient regimes were represented as spectra of indicator species  groups. The synoptic effect of individual measures of moisture and nutrients were expressed as site associations and assumed to be represented by plant associations. The relationship between the categorical and continuous variables for soil nutrients was also examined.  CHAPTER 2. THE STUDY AREA  The Queen Charlotte Islands are located off the coast of British C o l u m b i a between A l a s k a and Vancouver Island i n the Coastal Western Hemlock (CWH) biogeoclimatic zone (Meidinger and Pojar 1991).  This study was situated i n  the Submontane Wet Hypermaritime C W H variant (CWHwhl) which occurs on the drier leeward side of the Queen Charlotte Islands mountains.  2.1 Physiography  The Islands are divided into three distinct physiographic regions; the Masset Lowlands, the Skidegate Plateau, and the Queen Charlotte Ranges (Holland 1976). The Queen Charlotte Ranges are rugged and range i n elevation from 1200 m to sea level. The Skidegate Plateau ranges from 840 m to sea level.  The Masset Lowlands are less than 160 m i n elevation. A distinctive  group within the latter is the Argonaut plain, a n extensive glacial outwash plain on the northeastern tip of Graham Island.  2.2 Geology and Surficial Materials  Three major bedrock types (volcanic, sedimentary and intrusive) occur on the Islands (Banner et al. 1983). Volcanic rocks are the major type underlying most of the productive forests of the Skidegate plateau and Queen Charlotte Ranges.  The Masset Lowlands are underlain by sedimentary rocks. Intrusive  bedrock largely occurs i n the Queen Charlotte Ranges, which were not included i n this study. Lewis (1982) divided the volcanics into hard and soft, i n relation to their ability to weather. The hard volcanics are basalts, found largely i n the Queen Charlotte Ranges. The softer volcanics are deeply  fractured and chemically weathered. Sedimentary rocks, which include sandstone and conglomerate, release few nutrients. The finer-grained elastics (siltstones, shales and argillites) weather more easily and are more base-rich than the former. Areas of limestone do occur i n the Islands, but not i n the area of this study. It is difficult to determine the contribution of bedrock type to nutrient status, since its effects can be influenced by the nature and depth of the surficial material (op cit.).  Morainal and coUuvial materials are widespread i n the forested environments (Banner et al 1983). Fluvial and organic materials also occur to a lesser extent, with marine, eolian and exposed bedrock the least common.  2.3 Climate  In the Koppen classification, the climate is Cfb (Pojar and Banner 1984, Trewartha 1968): mesothermal (C), with no distinct dry season (f) and a cool summer season with the temperature of the warmest month under 22^C (b). Selected climatic characteristics for stations occurring within the study area are presented i n Table 1.  The climate of the Queen Charlotte Islands is determined by four factors (Williams 1968): (1) the proximity of the Pacific Ocean to the west, (2) the barrier of the mountains of the mainland to the east, (3) the rugged topography of the Islands, and (4) the behaviour of three prominent features of atmospheric circulation - the prevailing westerlies, the Aleutian Low and the North Pacific High.  Table 1. Selected climatic characteristics for the C W H w h l variant^ Climate station  Masset  Tlell  Sandspit  Sewell Inlet  1444  1152  1281  4168  Mean annual precipitation (mm) Mean annual snowfall(%MAP) Mean monthly precipitation May-Sept, (mm)  6.1  5.3  6.7  35  76.9  64.2  85.4  143.9  Mean monthly precipitation Oct.-April (mm)  150  111  142  473  Mean precipitation of the driest month (mm)  55  51  43  108  Mean precipitation of the wettest month (mm)  208  172  194  671  Mean annual temperature (^C)  7.6  7.4  7.9  7.6  Mean temperature of the warmest month (OC)  14.4  14.2  14.7  15.1  1.4  1.3  2.0  1.1  Mean temperature of the coldest month iPc)  ^ Source : Environment Canada (1980)  M u c h of t±ie air masses that move over the Charlottes are carried by the prevailing westerlies from the direction of the open ocean (since the ocean tends to be warmer t h a n the land i n winter and cooler than it i n summer) Williams (1968). The mainland Coast Range protects the Islands from continental air masses, which are cold i n winter and hot i n summer. Winters are thus very mild and summers rather cool.  There is an orographic effect created by the Queen Charlotte Ranges. Moist air moving i n from the ocean on the west is forced to rise thereby losing most of its moisture on the windward slopes. The precipitation is heavier on the westside of the Islands; this is more pronounced during the wettest (i.e., winter) months. This effect creates a large variation between the west and east coast.  The Aleutian Low is generated by the sharp temperature differences between the relatively warm ocean and the cold continental land mass during the part of the year when the s u n is south of the equator. This unstable low pressure system dominates the weather beginning i n late September.  It  creates a series of storms that give dense cloudiness, heavy precipitation and strong winds during the fall and Avinter.  The North Pacific High creates rather dry sunny conditions on the coast from mid-May to mid-September. Although the summers are still relatively cool and wet, warm, dry spells are not uncommon especially on the east coast.  The moderating influence of the ocean has other effects. The number of growing degree days is high. Frost is rare. The snowpack if any, rarely persists, except i n the higher elevations. Cloud cover and fog are frequent, so evapotranspirational demand even during the summer can be quite low. Variations i n temperature are more closely related to degree of protection from the open ocean than latitude, and generally do not vary widely - daily, seasonally or annually.  Conifers are thought to be competitively favoured by mfld, wet growing conditions that persist year round with the possibility of photos5mthesis during the winter months (Waring and Franklin 1979). The climate i n the Charlottes is an optimum one for conifers.  2.4 Forest Vegetation and Soils  The forests of Queen Charlotte Islands are described i n numerous publications, such as Lewis (1982), Banner et ai. (1983), Banner and Pojar (1984) and Banner et al. (1989). This discussion is largely based on the last, most recent work.  Climatic climax or zonal ecosystems occur on mesic to mesotrophic (i.e., fresh and nutrient medium) sites, usually on middle and upper slopes, on moderately well-drained moralnal and colluvial deposits. Loamy Humo-Ferric and Ferro-Humic Podzols with Mor humus forms are the typical zonal soil. Typic Folisols and Dystric Brunisols also commonly occur.  Banner et al. (1989) identified edaphic climax forests on moist, productive sites, which have an intermittent or constant lateral flow of mineral  seepage water providing a turnover of nutrients and sufficient soil aeration. These forests occur on active alluvial sites and middle to toe slopes of steep coUuvial or coUuviated morainal landforms. Mixing of soil from colluvial action or windthrow is also often characteristic. Banner et al (1989) considered these forests the most productive i n the C W H on the Islands, especially on recently deposited alluvial material. Less productive spruce forests occur on stabilized sand dunes and rocky marine headlands, both of which are influenced by sea spray.  Typical soils on active alluvial landforms include Dystric and Sombric Brunisols and Regosols with Mormoder, Rhizomoder, MuUmoder, and RhizomuU h u m u s forms. Podzols with Humimors occur on inactive alluvial landforms. O n moist, rich sites of colluvial or colluviated morainal deposits, soils are Dystric Brunisols cind Folisols with Humimor, Hydromor and Hydromoder h u m u s forms.  Low productivity forests are found on imperfectly to poorly drained gently undulating to flat terrain, such as the Masset Lowlands. Non-tidal wetlands are also common, but rarely contain spruce unless it is on raised hummocks.  Sitka black-tail deer were introduced to the Islands i n the early 1900s (Carl and Guiget 1972). They have severely altered the forest vegetation and depleted shrub and herb layers (Pojar and Banner 1984). Understorey species generally cannot persist upon stand closure (at age 25-35 years) because of severe competition, especially for light (Alaback 1982). The deer no doubt exacerbate this situation. A sparse understory on the forest floor with luxuriant growth on old stumps that are too high for the deer to reach is a common sight. Banner and Pojar (1984) caution that care must be taken i n  blaming the deer as the sole reason for a depauperate flora since overbrowsing varies i n its severity and there is little pre-and post-deer vegetation data.  CHAPTER 3. MATERIALS AND METHODS  3.1 Field Methods  Fifty-five stands of naturally regenerated second-growth Sitka spruce were chosen for study, representing a wide range of ecological site quality and geographic location (Fig. 1). In each stand, a 400m2 plot was selected that was 1) homogeneous with respect to soils and vegetation, 2) contained trees close to 50 years i n age with dominants free of disease or damage, and 3) had not been altered by silvicultural practices.  Site description for each sample plot followed the standard methods used by the B.C. Forest Service (Walmsley et al. 1980). Elevation, slope position, aspect and gradient, ground surface cover by forest floor, deca5àng wood, mineral soil and coarse fragment/bedrock were described. Vegetation on each sample plot was described using the standard methods of the B E C system (Pojar et al. 1987). Nomenclature followed Taylor and MacBride (1977) for the vascular plants, and C r u m et al. (1973) for the bryophytes. Percentage cover (on a significance scale) of each species was estimated. Percentage cover by recognized vegetation strata: (A (tree), B (shrub), C (herb) and D (moss)) was also estimated. Epiphytic species were not included.  In each plot, five dominant spruce trees were measured for age and height. Site index was determined from these measurements using site index curves for second growth Sitka spruce i n the Queen Charlotte Islands (Barker and Goudie 1987). Four soil pits, randomly located within the plot, were dug. One was dug down to parent material or root restricting layer to determine depth of rooting zone, the presence of gleyed soil horizons and the depth to groundwater table. Rocks greater than 2 cm diamter were weighed for coarse fragment determination. Soils were described following the Canadian Soil Survey Committee (1987) and D u m a n k s i (1978). H u m u s forms were determined from K l i n k a et al (1981).  Samples for mineral soil and forest floor bulk density were collected from each of the four pits. The volume of soil removed was estimated by measuring the amount of water required to fill the hole created. Samples for chemical analysis of forest floor and of mineral soil for 0-50 cm and 50+ cm were taken from the centre of the three undisturbed sides of the soil pit. These were composited into one forest floor sample per plot and one mineral soil sample for 0-50 cm and 50+ cm per plot.  3.2 Laboratory Analysis  The mass component of bulk density was determined by weighing the mass of soil removed from the bulk density pit after oven-drying at 105°C to a constant mass. Physical variables (bulk density, coarse fragment content and mineral sofl porosity) were calculated following K l i n k a et al (1981). Sofl nutrient variables were expressed i n concentration (% or ppm) and on a per hectcire basis, calculated following the equations given i n Green (1989).  After air drying to a constant mass, composite forest floor and mineral soil samples for chemical analysis were ground or crushed, if necessary, then passed through a 2 m m sieve. The coarse fraction (>2 mm) was weighed. Chemical analyses were undertaken on the fine fraction (<2 mm). Mineral soil and forest floor p H were determined with a p H meter after suspension i n water 1:5 for mineral soil, and 1:20 for forest floor. Total C was measured u s i n g an Leco Induction Furnace and C analyzer (Bremner and Tabatabai 1971). Total N was measured colourmetrically using semimicro-Kjeldahl digestion and N H 4 estimated with a Technicon Autoanalyzer (Anomymous 1976). Mineralizable N was measured colourmetrically with the autoanalyzer after an anaerobic incubation procedure (Powers 1980). Total P (forest floor) and extractable P (mineral soil) were estimated using the extraction procedure of Mehllch (1978). Total S for forest floor was determined using the Fisher Model 475 Sulphur Analyzer. Available S O 4 - S for mineral soil was extracted using an ammonium acetate extraction then reduced to sulphide by HI and the amount determined colourmetrically (Kowlenko and Lowe 1972). Extractable Ca, Mg, and K were determined AA^th a Morgan's solution of sodium acetate (pH 4.8) (Greweling and Peech 1980) and then measured by absorption spectrophometry (Price 1978). Textural analysis was determined by the pipette method, with sand content being determined by wet sieving (Lavkulich 1978).  3.3 Water Balance  Growing season water balance was calculated using the Energy/SoilLimited Water Balance Model developed for Douglas-flr on the south-east of Vancouver Island (Spittlehouse and Black 1981). Thirty year climate normals (Environment Canada 1980) were used for solar radiation, average monthly  temperature and total monthly precipitation. Percentage tree cover was used for the rain interception coefficient (the amount of rain intercepted by the canopy). A solar radiation coefficient of 0.85 (out of a maximum of 1.00) was chosen (Carter, pers. comm. 1). The lower value accommodates for fog which is frequent i n the Islands i n summer. Mineral soil rooting depth was used for rooting depth, except i n the case of Lignohumimors where the depth of forest floor was used with the soil coefficient for clay. The model assumes no recharge due to subsurface flow from higher elevations and/or no restricted drainage (Carter and K l i n k a 1990) and therefore cannot be applied for sites with those conditions.  Actual évapotranspiration and moisture deficit were also calculated following Thornthwaite et al. (1957). Thirty year climate normals (Environment Canada 1980) were used for average monthly temperature and total monthly precipitation. Soil rooting depth and soil texture were used to determine the appropriate coefficients for water holding capacity from which actual évapotranspiration and moisture deficit were calculated.  1. Reid Carter, Resource Analysis/Evaluation Forester, Fletcher Challenge Canada Ltd.  3.4 Ecological Analysis and Classification  3.4.1 Vegetation Analysis and Classification  Plots were sorted into groups using Reciprocal Averaging (RA) i n a progressive fragmentation fashion (Gauch 1977; Peet 1980).  With each  iteration, outliers were removed and the analysis performed again until there were no distinct groups remaining within the data. Tabular analysis was also used i n the group formation process. The associations were then formed i n a hierarchical classification following Pojar et al. (1987) using the VTAB computer programme (Emanuel 1985).  The classification was also examined using a centered, covariance Principal Component Analysis (PCA) (Gauch 1977). Degree of separation was examined at both the association and alliance level. Component scores were analyzed for Axes 1-4 and for all species, understory species, with rare species (those occurring only once) removed, and for diagnostic species identified i n the classification.  Although the first PCA axis, by definition, explains the most variation i n the data, it is possible that ubiquitous species are contributing to most of that variation, thus differences between groups may not be revealed on the first or second axis because they are mainly influenced by the more dominant, common species shared by all groups. Therefore the scores from the third and fourth axis were also explored.  3.4.2 Site Analysis and Classification  Soil nutrient regime (SNR) was estimated from the field key (Banner et al. 1990), based on soil morphological properties, assisted by spectral analysis of indicator plants (Emanuel 1985). Relative soil moisture (SMR) regime was estimated from the field key, then converted into actual moisture regime from the conversion table, based on biogeoclimatic subzone (Banner et al. 1990). Final assignment was further refined by spectral analysis. Using the actual soil moisture and nutrient regimes identified, site associations were determined from Banner et al. (1990). Determination of beach side, ocean-spray affected sites was guided by estimates of the area of edge influence of salt spray (Cordes 1972).  Since classification based on field estimates is potentially subject to observer bias or error, the degree of agreement between the soil chemical nutrient data and the SNR classification was explored. No independent measure of soil moisture was available to test the soil moisture regime classification. A Ward m i n i m u m variance cluster analysis (Ward 1963; Wilkinson 1990) was performed to reveal natural groupings i n the data and whether they related to assigned classes. This type of clustering, also known as the sums of squares method, minimizes the total within group sums of squares about the group centroid (Gordon 1981). The relative branch lengths indicate the joining distance between pairs of clusters.  A discriminant analysis (DA) was used to test the ability of the nutrient data to correctly place each plot i n the assigned soil nutrient regime (Wilkinson 1990). D A is a procedure for combining variables so that the resulting  composites have optimal properties for distinguishing the groups i n the smallest possible number of dimensions (Gittins 1979). It identifies the linear combinations of the responses which best separate the groups and the variables that contribute the most to separating the groups. DA requires that the input variables have homogeneous variances and are multivariate normal. Homogeneity of variances of input variables was tested Avith a Bartlett's test (Zar 1984, Wilkinson 1990). Multivariate normality was assumed.  3.5 Relationships Between Site Index and Ecological Variables  Using the conceptual model that plant growth is a function of soil moisture and nutrients, relationships between site index and categorical or continuous variables representing soil moisture and nutrients were explored using simple and multiple linear regression analysis. Relationships were also explored between spruce site index and vegetation; using both plant associations and spectra of indicator species groups, and site, represented as site associations. For linear regression, the variances must be homogeneous. The presence of heterogeneity was was tested using a Bartlett's test (Zar 1984)  Multiple regression analysis requires that the explanatory variables are not strongly interrelated (Chatterjee and Price 1977). If this condition is not met, then the result of the regression can be ambiguous since the coefficients estimated become very sensitive to slight changes i n the data or addition or deletions of variables i n the equation. Nutrients are often interrelated.  A Pearson correlation matrix was calculated for the sets of variables of interest (Wilkinson 1990) (Appendix D) and only variables that were not highly correlated were used i n a stepwise regression. The stability of the nutrient  equations generated was tested by randomly removing five plots and comparing the coefficients with the original equation. If the coefficients did not vary greatly, then the variables are not multicollinear (Chatterjee and Price 1977).  CHAPTER 4 RESULTS  4.1 Vegetation Classification  The progressive fragmentation, i n conjunction with the tabular analysis, resulted i n ten floristically similar groups (Table 2; Appendix B). These groups were then organized into a hierarchical classification, consisting of two alliances each with five associations which were arranged on a diagnostic table (Table 2). Only the Picea-Calamagrostis,  -Claytonia, -Gymnœarpium and -  Gaultheria associations were clearly distinct A v i t h the fragmentation technique (Appendix B). The other associations were recognized through tabular analysis.  Of all the combinations of species used to test the degree of separation of the vegetation groups with principal component analysis (all species, no rare species, diagnostic species, understoiy species), understory species at the alliance level showed the most separation, which was i n itself marginal (Fig. 2; Appendix C). This lack of distinctness is further confounded at the association level by the small sample size of some of the associations. The third versus second, and fourth versus third axes showed no differences between groups, associations or alliances.  Table 2. Diagnostic combination of species for plant associations distinguished i n the study plots. Vegstot  ton  unit  Number of Veaetatlon  1  ptots units  a n d species  Dlagnostlc va lue'  3  2  l  4  5 Presence  5  1  G  1 B c l a s s ' a n d m ean  s p e c i 'S  3  ctqn  G f 1cance'  B  â  10  16  G  7  PIcaa-Ga1lum ©11. Gai  fum  tri  fforum  2  8  IV 3 . 3  3  5  IV  1 0  3  B  IV  J 3  I I  2  8  5 8  3  0  • 0 3 5  3 7 3 2  0 2 0 5  Picea-Calamogrost Is ts  Cel fimagrost is nuthaen» Oicentra formosa Hypochocrfs radical a Rubus parviflorua Vicia americana P I c e a - C l a y t o n i a  (d.cd) (d.c) (d.c) (d.c) (d.c)  5 5 5 5 5  9 *  5 5 5 5 5  3  a .  A\nu% rutyra Cl aytoni a sltji rica Gailum Bparine Ci»''L/m triflorufn Poa marc Ida Polypodtum glycyrrhlza S t f l 1 a r l a cri spa PIcaa-Rhytidiada1phus Rhy(tdlade(pHus  (dd) (d.cd) (d.c) (dd) (d.cd) (d.cd) (d.c) t r1 q u e s t r j s  trtqufftrus  PIcea-Gymnocarplum  S  •  5  5 5 5 5 5 5 5  7 8 4 . 7 , 8 . 6 . 4 ,  5 5 5 5 5 5 5  2  3 2,4  7.0 m * .0  11  1 , -1  a . (d)  a .  Conocepha1um conicum Cymnocsrplum dryopteris Hy 1 ocom lyjm spl cndens Osmorhtza chllensis PIagiomnlum Insigne Rh 1 zomn 1 i / m g 1 abrescens P i c e o - P o l y s t i c h u m  (d.c) (d.c) (d.c) (d.c) (d.cd) (d.cd)  I I  I  5,7  1 5  1 1 4 8 I 1 3 8 5 5 5 7  5 5 5 5 5 5  III  I  *  0  IV 1 I IV  4  5 0 1  3 a  2  I V  ' 4  0 8  1 . 1  III III  3.3  a . (d) (dd) (d) (d)  Dryopt er1 s expansa Po 1 y^t 1 chum m\jn i t um Pter1dium equ1t1 nom VaccInlum parvifollum  5  1 1 I I I  3 . 5  1 * • *  0 0 G .0  3  IV V IV IV  0  ^ 0 * .0  i V 1 I  *  0  iir  l 5  5 5 G 5  III  3 .0 * .0 * 0  Plcea-Hylocomlum a M .  P f c e a - G a u l t h e r l a Caultnerla  ThuJs  plI  (d.cd)  on  1 l 2 8 1 5 4 G  I I IV IV IV  2 5 5 3  , 1 III V 3 V . 1 V , 3  G  1 . 1 5.8 1 . 3 r. t  V IV V  IV IV V V  2 A  2  2 5 2 7  V V V IV  1 . 8 4 , 8 5,7 3 4  5  +.5  cat a a .  rutjra  PIcea-KIndborgla Plagi Polyst  1 I 1 I I I I IV  a .  P1cea-Hy1ocomlura  Alnus  3 3  G 1 5 5  a .  shalI  Picea-Thoja  3 5 3 3  (d) (d.cd) (d) (d.c)  Blechnum spicant Hylocomlum sp/tfndens R f t y f fdi a d « ( p h u s ( oreus Tsuga heterophylI a  omnium ichum  'Spoclea dominent,  a .  c  -  (d) (d.c)  valuaa:  constant,  d -  l e -  5  d I f f e r o n t I aI.  important  1  3 , 5  d d -  companion  dominant (Pojar  •Presence classes a s percent o f frequency; I - 1-30. II V - 81-100. I f 5 p l o t s o r less, presence class Is araolc 'Species s i g 0 7 ( 0 . 4 15.0 ( 1 0 . 1 70.0). 9 -  3 5  IV  1 0  5 3  5 5 • .0  V  7 2  I!  « 5  V  2 8  1  *  1  •  Ol  V  I  * . 0  1  J  * . 0  II  7.o]lII  1.4  ^  Iniigne muni turn  d i a g n o s t i c  5  2-  n i f i c a n c e class midpoint percent 1 . 0 ) , 1.6 (1.1 - 2.1), 3 - 20.0), 6 - 3 6 . 5 (30,1 - 33.0), 8 5 . 0 ( 7 0 . 1 - 100).  « .0  d i f f e r e n t i a l ,  s t a l .  1987).  21-40, value  111 (1-5).  c d -  41-60,  constant  IV -  61-80.  c o v e r a n dr a n g e : • • 0 . 2 (0 I - 0 3 ) . i • 3 . 6 ( 2 . 2 - 5 0 ) , 4 . 7 . 5 (5.1 1 0 0 ) . 5 7 - 4 1 . 5 (33.1 - 50.0), 8 - 6 0 . 0 ( 5 0 . 1 -  0  3  .  1  + 0  m V  1.1 2 . 0  Fig. 2. 8 0 % confidence ellipses for principal component scores for plant alliances along first two principle axes. l=Picea-Galiam alliance, 2=PiceaHylocomium alliance from Table 2.  Synoptic characteristics for plant associations are presented i n Table 3 and box plots of site index i n Fig. 3. There is a trend of increasing site index within the associations, but it is not related to the order of the associations i n Table 2.  Table 3. Means and standard deviations (in parentheses) for selected environmental characteristics for plant associations. Plant  1  2  3  4  5  6  7  8  9  10  1  1  5  2  8  3  6  16  6  7  association ^ N u m b e r of plots  21  32  23  33  31  23  27  37  32  35.1  (0.0)  (0.0)  (3.9)  (4.2)  (5.4)  (7.2)  (3.9)  (3.7)  (6.8)  (3.4)  SMR  SD  SD  SD-W  F-M  SD-W  SD-M  SD-W  F-VM  SD-VM  SD-W  SNR  M  R  P-VR  M-R  M-VR  P-M  M R  M-VR  P-VR  R-VR  S S SI  Gleying2  Water  TOTN^  MINN  FFCN  table^  0  0  20  0  32  0  12  67  18  24.5  (0)  (0)  (0)  (0)  (0)  (0)  (0)  (21)  (0)  (1)  0  0  0  0  45  0  54  10  66  35  (0)  (0)  (0)  (0)  (17)  (0)  (9)  (21)  (0)  (16)  7012  10918  4672  4748  10245  3878  5696  8635  (0)  (0)  (2930)  (605)  (3106)  (1699)  (1640)  106  109  127  113  199  120  157  (0)  (0)  (54)  (23)  (149)  (25)  (50)  51  35  40  36  36  52  40  (0)  (0)  (8)  (18)  (9)  (7)  (4)  8626  8519  (2719)  (5305)  175  192  180  (68)  (70)  (80)  42  39  27  (8)  (8)  (6)  (3337)  ^ Plant associations from Table 2. 1-Calamagrostis, 2-Claytonia, 3-RhytidiadelphiLS, 4-Gymnocarpium, 5-Polystichum, 6-Gaulthena, 7-ThiLja, 8-Hylocomium, 9-Alnus, lO-Kindbergia ^ Depth to gleyed horizon (cm) measured from the soil surface. Average is based on plots where gleying occurred. Depth to water table (cm) measured from the mineral soil surface. Average is based on plots where water table occurred. ^ A l l nutrient values in kg/ha and combined forest floor and mineral soil unless otherwise indicated.  50  1  1  1  1  \  \  r  40  X CD  TD  -  30  20  o  L o  10  J  6  J  _L  3  7  5  9  \  4  I  L  10  8  Plant association  Fig. 3. Box plots for Sitka spruce site index by plant association (PA) for associations with greater than one plot. Numbers designating PAs are explained i n Table 2.  4.2 Soil Moisture Analysis  Selected characteristics for S M R are presented i n Table 4. Site index shows a significant trend (p<0.05) of higher i n F - M - V M and lower i n S D and W SMRs. Soil nitrogen shows no significant trends but the means are highest i n the moist sites and lowest i n the wet sites. A s would be expected, depth of gleying and depth to water table are lowest i n V M and W.  Values generated from the Energy/Soil Limited Water Balance Model (Spittlehouse and Black 1981) are not presented.  Model results could not be  confidently applied due to poor weather data and the lack of calibration of model coefficients for the North Coast. This finding is further discussed i n Section 4.5 and Chapter 5.  Values from the Thornthwaite model are also not presented (Thornthwaite et al. 1957). For the plots at Sewell Inlet, precipitation exceeded the actual potential évapotranspiration calculated by the model, therefore, no deficit occurred. The model calculated a growing season moisture deficit for all other plots of between 45 and 105 mm, giving an actual/potential évapotranspiration ratio between 70-85% and deficit for 3 months of the growing season. Excluding the plots where a water table was present so no deficit occurred, the Thornthwaite model would place all the remaining plots as moderately dry (Klinka et al. 1989), which is unlikely to be accurate. These results are discussed further i n Chapter 5.  Table 4. Means and standard deviations {in parentheses) for Sitka spruce site index and selected soil characteristics by soil moisture regimes. Rows w i t h no values superscripted i n the same row indicate no significant differences between means. SMR  Number of plots SS SI  TOTNI  MINN  Slightly Dry  Fresh  Moist  Very Moist  Wet  14  11  15  10  5  25.1^  33.6^  36.7^  33.1^  26.4A  (5.0)  (4.7)  (3.5)  (5.7)  (4.4)  7077  8323  8705  8187  5006  (3349)  (4121)  (3444)  (3869)  (2833)  158  158  203  157  126  (60)  (73)  (111)  (70)  (41)  GLEYING^  oA  QA  30B  40^  30^  DEPTH  (0)  (0)  (0)  (7)  (5)  WATER3  oA  QA  oA  44B  27B  DEPTH  (0)  (0)  (0)  (27)  (16)  ROOTING DEPTH  57  55  65  50  26  (21)  (16)  (29)  23  (15)  ^ A l l n u t r i e n t v a l u e s i n k g / h a a n d c o m b i n e d forest floor a n d m i n e r a l s o i l . 2 D e p t h to g l e y e d h o r i z o n (cm) m e a s u r e d f r o m t h e s o i l s u r f a c e (average o f w h e r e g l e y i n g o c c u r r e d ) . D e p t h to w a t e r table (cm) m e a s u r e d f r o m t h e m i n e r a l s o i l s u r f a c e (average o f p l o t s w h e r e w a t e r t a b l e o c c u r r e d ) . V a l u e s w i t h different letters s u p e r s c r i p t e d i n the s a m e r o w i n d i c a t e s i g n i f i c a n t l y different m e a n s (P< 0 . 0 5 , T u k e y ' s test).  4.3 Soil Nutrient Analysis  Selected characteristics for soil nutrient regimes (SNR) are presented i n Table 5. Site index increases significantly (p<0.05) from poor to very rich SNR. While not significantly different, the mean values of F F P and FFS are highest for the very rich sites. There is a trend of decreasing forest floor C/N and increasing total and mineralizable N with richer sites. In comparison with the classes proposed by K l i n k a et al (1989) for actual soil nutrient regimes, those identified i n this study have much higher values for nitrogen (both total and mineralizable), while p H and forest floor C/N have a relatively smaller range. However, the classes of Klinka et al (1989) were developed for southwestern B.C. so direct comparisons are not likely applicable. Percentage of indicator species indicating nitrogen rich soils (NITR3) do increase with richer nutrient regimes, although it is not significant (p>0.05).  Table 5. Means and standard deviations (in parentheses) for Sitka spruce site index and selected characteristics by soil nutrient regimes. Rows with no values superscripted i n the same row indicate no significant differences between means. SNR Number of plots  Poor  Medium  Rich  Very R i c h  4  18  22  11  SS SI  I9.5A  28.9^  34.8^  33.6^  (3.7)  (5.8)  (3.7)  (6.2)  FFPH  4.0  4.1  4.1  4.2  (0.4)  (0.4)  (0.4)  (0.4  49A  43A  35B  32B  (8)  (7)  (9)  (4)  2777A  5770^ (1429)  8811^  10844^ (3262)  14lA (48)  I65A  (7)  (62)  243^ (116)  414 (94)  444  (288)  437 (232)  652 (258)  FFS  308 (188)  383 (273)  397 (215)  573 (256)  SUMCAT  1115 (763)  2040 (2136)  2287 (2550)  2087 (939)  3  10 (18)  27 (34)  30 (30)  FFCN TOTN^  (857) MINN FFP  NITR32  87A  (5)  (3578)  ^ A l l nutrient values combined forest floor and mineral soil i n kg/ha unless prefaced by F F (forest floor only). ^ Spectral frequency of indicator species groups indicating nitrogen rich soils. Values with different letters superscripted i n the same row indicate significantly different means (P< 0.05, Tukey's test).  The patterns of the combined soil moisture and nutrient regimes with respect to site index (Table 6) are similar to that of individual regimes (Tables 4 and 5). Site index increases with site richness. Moist sites show the highest site index followed by fresh and very moist sites.  Table 6. Means, standard deviations (in parentheses) and number of plots for Sitka spruce site index by soil moisture and nutrient regimes. Soil moisture regime  Poor  Soil nutrient regime Medium  Rich  Very Rich  16.5 (1.5) n=2  22.6 (2.1) n=5  30.3 (1.5) n=3  28.5 (0.6) n=4  Fresh  29.7 (2.1) n=3  35.5 (5.3) n=6  34 (0) n=2  Moist  34.6  37  41  (3.6)  (0.7)  (1)  n=7  n=5  n=3  Slightly Dry  Very moist  22 (0) n=l  26 (0) n=l  34.7 (3.0) n=7  40 (0) n=l  Wet  23 (0) n=l  25 (1.4) n=2  34 (0) n=l  25 (0) n=l  4.3.1 Cluster Analysis for Soil Nutrients and Soil Nutrient Regime  The Ward m i n i m u m variance cluster analysis (Wilkinson 1990) used F F p H , F F C N , combined forest floor and mineral sofl total N, mineralizable N, forest floor total S, and s u m extractable cations (Ca, Mg, K) (Fig. 4). A l l nutrients were measured i n kg/ha. These measures were chosen as they showed the most variability within the data and/or have been used to previously characterize SNR (Klinka et al. 1989).  The cluster analysis shows 4 broad groupings within the data which generally exhibit a trend from poor to very rich. There are however some exceptions. These were either extremely high or low i n one nutrient with respect to other plots i n the same SNR or they are at the boundary between classes. Generally, cluster analysis supports the SNR classification.  DISTAWCE METRIC IS EUCLIDEAN DISTANCE WARD MINIMUM VARIANCE METHOD TREK DIAGRAM P  0.000  DISTANCES  P P M M  h  M M P R R R R R  M M M M H M R R R R  M  H M  M H  1^  M M R R R  11-  R  VR R  VR VR VH VR  R R R M R R R •/R •JH  R  •/R  R VR VR VR  Fig. 4. Dendrogram produced by cluster analysis for selected soil nutrient variables and soil nutrient regimes. P=poor. M=medium, R=rich, VR=very rich  4.3.2 Discriminant Analysis for Soil Nutrients and Soil Nutrient Regimes  Forest floor p H and C/N, mineral soil p H and C/N, forest floor P and S, and combined forest floor and mineral soil total N, mineralizable N and s u m extractable cations (all i n kg/ha) were used i n the discriminant analysis. Nutrients i n kg/ha were transformed to log (10) to homogenize the variances. These measures were chosen since values for individual nutrients i n kg/ha or concentration could not be transformed to make the variances homogeneous. Of all the variables tested, only log total N, log mineralizable N, and F F C N contributed significantly to the discrimination (p< 0.05). The canonical correlations of the derived functions were 0.87 for the first function and 0.34 for the second. The first function is, therefore, more strongly related to the nutrient regime classes than the second. With 8 0 % confidence ellipses, there is overlap between the groups (Fig. 5) but the centroids are all significantly different from each other (p < 0.01 Wilks' Lambda F-statistic). The ellipse was omitted for the poor group because of too few observations.  As with the cluster analysis, the cases of where the proposed classification did not agree with the DA, the plots were located at class limits, and/or had one nutrient that was exceptionally low or high compared the other plots i n the same class. Generally the D A supports the classification (Table 7). Standardized coefficient functions are presented i n Table 8.  Fig. 5. Ordination of the study plots as a function of the first canonical yanates (determined by discriminant analysis) showing 8 0 % confidence el for soil nutrient regime classes. P=poor, M=medium, R=rich, V=very rich  Table 7. Group membership predicted by discriminant analysis SNR  P  M  R  VR  Total  %Correctly classified  P M R VR  3 0 0 0  1 17 4 0  0 1 16 1  0 0 2 10  4 18 22 11  75 94 73 91  Total  3  22  18  12  55  83 (average)  Table 8. Constants and coefficients of classification functions for soil nutrient regimes determined by discriminant analysis. SNR Constant  P  M  R  VR  -608.03  -674.05  -722.90  -770.24  0.12  -0.08  -0.32  -0.44  LGTOTN  235.65  251.63  264.00  268.48  LGMINN  -47.59  -44.21  -44.91  -40.14  FFC/N  4.4 Site Classification  Thirteen site associations were distinguished following Banner et al. (1990). Selected characteristics by site association are presented i n Table 9, and for site index i n Fig. 6-8. There is a trend of increasing site index with rich-very rich and fresh-moist-very moist site associations (Fig. 8). The nutrient values are less clear (Table 9). The slightly dry rich-very rich associations (11 and 5) have as m u c h total N as 6, but site index is less. Mineralizable N is highly variable. A Tukey's test (Zar 1984) was attempted to test for significant differences i n site index among the site associations (SA). However, the results were too difficult to interpret. While the extremities are significantly different (SA 10,4,11 versus 2,7,6,9 i n Table 9), the associations i n the middle overlap. The site index for SA8 is not significantly different from any other association, likely due to its high standard deviation. A box plot of site index versus site association was chosen instead for presentation (Fig. 8). Associations with only one plot were not included.  Table 9. Mccins and standard deviations (in parentheses) for selected characteristics by site associations i n order of increasing average site index. Site association^ Number of plots  10* 1  13 5  3 1  11* 1  4 3  12* 1  5 7  8 2  1 3  2 7  7 8  6 14  9 2  SS SI  15.0 (0)  22 (2.5)  22.0 (0)  21 (0)  24.3 (1.5)  26.0 (0)  29.3 (1.4)  29.5 (6.4)  29.7 (2.1)  34.6 (3.7)  35.4 (3.4)  36.6 (4.1)  38.0 (2.8)  SMR SNR  SD* P-M*  SD P-M  VM P-M  SD-F* R-VR*  W P-M  M-VM* P-M  SD-F R-VR  W R-VR  F P-M  M P-M  VM R-VR  F-M R-VR  HB R-VR  FFPH  4.5 (0)  4.1 (0.3)  3.7 (0)  4.0 (0)  3.9 (0.1)  4.2 (0)  4.0 (0.3)  4.6 (0.4)  4.2 (0.1)  3.9 (0.5)  4.3' (0.5)  4.0 (0.3)  4.1 (0.4)  FFCN  59 (0)  39 (8.4)  41 (0)  51 (0)  45 (1)  45 (0)  33 (9)  31 (5)  46 (4)  49 (5)  38 (8)  33 (9)  33 (2)  2448 (0)  4044 (1289)  4030 (0)  7012 (0)  4051 (1821)  5980 (0)  9914 (1598)  6441 (4315)  5716 (1200)  6569 (1454)  8982 (3919)  10068 (4265)  9025 (754)  91 (0)  123 (21)  91 (0)  107 (0)  110 (32)  48 (0)  199 (57)  149 (54)  171 (32)  155 (59)  179 (59)  203 (124)  169 (56)  Total N 2  Mineralizable N  1 S i t e a s s o c i a t i o n codes: (*=beach a s s o c i a t i o n ) 1 - H w S s - L a n k y m o s s , 2 - C w H w - B l u e b e r r y , S - C w Y c - G o l d t h r e a d , A-PlYc-Sphagnum, 7 - C w S s - C o n o c e p h a I u m , 8 - C w S s - S k u n k cabbage. 9-Ss-Uly-of-the-valley,  10*-Ss-Salal, 11*-Ss-Reedgrass,  ^ N u t r i e n t v a l u e s i n k g / h a c o m b i n e d forest floor a n d m i n e r a l s o i l u n l e s s o t h e r w i s e i n d i c a t e d .  5-CwSs-Swordfem,  12* S s - S w o r d f e m , 13 S s - S a l a l  6-CwSs-Foamflower,  Wet Hypermaritime C o a s t a l Western Hemlock(CWHwh)  P  SOIL NUTRIENT REGIME M R  SD Moss  VM  C w - S s - F o a m f lower 37(4,1) n=14  35(3.7) n=7  ISs-Lily-of-^the-valleyl* 138(2 &) n=?!* Ifloodplain-high bench!'  Cw-Yc-(3oldthreaci  Cw-Ss-Conocephalum  22(0)  n=1  Pl-Yc-Sphagnum  W  29(1.5) n=5  3 0 ( 2 1) n=3 Cw-Hw-Blueberry  M  VR  Cw-Ss-Swordfern Hw-Ss-Lanky  F  Subzone  24(1,6) n=4  35(3.4) n=8 Cw-Ss-Skunk  cabbage  30(6.4) n=2  Fig. 6. Mean Sitka spruce site index, standard deviations i n parentheses and number of plots for site associations. n=number of plots. Site associations from Banner et al. (1990).  Wet Hypermaritime C o a s t a l Western Hemlock(CWHwh)  subzone  Beach associations SOIL N U T R I E N T P  M  REGIME R  VR  S s - î 3alal  SD LU  1  (3 LU ce LU  21(3.7') n=6  Ss-Reedgrass 29(5.6) n=2  F  ce Z)  o  Ss-Swordfern  M  o  2 6 ( 0 ) n=1  CO  <  ^  VM  O  < W  Fig. 7. Average Sitka spruce site index, standard deviations i n parentheses, and number of plots for beach side, ocean-spray affected site associations. Site associations from Banner et al. (1990).  40  -  X CD  30 CD  90 L  10 10  4  11  5  8  1  2  7  6  9  Site association  Fig. 8. Box plots of Sitka spruce site index for site associations with greater than one plot. See Table 9 for explanation of site association numbers.  4.5 Relationships between Sitka Spruce Site Index and Measures of Ecological Site guality  There was a very weak relationship between spruce site index and individual continuous soil variables. Mineral soil TOTN, MINN, forest floor Mg, and K (logged to homogenize the variances) showed a weak relationship with site index (adjusted  = 0.25, 0.10, 0.27, and 0.23, respectively). TOTC, C/N,  pH, C a , F F P and S showed no relationship with site index. M S P could not be transformed to make the variances homogeneous.  Soil physical variables  (coarse fragment content, rooting depth and mineral soil porosity) also showed no relationship with site index.  Multiple linear regression equations were more successful than single variable models i n explaining variation i n site index (Table 10). Soil nutrients were able to explain 35-45% of the variance i n site index, either as continuous or categorical variables or principal component analysis scores. The nutrient analytical models (models 3 and 4) and the categorical model (model 2) are roughly equivalent. Various permutations of principal component analysis scores of nutrient data (covariance versus correlation; data i n concentration versus kg/ha) were essentially equivalent i n explaining i n the range of site index.  The variables that were successful i n discriminating the SNR classes (LOGTOTN, LOGMINN and FFCN) could not be used together i n a multiple regression model since the nitrogen terms are highly correlated (Appendix D). Adding F F C N to either LOGTOTN or LOGMINN did not increase the amount of variation explained by either nutrient used alone.  Table 10. Models for regression of Sitka spruce site index on selected continuous and categorical variables, n = 55 1) SI = 29.352 - 8.352(PA1)1 + 2.648(PA2) - 6.352(PA3) + 3.648(PA4) + 1.523(PA5) - 6.352(PA6) - 2.532(PA7) + 7.148(PA8) + 2.648(PA9) adjusted  = 0.49 S E E = 4.6 m.  2) SI = 33.636 - 14.136(P)2 - 4.747(M) + 1.182(R) adjusted R^ = 0.40 S E E = 5.0 m. 3) SI = -20.703 + 19.3(LOGTOTN)3 - O.OKLOGK)* adjusted R^ = 0.37 S E E = 5.1 m 4) SI = 29.567 + 0.0I2(FFMINN)5 + 0.002(FFCA)6 - 18.923(LQGFFK)'' + 3.793(LOGMSMINN)8 adjusted R^ =.42 S E E = 4.9 m 5) SI = 27.884 - 2.772(SD)9 + 5.034(F) + 8.341(M) + 5.256(VM) adjusted R^ = 0.45 SEE= 4.8 m. 6) Ln(SI)10 = 3.351 . 0.179(SD)11 + 0.048(F) + 0.174(M) + 0.039(VM) - 0.294(P) - 0.025(M) + 0.157(R) adjusted R^ = 0.80 S E E = 0.1 m. l212 = 0.79 Empirical S E E 13 = 7 . 8 m . 7) Ln(SI) = 3.296 + 0.093(SA1)14 + 2.242(SA2) - 0.205(SA3) - 1.105(SA4) + 0.080(SA5) + 0.300(SA6) + 0.266(SA7) + 0.077(SA8) + 0.340(SA9) - 0.588{SA10) - 0.251(SA11) -0.038(SA12) adjusted R^ = 0.80 S E E = 0.1 m. I^ = 0.79 Empirical S E E = 6.0m. ^ D u m m y v a r i a b l e s r e p r e s e n t i n g p l a n t a s s o c i a t i o n s f r o m T a b l e 2. ^ D u m m y v a r i a b l e s r e p r e s e n t i n g s o i l n v i t r i e n t regimes. c o m b i n e d forest floor a n d m i n e r a l s o i l t o t a l n i t r o g e n i n k g / h a (loglO) ^ c o m b i n e d forest floor a n d m i n e r a l s o i l p o t a s s i u m i n k g / h a (loglO) ^ forest floor m i n e r a l i z a b l e n i t r o g e n i n p p m ^ forest floor c a l c i u m i n p p m ^ forest floor p o t a s s i u m i n p p m (log 10) ^ i n i n e r a l s o i l m i n e r a l i z a b l e n i t r o g e n i n p p m (loglO) ^ D u m m y v a r i a b l e s r e p r e s e n t i n g s o i l m o i s t u r e regimes, ^ " ^ n a t u r a l l o g o f site i n d e x ^ ^ D u m m y v a r i a b l e s a s i n 1 a n d 10 a b o v e . E m p i r i c a l R^, a n e s t i m a t e d  based o n antilog values.  ^•^ E s t i m a t e d s t a n d a r d e r r o r of tlie e s t i m a t e , b a s e d o n a n t i l o g v a l u e s 1^ D u m m y v a r i a b l e s r e p r e s e n t i n g site a s s o c i a t i o n s . S y m b o l s e x p l a i n e d i n T a b l e 9.  Since nutrients are often interdependent which could potentially violate the assumptions of multiple linear regression analysis, the two chemical nutrient equations (models 3 and 4) were tested for multicoUinearity. Five plots from the data set were randomly removed and the coefficients examined for any large differences (Table 11). If large differences were found, the equations were unstable, a sign of multicoUinearity (Chatterjee and Price 1977). The equations developed were relatively stable when five plots were randomly deleted and, therefore, were not multicoUinear (Table 11). Table 11. Coefficients and R-^ for best multiple linear regressions of continuous nutrient variables for all plots and 5 plots removed. A l l Plots  5 Plots Removed  Variable  Coefficient  R2  Coefficient  R2  FF+MS (kgha) (model 3) CONSTANT LOGTOTN LOGTOTK  -20.703 19.300 -10.010  .37  -19.367 18.363 -9.026  .35  Concentration (model 4) CONSTANT FFMINN FFCA LOGFFK LOGMSMINN  66.166 0.012 0.002 -18.923 3.793  .42  66.271 0.010 0.002 -19.091 4.148  .44  S M R was the only successful moisture variable, with an adjusted R"^ of 0.45 (model 5). Models using depth to gleying and depth to water table and water deficit predicted by the water balance models (Thornthwaite et al. 1957; Spittlehouse and Black 1981) showed no relationship with site index.  A further examination of the Energy/Soil Limited Water Balance Model (Spittlehouse and Black 1981) results showed that the major discrepancy  between slightly dry sites predicted by the model versus those of the S M R key occurred consistently i n two conditions. First, the model did not predict a moisture deficit for the beach sand dune sites which are very coarsely textured and rapidly drained but have a large rooting depth.  Second, the model predicted a deficit for some fresh sites i n the interior of the Islands, near Moresby Camp and south of J u s k a t l a (Fig. 1). Climate data from the nearest stations, Sandspit and Masset, are unlikely to be applicable to these sites since the inlets receive more precipitation than the coast. Sewell Inlet is at the head of an inlet abutted by the San Cristoval Ranges at the border between the CWHwh and CWHvh subzones and has a climate record for only 7 years (Environment Canada 1980). Therefore the data from the Sewell Inlet station could not be used as a substitute. It is unlikely that the model developed for south coast Douglas-fir, is applicable in the Queen Charlottes, a point that is further elaborated upon i n Chapter 5.  For models that expressed a combination of moisture and nutrients, both the SMR+SNR model (model 6) and the site association model (model 7) were equally successful i n explaining the most variation i n the logarithm of site index (adjusted R 2 = 0.80) with similar M S E (0.466 and 0.432 respectively). However, the variances of the site association model were not homogeneous even when logged as compared to the SMR+SNR model (Fig. 9 and 10). The categorical regression of plant associations was the most successful vegetation model (adjusted R'^=0.49). Models using vegetation PCA scores only explained 3 0 % of site index variation. There was no relationship between site index and frequency of indicator species groups for soil moisture or soil nitrogen.  3,0  3,5  4,0  ACTUAL LN (SITE INDEX)  20  25  30  35  40  ESTIMATED LN (SITE INDEX)  Fig. 9. Predicted versus actual values of ln(site index) and residual versus estimated values of ln(site index) for model 6 (soil moisture plus soil nutrient regime). Adjusted = 0.80  30  35  40  ACTUAL LN (SITE INDEX)  -0.2  2.5  3,0  3.5  4.0  ESTIMATED LN (SITE INDEX)  Fig. 10. Predicted versus actual values of ln(site index) and residual versus estimated \jalue of ln(site index) for model 7 (site association). Adjusted R-^ = 0.80  The relationship between SNR, S M R and site index can also be expressed graphically with clines of site index on an edatope (a grid of S M R and SNR) seasu Krajina (1969) (Fig. 11). Model 6 (SMR and SNR) was used with least squared smoothing to construct the clines. The grid reveals that the optimum growth for Sitka spruce is in the region of moist and very rich sites, which concurs with K l i n k a et al (1989). This differs with Krajina (1969) which considered that very moist and very rich were the optimum conditions for spruce.  Fig. 11. Edatopic grid showing Sitka spruce site index i n relation to soil nutrient and soil moisture regimes. P=poor, M=medium, R=rich, VR=veiy rich; SD=slightly dry, F=fresh, M=moist, VM=very moist, W=wet.  CHAPTER 5. DISCUSSION  The objective of this study was to test the ability of biogeoclimatic ecosystem classification units to represent ecological site quality and be able to predict growth of Sitka spruce, measured as site index. For B E C variables to be good predictors, they must explain a significant amount of variation i n site index and be equal or superior i n that ability with respect to the analogous continuous measures. Therefore relationships between categorical and continuous variables were also investigated.  The B E C classification units were successful at predicting site index of Sitka spruce. The best models developed i n this study used either soil moisture plus nutrient regime or site association. Both had an adjusted R-^ = 0.80 (I-^ = 0.79), and similar M S E s although the variance of the site association model was not homogeneous; therefore it should not be used for predictive purposes. This result concurs  Avith  previous studies that have used units of  the B E C system to predict site index for second-growth Douglas-fir and found them to be successful (Green et al 1989; K l i n k a and Carter 1990).  Carter and K l i n k a (1990) who found a good relationship with Douglas-fir site index and water deficit as predicted by the Energy Limited Water Balance Model (Spittlehouse and Black 1981). This study showed no relationship between site index and water deficit as predicted by the model. A similar study investigating ecological site quality and site index relationships for western hemlock on the west coast of Vancouver Island also had the same result (Kayahara 1992).  There are two sources of problems with the water balance model. First is the input data for weather which consists of monthly time steps of 30 year climate normals. Second are the coefficients used to calculate model parameters and the assumptions made about when évapotranspiration becomes soil moisture limited. Spittlehouse (1981) concluded that it was "obvious" that the model parameters needed to be calibrated for different species and ecosystems before the model could be applied outside the range of south coast Douglas-fir ecosystems where it was developed. The study of Carter and Klinka (1990) occurred within the range for w h i c h the model was developed.  The model is especially sensitive to precipitation which is essentially its driving variable (D. Spittlehouse pers. comm. ^) The climate stations i n the Charlottes are i n the driest sunniest part of the Islands and are not representative of the range of rainfall patterns that occur. Frequently the east coast (Sandspit and Tlell) will be clear and sunny while it is raining or foggy i n Skidegate Inlet. The fresh sites predicted by the model as slightly dry were located away from the east coast. For stands not along the east coast, the forest is likely experiencing more rainfall than the weather data would predict. In the Vancouver Island study (Kayahara 1992), for sites that were considered to be dry or slightly dry by morphological characteristics, the model predicted no deficit which Kayahara attributed to very high precipitation i n the climate data. Finally, daily rather than monthly time steps may be more appropriate; however, an investigation of this was beyond the scope of this study.  David Spittlehouse, Research Scientist, B.C. Ministry of Forests Research Branch.  Several coefficients are used i n the model (Spittlehouse and Black 1981; Carter and K l i n k a 1990). Soil depth and coefficients for texture are used to calculated available water storage capacity. Evaporation is calculated using the Priestley-Taylor coefficient which is based on tree physiology. The model assumes that évapotranspiration is energy limited, until soil water reaches 60%, then it becomes water limited.  It is unknown whether the solar coefficient chosen sufficiently accounted for fog, a common phenomenon i n the Charlottes and likely a n important component of water balance. It is also unknown whether a n evaporation coefficient for Douglas-fir can be applied to Sitka spruce and whether the assumption of évapotranspiration becoming water limited at 6 0 % of soil water capacity is valid.  The coefficients used for soil are probably accurate since they were measured empirically i n the field and are based on soil texture which is independent of other factors. Water characteristics as a function of texture class are probably relatively consistent (A. Black pers. comm.  The model did  not predict a deficit for the very coarse textured rapidly drained soils on the beach side sand dunes. These forests occur i n an exceptional edaphic situation and may be difficult to incorporate i n a model. It is possible that aside from such edaphic situations, (i.e., forests growing on beach dunes and bedrock), trees do not experience a moisture deficit on the coast.  Andy Black. Professor, Dept. of Soil Science, U.B.C.  The Thornthwaite model (Thornthwaite et al. 1957) is also inadequate for describing water balance. It derives potential évapotranspiration from temperature, latitude and precipitation and its accuracy is a function of the climate data and therefore may be suspect. The most serious problem occurs with the calculation of actual évapotranspiration which is based on coefficients for combinations of rooting depth, soil texture and vegetative cover. There are no appropriate coefficients for second-growth forests. The only coefficients given for forests assume a minimum rooting depth of 1.5 m. Rooting depth i n this study ranged from 0.25 to 1.00 m. The only coefficients with appropriate rooting depths given were for agricultural crops which have different moisture behaviour from closed canopy second-growth forests. The higher transpiration and evaporative loss associated with food crops i n bare soils i n comparison to forests likely accounts for all the sites being classified as moderately dry, which is not realistic.  A n independent field based measure of soil moisture is required to assess the performance of its categorical equivalent, soil moisture regime, i n predicting site index, to give an independent evaluation of the relative soil moisture key and actual conversion table (Banner et al 1990) and provide more accurate, site specific climate data for the Charlottes. The EnergyLimited Water Balance model does predict deficits for some sites i n the Charlottes. However, they show no relationship with site index. Inspection of the results indicates that the soil moisture regimes are more realistic i n terms of their assessment of moisture status, but that conclusion is not supported by any data. Therefore, the possibility still exists that the model is correct and the key is wrong with respect to moisture deficit. It is beyond the ability of this author to recommend specific techniques, but instruments do exist that can measure both the soil moisture content, its change over the growing season  and rainfall on the site (A. Black pers. comm.). With the poor climate data, even if the coefficients were properly Ccilibrated, the model may still be inadequate.  K l i n k a and Carter (1990) concluded that categorical variables were slightly better than their analytical counterparts i n explaining variation i n Douglas-fir site index. It is difficult to say whether this study supports this finding since there was no valid analytical variables for soil moisture, therefore no analytical model of moisture and nutrients could be developed. Analytical models may be valuable for describing Sitka spruce site index and the poor results obtained i n this study may be due to a poor representation of soil water balance.  The S N R model was not clearly superior to the analytical multiple regression models using soil nutrients. However, the former does have the advantage of being more easily measured i n the field. Also, given the variability i n nutrient equations reported i n previous studies (e.g. Blyth and MacLeod 1981b) the SNR model may be more reliable and applicable over a wider range of conditions.  In terms of nutrient studies, previous work has found nitrogen to be important (Blyth and MacLeod 1981b; Green 1989). This study concurs with that finding, which is not surprising since nitrogen is considered the most limiting factor for any conifer i n the Pacific Northwest (Radwan et ai 1986). The importance of C/N ratio and F F p H is no doubt related to their effect on nitrogen availability. Considering the autecology of spruce, Mg, Ca, and P are also potentially significant (Krajina et al. 1982). Blyth and Macleod (1981b) found "humic" P to be important. This study found no relationship with F F P  and no conclusions could be made about M S P since the variances were not homogeneous.  C a was shown to be related to site index, and K rather than Mg.  Due to the correlation of the cations (Appendix D), regression models could not be constructed with both C a and M g and the models with C a and K performed the best. With the interrelationships between the cations, Mg is likely indirectly represented. With respect the studies i n Great Britain,(e.g.Blyth and Macleod 1981a; 1981b) caution should be applied i n extrapolating results to North America. Sitka spruce i n Britain is an exotic, planted and managed under an intensive silvicultural regime and therefore may be established on sites where it might not occur i n its native conditions.  Previous studies on productivity of Sitka spruce, as reviewed i n Chapter 1, have had inconclusive results as to the most important factors for spruce growth. Some studies found a weak relationship to soil physical and taxonomic variables (Stephens et al 1968; Worrell 1987; Ford et al 1988), but such models have performed poorly. This study found no relationships with soil physical variables.  The SNR classification was well supported by the discriminant analysis (DA), and to a lesser extent the cluster analysis. The DA indicated that of all the nutrient variables, only total N, mineralizable N and forest floor C/N ratio are important i n discriminating the classes. These variables were found previously to be important i n delineating soil nutrient regimes (Courtin et al 1988). However, this study found fewer variables useful i n the discrimination than the latter study. There were no independent data to assess the accuracy of the soil moisture key. As stated previously, this is essential. Until such a calibration is done, the key should be applied with caution.  This study was conducted i n second-growth forests i n order to limit the error i n predicting site index (metres @ 50 years breast height age) However, for forest management purposes results need to be applicable to other successional stages. Green et al. (1989) felt that their model for immature Douglas-fir using site variables was applicable to other successional stages, since site units are identified by environmental variables which remain relatively constant over time. These models are therefore stronger than the ones based solely on current vegetation on the site which change with successional processes.  There were no relationships between indicator species groups and site index. This result contrasts with Green et al. (1989) which found an  of 0.87  with frequency of indicator species groups and second-growth Douglas-fir site index. Forests at or near canopy closure have poorly developed understorey (and indicator) flora (Klinka et al. 1989) and the lowest species diversity (Harris 1984). In the Queen Charlottes, this lack of understory is further exacerbated by the intense browsing pressure of introduced deer. In addition, immature Douglas-fir stands may remain more open so there is more light for understory species. Green (1989) found a poor functional relationship with indicator species groups and site index of western redcedar i n the Charlottes which he attributed to a closed canopy and the "non-site" variation of deer browsing. The predominant influence on understorey vegetation i n the Charlottes is most likely not site characteristics. Rather, it is intense deer browsing. Therefore any relationships with vegetation will be marginal at best, such as the degree of separation of plant alliances and associations.  The PCA shows that the plant associations are weakly distinguished. Vegetation occurs on a continuous gradient, therefore some overlap could be  expected. While it is admittedly arbitrary to place more weight on tabular analysis, given the paucity of the understorey, it is justified. Franklin et al. (1982) also found P C A a poor tool for delineating plant communities i n comparison to tabular analysis.  Use of site index as a measure of growth potential of the moisture and nutrient conditions of a site assumes that height growth only reflects those site conditions and other factors such as mechanical forces are a constant. In the case of the beach ecosystems, this assumption may not be true i n that height growth may actually be more influenced by mechanical forces i.e. wind. Trees may allocate more energy to diameter growth and windfirmness, than to height growth i n comparison to non-beach sites. This is further confounded by loss of height growth from broken tops, which is often difficult to detect when determining site index i n a dense second-growth stands. Barker reported encountering broken tops i n his stem analysis study (J. Barker pers. comm.*^). Blyth and Macleod (1981a) also found this to be the case. Since the site association classification recognizes beach ecosystems as distinct, it is likely able to account for this phenomenon and variation i n mechanical forces.  J o h n Barker. Research Scientist, Western Forest Products Ltd.  CHAPTER 6. CONCLUSIONS  1) The classification units of the biogoeclimatic ecosystem classification system are good predictors of site index of Sitka spruce and therefore good indirect measures of ecological site quality. Multiple linear regression models using either soil moisture plus soil nutrient regime or site association explained the most variation i n site index of the models developed i n this study (adjusted  = 0.80 for ln(site index; 1^ = 0.79). The former model is the best  for predictive purposes since the variances of the latter were not homogeneous.  2) The soil nutrient regime classification, based on qualitative field descriptors, was well supported by the analjrtical soil nutrient measures. A discriminant analysis using combined forest floor and mineral soil log total N, log mineralizable N and forest floor C:N ratio was able to classify the plots correctly in, 8 3 % of the cases, on average. There was no suitable soil moisture analytical values to test the soil moisture regime classification.  3) Forest soil moisture relationships still need to be determined for S i t k a spruce i n the Charlottes. Neither water balance model used i n this study (Thornthwaite et ai 1957; Spittlehouse and Black 1981) were adequate. It is stiU unclear to what degree spruce growth is influenced by moisture deficit and the soil moisture regime classification needs to be assessed with independent data. Due to the inadequate representation of climate i n the Charlottes from the location of weather stations, field measurements i n the forest stand are likely necessary.  LITERATURE CITED Alaback, P. B . 1982. Dynamics of understorey biomass i n Sitka sprucewestern hemlock forests of southeast Alaska. Ecology 63:1932-1948. Anonymous. 1976. Techicon Autoanalyzer. II. Methodology: individual/simultaneous determination of nitrogen and/or phosphorus i n B D acid digest. Industrial Method. No. 329/74W/A, Technicon Corp., Tanytown, NY. Banner, A., R.N. Green, K. Klinka, D.S. McLennan, D.V. Meidinger, F.C. Nuszdorfer and J . Pojar. 1990. Site classification for coastal British Columbia: a first approximation. B.C. M i n . For., Victoria 2pp. (a coloured pamphlet). Bcinner, A., J . Pojar, J . W . Schwab and R. Trowbridge. 1989. Vegetation a n d soils of the Queen Charlotte Islands: recent impacts on development pp. 261-280 In: G.G.E. Scudder and N. Gessler (eds.) The Outer Shores. Queen Charlotte Islands M u s e u m Press, 1989. Banner, A., J . Pojar, and R. Trowbridge. 1983. Ecosystem classification of the Coastal Western Hemlock Zone, Queen Charlotte Island Subzone (CWHg), Prince Rupert Forest Region, British Columbia. B.C. M i n . For. Res. Branch, Smithers B . C . Draft Report. 255pp. Barker, J . E . and J . W. Goudie. 1987. Site index curves for Sitka spruce. B.C. For. Serv. Br. Unpubl. Rep., Vancouver, B.C. Bakuzis, E.V. 1969. Forestry viewed i n a ecosytem perspective, pp. 189-258. In: G . M . V a n Dyne, (ed.) The Ecosystem Concept i n Natural Resource Management. Academic Press, New York. Blyth, J . F . and D.A. McLeod. 1981a. Sitka spruce (Picea sitchensis) i n northeastern Scotland. I: Relationships between site factors and growth. Forestry 54:41-62. Blyth, J . F . and D.A. McLeod. 1981b. Sitka spruce (Picea sitchensis) i n northeastern Scotland. II: Yield prediction by regression analysis. Forestry 54:63-73. Bremner, J . and M.A. Tabatabai. 1971. Use of automated combustion techniques for total carbon, total nitrogen, and total sulfur analysis of soils, pp. 1-16 i n L . M . Walsh (ed.). Instrumental Methods for Analysis of Soils and Plant Tissue. Soil Sci. Soc. Am., Mad., Wis.  Canadian Soil Survey Committee. 1987. The Canadian system of soil classification. C a n . Dept. Agric. Publ. 1646. 164pp. Cajander, A.K. 1926. The theory of forest types. Acta. For. Fenn., 2(3) 11-108 Carmean, W.H. 1975. Forest site quality evaluation i n the United States. Adv. Agron. 27: 209-269. Carl, G.C. and C. J . Guiget. 1972. Alien Animals i n British Columbia. Brit. Col. Prov. M u s . Handb. 14. Victoria, B. C. 103pp. Carter, R.E. and K. Klinka. 1991. Use of ecological site classification i n the prediction of forest productivity and response to fertilization, unpubl. rept. 11pp. Carter, R.E. and K. Klinka. 1990. Relationships between growing-season soil water-deficit, mineralizable soil nitrogen and site index of coastal Douglas-fir. For. Ecol. Manage. 30:301-311 Carter, R . E . and K. Klinka. 1988. Douglas-fir fertilization decision-making for industrial use: an establishment report. Faculty of Forestry, Univ. of B.C., Vancouver, B.C. 33pp. Chatterjee, S. and B. Price. 1977. Regression analysis by example. Interscience, New York. 228 pp.  Wiley-  Corns, l.G.W. and D . J . Pluth. 1984. Vegetational indicators as independent variables i n forest growth prediction i n west-central Alberta, Ccinada. For. Ecol. M a n . 9:13-25. Cordes, L.D. 1972. A n ecological study of the Sitka spruce forest on the west coast of Vancouver Island. Phd. thesis, Univ. of British Columbia. 268pp.D Courtin, P . J . , K. Klinka, M.C. Feller, and J . P . Demaerschalk. 1988. A n approach to quantitative classification of nutrient regimes of forest soils. C a n . J . Bot. 66:2640-2653. C r u m , H.A., W.C. Steere and L.E. Anderson. 1973. A new list of mosses of North America north of Mexico. The Biyologist 76:85-130. Daubermire, R.F. 1976. The use of vegetation i n assessing the productivity of forest lands. Bot. Rev. 42:115-143.  Day, W.R. 1957. Sitka spruce i n British Columbia - a study i n forest relationships. Forestry Ccommission bulletin No. 28, Imperial Forestry Institute, Oxford. Dillon, W.R. and M . Goldstein. 1984. Multivariate Analysis: Methods and Applications. J o h n and Wiley and Sons. New York. 587pp. Dumanski, J . (ed). 1978. M a n u a l for describing soils i n the field; The Canadian Soil Information System (CanSIS). Land Resource Institute, Agric. Can. Ottawa, Ont. 92pp. Emanuel, J . 1985. A vegetation classification program. Unpublished manuscript. Faculty of Forestry, Univ. of B.C. 19pp. Environment Canada 1980. Canadian Climate Normals, 1951-1980, British Columbia. Atmospheric Environment Service. Downsview, Ont. 268pp. Farr, W. A. and A.S. Harris. 1979. Site index of Sitka spruce along the Pacific coast related to latitude and temperature. For. Sci. 25:145-153. Farr, W. A. and A.S. Harris. 1983. Site index research i n the western hemlock-Sitka spruce forest type i n coastal Alaska. In: Ballard R. and S.P.Gessel, (eds). I.U.F.R.O. S5miposlum on forest site and continuous productivity. Proceedings of a symposium, Aug. 22-28 1982; Seattle WA. Gen. Tech. Rep. PNW-163, U.S.D.A. Pac. Nor. West. Res. and Exp. Station. Ford, E.D., W.A. Farr and C. Lu-Ping. 1988. Preliminary analysis of four soil variables and their relation to site index of S i t k a spruce i n Southeast Alaska, pp. 84-89 In: Slaughter, C. W. and T. Gasbarro (eds.) Proceedings of the Alaska forest soil productivity workshop; April 28-30; Anchorage AK. Gen. Tech. Rep. PNW-219. U.S.D.A. Pac. Nor. West. Res. and Exp. Station. Franklin, J . F . , W.H. Moir, M.A. Hemstrom, S.E. Greene and B . G . Smith 1988. The forest communities of Mount Rainier National Park. Scientific Monograph. Ser. No. 19. USDI Nat. Park Serv. Washington D.C. Gauch, H.G. 1977. O R D I F L E X - A flexible computer program for four ordination techniques: weighted averages, polar ordination, principal components analysis, and reciprocal averaging. Release B. Ecology and Systematics, Cornell University, Ithaca, New York. 185pp. Gittins, R. 1979. Ecological applications of canonical analysis, pp. 309-353 In: L. Orloci, C.R. Rao and W. M . Stiteler (eds.) Multivariate Methods i n Ecological Work. Burton Md: International Co-operative.  Gordon, A. D. 1981. Classification. Methods for the exploratory analysis of multivariate data. Chapman and Hall, New York. Green, R.N. 1989. Site - forest productivity relationships and their management implications i n coast lowland ecosystems of east G r a h a m Island, Queen Charlotte Islands. MSc. thesis, Univ. of British Columbia. 169pp. Green, R.N., P.L. Marshall and K. Klinka. 1989. Estimating site index of Douglas-fir (Pseudotsuga menziesii [Mirb.] Franco) from ecological variables i n southwestern British Columbia. For. Sci. 35:50-63. Greweling, T. and M . Peech. 1960. Chemical soil tests. Cornell. Agric. E x p . Sta. B u l l . 960. Hagglund, B. 1981. Evaluation of forest site productivity. Comm. For. Bureau, For. Abstracts 42(11): 515-527. Harris, L.D. 1984. The Fragmented Forest. Univ. of Chicago Presss. 211pp. Hills, G.A. 1952. The classification of evaluation of site for forestry. 24, Ontario Dept. of Lands and Forests, Toronto, Ont. 41pp.  Res. Rep.  Holland, S.S. 1976. Landforms of British Columbia. Bulletin 48, B.C. Dept. of Mines and Petroleum Resources, Victoria, B.C. 138pp. Kayahara, G . J . 1992. A n investigation of relationships between western hemlock ecosystems, ecological site quality and productivity i n the Coastal Western Hemlock zone of B.C. M S c . thesis, Univ. of British Columbia. 127pp. Klinka, K., and R.E. Carter. 1990. Relationships between site index and synoptic environmental factors i n immature coastal Douglas-fir stands. For. Sci. Vol. 36(3):815-830 Klinka, K., V . J . Krajina, A. Ceska and A . M . Scagel. 1989. Indicator Plants of Coastal British Columbia. Univ. of British Columbia Press, Vancouver B.C. 287pp. Klinka, K., R.N. Green, R.L. Trowbridge and L.E. Lowe. 1981. Taxonomic Classification of H u m u s Forms i n Ecosystems of British Columbia. First Approximation. Land Manage. Rep. 8. B . C . Ministry of Forests, Victoria B.C. 54pp.  Kowalenko, G.D. and L.E. Lowe. 1972. Observations on the b i s m u t h sulphide colourmetric precedure for sulphate analysis i n soil. Comm. i n Soil. Sci. Plant A n a l . 3:79-86. Krajina, V . J . , K. Klinka, and J . Worrall. 1982. Ecological characteristics of trees and shrubs of British Columbia. Fac. of For., Univ. of British Columbia, Vancouver, B.C. 131 pp. Krajina, V . J . a n d R.C. Brooke 1969 . Ecology of forest trees i n British Columbia. Ecol. West. N. A m . 2(19) : 1-146. Lavkulich, L . M . 1978. Methods manual. Pedology Laboratory, Soil Sci. Dept., Univ. British Columbia, Vancouver, B.C. 224 pp. Lewis, T. 1982. The Ecosystems of Tree F a r m License 24, Queen Charlotte Islands, B.C.. A contract report to Western Forest Products Ltd., Vancouver, B.C. 191pp. Livingston, B . C . and F. Shreve. 1921. The distribution of vegetation i n the United Stats, as related to climatic conditions. Carnegie Inst. Wash. Publ. 285:1-585. Major, J . 1963. A climatic index to vascular plant activity. Ecol. 44: 485498. Mason, H.L. and J . H . Langeheim. 1957. Language analysis and the concept of environment. Ecol. 38:325-339. McLean, C D . and C.L. Bolsinger. 1973. Estimating Dunning's site index from plant indicators. U S D A Forest Serv. Res. Note PNW-152, Pac. Nor. West. Range and Exp. Station, Portland OR Mehlich, A. 1978. New extractant for soil test evaluation of phosphorus, magnesium, calcium, sodium, manganese and zinc. Comm. Soil Sci. Plant A n a l . 9: 477-4922 Meidinger, D.V. and J . Pojar. 1988. Biogeoclimatic and ecoregion units of the Prince Rupert Forest Region. Map published by the B.C. Ministry of Forests and Lands Research Branch, Victoria B.C. Monserud, R.A., U . Moody, and D.W. Breuer. 1990. A soil-site study for inland Douglas-fir. C a n . J . For. Res. 20:686-695.  Monserud, R.A. 1984. Problems with site index: A n opinionated review, pp. 167-180 In: J . G. Bockheim (ed.) Forest land classification: experience, problems and perspectives. Proc. North Central Forest Soils Committee symposium. NCR-102, Univ. of Wisconsin, U.S.D.A. For. Serv. North Central For. Exp. Stn. Peet, R.K. 1980. Ordination as a tool for analyzing complex data set. Vegetatio 42:171-4. Pojar, J . , K. Klinka, and D.V. Meidinger. 1987. Biogeoclimatic ecosystem classification i n British Columbia. For. Ecol. Manage. 22:119-154. Pojar, J . and A. Banner. 1984. Old-growth forests and introduced deer on the Queen Charlotte Islands, British Columbia, pp. 247-257. In: W.R. Meehan, T.R. Merrel J r . and T.A. Hanley (tech. eds.) F i s h and Wildlife Relationships i n Old-growth Forests: Proc. Symp. (Juneau, Alaska, 12-15 April 1982). Amer. Inst. Fish. Res. Biol. Powers, R.F. 1980. Mineralizable soil nitrogen as a n index of nitrogen availability to forest trees. Soil Sci. Soc. A m . J . 44: 1314-1320. Price, W . J . 1978. Anal5^c atomic absorption spectrophotometry. and Son Ltd., London. 239 pp.  Heydon  Radwan, M.A. and Brix. 1986. Nutrition of Douglas-fir. In: C D . Oliver, D.P. Hanley and J . D . Johnson (eds.) Douglas-fir: Stand management for the future. College of Forest Resources, Univ. Washington, Contrib. 55, 388pp. Rogers, G.L. 1988. Estimations of site index i n forest inventories of coastal Alaska—and other confessions, pp. 77-80 In: Slaughter, C. W. and T. Gasbarro (eds.) Proceedings of the A l a s k a forest soil productivity workshop; April 28-30; Anchorage AK. Gen. Tech. Rep. PNW-219. U.S.D.A. Pac. Nor. West. Res. and Exp. Station. Rowe,J.S. 1956. Uses of undergrowth plant species i n forestry. Ecol. 37(3): 462-473 S A S Institute Inc. 1982. SAS User's Guide: Statistics, 1982 edition. Caiy, NC. SAS Institute Inc. Schonau, A.P.G. 1988. Problems i n using vegetation or soil classification i n determining forest site quality, pp 3-11 In: Cole, D.W. and S.P. Gessel. Forest Site Evaluation and Long-term Productivity. Univ. of Washington Press. 196pp.  Spies, T.A. and B.C. Barnes. 1985. A multifactor ecological classification of the northern hardwood and conifer ecosystems at Sylvania Recreation Area, Upper Pennisula, Michigan. C a n J . For. Res. 15:949-960. Spittlehouse, D.L. 1981. Measuring and modelling évapotranspiration from Douglas-fir stands. Phd. thesis. Univ. of British Columbia. Spittlehouse, D.L. and T.A. Black. 1981. A growing season water balance model applied to two Douglas-fir stands. Water Resour. Res. 17: 16511656. Spurr, S. H . 1952. Forest Inventory. Ronald Press, Co., New York. 476pp. Spurr, S. H., and B. V . Barnes 1980. Forest ecology (3rd edition). J o h n Wiley & Sons, New York. Stephens, F.R., C. R. Gass, and R. F. Billings. 1968. Soils a n site index i n southeast Alaska. Report No. 2 of the Soil-site Index Administrative Study, A l a s k a region. U.S.D.A, Forest Service 17pp. Taylor, R.L. and B . MacBiyde. 1977. Vascular plants of British Columbia. Tech. B u l l . No. 4, Univ. of B.C. Press, Vancouver, B . C . 659pp. Thornthwaite, C.W., J . R . Mather and D.B. Carter. 1957. Instructions and tables for computing potential évapotranspiration and the water balance. Publ. Climatol. Vol. 10, No. 3, Drexel Inst. Technol.,Lab. Climatol., Centerton, New Jersey Stone, E.L. 1978. A critique of soil moisture - site productivity relationships. pp.41-52 In: W.E. Balmer. (Ed.) Soil moisture - site productivity symposium, proceedings, 1-3 Nov. 1977. M5atle Beach, S C . U S D A Forest Service, Atlanta G A Thrower, J . S . 1989. Site quality evaluation using site index. A presentation to the Silvicultural Institute of B.C. Module III training course at Surrey B.C., March 8, 1989. 11pp. Trewartha, G.T. 1968. A n Introduction to Climate. 4th edition. McGraw-Hill, New York. Walmsley, M . , G. Utzig, T. Void, D. Moon, and J . van Barnveld (eds.). 1980. Describing ecosystems i n the field. B . C . M i n . Env., R A B Technical Paper 2, B.C. M i n . For., Land Management Rep. No. 7, Victoria, B.C.  Ward, J . H . 1963. Hierarchical groupings to optimize an objective function. J . Amer. Stat. Assoc. 58:236-244. Waring, R. H. and W. H . Schlesinger. 1985. Forest Ecosystems Concepts and Management. Academic Press, London. 340pp. Waring, R. H. and J . F. Franklin. 1979. Evergreen coniferous forests of the Pacific Northwest. Science 204(4400): 1380-1386. Waring, R.H. and J . Major. 1964. Some vegetation of the California coastal redwood region i n relation to gradients of moisture, nutrients, light and temperature. Ecol. Monog. 34(2) pp. 167-215 Williams, G. 1968. Climate of the Queen Charlotte Islands, pp. 15-54. In Calder. J . and R. Taylor. 1968. Flora of the Queen Charlotte Islands, Part 1. Systematics of the vascular plants. Res. Branch. C a n . Dept. of Agric. Monograph No. 4. Part 1. 659pp. Wilkinson, L. 1990. SYSTAT. The system for statistics. S Y S T A T I n c , Evanston. 111. Worrell, R. 1987. Predicting the productivity of S i t k a spruce on upland site i n Northern Britain. For. Comm. B u l l . No. 72. Worrell, R. and D.C. Malcolm 1990a. Productivity of Sitka spruce i n northern Great Britain. 1. The effects of elevation and climate. Forestry 63(2)104117 Worrell, R. and D.C. Malcolm 1990b. Productivity of Sitka spruce i n northern Great Britain. 2. Prediction from site factors. Forestry 63(2)119-128 Zar, J . H . 1984. Biostatistical analysis. 2nd edition. Prentice-Hall. Endglewood Cliffs, N . J .  APPENDIX A. Alphabetical list of plant species Alnus rubra AthyriumJUix-feniina  Bong. (L.) Roth  red alder lady fern  Blechnum  (L.) Roth  deer fern  Calamagrostis nutkaensis Calypso bulbosa Claytonia sibrica Conocephalum conicum Coptis asplenifolia  (Presl) Steud. (L.) Oakes (L.) (L.) Llndb Salisb.  Pacific reedgrass fairy-slipper Siberian miner's lettuce  Dicentraformosa Dryopteris expansa  (Andr.) Walp. (Presl) Fraser-Jenkins  bleeding heart spiny wood fern  Galium aparine Galium, triflorum Gaultheria shallon Goodyera oblongfolia Gymnocarpum dryopteris  (L.) Mlchx. Pursh Raf (L.) Newm.  cleavers sweet-scented bedstraw salal rattlesnake-plantain oak fern  Hylocomium Hypochoeris  (Hedw.)B.S.G. (L.)  hairy cat's ear  Listera caurina Lysichitum americanum  Piper Hult.& St.John  northwestern twayblade s k u n k cabbage  Maianthemum dilatatum Moneses uniflora  (How.) Nels & Macbr. false lily-of-the-valley single delight (L.) Gray  Osmorhiza  Hook. & A m .  mountain sweet cicely  Peltigera aphthosa Plagiomnium insigne Poa marcida Pogonatum alpinum Polypodium glycyrrhiza Polystichum munitum Pteridium aqidinium  (L.) Willd. (Mitt.) Koponen Hitchc. (Hedw.) Roehl (L.) (Kaulf.) Presl (L.) K u h n i n Decken  licorice fern sword fern bracken fern  Rhizomnium glabrescens Rhytidiadelphus loreus Rhytidiadelphus triquestrus Rubus parvtflorus  (Kindb.) Koponen (Hedw.) Wamst. (Hedw.) Wamst. Nutt.  thimblebeny  Stellaria crispa Sphagnum spp.  Cham. & Schlect  crisp starwort  Tiarella  L.  foamflower  Smith Sm. i n Ress Muhl.  oval leaf blueberry red huckleberry American vetch  spicant  splendens radiacata  chilensis  trifoliata  Vaccinium oualifolium Vaccinium parvijlorum Vicia americana  goldthread  lanky moss  Appendix B. Ordination of reciprocal averaging plots scores for second versus first axis showing vegetation groups delineated through progressive fragmentation. Small numbers represent plots. Large numbers represent plant associations from Table 2.  1 Ss-Calamagrostis 2 Ss-Claytonia 3 Ss-Rhytidiadelphus 4 Ss-Gymnocarpixum 5 Ss-Polystichum 6 Ss-Gaultheria 7 Ss-Thiya 8 Ss-Hylœomium 9 Ss-Alnus 10 Ss-Kindbergia  Aii;.Lii.SA OlllllMiTE  IX (r  IS IS  SM.U'ltS S»I.1IMES  SCOIIES SCOIIES  KlUU f IIOI.I  (IECIP»VE IIECll'ivE  o r i n I 111 I 1 O i l OIIDI l U 1 1 O i l  1 1 AXIS *XIS  1 2  l l l l l 2 2 2 2 2 3 ] 3 3 ] 4 4 < 4 < 5 5 5 5 5 6 6 6 6 6 7 7 7 7 7 8 a O B 6 9 0 9 0 9 0 U i ) 2 4 B a 0 2 4 6 a û 2 4 6 a 0 2 < 6 a 0 2 < 6 a 0 2 < B B 0 2 4 6 6 0 2 4 6 B 0 2 4 B 6 0  * x i b ) AXISI  p  o  nr.  -I  rt)  cn  !31  AOSCISSA OIIOINAIE  '  "  " "  IXAXIS) lYAXISl  SCORES F n O U R E C I P * V E SCailES FIIOU R E C I I U V E  OtlDIMUIOII OIIDIIUIIOII  \  1 1  /  \  /  AXIS AXIS  1 2  II  II  I  2 I I I I  I  I  I  I  I  1 I  I I 2 I I I I I I I I  I  I 1  I  I  I  1 1 2 2 2 2 2 3 3 3 3 3 4 4 2 ^ " " 0 2 < 6 a Q 2 4 6 a 0 2 4 5 a Q 2 4 G a 0 2 4 C 8 Û 2 4 6 0 0 2 4 6 B 0 2 4 C 8 n  ISSUIPIES I S SAI.II'IES  13  nr. o  r-t-  Cl  o  (J) a> n  0  6 4 2  D  •  B4 B2 80 78 7B U 72 70 SB 66 64 62 60 5B 5B 54 52 50 48 48 44 42 40 38 36 34 32 30 28 28 24 22 20 18 16 14 12 ID  es 69  96 g 4 92 90  lun 98  AUStl'JSA OflUIMAIE  1  .!  4  \  \  o  I S S I M P I E S I S SAMPLES  SCORES SCORES  FnOU (ROM  1  REClPiVE RECIPAvE  1  OnDllUIIOII O R O I I I * I I Oil  1  1  I  1 I  V  \  1  •  1 2  2  6  1  AXIS A X I S -  t  1  I  2  1  1 1  1 1 1 1I 2 2 2 2 2 3 3 3 3 3 4 4 4 4 4 5 5 5 5 5 5 B B B 6 7 7 7 (i U 0 2 4 6 8 0 2 4 6 8 0 2 4 6 8 0 2 4 6 8 0 2 4 6 8 0 2 4 6 8 0 2 4  /  \  \  (X-AXISl lY AXIS!  1  1  1  I I  1  \  1  1  1  /  1  1  1  1  1  1 7 7 B B B 8 B 9 9 B 9 !) 0 6 8 0 2 4 6 8 0 2 4 6 1 ) 0  n  —-  \  AllSCISS» OIIOINAIE  4(IU  (X-AXISl lY-AXlSI  I  SAMCI ES S A M r i ES  SCORES SCORES  FROM P E C I P A V E FROM H E C I I ' A V E  OnOIH»IIOtl OROItlAlIOM  I I  3  AXIS AXIS  1 2  I  I I  I  I  I  n I I 11  I 1 1 1 II?  I  I 12  I I I 1 2 2 2 2 2 3 3 3 3 . ' < 4 4 4 < 5 5 5 t i 0 6 6 B 6 B 7 7 7 7 7 0 O B B a 9 g 9 9 9 O 0 2 4 6 n O 2 4 6 B 0 2 4 6 d 0 2 4 8 a O 2 4 6 B 0 2 1 6 8 O 2 4 6 ( j n 2 4 6 8 O 2 4 6 8 O  IS IS  o = !1  o  Fift±i iteration  -oc o)  œ  O)  o  a i ru  C7) O m  o  CD  03 O  -  -  03 m  to to Œ3 • »  113 nj  ^  /  s  m 03  \  i n ID  \  i n "«r i n CM  1  in  -  o  •a- IM  -  -  /  -"^ / . _ -\ ^  t\j œ  If  ru C3  oj nj ru  O  — OJ  \  CO  — C3  \  ~  \  — fVJ —  -  O t D U 3 ^ r y o a ) o - « ' r > j O ( 3 O * ' ' ^ O a ) O ' w r u o c 3 O - ' ' ^ ' O C D o - » r a O 0 3 c r - * r y o o i O ' W ' r i ; o œ i O ' v r ^ O C l C o a l C l C l œ o 3 a D a 3 C 3 ^ ^ ' ^ ' ^ ' ^ ' ^ o o ^ û o o ^ n u ^ l n l n l n ' T - » ^ • v • v * * l | - ^ ' ^ ' ~ ' ^ —  —  O  Appendix C 8 0 % confidence ellipses for principal component analysis scores for plant associations. 80% confidence ellipses for plant associations with greater than 5 plots for first and second PCA axes (Y2 verses Y l ) and second and third axes (Y3 verses Y2). Understorey species = all species occurring i n the understorey. Diagnostic species = those species that are diagnostic for the plant association from Table 2. Numbers represent plant associations from Table 2.  1 2 3 4 5 6 7 8 9 10  Ss-Caiamagrostis Ss-Claytonia Ss-Rhytidiadelphus Ss-Gymnocarpium Ss-Polystichum Ss-Gaultheria Ss-Thuja Ss-Hylocomium Ss-Alnus Ss-Kindbergia  80% confidence ellipses for principal component analysis scores for understorey species for second versus first axis and third versus second axes.  80% confidence elHpses for principal component analysis for diagnostic species for second versus first axis and third versus second axes.  Appendix D Pearson correlation matrix for multiple linear regression equations Symbols used for individual forest floor and mineral soil nutrients i n concentration FF MS PH TN MNN P S CA MG K LOGFMG LOGFK LOGMSP LOGMSMN  forest floor mineral soil pH total nitrogen mineralizable nitrogen phosphorus sulphur calcium magnesium potassium log (10) forest floor magnesium log (10) forest floor potassium log (10) forest floor phosphorus log (10) mineral sofl mineralizable nitrogen  Symbols used for combined forest floor and mineral soil nutrients i n kg/ha TOTC TOTN TOTMN TOTP TOTS SUMCAT FFPH MSPH FFCN MSCN LGTOTN LGTOTP LGTOTS LGSUMC  total carbon total nitrogen mineralizable nitrogen forest floor phosphorus forest floor sulphur s u m calcium, magnesium and Dotassium forest floor p H mineral soil p H forest floor carbon to nitrogen ratio mineral sofl carbon to nitrogen ratio log (10) total nitrogen log (10) forest floor phosphorus log (10) forest floor sulphur log (10) sum calcium, magnesium and potassium  FCfi i m m w À L îms: FLOOR ÀÏD K i i m SOIL I V I S I E T S IH coHCEmiTios FFPB  1.000  FTC  mt nm m m îm mfô m Nspa use  HSIffl  xsi:  KSP MSS  LOGfîfô Locn;  LOGXSP LOOQffl ÎÎS  fîa îm îîi KSC  KSVi KSCA KSHG KSI HSP HSS  Locm;  LOGfT LOCXSP LOGXSMI  -0.307 o.in  Û.303 -0.383 0.653 0.0<3 0.00< 0.363 0.003 0.085 0.056 0.331 0.U5 0.0<5 -0.062 O.Hl 0.010 -0.003 -0.039 D.031 1.000 -0.166 0.(65 0.2(( -3.303 0.03! 0.109 -0.091 -0.182 0.271 0.391 O.IH -0.163 0.(92 0.257 0.17( -0.115  îm  FTC  1.000 0.371 0.071 -0.358 0.611 -0.300 0.(03 0.3U -0.187 0.060 -0.008 -0.05< -0.215 0.126 0.23( 0.182 -0.133 0.<<7 0.337 0.153 -0.11!  1.000 0.<89 0.3H 0.63< -0.172 0.168 0.223 -«.322 0.128 0.215 0.083 -0.253 0.015 0.168 0.057 -0.202 0.201 0.211 0.072 0.071 FÎTC  Fîa 1.000 0.097 -0.072 0.311 -0.1J8 -O.U( -0.200 0.151 0.162 -0.01! -0.070 -0.128 0.125 -0.150 -0.108 -0.16(  1.000 0.(91 0.183 -0.321 -0.277 -0.323 -0.115 0.612 0.607 0.(78 -^1.172 0.969 0.(99 0.530 -0.(62  îm  1.000 0.320 0.333 0.032 -0.07< 0.137 -0.272 0.180 0.269 0.115 -0.10( -0.071 -0.002 -0.068 H1.09( -fl.008 0.2<5 -0.090 n ISS  Fît  KSPB  1.000 0.1(1 -0.102 -0.11! -0.095 -0.063 0.073 0.165 0.388 -0.082 0.538 0.965 0.(11 -0.319  Ksa HSMG KSI KSP .KSS LOGFMC LOGFI LOGXSP LOGHSKil  1.000 • 0.328 0.181 0.525 0.093 0.211 -O.075 0.517 -0.3(7 -0.02( -0.0(3 0.735 KSI  KSI KSP .KSS LOGTWG LOGFI LOGXSP LOGMSNI  1.000 0.260 -0.051 •3.525 0.257 0.278 0.078  LOCXSP LOGHSKÏ  1.000 -O.303  LOGKS?  1.000 0.776 0.(89 0.137 0.285 -0.0(8 0.(95 -0.30( -0.033 -0.038 0.801 KSP  1.000 -0.027 0.(60 0.382 0.929 -3.2(2 LOGXSM  1.000  l.OOO 0.(01 -0.070 0.092 -0.0(7 0.(50 -3.337 -0.017 -0.065 0.876 <SS  1.000 -3.253 -0.059 0.0(0 0.308  i.ooe -0.3/6 -0.38( -0.26( 0.03! 0.090 -0.08! 0.010 -0.12! 0.1(8 0.056 0.05S -0.367 KSMG  XSîffl KSC KSTX  1.000 5.052 0.033 -0.312 -O.Û7< 0.022 0.223 0.368 0.231 0.15< -0.132 -0.137 -0.051 0.184 -0.351 -0.12( -0.039 0.306  1 0 0 0 0 -0 -0 0 0  000 312 069 003 !7( 197 059 10( 252  Locm;  ) 000 0 5(6 0 500 -0 (75  1.000 0.866 0.230 0.129 0.509 0.132 0.296 -0.108 LOOT  1.000 0.(07 -0.226  mSOi CORKELATIOI « A Î S I I FOB CMffllKD FOfiKT FLOÛfi À8D MIlESiL SOIL imiEÏTS l ï IQ/U TOTC TOTÏ TOTM  TOTS s m T Ffpe KSPR FfCH «SOf LCTOn LGTOTWi L (J T I W  LGTOTS LGSIKC  simt  ÏTPB NSPH FFCK KSOI LGTOTH LCTOTîffl L'JTOT LGTOTS LGSUMC  LGTOTI LGTOTM LGTOTP LGTOTS LGSUMC  TOTC 1.000 0.122 0.(60 0.337 0.540 0.631 -0.235 -O.IM 0.0!5 0.253 0.122 0.710 0.270 0.521 0.111  TOTÏ  SUMQT 1.000 8.335  FFPH  e.225 -0.341 0.161 -O.067  -1.¥J  -0.060 0.903  1.000 0.644 0.150 0.238 0.160 -0.041 -0.121 -0.261 -0.124 0.949 0.652 0.073 0.266 0.255  i.ÛOO 0.363 -0.149 -0.2!4 -0.047 -0.010 -0.270 -0.326 0.366  LGTOTH LGTOIlf 1.000 1.000 0.655 Û.394 0.106 0.317 0.439 0.011 0.261  TOTM 1.000 0.409 0.352 -0.049 0.051 H3.104 -0.166 0.026 0.611 0.940 0.349 0.371 0.020 «SPB l.OOO 0.160 -0.314 -0.073 -0.092 0.046 -0.172 0.464 LGTOTP 1.000 0.633 -0.269  TOTP  1.000 0.736 -0.151 -0.159 0.OJ4 -0.014 -0.062 0.198 0.449 0.905 0.715 -0.073 FFd  1.000 0.153 -0.295 -0.188 -0.106 0.007 0.180 LGTOTS  1.000 -0.043  TOTS  1.000 0.069 -0.236 -0.076 0.131 O.OI! 0.300 0.411 0,564 0.932 0.109 Hsa  1.000 -0.041 0.085 -0.0!0 0.065 -0.39Î LGSUWC  1.000  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0075211/manifest

Comment

Related Items