Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Structure and function of western red cedar and western hemlock forests on northern Vancouver Island Keenan, Rodney 1993

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

Item Metadata

Download

Media
831-ubc_1993_fall_phd_keenan_rodney.pdf [ 7.81MB ]
Metadata
JSON: 831-1.0075208.json
JSON-LD: 831-1.0075208-ld.json
RDF/XML (Pretty): 831-1.0075208-rdf.xml
RDF/JSON: 831-1.0075208-rdf.json
Turtle: 831-1.0075208-turtle.txt
N-Triples: 831-1.0075208-rdf-ntriples.txt
Original Record: 831-1.0075208-source.json
Full Text
831-1.0075208-fulltext.txt
Citation
831-1.0075208.ris

Full Text

STRUCTURE AND FUNCTION OF WESTERN RED CEDAR AND WESTERN HEMLOCKFORESTS ON NORTHERN VANCOUVER ISLAND.byRodney KeenanB.Sc. (Forestry), Australian National University, Canberra, 1979.A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FORTHE DEGREE OF DOCTOR OF PHILOSOPHYINTHE FACULTY OF GRADUATE STUDIESDepartment of ForestryWe accept this thesis as conforming to the required standardTHE UNIVERSITY-OF BRITISH COLUMBIAAugust 1993.© Rodney John Keenan, 1993In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature)Department of  rotes-re `-JThe University of British ColumbiaVancouver, CanadaDate  6 067706E-A t4q3DE-6 (2/88)ABSTRACTTwo forest types predominate on middle or upper-slope situations on northern VancouverIsland: an old-growth type dominated by western red cedar (Thujaplicata Donn) and westernhemlock (Tsuga heterophylla (Raf.) Sarge) (the CH type), and a windstorm-derived, second-growthtype dominated by western hemlock and amabilis fir (Abies amabilis (Dougl.) Forbes) that largelyoriginated following a widespread windstorm in 1906 (the HA type). Distribution of the two typeshas no obvious relationship with geology, topography or mineral soil. However, seedlingsregenerated following cutting on the two types exhibit large differences in growth. Seedlings oncutovers in the CH type grow slowly and have symptoms of nutrient defiency; those in the HA typegrow relatively rapidly with no sign of nutrient deficiency. This difference in productivity is partlydue to lower nutrient availability in the forest floors of the CH type, and to competition from theericaceous shrub salal, that dominates the CH type following clearcutting.The objectives of this study were to describe the structure of the two forest types, and toinvestigate aspects of their functioning, in particular the cause(s) of differences in rates of nutrientmineralisation between the two types. It was hypothesised that nutritional differences may have beendue to: (i) regular windthrow in the HA type, (ii) a greater accumulation of forest floor organicmatter in the CH type, which may lead to detrimental changes in soil physical properties; or (iii)differences in foliar and woody litter quality of the dominant coniferous species, andthe way that this affects rates of decomposition and nutrient mineralisation.The diameter-class structure of the CH type suggested it was a self-replacing, climaxcommunity. The diameter-distribution of western hemlock indicated continuous recruitment, whilethat of western red cedar suggested more periodic recruitment, at a slower rate than hemlock. Thediameter class distribution of the HA type was unimodal, suggesting an even-aged stand, but asample of tree ages indicated that many trees established some time before, or after, the 1906windstorm.There were substantial quantities of detrital biomass in and on the forest floor in both foresttypes (642 Mg/ha in the CH type, and 436 Mg/ha in the HA type). This suggested that differences111in nutrient availability were not due to changes in soil physical properties brought about by higherorganic matter accumulation. Detrital biomass was generally greater in the CH type, but the totalamount of N contained in detrital biomass was similar, because of higher N concentrations in theHA type.To investigate the windthrow disturbance hypothesis, soil properties were measured in anexperiment that partly intended to simulate the effect of windthrow, by mixing organic and mineralsoil horizons. The mixing treatment had no significant effect on soil nutrient availability in the HAtype, and generally reduced nutrient availability in the CH type. Thus, windthrow was probably notthe direct cause of higher nutrient availability and productivity in the HA type.Above-ground litter fall was lower in the CH than the HA type. Cedar resorbed higherproportion of foliar N at the time of foliage senescence (76%) than hemlock in the CH (64%), orhemlock in the HA type (51%), resulting in poorer quality of cedar foliar litter. Decomposition rateof a standard substrate, lodgepole pine needles was almost the same in the two types, suggesting thatenvironmental conditions for decomposers are similar in the two types. Cedar foliar litter in the CHtype decomposed more slowly than the foliar litter of other species in either type, and salal leaves lostmass significantly more rapidly than coniferous foliage.Using an ecosystem-level computer model, it was found that much of the difference in Nmineralisation between the two forest types could be explained by differences in litter quality of thedominant species. Structural and functional differences between the two forest types can be ascribedto differences in life history, physiological, and biochemical characteristics of the dominant tree andunderstorey species, and way these features interact with the environment and the disturbanceregime. In particular, differential responses to N availability may be an important mechanism in thesuccessional dynamics that determine forest composition in this environment.TABLE OF CONTENTSABSTRACT^ iiTABLE OF CONTENTS^ ivLIST OF TABLES viLIST OF FIGURES viiACKNOWLEDGEMENTS^ ixDEDICATION^ xChapter 1. Introduction  1Approach 3Objectives^ 9Thesis Structure  10Chapter 2. Study area and design^  12Study sites^  14Chapter 3. Population structure  15Introduction  15Methods^  16Results  18Discussion 26Conclusions^ 39Chapter 4. Mass and nutrient content of forest floor and woody debris^ 40Introduction 40Methods^ 41Results and Discussion^ 44Conclusions  55Chapter 5. Mineral soil properties  56Introduction^ 56Methods 57Results and Discussion^  58Conclusions^ 69Chapter 6. Effects of soil mixing on soil properties and understorey vegetation^ 70Introduction 70Methods^ 71Results and Discussion^ 74Conclusions 81Chapter 7. Litter production, nutrient resorption and decomposition^ 82Introduction^  82ivMethods^ 84Results  87Discussion^ 97Conclusions  100Chapter 8 Modelling organic matter and nutrient dynamics^  102Introduction^  102Model structure and calibration^  104Results  115Discussion^  123Conclusions  128Chapter 9. Conclusions^  130Introduction  130Causes of differences in functioning^  132Competition theory^  139Succession^  141Management Implications^  147Future research^  150References^  153viLIST OF TABLESTable 3.1 Abundance of seedlings and saplings less than 4 m tall by height class for different species22Table 4.1 Density and nutrient concentrations of wood samples of each species^ 45Table 4.2 Nutrient concentrations (mg/g) of forest floor layers^ 46Table 4.3 Mass (Mg/ha) of dead standing trees, downed logs and small woody debris^ 48Table 4.4 Mean mass of forest floor layers^ 50Table 4.5 Mass (in Mg/ha) of fine (< 2 mm) and coarse (>2 mm) roots in the forest floor layers ^ 52Table 4.6 Mean mass and nutrient content of four detrital pools.^ 54Table 5.1 Morphology of mineral soils under the CH and the HA type at the Beaver Lake site. ^ 59Table 5.2 Morphology of mineral soils under the CH and the HA type at the Rupert site.^ 60Table 5.3 Morphology of mineral soils under the CH and the HA type at the SCHIRP site. ^ 61Table 5.4 Bulk density coarse fragment content, and particle size distribution (in the fraction smallerthan 2 mm in diameter) of mineral soils under the CH forest type. 64Table 5.5 Bulk density coarse fragment content, and particle size distribution (in the fraction smallerthan 2 mm in diameter) of mineral soils under the HA forest type.^ 65Table 5.6 Chemical properties of mineral soils under CH forest type. 67Table 5.7 Chemical properties of mineral soils under HA forest type^ 68Table 6.1 Soil physical properties in the uncut forest, after clearcutting, and after clearcutting andsoil mixing treatments.^ 75Table 6.2 Soil chemical properties and microbial activity in the uncut forest, after clearcutting, andafter clearcutting and soil mixing treatments^ 77Table 7.1 Annual above-ground litterfall (kg/ha) for the CH and Ha types.^ 88Table 7.2 Percent N and P, percentage resorption, and the relative leaf density for four types offoliage^ 91Table 8.1 Parameters used for each species in the growth component of the model.^ 108Table 8.2 Parameters used in decomposition component of the model. ^  113viiLIST OF FIGURESFig. 3.1 Relationship between age at germination height , adjusted to germination height, anddiameter o.b. at 1.3 m for (a) western red cedar and (b) western hemlock.^ 19Fig. 3.2 Diameter-class frequency diagrams for CH stands.^ 21Fig. 3.3 Height class distributions for CH stands.^ 23Fig. 3.4 Relationship between diameter (over bark) and age at 1.3 m above ground for second-growth HA stands.^ 24Fig. 3.5 Diameter-class frequency diagrams for three HA stands on northern Vancouver Island^ 25Fig. 3.6 Reconstructed diameter-class distribution of three HA stands prior to the 1906 windstorm.27Fig. 3.7 Stem plot of the CH stand at the Beaver Lake site^ 28Fig. 3.8 Stem plot of the CH stand at the Rupert site. 29Fig. 3.9 Stem plot of the CH stand at the SCHIRP site.^ 30Fig. 3.10 Stem plot of the three HA stands^ 31Fig. 3.11 Distribution of established seedlings and saplings (5 cm - 4 m tall) by growing substrate inthree CH stands. 32Fig. 6.1 Above-ground biomass of understorey species in the uncut forest, and after clearcutting andclearcut and soil mixing^ 80Fig. 7.1 Mean conifer foliage litterfall (in kg/ha/day) from CH and HA stands.^ 90Fig. 7.2 Mass remaining versus time for decomposing lodgepole pine needles. 93Fig. 7.3 Mass remaining versus time for decomposing western red cedar, western hemlock and salalfoliar litter in the CH type, and mixed western hemlock and amabilis fir foliar litter decomposing inthe HA type.^ 94Fig. 7.4 Relationship between percentage mass remaining and percent nitrogen for western redcedar, western hemlock and salal foliage decomposing in the CH type, and mixed western hemlockand amabilis fir decomposing in the HA type.^ 96Fig. 8.1 LINKAGES structure, rectangles represent subroutines, and arrows indicate important flowsbetween subroutines.^  105viiiFig. 8.2 Growth response functions for western red cedar and western hemlock to available N^ 112Fig. 8.3. Actual (bars) and simulated (dotted lines) diameter distributions for western hemlock andwestern red cedar in CH stands^ 116Fig. 8.4 (a) and (b) Simulated above-ground biomass for four different scenarios, and (c) meanabove-ground biomass calculated from 3 stands in each of the CH and HA forest types. ^ 117Fig. 8.5 (a) and (b) Simulated nitrogen availability (annual net N mineralisation, not including treeuptake) in the forest floor of the four scenarios.^  118Fig. 8.6 Simulated forest floor accumulation by forest floor layers for the CH and HA scenarios, andmean measured forest floor accumulation in three stands of the two forest types.^ 121Fig. 8.7 Simulated growth of cedar and hemlock planted at 3000 stems per ha on sites previouslydominated by cedar (CH), and hemlock^ 122ixACKNOWLEDGEMENTSThere was a time, back before we started all this, when Deb said that having kids wouldn'tstop us doing anything that we really wanted to do. She may not have said it so easily if she had anyidea of what she would have to go through over the last four years, but firstly I really want to thankher for her love, support, and her tolerance of all that getting this degree has demanded. To my kids,Henry, Tess, and Joe, thanks for putting up with my time away, and for continually helping me toremember that there are are lot more important things in life than academic pursuits.It was Hamish Kimmins' boundless enthusiasm that encouraged me to come to UBC.Initially, I found the breadth of his ideas as complex and confusing as the forests I was trying tostudy, but eventually I did reach some understanding of both. While I have tried to capture myunderstanding of the forests in this thesis, I will take Hamish's positive attitude to life and hisfriendship with me in my future career. Thanks. Cindy Prescott's assistance in getting this project ofthe ground was invaluable. Her help in designing and implementing the initial field work (even tothe detriment of her knees), and her willingness to share her time for discussion and her library aregreatly appreciated. The success of the project was largely due to her early guidance. The othermembers of my supervisory committee, Drs. Feller, Le May, and Weetman gave considerableassistance in various stages of the project and I thank them for their help.From helping us settle into Vancouver, to collaboration in the latter stages of this thesis,Christian Messier was a constant source of friendship, advice, and discussion. I look forward tocontinuing the collaboration that began during our time at Port McNeil. To the rest of the gang atPonderosa, and those from across campus who contributed to my learning experience at UBC,thanks for your help, your friendship, and the good times we had together.A large number of people were involved in the logistics of the study. I would particularly liketo thank Bill Dumont, Steve Joyce, Paul Bavis, and Cindy Fox and their staff at Western ForestProducts Ltd. in Port MacNeill for providing field assistance, equipment, advice, and accomodationduring the study. Terry Lewis provided much of the initial motivation for this work. During thecourse of the study he reviewed various aspects of the design, and regularly considered my ideas inthe light of his unparalleled knowledge of the soils and vegetation of Vancouver Island. HeatherJones was a great help in the field early on, and later Ron Burlison had to put up with my dumbjokes for hours in the field (although he did laugh... occasionally). Arlene Gammell at theMacMillan Bloedel laboratory in Nanaimo, and Min Tsze and his staff in the Forest Ecologylaboratory at UBC did most of the chemical analyses and other lab. work. John Pastor, at theUniversity of Minnesota, readily made his computer model available for the study, and gave up histime to help me understand it, and adapt it for the north Island forests.Finally, I would like to thank the people of Canada, through the National Sciences andEngineering Research Council, and the corporations: Western Forest Products Ltd., MacMillan-Bloedel Ltd., and Fletcher-Challenge (Canada) Ltd.; for the financial assistance that provided theopportunity to live and study in your country. It has been a time of tremendous personal growthand understanding for me, and I hope that I have returned the contribution, in a small way, byfurthering our knowledge of your magnificent forests.DEDICATIONTo my parents,Wally and RaiChapter 1. IntroductionThe natural landscape along the west coast of North America, from northern California tothe top of the Alaskan panhandle, is dominated by coniferous forests of globally unsurpassed formand stature. Most of the genera composing these forests differentiated from earlier gymnospermsduring the Jurassic period from 140-160 million years ago (Scagel et al. 1965). Therefore, conifershave had a considerable period in which to adapt to a variety of environmental conditions and nowoccur in a wide range of climatic environments. However, Pacific coastal forests are thought to beremnants of vegetation types that once dominated the land masses of the northern hemisphere.Their current range is restricted to areas with a temperate, wet winter climate and mild to warmsummers (Waring and Franklin 1979).The ranges of individual species along the Pacific coast have shifted continuously overgeological time in response to glaciations and other climatic fluctuations. The most recent glaciationretreated from northern Vancouver Island about 14,000 years ago, and over the last 3,000 years theforest vegetation has become dominated by western red cedar ( Thuja plicata Donn), westernhemlock (Tsuga heterophylla (Raf. Sarge) and amabilis fir (Abies amabilis Dougl.), perhaps inresponse to higher precipitation and cooler temperatures (Hebda 1983). This expansion of westernred cedar has been connected to the evolution of massive woodworking technology in the culture ofthe local native communities who used the large cedar trees for shelter and transport (Hebda andMathewes 1984).Northern Vancouver Island in British Columbia (B.C.) is located in the centre of this bandof Pacific coastal forest, and its temperate climate has mild winters and cool wet summers. Thedistribution of forest vegetation across this area varies with topography, geological substrate, and thetype and frequency of natural disturbance. On well-drained to somewhat imperfectly-drained middle12or upper-slope situations the forests form two distinct types: (i) an old-growthl type dominated bywestern red cedar with a smaller component of western hemlock (the CH type), and (ii) a second-growth type dominated by western hemlock and amabilis fir (Abies amabilis), that appears even-agedand to have originated following a widespread windstorm in 1906 (the HA type). The forest floorsof both types are deep mor humus, generally of greater depth in the CH than the HA type (Germain1985). Mineral soils are duric or orthic Humo-Ferric Podzols.In classifying the ecosystems of this area, Lewis (1982) could not distinguish between thetwo types on the basis of topography or mineral soil characteristics, and included them in the sameecosystem association. He further hypothesised that they were different stages (or phases) of asuccessional sequence.Since the 1930's, these forests have been extensively harvested for timber. Followingclearcutting and slash-burning, major differences in the productivity of planted and naturally-regenerated seedlings have been observed in the two forest types (Weetman et al 1989a and b).Seedlings regenerated after cutting in the CH type grow slowly and exhibit symptoms of nutrient(particularly nitrogen) deficiency, while those regenerating on the HA type grow relatively rapidlyand exhibit no signs of nutrient deficiencies. This slower seedling growth on the CH type is partlydue to lower forest floor nutrient availability, and partly to competition for nutrients from theericaceous shrub salal (Gaultheria shallon Pursh) which resprouts rapidly from rhizomes followingclearcutting and slash-burning (Weetman et al 1990, Messier and Kimmins 1991, Messier 1991).The difference in forest floor nutrient availability (particularly of N) that leads to thedifference in productivity of regeneration on cutovers sites has also been found in the uncut forest(Prescott et al. 19936). The magnitude of this difference, in terms of annual rates of N-mineralisation in the field, has not been determined. However, a variety of indirect estimates areavailable. Germain (1985) reported that total N was 0.9% in soils from clearcut and burned CH1 Old-growth is a difficult term to define, but for these sites the oldest trees are over 500 years old and very large in diameter. These stands alsohave the diversity of height, diameter, age, and understorey structure considered characteristic of old-growth stands of the Pacific North West byFranklin et al. (1981).3sites, and 1.17% in similarly treated sites in the HA type. Extractable N was about 50% higher (191vs. 131 ppm). Extractable N was 20 to 30% higher in forest floor material from intact HAcompared with CH stands (Prescott et at 1993b). In the same study the L layer from intact standsin both types immobilised N in a 30 day anaerobic incubation, the F layer in the HA type releasedabout 120 peg of N, but the F layer from the CH showed no net mineralisation. The biomass ofseedlings grown in pots containing F layer material for 1 year in the greenhouse was about 50%higher in material from the HA than the CH type. The height of Sitka spruce (Picea sitchensis(Bong.) Carr), a relatively nutrient-demanding species, 14 years after planting on the HA type wasalmost double that on the CH type (Weetman et al 1990). Thus, it appears that the rate of Nmineralisation is about 50% greater in intact stands of the HA compared to the CH type, and thatthis difference may increase to almost 100% after clearcutting and burning (Weetman et al. 1990,Prescott et al. 1993b).The aims of this thesis were to characterise the stand structure, detrital biomass distributionand ecosystem functioning of these CH and the HA forest types; to explore the factors that canpotentially explain the observed differences in nutrient availability and forest productivity; and toinvestigate how these factors may contribute to the mechanisms associated with the hypothesisedsuccessional development from the HA to the CH type.ApproachDifferences in ecosystem functioning can result from several interacting factors. Foreststructure, functioning and productivity are strongly influenced by climate, soil parent material andtopographic position, and the variability in mineral soil properties imposed by these characteristics.Topography is a major influence on soil development (Jenny 1981, Haywood 1983,Van Cleve et al.1991). Disturbances such as fire, windstorms, and landslides can strongly influence forestcomposition and functioning (Pickett and White 1985), and changes that occur within theecosystem during the course of secondary succession following disturbance can markedly affectecosystem properties such as nutrient cycling and availability (Sprugel 1986, Vitousek et al. 1988).4The influence of the vegetation itself on ecosystem functioning has been apparent toecologists and foresters for some time (Rommell 1935, Handley 1954). These effects come aboutthrough feedbacks between the way in which different species store and utilise the carbon fixed inphotosynthesis, and the rate at which nutrients are released in the forest floor during decomposition.Ultimately, this can affect the level of production that the forest sustains (Hobbie 1992). In recentyears, it has increasingly been recognised that differences in functioning due to species compositioncan occur at local scales (Pastor et al. 1984, Zak et al. 1986), and changes in rates of functioningcan occur over relatively short periods of time (Wedin and Tilman 1990).Determining which of these factors are producing functional differences in any particularcase is problematical. Changes due to topography or disturbance are confounded with species effects,because species which are better adapted to a particular topographic position or type of disturbancecan have attributes which markedly alter the functioning of the ecosystem. Changes in functioningthat occur in the course of secondary succession following disturbance are also tightly linked withshifts in composition to species with different regeneration strategies and chemical characteristics.Many of these ecosystem changes occur over long time periods and are the result of naturalevents that occur infrequently. Natural events, such as fire or flood, have varying impacts that areinfluenced by many chance factors. Because of this complication of factors, the concept of cause andeffect is difficult to apply in the study of ecological systems (Ulanowicz 1990). It is thereforedifficult to investigate the influence of these factors through traditional scientific approaches such asobservation and analysis, or experimentation and hypothesis testing. The goal of scientific research isoften considered to be the development of simple hypotheses that can be tested through experimentsthat vary one factor at a time. However, this approach can be impossible in ecological studies whereit is difficult to find, or devise, situations where single potential contributing factors can be variedindependently. Furthermore, if a process is investigated assuming there is only one cause, then otherimportant causes may be ruled out, and this may ultimately slow down the process of scientificdiscovery (Hilborn and Stearns 1982).5Positive feedbacks may also reinforce certain conditions. For example, the proximal cause oflow N availability may be the chemical characteristics of a dominant plant species, but the presenceof the species may be due to the type or pattern of disturbance, or the effect of topography.Assigning a 'cause' is therefore a problem, and it may be more appropriate to consider many of ourobservations of natural ecosystems in a more holistic way, as attributes of the system and the way itinteracts with its abiotic environment, and not look to ascribe causes to any particular observedeffects.Because of these interactive effects, the process of ecological investigation has been calledinto question in recent years. Various authors have argued for stronger linkages between populationor community studies, and between process- and functionally-oriented studies that use the ecosystemas the basis for investigation (Carney 1989, Allen and Hoekstra 1989). Chapin et al. (1987) suggestthat collaboration between physiological, ecosystem and community ecologists will be a key todeveloping a mechanistic understanding of the dynamics of complex natural communities.Thus, in describing the structure and functioning of these two forest types, an attempt wasmade in this study to link together information on population structure and functional processes, tofurther understand factors affecting the plant-soil system. At the beginning of the study three broadexplanations were put forward for observed differences in functioning between the two forest types.These were developed after reviewing the literature and following discussion with other researchers.1.Site factorsThe differences in functioning between the CH and HA types may be due to the two foresttypes occupying sites with different edaphic, topographic, or hydrologic conditions.Topography within the study area is quite subdued compared to much of coastal B.C. Thelarger scale influences of topography on vegetation evident in other areas are not generallyexpressed there, except for low-lying, poorly-drained situations where Sphagnum spp. andother species predominate. Lewis (1982) could not differentiate between the two forest typeson the basis of topography, geology or macromorphic mineral soil characteristics, and heincluded them within the same ecosystem association. Other observers have suggested that6the HA type tends to occupy more wind-exposed situations (de Montigny 1992). Germain(1985) considered that the CH type tended to occur on more poorly-drained situations,although he also observed gleying and evidence of periods of reducing conditions in mineralsoils under the HA type.2. DisturbanceThis explanation has two components.(i) Repeated wind throws in the HA type may increase nutrient availability by periodicallymixing organic horizons with mineral soil and bringing buried organic material to thesurface where it can decompose more rapidly (Lewis 1982, de Montigny 1992).Pedoturbation due to windthrow is a widespread phenomenon in forests (Lyford andMacLean 1966, Dunn et al. 1983, Cremeans and Kalisz 1988), and windthrow can have asignificant impact on many aspects of forest ecosystem functioning (Schaetzl et al. 1989,Nakashizuka 1989, Peterson et al. 1990). Tree falls create a variety of soil microsites withina stand, depending on whether the trees snap off, the roots are uplifted and form a hinge, orthe root ball rotates within the soil profile (Beatty and Stone 1986). Uplifting or rotation ofthe roots of larger trees results in a major disturbance to the soil profile, and can have aneffect similar to mechanical ploughing (Armson 1977). Windthrow has been suggested as animportant mechanism for maintaining forest productivity in some areas. For example, in thecoastal forests of south east Alaska it is considered to be the only form of disturbanceoccurring often enough to keep soils in a juvenile, or semi-mature stage of development.Mature soils in south-east Alaska are not as favourable for tree growth as less-developed soils,because the extensive podzolisation that occurs in this wet, cold environment can lead tonutrient immobilisation, and the development of impermeable horizons (Ugolini et al.1990).(ii) Changes in forest floor physical and chemical properties brought about by anaccumulation of organic matter in the forest floor over the extended period of time since amajor disturbance may lead to a lower rate of nutrient mineralisation in the CH type. This7may be a function of climate, because in cooler areas the forest floor can continue toaccumulate over time since disturbance. Deeper organic matter alters soil physical properties,slowing spring warming, and reducing mean temperatures below the soil surface. Forest floorbulk density and water holding capacity can also be affected, adversely impacting onaeration. These changes lead to slower decomposition, and nutrients in litterfall are not madeavailable as rapidly to the trees, slowing tree growth. In extreme situations these changes canlead to soil paludification and the replacement of forests by less productive bog-woodlandcommunities (Van Cleve et at 1991).Alternatively, organic matter accumulation may be a function of changes in allometry due tochanges in stand structure during the course of successional development. Vitousek et al.(1988) suggested that, as succession proceeds, the proportion of wood compared to foliage inthe standing biomass and in the total litterfall increases. Because of this, the ratios of carbonto other nutrients in the standing forest and in the litterfall increase, and the quality andquantity of resources available to decomposer communities is reduced. The decomposeractivity that goes into breaking down these carbon compounds in wood can lead to short-term nutrient immobilization and lower rates of forest floor nutrient supply. These changesmay be exacerbated by differences in the wood chemistry of species occurring later in asuccessional sequence.3. SpeciesDifferences in functioning of the two types may be due to differences in the chemicalattributes of the tree species dominating each site. N is the element most limiting treegrowth in this region, and its availability is determined by the rate of decomposition and thepattern of accumulation and release in decomposing detritus. If climatic conditions aresimilar, the rate of decomposition is a function of the 'quality' of the detritus as a substratefor decomposers. Quality is a general term, and is a combination of the availability todecomposers of the energy contained in carbon compounds and nutrients in detritus. Intemperate forests, studies of the decomposition process have found two factors to be of8primary importance in determining both the rate of decomposition and the pattern ofnutrient mineralisation and release: (i) the concentrations of recalcitrant, secondary carboncompounds, generally measured as the acid-insoluble fraction and considered functionallyequivalent to lignin, and (ii) the concentration of N (Aber and Melillo 1982, Melillo et al.1982, Flanagan and Van Cleve 1983, Berg and McClaugherty 1989, Upadhyay et al. 1989,Aber et al 1990, Harmon et al. 1990b, Taylor et al. 1991).Lignin concentrations of western red cedar and western hemlock foliar litter in this regionare high (about 22%) but generally similar (Harmon et al. 1990b). In studies of foliarnutrient content, cedar had consistently lower N concentrations (about 0.4-0.5%, Tarrantand Chandler 1951, Beaton et al. 1965, Ovington 1965, Harmon et al. 1990b), thanhemlock (about 0.7-0.8%) or amabilis fir (about 1%). Much of the investigation of theeffect of differences in foliar litter quality on soil properties has related to the investigation ofspecies with markedly different foliar characteristics, for example deciduous versusconiferous species, and the resulting major differences in the character of mull and morhumus forms. However, more recent studies have indicated that even within the conifersmore subtle differences in indices such as N concentration can have a bearing on nutrientmineralisation rates (Stump and Binkley 1993).Lower nutrient availability has been reported in forest floors under western red cedar inother localities. In a comparison of soil properties under cedar and hemlock in easternWashington, Turner and Franz (1985b) found that while microbial numbers andnitrification were greater under cedar, there was a greater biomass of fungi and substantiallyhigher rates of N-mineralised in forest floor material under hemlock.However, cedar has also been regarded as a 'soil improver', with higher foliar calciumconcentrations and higher pH in soil under cedar than under other conifers, and it isconsidered an indicator species for 'nutrient rich', highly productive sites on slightly dry towet or ground water enriched soils (Krajina 1969). Stone (1975) suggested that these9influences on Ca and pH levels are generally characteristic of drier climates, and are lessapparent in areas of high rainfall.Western red cedar wood contains thujaplycins, methyl thujate and thujaplicatins, classes ofnatural preservatives that are highly toxic to decomposers (Minore 1983). These defensivecompounds contribute to cedar individuals growing to large sizes, and the death and collapseof these large trees produces a large amount of woody material on the forest floor thatdecomposes slowly and leads to a high forest floor C:N ratio, and this may influence nutrientmineralisation rates.Salal, or other species, may be contributing tannins and polyphenols to the forest floor inlitterfall or root exudates that reduce nutrient availability. For example, Haynes (1986)suggests that the lignin in wood degrades to phenolic compounds and combines with thepolyphenols, plant proteins and amino acids to form humic polymers that are resistant todecay.Topographic differences were not considered to be the sole explanation for the observedfunctional differences between the two forest types. In frequent field trips to the area I have seenexamples of both types in similar topographic positions. Other observers have also seen examples ofthe two different forest types on different topography and geology in other locations (Terry Lewispers. comm.). The potential effects of salal on forest floor chemistry between the CH and HA typeshave been investigated by others (de Montigny 1992, de Montigny et al. 1993), and theimplications of their findings are discussed in the context of the results of this study in Chapter 9.ObjectivesThe objectives of this thesis were: to describe the structure and functioning of the CH andHA forest types; to investigate the cause, or causes, of lower nutrient availability (particularly Navailability) in the CH type; and to investigate the mechanisms associated with the successionaldevelopment from the HA to the CH forest type.10Three major hypotheses were tested in investigating the causes of differences in nutrientavailability: (i) that the HA type has higher nutrient availability because of the physical effect ofrepeated windthrow; (ii) that the CH type has lower nutrient availability because of a greateraccumulation of organic matter; and (iii) that the difference is due to the litter quality of thedominant coniferous species, and the way that this affects rates of decomposition and nutrientmineralisation.Thesis StructureThe thesis consists of 9 chapters, including this introductory and the concluding chapter.Chapter 2 describes the physiography, climate, and vegetational history of the study area, and thethree locations where intensive investigations of the structure and functioning of the two forest typeswere carried out. Chapters 3-8 comprise separate studies to investigate the hypotheses describedabove. Investigations were undertaken at three sites. At each site, intact stands of the CH and HAtype occurred in close proximity on similar topography. There was little information available on thestructure of natural stands of either type, and therefore, the age-diameter relationship, and size classstructure were described. This structural description provided a basis for estimating above-groundbiomass, and gave some indication of the mode and patterns of regeneration of the dominantspecies. This investigation is described in Chapters 3. Mineral soil properties are described inChapter 5.The amount, distribution and nutrient content of detrital biomass was quantified. This wasused to test the hypothesis that lack of disturbance in the CH type was leading to a greateraccumulation of forest floor biomass, and slower rates of nutrient cycling and mineralisation.Knowledge of the nutrient content of detrital biomass and its distribution in the forest floor alsocontributed to an understanding of the dynamics of organic matter in the two types. This study waspreviously published in Keenan et aL (1993), and is reported in Chapter 4.Chapter 6 reports a study of the effects of an experiment designed to simulate windthrow bymixing mineral soil and organic matter on indices of nutrient availability and microbial activity in11the two types. This was a test of the hypothesis that the physical effects of windthrow couldstimulate decomposition and nutrient mineralisation.Rates of litterfall, nutrient resorption at the time of leaf senescence, and the rate of mass lossand nutrient dynamics in decomposing foliar litter of different species in the two forest types weremeasured. This allowed the investigation of the hypothesis that differences in the characteristics ofthe dominant coniferous species could contribute to differences in nutrient availability. This study isreported in Chapter 7.Lastly, this structural and functional information was used to calibrate a computer model offorest growth and nutrient dynamics. This model was used to quantify the association betweenspecies, nutrient availability and site productivity, and the investigation is reported in Chapter 8. InChapter 9, the results of these individual studies are summarised and considered in the context ofcurrent ecological theories of competition and succession. In most cases, chapters were written injournal publication format. This has led to some repetition in the introductory description of eachchapter, and to the repetition of some ideas in the discussion sections.12Chapter 2. Study area and designThe study area is a gently undulating coastal plain generally less than 300 m in elevationsituated near the town of Port McNeill on the northern part of Vancouver Island (latitude50°60'N). Administratively, this area is Provincial Forest and is the major part of Block 4, TreeFarm Licence 25, which is operated by Western Forest Products Ltd. The area is within the very wetmaritime subzone of the Coastal Western Hemlock (CWH) biogeoclimatic zone (Pojar et al. 1991),that occupies the lower and middle altitudes of Vancouver Island and the coastal mainland of BritishColumbia. The vegetation in this locality has been described by Lewis (1982). Two forest typesoccupy well-drained to somewhat-imperfectly-drained situations (see Chapter 1), and Lewis includedthese in the same ecosystem association, the Thuja plicata - Tsuga heterophylla - Abies amabilis -Gaultheria shallon - Rhytidiadelphus loreus (the salal moss, Si association). The CH type wasconsidered to be the climatic climax vegetation type in this environment (Lewis 1982). Maturestands are relatively open, and contain large western red cedar, up to 260 cm dbh (diameter at breastheight, over bark, at 1.3 m from the ground), with a smaller component, in terms of basal area, ofwestern hemlock, and occasional amabilis fir. The understorey consists mainly of salal, an ericaciousshrub which forms a dense cover up to 2 m tall. The forest floor is occupied largely by the mossesRhytidiadelphus loreus and Hylocomium splendens, with occasional ferns.HA stands are more uniformly-sized, densely-stocked western hemlock and amabilis fir. Theunderstorey is dominated by advanced growth of both species, with small patches of salal andVaccinium sp., and a ground cover of mosses and ferns. The transition between the two types isquite abrupt. Large cedar trees are occasionally found in the HA stands along this transition, butthere was no evidence of extensive occupation by cedar prior to the 1906 windstorm. Contiguouspoorly-drained areas were occupied by widely-spaced cedar and lodgepole pine (Pinus contorta var.contorta), Sphagnum and Myrica spp.The climate is wet with mild winters and relatively cool summers (Chilton 1981). The meanannual precipitation at Port Hardy airport, 15 km from the study area, is approximately 1700 mm,1365 percent of which occurs between October and February. The summer months experience lessrainfall than the winter months, but rainfall during the growing season prevents any soil moisturedeficit in most years (Lewis 1982). The small amount of snowfall occurs from December toFebruary and melts quickly. Hours of sunshine range from an average of 6.4 h/day in July to 1.5h/day in December, reflecting the frequent occurrence of fog in the summer and frontal clouds inthe winter. Extremes in temperature are rare, and there is a long frost-free period (175 days). Meanannual temperature is 7.90C and the daily average ranges from 2.4 0C in January to 13.80C inAugust. Because of relatively wet, cloudy conditions and morning marine fogs in summer, wildfire isuncommon in this locality and the predominant source of disturbance is windstorms.The surface geological material consists of deep (> 1 m in many places) unconsolidatedmorainal and fluvial outwash material overlying three types of bedrock: gently dipping sedimentaryrocks of the Cretaceous Nanaimo formation, relatively soft volcanics of the Bonanza group, and asmall area of harder, Karmutsen formation basalt which protrudes through the morainal cover inthe north-west of the basin.This locality is considered to be one of the first places along the British Columbian coast thatbecame free of ice after the most recent (Fraser) glaciation, with vegetation becoming establishedaround 14,000 years ago (Hebda, 1983). From a paleobotanical investigation of a bog near PortHardy, Hebda (1983) recognised five distinct periods of vegetation occupation:1. Pinus contorta/Alnus/Pteridium, 14,000 - 11,500 BP, a pioneer community on therecently deglaciated landscape, indicating a relatively dry climate.2. Picea/Tsuga mertensiana/Alnus, 11,500 - 8,800 BP, an assemblage indicative of a coolerclimate than today's, possibly with Picea engelmannii.3. Picea/Pseudotsuga/Alnus/Pteridium, 8,800 - 7000 BP, this assemblage has no modern dayequivalent but suggests a warmer, drier period which has been recognised in otherpaleobotanical studies.4A. Tsuga heterophylla/Picea, 7,000 - 3,000 BP, indicating a gradual climatic cooling andincrease in precipitation, with an increasing predominance of hemlock.144B. Tsuga heterophylla/Cupressaceae, 3,000 - present, a continued cooling to today'smesothermal climate, with spruce being replaced by cedar.From this investigation it is apparent that the current pattern of vegetation (dominated byold-growth cedar-hemlock and/or hemlock-amabilis fir) became established about 3,000 years ago.Mathewes (1973) considered the development of this assemblage to be indicative of today's wet,mesothermal climatic conditions, while Hebda (1983) suggested that the tendency of forest humusto accumulate under this regime causes edaphic changes which favour cedar.Study sitesThree sites were selected from within the study area for detailed investigations of structureand function. These were located within about 10 km of each other, and had stands considered to berepresentative of the two forest types on similar aspects and topography separated by a transition ofno more than 150 m. The sites were established after consultation with local foresters and otherresearchers familiar with the area.The first site was 1 km west of Beaver Lake, and had a westerly aspect with a slope of 0-5°(the Beaver Lake site). The second site, 3 km east of Beaver Lake and close to an experimentdesigned to investigate the effect of soil mixing on tree growth (the SCHIRP site), had a slightnortherly aspect and a slope of 0-3 0. The third was in more undulating topography about 6 kmnorth-east of Beaver Lake, near the logging road Rupert 472A (the Rupert site), and the slope variedfrom 0-10°. The plot in the HA type at this site had a northerly aspect and the CH type wasrelatively flat. At each site a 50 x 50 m (0.25 ha) plot was located in each type. This was consideredlarge enough to encompass the variation found in mature, temperate coniferous forests (Mueller-Dombois and Ellenberg 1974). All field investigations were carried out in these plots, apart fromthose carried out in the soil mixing experiment described in Chapter 6.15Chapter 3. Population structureIntroductionWestern red cedar is a widespread species in the coastal forests of Alaska, British Columbia,Washington and Oregon, and in the interior wet belt forests of British Columbia, easternWashington and Idaho (Minore 1983). Individuals of this species can live up to 1000 years, andattain diameters of over 3 m, and its decay resistant wood has a high value for timber. In wetter areasof coastal British Columbia, cedar is a common dominant or co-dominant species, generally formingmixed stands with western hemlock, amabilis fir, grand fir, Douglas-fir (Pseudotsuga menziesii(Mirb.)Franco) or Sitka spruce. Despite their economic importance there has been littledocumentation of the structure of naturally-occurring populations of these two species (Schmidt1955, Gregory 1957, Turner and Franz 1985a).Because of their longevity, the study of forest population dynamics by direct observation ofthe progression of individuals or cohorts from birth to mortality is difficult. However, in temperateregions the accumulation of woody tissue in tree boles forms annual rings, and used to estimate theage of individuals. Current diameter or age class distribution can be used to draw inferences aboutpopulations dynamics of forests (eg. Turner and Franz 1985a, Stewart 1986a and b).Many factors interact to determine forest age or size distributions. Foresters have oftendistinguished shade 'tolerant' from shade 'intolerant' species on the basis of stand structure, with theformer able to develop as mixed-age stands with higher numbers of trees in younger age (smallersize) classes, and a reverse-J age or size structure (Stewart 1986a). However, many other factors caninfluence tree population structures, including life history characteristics such as establishmentrequirements and survivorship (Harper 1977) and gap formation due to the death of overstorey trees(Runkle 1981); physical or biological disturbances (Lorimer 1985); and environmentalheterogeneity, such as the variation in soil properties created by the trees themselves (Turner andFranz 1985b).16Both cedar and hemlock are considered shade-tolerant (Franklin and Dyrness 1973), but canhave markedly different diameter-class distributions (Schmidt 1955). Hemlock is considered theclimatic 'climax' species for most of coastal B.C. (Pojar et al 1991), because it regeneratesprolifically under its own canopy, usually on decaying wood (Christy and Mack 1984). There is aconsensus that cedar can form part of the climax association on more poorly-drained sites (Franklinand Dyrness 1973), but its successional status on better-drained sites capable of higher productivityis a matter of some uncertainty (Franklin and Hemstrom 1981, Minore 1983).The objectives of this study were to (i) establish if there was a significant relationshipbetween age and diameter in the CH type, and the range of tree ages in the HA type; (ii) describethe size-class structure in the two forest types; (iii) determine whether the CH type has a diameter-class structure indicative of a climax community; and (iv) determine the number and location ofcedar advanced regeneration in the HA type.This information provides a benchmark against which to compare the effects of the extensivehuman impact that is currently affecting these forests, and it provides structural information for laterstudies investigating the causes of differences in ecological functioning between these two foresttypes.MethodsField measurementsIn studies of forest populations, it is generally preferable to undertake a full census of treeages within study plots of sufficient size to characterise the within-stand variation at the studylocation (Stewart 1986a). However, tree size has often been used as a surrogate for age, because it ismore easily measured (Schmelz and Lindsey 1965, Harcombe and Marks 1978). Stewart (1986a)considered this a valid approach if a strong relationship between size and age can be established. Thelarge size of the trees in this area and the presence of decay in the centre of many boles prevented acomplete census of tree ages, and a combination of size and age information was therefore used inthis study.17At the three sites described in Chapter 2, the following measurements were made on all livetrees greater than 4 m in height within a 50 x 50 m plot in each type. Height, diameter at breastheight (dbh, at 1.3 m, over bark), crown radii on opposite sides (using a tape measure and aclinometer to check vertical projection) and species were recorded . Position within the plot (in x,ycoordinates) was estimated to the nearest 0.5 m by setting up a 50 m tape at 10 m intervals along theplot, and measuring perpendicular distances from the tape to each tree. Distance along the tape andto each tree were recorded and converted to x,y coordinates in the laboratory. The positions of largefallen boles and root mounds in the CH stands were mapped using the stem map and a 10 x 10 mgrid of wooden stakes as reference points.In the HA type, height, diameter (at mid-point for stumps and at breast height for deadstanding trees), species (where possible) and decay class (Sollins 1982) of all stumps and deadstanding trees in each plot were recorded. The diameter at breast height of stumps was estimatedusing height and diameter measurements in the equations developed by Demaerschalk and Omule(1982). No allowance was made for shrinkage since death. A number of stumps and trees in decayclasses III and IV were considerably larger than the live trees, and stumps in these classes weretherefore considered to be the remains of stems standing prior to the 1906 windstorm. Theestimated dbh of those stems were used to 'reconstruct' the diameter class distribution of the standprior to the windstorm.To determine the relationship between age and diameter for cedar and hemlock in the CHtype, a 200 m long transect was randomly located in a clearcut area within 1 km of each site. Theage of all stumps within a certain distance of the line was determined by counting the annual ringsalong a given radius. The distance either side of the line varied with the size class of the stump (10 mfor stumps <50 cm diameter, 20 m for 50-150 cm, and 40 m for >150 cm). Diameter o.b. andstump height were recorded. 'Where large stumps had rotted centres, the diameter of the rottedportion was recorded and the age of this portion estimated using a regression equation between ageand diameter derived from sound stumps. dbh of the tree prior to the cutting which formed thestump, was estimated by multiplying the height of the stump above the germination point by the18factors in Demaerschalk and Omule (1982). To adjust the age of the tree from the annual ringscounted on the stump to the age at germination height, a sample of 25 saplings was taken from theCH stands. The average age at 1 m height for these saplings was 26.7 years, therefore age wasadjusted to germination point using the relationship:age (in years) = annual ring count + stump height (in m) x 26.7.In the HA stands, age was determined by counting annual rings on increment cores taken atbreast height from approximately two trees in each 10 cm diameter class up to 80 cm. These ageswere not adjusted to germination age.Saplings and seedlings less than 4 m tall were surveyed on 100, 1 m 2 subplots located on two1 x 50 m transects in each plot. The number of seedlings of each species in four height classes(germinants, 5-30 cm, 30-100 cm, 100-400 cm) was recorded. If more than 20 seedlings wereencountered on a sub-plot in any class, a common occurrence in the HA stands but rare in the CH,20+ was recorded. The growing substrate of each seedling or sapling was recorded according to thefollowing classes: forest floor, mineral soil, live tree bole, and log or tree stump by decay class (Sollins1982).ResultsStand structureCH type.The relationship between age and diameter was significant at all three sites (Fig. 3.1). r 2values ranged from 0.60 to 0.89 for western red cedar and from 0.61 to 0.83 for western hemlock.Diameter growth rates were similar for both species, but cedar had a considerably greater longevityand therefore achieved larger sizes. The ages of the oldest cedar ranged from 600 to 800 years (withone exceptional individual estimated to be just over 1000 years old), and the diameter of the largesttree measured was 260 cm. The maximum life span of hemlock was 300 to 400 years, with acorrespondingly smaller maximum diameter of around 90 cm. The y-intercept of 100 years,AGE - 115.5 + 3.2•DBH -R2 - 0.77SEE - 78.4AGE - 109.8 + 3.8•DBH _R2 - 0.83SEE - 24.97110010009008007 7006005004003002001000500400300•1110 200•1000500110010004009008007 7007-3 300•6001U 5000 0 200•100AGE - 108.2 • 3.8•DBHR2 - 0.60SEE - 108.21010080200300100^200DBH (cm)< 40030020010000 40^60DBH (cm)AGE 89.8 + 3.2•DBHR2 - 0.89SEE - 82.78AGE - 101.8 + 3.5•DBHR2 - 0.81SEE - 38.920 300100^200DBH (cm)Fig. 3.1 Relationship between age at germination height , adjusted to germination height, anddiameter o.b. at 1.3 m for (a) western red cedar and (b) western hemlock in the CH type at threesites on northern Vancouver Island. Ages estimated from counting annual rings on stumps oncutovers All regression equations were significant at alpha<=0.001.Thuja plicate^Tsuga heterophylla190^100^200^300^0^20^40^80^80^100DBH (cm)^DIM (cm)110010009008007 700to 600300400300200100020compared with the average measured age of seedlings at 1 m of about 27 years, suggests that the first100 years of growth does not fit the linear relationship exhibited by older trees.Diameter-class frequency diagrams for the three species at each site are shown in Fig. 3.2,and the numbers of seedlings and small saplings recorded on the line transects are shown in Table3.1. Hemlock had a much higher stem density than cedar, but the large diameter of some cedarmeant that it dominated in terms of basal area. Hemlock had a reverse-J shaped diameterdistribution and high numbers of seedlings and saplings recorded on the line transects: a mean ofabout 13,000 established seedlings (those > 5 cm tall) per hectare. There were lower numbers oftrees in the 0-10 cm diameter class at two of the sites, but this was due to the 4 m height (around 5cm dbh) delimitation.The diameter-class distribution for cedar was generally flat with many gaps, and the numberof seedlings recorded on the line transects, 4100 established seedlings per hectare, was considerablylower than for hemlock. All the cedar reproduction observed on the line transects appeared to haveestablished from seed, rather than through vegetative propagation. Although, in other areas therewere saplings that had established on elevated substrates, fallen over, and continued to grow anddevelop roots where the stem came in contact with the forest floor.The height-class distributions at each site (Fig. 3.3) exhibited a similar pattern to that ofdiameter, with cedar represented in low numbers in each canopy strata while hemlock had largenumbers in the lower strata at each site and relatively few in the upper strata.Second-growth, HA stands.The relationship between age and diameter for hemlock and amabilis fir in the three HAstands is shown in Fig. 3.4. Because of the poor correlation between age and diameter no linearrelationship is shown. The age at breast height of canopy trees in these stands ranged from 30 to 160years, with most of the trees between 70 and 100 years old. The diameter class distributions in thesestands (Fig. 3.5) were unimodal. The numbers of seedlings present in the understorey were veryhigh: 73,000 to 140,000 per hectare (Table 3.1), but there were few stems greater than 1 m tall.0 all so so 40 so 0 10 50 000(c) amabilisfir(n - 22)BA - 1.9O 10 20 20 40 SO SO(d) western redcedar50- (n = 31)BA = 67.6 m2/ha40• 30100 L11611 ' "• l•I 1.= r or II I r i i MI If. 1r •• w 1•• 1110 w MO 1.• w^ .71 MI(g) western redcedar50- (n - 32)200z100(h) westernhemlock(n - 97)BA - 22.9(i) amabilisfir(n - 2)BA 0.2v-taBA - 121.5 m2 /ha4021Fig. 3.2 Diameter-class frequency diagrams for CH stands at three sites on northern VancouverIsland. Number of trees > 4 m tall (around 5 cm dbh) in a 0.25 ha plot. (a-c) Beaver Lake site, (d-ORupert site, and (g-i) SCHIRP site. BA is basal area in m 2/ha, diameter measured over bark at 1.3 mabove ground. n is total number of trees.(a) western redcedar^(b) westernhemlock(n - 106)BA - 21.9IMO I 10 ••^1•• I•11 NO WO NO r w m50 _^(n - 26)40 _^BA - 63.7 m2/hao 30toi20z100 1 1 1^1 1^1(e) western0 .0 sOs0i 0 so 010 so 030.0(f) amabilis?,0)5.3nemiock^firL O 10 20 30 40 30 SO ND^••^1•11 •• w TM. o•• •• 0.000 0 so 0 10 so sotoo 0 10 20 30 40 SO SOSize class (diameter at 1.3 m, in cm)22Table 3.1 Abundance (per hectare) of seedlings and saplings less than 4 m tall by height class fordifferent species in adjacent old-growth cedar-hemlock (CH) and second-growth hemlock-amabilisfir (HA) forests at three sites on northern Vancouver Island (BL = Beaver Lake site, RU = Rupert,and SC = SCHIRP).Height(cm) BLCHRU SC BLHARU SCwestern red cedar< 5 0 0 1 800 0 0 05-30 600 5 800 2 100 0 0 030-100 200 3 000 300 0 0 0100-400 0 300 100 0 0 0western hemlock< 5 3 300 300 3 000 88 900 1 000 72 2005-30 7 200 11 600 8 300 65 600 120 100 121 20030-100 2 200 4 100 2 300 7 300 2 600 18 600100-400 1 000 900 1 300amabilis fir700 0 200<5 0 0 0 0 0 05-30 100 0 100 5 200 8 700 5 90030-100 0 0 0 200 200 600100-400 100 0 0 200 0 100E25i1 20Z150 50 03010 20 405045403530105O 1 I^'^I^I^IO 1 0^20^30^40^50^II II I^,^,^i^1^10^20^30^40^50504640..,. 3530a 2.I 201510210^20^30^40^50 0^10^20^30^40^50 0^10^20^30^40^50Fig. 3.3 Height class distribution for CH stands at three sites on northern Vancouver Island. (a-c)Beaver Lake site, (d-O Rupert site, and (g-i) SCHIRP site.(a) western redcedar (b) western hemlock (c) amabilis fir504540351305 25201161020 ^ I ' ' I I A ' JO 10^20^30^40^50 0^10^20^30^40^50 0^10^20^30^40^50(d) western redcedar (e) western hemlock (f) amabilis fir23(g) western redcedar (h) western hemlock (i) amabilis firNo. of stems per 0.25 ha plotBeaver Lake•• 0•• 00 •- •^•o^0^e •• •0 • 0o^•Rupert00 •0•00,••••200 ^024Fig. 3.4 Relationship between diameter (over bark) and age at 1.3 m above ground for second-growth HA stands at three sites on northern Vancouver Island. (a) Beaver Lake, (b) Rupert and (c)SCHIRP.20^40^80^SO^100DBH0 ^0 20^40^50^80^100DBH200150N 100.4500 ^0200150N 10050200150 SCHIRP00.^•. i „ • .QN 100•• hemlock500 • •^•a o fir40^80DBH80^100BO r^ 113) (n - 8)70SOI 80• 40Oz• 32010ao r (n = 149)706050403020100 O 10 20 30 40 50 80 70 80 90 100_ (n = 27)O 10 20 30 40 80 SO 70 80 90 100(e) western hemlock^(f) amabilis fir1..w^I O 10 20 30 40 80 00 70 80 90 100^0 10 20 30 40 80 60 70 80 90 100Fig. 3.5 Diameter-class frequency diagrams for three HA stands on northern Vancouver Island.Number of trees > 4 m tall on a 0.25 ha plot. (a-b) Beaver Lake, (c-d) Rupert, (e-f) SCHIRP.Diameter measured over bark at 1.3 m above ground. n is total number of trees.(a) western hemlock^(b) amabilis fir2580 r (n - 88)7060I 50• 400g 3020100 O 10 20 30 40 80 80 70 80 90 100(n 39)■^1 1O 10 20 30 40 80 00 70 SO 90 100(c) western hemlock^(d) amabilis firSize class (diameter at 1.3 m, in cm)26The 'reconstructed' diameter-class distribution of each HA stand prior to the initiation ofthe current stand is shown in Fig. 3.6. The stumps at the Beaver lake site had a similar diameter-class distribution to the current stand. At the other two sites, the reconstructed distributions had areverse-J shape, and trees in larger size classes up to 120 cm in diameter, and were quite different tothose of the present stands.Spatial PatternMaps of canopy trees (defined as 20 m or taller), trees less than 20 m, and the position ofmajor structural features in the old-growth CH stands are shown in Figs. 3.7-3.9. Stem position andcrown cover maps for the HA stands are shown in Fig. 3.10. The crown cover maps for the CHstands indicate that the canopy cover was highly variable, with many gaps. This suggests that therehas been a considerable amount of disturbance in these stands. Smaller stems tended to be associatedwith these gaps, although this was not exclusive and hemlock and amabilis fir in particular were ableto establish under the canopy of existing trees. Other large canopy gaps had no regeneration. Therewas a tendency for smaller trees to be associated with elevated surfaces (fallen boles, upturnedmounds, and stumps, see also Fig. 3.1 1).Seedling substrateThe distribution and number of seedlings and saplings less than 4 m tall by rooting media isshown in Fig. 3.11. Amabilis fir seedlings were poorly represented and did not provide enoughevidence to draw any inferences. A low number of cedar seedlings and saplings were surveyed at theBeaver Lake site, and these were growing on a range of substrates. At the Rupert site, all the cedarseedlings and most of the hemlock, were found on decay class III logs or stumps (Sollins 1982). Atthe SCHIRP site, virtually all seedlings of both species were found on decay class III and IV logs andstumps.DiscussionThe strength of the relationship between age and diameter for both cedar and hemlock inthe CH type suggested that competition has been a relatively uniform influence on the growth of27Fig. 3.6 Reconstructed diameter-class distribution of three HA stands prior to the 1906 windstorm,using a survey of decomposing stumps and dead standing trees. Diameter estimated over bark at 1.3m above ground. Count is number of trees on a 0.25 ha plot.Beaver Lake20158 10050 O 10 20 SO 40 40 00 70 00 40 100 110 120 120 140DBHRupert2015= 10380O 10 10 SO 40 00 SO 70 00 90 100 110 120 120 140DOHSCHIRP2015ii8 10050O 10 NO 20 40 SO SO 70 00 90 100 110 120 120 140DBH28Fig. 3.7 Stem plot of an old-growth Thuja plicata-Tsuga heterophylla (CH) stand at the Beaver Lakesite showing: (a) canopy trees (=> 20 m tall), (b) sub-canopy trees (< 20 m tall), and (c) majorstructural features such as large boles, root mounds, stumps and snags. Stumps and snags are filledcircles in (c).(a)(b)29Fig. 3.8 Stem plot of an old-growth Thuja plicata-Tsuga heterophylla (CH) stand at the Rupert siteshowing: (a) canopy trees (=> 20 m tall), (b) sub-canopy trees (< 20 m tall), and (c) major structuralfeatures such as large boles, root mounds, stumps and snags. Stumps and snags are filled circles in(c).30Fig. 3.9 Stem plot of an old-growth Thuja plicata-Tsuga heterophylla (CH) stand at the SCHIRP siteshowing: (a) canopy trees (=> 20 m tall), (b) sub-canopy trees (< 20 m tall), and (c) major structuralfeatures such as large boles, root mounds, stumps and snags. Stumps and snags are filled circles in(c). Cedar'4 HemlockFirhemlock firFig. 3.10 Stem plot of three windstorm-derived second-growth stands of Tsuga heterophylla-Abiesamabilis (HA) stands on northern Vancouver Island. Canopy cover is shown on the left, stemposition is shown on the left. (a) Beaver Lake site, (b) Rupert site, and (c) SCHIRP site.5000 000o 030r4^304kill■tijeie .:4Q 'IN."1.,040 ..Ai•-•-•;4. ;ie- tibtia1 .\ ....---ft. r ■^41,-4 , iI4jP; Cf, , silf4;* i; 04 I li ‘61 9 ,i4i^* 4: . ,4.4:•.• ' - ■ P •• OS- •*4;04 ,'; - . 1- . • . - .44•4 1 ' ••;,•• • ,^■ ■ • 4. •0 . A , ill kW,... irate ..../:".• .•.•• . i *Ikatio./Alk, .4ilielk. .4■•'4: )*:4:-7 - — .47:4•",,''s _7**" =Zits','204010o80000 000 00o^O 0 0 0O 00O q•o& o u 0O 00  00 0 ^-oo 0000100^ 00 (9 °00 0O002000000 0090O 00000 u0O 0O 0 0oe°o0040^0^co0000050203040105000• •• ,••^•• •• fr•.•• . ••• glr^g^0 •^•• •• a• • ••'ID •up • ••10••• •0••20▪ lb •00 •• oe goe•030s• 9I •• •0 • ■56.• 0.0• 0.•• •I • 0 n• so es eio • „• -•• .0• ••• ••40•• oo •, • ••5020 30 40 50O 10• hemlock^0^fir••• • 0 • 030504020100•0•.• 1st ••• "• • •^••„.IP  • ••• .^.:••• •• 55 0 ••• •SS•• • ••• • • "0• ••31Oa. 402001201008060VFIR^(n - 1)HEMLOCK(n - 119)■ CEDAR (n 25)32Fig. 3.11 Distribution of established seedlings and saplings (5 cm - 4 m tall) by growing substrate inthree old-growth Thuja plicata-Tsuga heterophylla (CH) stands on northern Vancouver Island. FFforest floor; I - V = logs, stumps and mounds varying from little (I) to highly (V) decayed; andBASE = live tree base. Percent is percent of the total seedling number for each species.Beaver Lake 120100806040200C•a.El FIR^(n - 2)HEMLOCK(n - 104)■ CEDAR (n - 8)FF^III^IV^V^BASERupert120100806040200C•2•a.Ei FIR^(n - 0)HEMLOCK(n - 166)■ CEDAR (n - 91)FF^IV^V^BASESCHIRPFF II^II^IV^ Decay classSUBSTRATEV BASE33these trees, and that different-aged individuals have developed with similar amounts of competingoverstorey. Maximum diameters and ages were consistent with values previously reported for thesespecies, although the largest individual diameter for hemlock in the CH type was about 90 cm, andthis is lower than the maximum of about 120 cm reported in other localities (Franklin and Dyrness1973). This suggests that hemlock may not be able to reach its full potential in the CH type.The relatively strong relationship between diameter and age, the presence of individuals inmost diameter classes at each site, and the relatively high numbers of seedlings and saplings (Table3.1) indicated that cedar populations in the CH type were broadly uneven-aged. The flat diameterdistribution suggested that once cedar became established, there was very little mortality until thetrees reach maximum age.This diameter-class structure is similar to that exhibited in studies of cedar on better-drained sites inother locations (Schmidt 1955, Gregory 1957, Turner and Franz 1985a), and similar diameter-classdistributions have been reported for other long-lived, shade-tolerant species in temperate rainforestsof the southern hemisphere, such as Lagarostrobus franklinil in Australia (Gibson and Brown 1991)and Nothofagus menziesii in New Zealand (Stewart 198613). Gaps in the diameter distribution, aswell as the number of boles on the forest floor, suggested that the canopy is being disturbed atintervals by the windstorms that occur in the area. Grouping of smaller trees (Figs. 7-9) suggestedthat younger trees enter the overstorey as cohorts within a canopy gap, and that gap-phasereplacement is the dominant process of regeneration. At the Rupert site, where more gaps in thediameter distribution occurred, cedar had larger numbers in lower height classes, and this suggestedthat past disturbance may have affected the canopy to a greater degree at this site than at the othertwo. The variability in both the diameter-class distribution and canopy height is characteristic of theold-growth stage of stand development in this region (Franklin et al. 1981, Oliver and Larsen 1990).Seedling establishment of cedar is less than that of hemlock, which may be indicative of adifference in reproductive strategy between the two species. The longevity of cedar, and its resistanceto disease and insect attack (Minore 1983), suggested that it is able to put less resources intoreproduction and still maintain its dominance in these stands, because proportionally more trees34survive once established, and they survive considerably longer, and grow much larger than mostother coniferous species found in the area. Hemlock, on the other hand, has higher early rates ofgrowth than cedar when sufficient nutrients are available (Weetman et al 1990), and considerablyhigher numbers of advanced regeneration, particularly in the HA type. However, it is moresusceptible to a range of diseases, and has a considerably shorter life-span and smaller maximumdiameter and, therefore, probably needs to put substantially greater resources into reproduction tobe maintained within the stand.The unimodal structure of hemlock in the HA stands, and its rapid early growth in this typefollowing disturbance, suggest that it might be classified as an 'r- strategist' (Mac Arthur and Wilson1967), i.e. a relatively short-lived species, that is capable of growing rapidly and fully exploiting thehigh level of available resources such as light, water and nutrients which exist following majordisturbances. However, a reverse-J diameter-class structure has commonly been reported forhemlock (Gregory 1957, Alaback 1982, Turner and Franz 1985a, Schmidt 1955), and it is generallyconsidered the dominant species in climax vegetation in this region (Pojar et al. 1991). Thissuggests that hemlock may also be considered a 'K-strategist, that is, a species with slowerdevelopment, longer life spans, and more tolerance of the lower levels of resources that exist later insuccessional development. Stewart (1986b) observed similar differential responses of Nothofagusmenziesii to different disturbance regimes in New Zealand, and suggested that the r-K continuumconcept is too simplistic when applied to forests that are being continually disturbed in space andtime. By maintaining a large bank of seedlings on the forest floor, hemlock is probably moreappropriately described as an opportunist species (Forcier 1975) capable of exploiting a wide rangeof regeneration niches in the space/disturbance-frequency/disturbance-type continuum.The preponderance of hemlock seedlings and smaller size class individuals compared to cedarmight indicate that these species cannot coexist indefinitely, and that over time cedar will decreaseand hemlock will increase in dominance. However, the greater longevity of cedar, and slow rates ofcommunity change relative to changing climatic conditions, could promote continued species35coexistence, as has been suggested for long-lived species in other forest ecosystems (White et al.1985, Lertzman 1992).The relatively wide range of ages in the HA stands has been documented previously for post-windstorm stands. For example, Henry and Swan (1974) found that a windstorm killed theoverstorey but left most of the understorey intact in a hardwood forest in North-eastern USA, andthe resulting stand was broadly uneven-aged stand. Stand replacement following a large wildfire mayalso take a long period of time (Franklin and Hemstrom 1981). The results therefore support thecontention of Lorimer (1985) that it is not really appropriate to use 'even-aged' as a generic term forstands of catastrophic origin.Although this kind of evidence is not conclusive, the 'reconstructed' diameter distributions(Fig. 3.6) suggested that the stands occupying the HA stands prior to the windstorm, had reverse-Jdistributions more typical of the old-growth stage of development (Oliver and Larsen 1990), with awider range of tree sizes, larger trees (up to 120 cm in diameter), and more canopy gaps.Investigation of the wood and bark on these remains indicated that the large trees were all hemlock.This suggests that the time interval between any previous, catastrophic, stand-initiating disturbanceand the one that created the current HA stands, at two out of the three sites, was well over 100years.Amabilis fir had a low representation in the three CH stands, making it difficult to drawinferences about patterns of regeneration. It is considered to have the highest shade tolerance of allthe three species present (Franklin and Dyrness 1973), but the low seedling numbers (Table 3.1)suggested that its establishment requirements were not being met, and that the regeneration of thisspecies is discontinuous in these particular stands.The importance of logs and other elevated sites for seedling establishment (Fig. 3.11) hasbeen reported in a wide variety of forest ecosystems. This has been attributed to lower litteraccumulation on logs (Christy and Mack 1984), the absence of soil pathogens, elevation aboveprolonged periods of standing water (McKee et al 1982), safe sites from browsing by animals(Stewart 1986b) and lower competition from understorey shrubs and herbs for light or rooting space36(Veblen et at 1980, Stewart 19866, and Nakashizuka 1987). In a study in Picea-Tsuga ecosystems incoastal Washington and Oregon, Harmon and Franklin (1989) eliminated many of these potentialexplanations and concluded that competition from herbs and mosses on the forest floor isresponsible for the high proportion of seedlings on logs.In this study, mosses form a dense layer across the forest floor in the CH type, but aredistributed more patchily in the HA stands. It is likely that they have some influence on seedlingestablishment, similar to that described by Harmon and Franklin (1989). The dense cover of theshrub salal also appears to play a major role in the distribution of seedlings in the CH type. Hemlockseedlings germinate in large numbers across the forest floor in the HA stands where the cover of salalis minimal, whereas few seedlings of any species were found on the forest floor in the CH stands.This may be due to the low light levels below the salal in the CH type. Smith and Clark (1990)found that salal leaf biomass peaked at intermediate light levels such as those found below the CHcanopies, and only 3 percent of the diffuse photosynthetically active radiation incident above the treecanopy was transmitted through the salal to the forest floor.The general absence of cedar in the HA stands, despite the proximity of potential seedsources, and the sharpness of the transition between the two types, which is not connected to anytopographic discontinuity, is particularly interesting. Once established, cedar is capable of persistingin dense shade, as long as moisture is freely available (Franklin and Hemstrom 1981, Adams andMahoney 1991). cedar seeds can also germinate under very low light conditions (Xiao Ji, Universityof British Columbia, pers. comm.). Therefore the density of the HA canopy cannot, alone, explainthe absence of cedar regeneration. The results of this study indicate that cedar is capable ofestablishing, albeit in lower numbers, on organic substrates such as decaying logs that also serve asgermination substrates for most of the hemlock advanced growth in the HA stands. Minore's (1983)indicated that cedar can germinate on organic substrates, and on disturbed, mineral soil seedbeds.It is possible that the general absence of cedar in the HA stands may be explained by thenature of cedar's mycorrhizae. Mycorrhizal fungi form mutually beneficial associations with manyplant species, and they provide considerable benefits to the plant host, improving water and nutrient37uptake (Pritchett and Fisher 1987). Some mycorrhizae can access organic forms of nutrients that areinaccessible for direct uptake by plants, and may therefore be particularly important for treesestablishing in organic substrates. In contrast to most other conifers in coastal B.C., which haveectomycorrhizal associates, cedar has vesicular-arbuscular (VA) mycorrhizae. While ectomycorrhizahave small spores that are widely dispersed by wind or other agents, the VA mycorrhizae reproduceusing relatively large spores, that do not disperse across any significant distance, but generally remainwithin the soil (Shannon Berch, B.C. Ministry of Forests, pers. comm.). Therefore, seedlings ofspecies such as cedar that form associations with VA mycorrhizae, depend either on alternative hostsfor mycorrhizal inoculation, or need to germinate near the roots of parent trees. Potential alternativehosts such as ferns are found in the HA type, but are relatively rare.In seedbeds on mineral soils with relatively high nutrient availability, there may be enoughlight and nutrient mineralisation, for cedar seedlings not to depend on mycorrhizal associates toestablish and grow. Therefore, wind-dispersed cedar seeds can probably become established in theseconditions. However, on organic substrates in undisturbed conditions light is lower, and nutrientsare generally contained in organic forms and mineralise relatively slowly. If cedar seedlings aredependent on mycorrhizal inoculation for establishment in organic substrates, then they may largelybe restricted to the rooting zone of established trees, which is generally a distance of about one treeheight.Because of the limitations on seedling establishment imposed by these slow-spreading VAmycorrhizae, one could speculate that the expansion of cedar on poorer soils, and without majorcanopy disturbance, is limited to advancement in a front from established stands. This would be arelatively slow process compared to the wind-dispersal of other conifers with ectomycorrhizalassociates. This expansion is probably facilitated by salal, which has ericoid mycorrhizal associatesand sets seed in fruits that are consumed and dispersed by animals. In the absence of disturbance thecanopy of the hemlock-dominated HA stands gradually become more open, and salal can getestablished. This would reduce the number of hemlock advanced regeneration through competition.38Occasionally, large cedar individuals will die and fall across the transition between the CH and theHA type providing elevated structures upon which conifer seedlings can establish.This expansionary process can be truncated if broadscale windthrow occurs while there arestill a large number of hemlock as advanced regeneration in the understorey of the HA stands. Thisadvanced regeneration is capable of responding rapidly to the increased availability of light andnutrients following the windstorm and hemlock dominance will be re-established. However, oncecedar has begun to replace hemlock as a dominant, a number of changes occur that prevent furtherwindstorms from causing the ecosystem to revert to hemlock dominance. Firsdy, although there ismore light reaching the understorey under the cedar, the number of hemlock advanced growth isreduced because of competition from salal. As a consequence, the seedling bank that can be used tore-establish hemlock dominance is no longer available. Secondly, cedar stems, particularly olderstems, have a pronounced taper (Oliver et al. 1988) and a broad, fluted base, and they appear to bemore wind-firm than hemlock. With a more variable canopy and size class structure, and more windfirm individuals, a catastrophic windstorm may not destroy the majority of the stand as it has in theHA type.This suggests that the absence of cedar in the HA type may be a result of the interaction ofthe mycorrhizal associates of each species, and their mode of dispersal, and the prevailingdisturbance regime. In the absence of broadscale windthrow in hemlock-dominated stands it ispossible for the successional development from the HA to CH type proposed by Lewis (1982) toproceed. However, given the limitations imposed on the rate of advancement of cedar, this is likelyto be a long process. Cedar is a relatively recent arrival in these forests compared to the otherconifers in the area, becoming dominant in the pollen record only over the last 3,000 years (Hebda1983). Prior to this, hemlock and Picea were the dominant species. This increase in cedardominance has been associated with a general cooler, wetter climate (Hebda 1983). Given itslongevity, this suggests that there have only been three to five generations of cedar dominance and,given the low rate of cedar advancement, interacting with the incidence of windstorms, that cedarsimply has not had time to spread across the entire landscape in this area.39ConclusionsThe results of this study indicated that there was a significant correlation between age anddiameter for cedar and hemlock in the CH type, and that the CH type had a diameter-classstructure indicative of a self-replacing, climax community. "While hemlock in the CH type has adiameter-class distribution that indicates relatively continuous seedling establishment, thedistribution and spatial pattern of cedar suggested that establishment occurred more periodically andat a substantially lower rate than hemlock. However, given the longevity of cedar, it is likely that itcan maintain its dominance in these stands, despite substantially lower establishment rates. Seedlingestablishment in the CH type is largely restricted to upturned root mounds and decaying logs,probably because of the density of the understorey and the moss cover. The openness of the canopyin the CH type and the number tree boles on the forest floor, suggested that these forests hadexperienced a substantial degree of disturbance during their recent development.The diameter class distribution of the HA type, entirely composed of hemlock and amabilisfir, was unimodal, which suggested that the 1906 windstorm had a catastrophic impact in this foresttype. A sample of tree ages indicated that many trees did not date directly from the windstorm.Some were present as advanced regeneration prior to the event and some probably establishedsubsequently. A 'reconstructed' diameter-class distribution from the stumps of trees that werethought to have snapped off in the windstorm, suggested that the pre-windstorm stands weregenerally older than the current stands, and that a disturbance of the intensity of the 1906 storm hadnot affected these stands for at least 150-200 years.The complete absence of cedar in the HA type suggested that it was not present as advancedgrowth prior to the windstorm. It is speculated that this absence may be a result of the interaction ofthe mycorrhizal associates of each species, their mode of dispersal and the prevailing disturbanceregime, which favours species such as hemlock and amabilis fir. These species also maintain a largebank of seedlings as advanced regeneration in the understorey, and can respond rapidly to higherlevels of resources created by the disturbance.40Chapter 4. Mass and nutrient content of forest floor and woody debrisIntroductionWoody debris and forest floor organic matter are important components of many forestecosystems (Harmon et al. 1986). Coarse woody debris (dead trees and fallen logs with a diametergreater than about 15 cm) is a major element of structural diversity in many old-growth forests(Franklin et al. 1981), and can be an important component of nutrient and organic matter dynamicsthat has often been overlooked in studies of carbon storage and nutrient cycling in forests (Harmonet al. 1990a, Harmon and Hua 1991). Downed logs can be important sites for seedlingestablishment (Scott and Murphy 1987, Harmon and Franklin 1989), and are habitat for insects,fungi and microorganisms that have a role in many aspects of ecosystem functioning (Maser andTrappe 1984).Accumulated woody debris is a visibly obvious component of the massive old-growth forestsof the Pacific Northwest region of the U.S.A., and several studies have quantified the biomass andnutrient content of woody debris and forest floor in these forests (e.g. Graham and Cromack 1982,Harmon et al. 1987, Spies et al. 1988, Bingham and Sawyer 1988). Recent studies have alsodocumented the quantities of these materials in smaller-statured forests in other parts of continentalNorth America (Arthur and Fahey 1990, Muller and Lui 1991).Forests of similar structure to those of the Pacific Northwest are found along coastal BritishColumbia. However, there have been few studies of biomass in woody debris or the forest floor inthese forests. This chapter provides information on the biomass and chemistry of these componentsin the CH and the HA types. The objectives were to: (i) quantify the biomass of standing dead anddowned logs by decay class, and the forest floor material by woody and non-woody L, F, and Hlayers; and (ii) determine the concentrations and total amounts of carbon, nitrogen, phosphorus,and potassium in this material, for each forest type. This study provided a test of the hypothesis thatthere were greater amounts of organic matter accumulated in the CH type, that could lead to41changes in soil thermal regime and slow down rates of nutrient cycling. It also provided data againstwhich to compare the output of the model of nitrogen cycling and organic matter dynamics used inChapter 8.MethodsField MeasurementsWithin the three 50 x 50 m plots described in Chapter 2, the following measurements weretaken. All downed logs greater than 1 m in diameter at mid point, and all dead standing trees andstumps were surveyed on the entire plot. The length and mid-point diameter (over bark, wherepresent) of each large downed log, the height and d.b.h. of each dead standing tree, and the heightand mid-point diameter of each stump were recorded.Fallen boles and branches between 1 cm and 1 m in diameter at mid-point were sampledusing the line intersect method (Brown 1974). In each plot, sixteen points were systematicallylocated on a grid pattern and used as mid-points for 32 randomly oriented lines (one line at eachpoint was 10 m long for pieces from 15 to 100 cm in diameter, and the other was 4 m long forpieces from 1 to 15 cm in diameter). The diameter over bark of each piece at the point ofintersection with the line, and the vertical angle between the ground and the log (used to correct thevolume calculation for logs not laying flat were recorded.Each dead standing tree, stump and downed log was classified into a decay class using amodified version of the five-class system of Sollins (1982). Decay classes were distinguished bystructural integrity, wood texture, the presence of twigs and branches, and the extent to which logshad become occupied by conifers and other vegetation. Modifications of Sollin's system related tothe presence and condition of bark which was often the portion of the bole most resistant to decayin the species surveyed during this study. Decay class I wood was intact material with no decay,whereas decay class V material had little structure and a mushy (when wet) or powdery (when dry)texture. Where possible, the species of log was also recorded, but this was not usually feasible beyonddecay class III.42From a subsample of logs in each decay class a section of wood (about 1 000 cm 3) was takenat two of the sites using a chainsaw to estimate wood density and nutrient content. Bark wasincluded when present on the sampled log. For decay classes I and II, four to six logs were selectedfrom each decay class-species combination; for decay classes III, IV and V, five logs were selected ateach site without any identification of species.30 cm x 30 cm samples of the forest floor were extracted at eight systematically-locatedpoints in each plot. These samples were sorted in the field into: L, L w (twigs less than 1 cmdiameter), F (non-woody), F es, (from wood), H (non-woody), and H am, (from wood). Five separatesamples of each forest floor component from each plot were analysed for nutrient concentration.Laboratory measurements and calculationsThe sample of wood (and bark, if present) cut from each log was divided in two. One partwas weighed, dipped in melted paraffin wax, weighed again, immersed in water and the volumemeasured by displacement (American Society for the Testing of Materials 1978). The volume of thewax was calculated from its mass and specific gravity and subtracted to obtain the volume of thewood sample. The other part of the sample was weighed, dried at 700C for 3 days and weighedagain to determine the moisture content. Density of each sample was calculated as dry massesdivided by wet volumes.These dried wood and forest floor samples were ground and digested using a modification ofthe method of Parkinson and Allen (1975). Percent N and P were determined colorimetrically witha Technicon auto analyzer, and percent K by atomic absorption spectrophotometer. Percent organiccarbon was determined by the titrimetric method of Walkley-Black (Nelson and Sommers 1982).The volumes of dead standing trees, down logs and woody debris were calculated for eachplot in the following ways. Volume of downed logs greater than 1 m diameter was estimated byusing mid-point diameter and the length in Huber's formula (Wenger 1984). Under-bark bolevolumes of dead standing trees were estimated from equations in British Columbia Forest Service(1976). No allowance was added for bark or branches. These were present on recently dead trees andbecause of this the mass of dead standing trees is probably slightly underestimated. For broken-43topped stems, the volume was adjusted using a regression equation developed between diameter andheight from live trees in each plot. Predicted volume was calculated using the predicted height fromthis equation, and this predicted volume was reduced by the ratio actual height/predicted height.Stumps were assumed to be cylinders and volumes calculated from the height and mid-pointdiameter measurements. The values for each dead tree, large bole and stump were summed and theplot totals multiplied by four to give volumes in m 3/ha .The estimated mean volume per hectare of logs between 0.15 and 1 m mid-point diameterwas calculated for each decay class using the formula in De Vries (1986), with the 16 lines per plottreated as independent samples. Volume of branches and twigs less than 0.15 m diameter wascalculated in the same way, but not separated into decay class. Mass was estimated by multiplyingthe volume of each class by the mean density of that class. For material less than 0.15 m, the meandensity of decay classes I and II was used.All roots greater than 0.5 mm diameter were removed from forest floor samples and sortedinto coarse (> 2 mm diameter) and fine roots, and forest floor material was weighed wet. Both rootsand forest floor were dried at 70°C for 48 h and re-weighed. If the wet mass of forest floor materialexceeded 1 kg, a subsample was dried to estimate water content, and this was used to correct thesample to oven-thy mass.The quantity of C, N and K contained in woody debris was estimated by multiplying themean mass in each decay class by the mean nutrient concentration for each decay class. Theconcentrations of P obtained in this study were below the detection limits of the chemical analysis,and P concentrations contained in decaying boles of the same species by Sollins et al (1987) wereused to calculate P content in woody debris. For the forest floor material the mean mass of eachlayer at each site was multiplied by the mean nutrient concentration.Differences between the forest types in density and mass of woody debris and nutrientcontent of woody debris and forest floor were investigated using t-tests (Sokal and Rohlf 1981). Ifthe variances were not homogeneous, the tests were done on log-transformed data. Differences inthe total mass, and the mass of the different layers of the forest floor, were investigated using a two44way ANOVA, with the eight replicates in each plot used to test the significance of differences amongthe three sites and the forest type-site interaction. All analyses were performed using SYSTAT(Wilkinson 1990).Results and DiscussionDensity and nutrient concentrationsValues for the density of different decay classes of logs are shown in Table 4.1. Density ofdecay class I hemlock wood was higher in the CH than the HA type, which may be due to a higherproportion of denser late-(summer)wood to early-(spring)wood. Undecayed cedar was less densethan hemlock which is consistent with the known timber characteristics of each species (Jessome1977). The density of decay class V logs was about 50% of the density of decay class I in both foresttypes. The pattern and magnitude of the decline in density with increasing decay generally concurwith other studies that used a similar system of decay classification (Sollins 1982, Sollins et al. 1987,Arthur and Fahey 1990). The exception was the value of 0.23 g/cm 3 for decay class V wood in theCH type which was higher than the values reported in a study of the same species by Sollins et al.(1987). This was probably due to the presence of less-decayed wood within some samples of logsthat were classified into class V on the basis of the external wood being in a highly decayedcondition.Nutrient concentrations in samples of downed logs are also shown in Table 4.1, and those inthe forest floor layers are shown in Table 4.2. Carbon concentrations were reasonably consistentacross all material sampled with an overall mean of 556.8 mg/g (SE 6.51 mg/g), indicating that C isgenerally being lost at the same rate as mass during the decay process. Nitrogen concentrations in thesamples generally increased with increasing log decay in both forest types, and this trend was alsoevident in the woody forest floor material. Phosphorus concentrations in the log samples weregenerally very low (<0.1 mg/g), and beyond the detection limits of the analytical method used.There were slightly higher, and therefore measurable, concentrations in decay class V logs and in thewoody forest floor material. The trend in this latter component was for P concentration to increase45Table 4.1 Density and nutrient concentrations of wood samples of each species from old-growth,cedar-hemlock (CH) and second-growth, hemlock-fir (HA) forests on northern Vancouver Island.Mean and (standard error).DecayClass Species nDensity(g/cmi)C(mg/g)N(mg/g)P(mg/g)K(mg/g)CHI cedar 6 0.36 561.0 0.92 nd 0.16(.039) (10.03) (.11) (0.05)hemlock 8 0.44 524.0 0.68 nd 0.35(.021) (8.57) (0.09) (0.08)II cedar 5 0.31 568.1 0.88 nd 0.35(0.046) (15.3) (0.07) (0.08)hemlock 9 0.42 492.9 0.71 nd 0.29(.067) (9.38) (0.18) (0.20)III 11 0.33 572.6 0.70 nd 0.15(.036) (16.61) (0.1) (0.03)10 0.25 575.2 0.80 nd 0.11(.028) (15.55) (0.09) (0.01)V 10 0.23 577.4 2.41 0.14 0.17(.034) (10.40) (0.72) (0.03) (0.04)HAI hemlock 9 0.34 563.7 0.58 nd 0.36(.034) (8.36) (0.08) (0.12)fir 7 0.34 576.2 0.50 nd 0.17(.036) (13.87) (0.06) (0.04)II hemlock 9 0.34 544.3 0.72 nd 0.31(.031) (6.13) (0.10) (0.15)fir 8 0.32 575.3 0.66 nd 0.19(.021) (18.98) (0.06) (0.06)III 10 0.35 553.9 0.67 nd 0.09(.030) (14.89) (0.11) (0.0)10 0.22 579.3 1.65 0.12 0.12(.032) (9.13) (0.44) (0.02) (0.01)V 10 0.18 571.3 1.84 nd 0.13(.018) (12.50) (0.18) (0.02)nd = none detected, i.e. below the measurement limits of the instrumentation used for analysis.46Table 4.2 Nutrient concentrations (mg/g) of forest floor layers in old-growth, cedar-hemlock (CH)and second-growth, hemlock-fir (HA) forests on northern Vancouver Island. Mean and (standarderror) of 15 samples taken from three sites.CH^ HAC^N^P^K^C^N^P^KNon-woodyL^541.4^6.7^0.6^0.6^534.3^8.3^0.8^0.8(7.96)^(0.14)^(0.00)^(0.02)^(7.36)^(0.13)^(0.04)^(0.01)F^530.4^9.0^0.6^0.7^530.6^9.9^0.7^0.6(9.83)^(0.65)^(0.03)^(0.15)^(6.61)^(0.18)^(0.03)^(0.05)H^522.6^9.7^0.5^0.6^518.7^11.8^0.5^0.3(6.07)^(0.24)^(0.02)^(0.09)^(5.10)^(0.58)^(0.03)^(0.02)WoodyFw^625.8^3.2^0.2^0.2^630.3^3.9^0.3^0.1(9.53)^(0.55)^(0.03)^(0.03)^(17.62)^(0.98)^(0.04)^(0.05)Hw^534.3^5.0^0.3^0.3^559.9^7.8^0.4^0.2(6.2)^(0.72)^(0.03)^(0.08)^(7.8)^(1.08)^(0.04)^(0.02)47with decomposition. P concentrations in non-woody forest floor were considerably higher than inthe woody forest floor, and, in contrast to the woody material, the trend was for the concentration todecrease with the increasing degree of decomposition.Several studies have shown increases in nutrient concentration, particularly N, with stage ofdecay (Lambert et aL 1980, Graham and Cromack 1982, Arthur and Fahey 1990). N is generallyretained by decomposers in this kind of substrate, while C is lost in respiration, and therefore Nconcentration tends to increase during decomposition. However, increases in N concentration havealso been attributed to transfers from other pools such as (i) the translocation of nutrients into thewood from throughfall, litterfall, and the adjacent forest floor by fungal hyphae (Grier 1978), and(ii) N-fixation by bacteria (Hendrickson 1991).Potassium concentrations declined with the stage of decomposition in both woody and non-woody substrates indicating that it is being lost more rapidly than mass. A similar trend has beenobserved in other studies (Lambert et al. 1980, Arthur and Fahey 1990) and this is consistent withthe relatively high mobility of this ion.Woody debris massThe mass of woody debris in each decay class in each forest type is shown in Table 4.3. Thetotal mass of dead standing and downed logs and branches material averaged 363 Mg/ha in the CHtype, significantly higher (p=0.037) than the 226 Mg/ha in the HA stands. Some of this differencewas due to the higher quantity of dead standing trees in the CH stands (80 versus 20 Mg/ha). Themean mass of downed wood in the CH type was 283 Mg/ha, not significantly higher (p = 0.268)than the 205 Mg/ha found in the HA stands.These values are among the highest reported in the literature. Agee and Huff (1987)estimated that the mass of dead standing trees and down logs was 550 Mg/ha in an old-growthPseudotsugahlemlock stand in the Olympic Mountains of Washington, USA; and Harmon et al.(1987) estimated a mass of 400 Mg/ha in a riparian Sequoiadendron giganteum stand in California. Acomparable figure of 200 Mg/ha of downed logs only was reported by Bingham and Sawyer (1988)for a Californian Sequoia/Pseudotsuga/Tsuga forest.Table 4.3 Mass (Mg/ha) of dead standing trees, downed logs and small woody debris in old-growth, cedar-hemlock (CH) and second-growth, hemlock-fir (HA) forests at three sites on northern Vancouver Island.TYPESite:CH HA P1BL RU SC MEAN BL RU SC MEANDead Standing 19.6 56.4 164.3 80.1 16.2 24.1 22.5 20.9 0.182Downed Logs(>0.15m)Decay Class: I 7.5 10.0 6.6 8.0 17.6 9.4 6.8 11.3 0.397II 1.3 27.2 11.1 13.2 8.3 17.2 5.0 10.2 0.736III 206.9 139.9 83.2 143.1 43.5 16.1 49.7 36.4 0.046IV 166.2 43.9 44.3 84.8 42.9 144.5 86.9 91.4 0.898V 15.6 33.2 29.9 26.2 16.1 44.6 56.0 38.9 0.386Woody debris(0.01-0.15 m) 2.1 13.2 6.4 7.2 17.0 21.3 11.0 16.4 0.105TOTAL 419.2 323.8 345.8 362.6 161.6 277.2 237.9 225.5 0.0371 level of significance of between forest type differences using a t-test.49Within the CH type, there were quite large differences in the mass of different categories ofmaterial between sites. At the "SCHIRP" site, the quantity of dead standing trees (mainly cedar) washigh while that of down logs was quite low. At the "Beaver Lake" site, on the other hand, the massof dead standing material was small but the mass of down logs was very high. The third site,"Rupert", had intermediate values between these two. These differences may reflect differences inthe way the catastrophic windstorm which created the HA stands affected the adjacent old-growthforest. Variability in quantity of woody debris in the HA stands suggests that the structure of thepre-1906 windstorm stands was quite variable. Although it was a small proportion of the totalwoody debris, the mass of small woody debris (<15 cm diameter) was higher (p= 0.105) in the HAthan the CH type. This is indicative of the greater input of this size material through branchmortality and stand-self thinning in the HA type. This is also reflected in the significantly higher(p=0.001) quantity of small woody material (<1 cm) in the HA type measured in the L layer as partof the forest floor material (Table 4.4).Forest floor massThe mass of the different layers of forest floor are shown in Table 4.4. The mean mass of thenon-woody L, F and H layers was significantly higher (p = 0.012) in the CH stands (113.7 Mg/ha)than in the HA stands (77.2 Mg/ha), and woody material composed about 60% in both types: 166Mg/ha in the CH stands and 134 Mg/ha in the HA stands.The overall (woody + non-woody) massof forest floor in the CH was 280 Mg/ha, not significantly higher (p=0.264) than the HA stands(211 Mg/ha).There have been few studies of the biomass of forest floor in similar forest types with whichto compare these figures, and none that used similar biomass categories to those in this study. Gesseland Balci (1965) reported an average value of 157 Mg/ha for the LFH of five old-growth coniferousstands in Washington state, USA. Vogt et al. (1983) reported a forest floor mass of 149.5 Mg/ha inan Abies amabilis stand in western Washington, which is probably comparable to the second-growthHA stand in this study.50Table 4.4 Mean mass of forest floor layers in old-growth, cedar-hemlock (CH) and second-growth,hemlock-fir (HA) forests at three sites on northern Vancouver Island.CH HAp1PBL RU SC MEAN BL RU SC MEANNon-woodyL 4.3 4.3 4.3 4.3 4.4 4.3 3.7 4.1 .667F 13.2 19.1 22.0 18.1 21.3 32.5 36.6 30.1 .001H 90.0 114.3 69.7 91.3 33.3 56.2 39.6 43.0 .013SUM 107.5 137.7 96.0 113.7 59.0 93.0 79.9 77.2 .012WoodyLw 1.1 1.5 2.7 1.7 3.6 3.0 4.6 3.7 .001Fw 0.3 88.9 0 29.7 7.7 19.3 50.6 25.9 .847Hw 177.4 91.6 134.9 134.6 125.8 84.1 102.4 104.1 .451SUM 178.8 181.9 137.6 166.1 137.1 106.4 157.6 133.7 .682Total 286.3 319.7 233.6 279.8 196.1 199.4 237.5 210.9 .2641 level of significance of differences between forest types using a two-way ANOVA.51In a broad survey of coniferous forest floor biomass in Oregon and Washington undertakenby Little and Ohmann (1988), the highest values, ranging from 181 to 277 Mg/ha, were found atthree sites in the Olympic National Forest in western Washington, and were comparable with thosein this study. Decaying wood accounted for 5-70% of the forest biomass at the full range of sites inthat study, indicating that the value of 63% found in this area is at the upper end of the range forthat geographic region.Root biomassThe mass of roots in different categories of the forest floor is shown in Table 4.5. The meanmass of fine roots (< 2 mm diameter) in the two forest type was similar, with an average of10.2 Mg/ha in the CH type and 9.0 Mg/ha in the HA. The mean mass of coarse root (> 2 mmdiameter) samples in the CH type (61.9 Mg/ha) was over double that in the HA type (28.3 Mg/ha),but the variability associated with these estimates was high, and the difference was not significant(p=0.120). About 90% of the coarse roots in both forest types were in the H and Hw layers. Seventysix percent of the fine roots were in the H and Hw layers in the CH type, but only 56% in the HAtype. There did not appear to be a strong preference for rooting in non-woody substrates to accessnutrients in the CH type - about 60% of fine roots were in non-woody F and H. In the HA type,there was more of a preference and about 70% of fine roots were found in non-woody substrates.The mass of roots was generally higher than the average live fine root values of 5.0 Mg/hafor 'cold temperate needle leaf evergreen' forest types reported in Vogt et al. (1986). No live fineroot figures were available in Vogt et al.'s review for the warm temperate needle leaf evergreen type,but the average total (live+dead) fine root biomass in that type was 19.1 Mg/ha. Total fine root massin the cold temperate needle leaf evergreen type was 7.3 Mg/ha, which is in the vicinity of the lowervalues obtained in this study. The roots sorted from the forest floor in this study were notscrutinised intensively to determine if they were alive or dead, and the total figures are probablymore comparable.52Table 4.5 Mass (in Mg/ha) of fine (< 2 mm) and coarse (>2 mm) roots in the forest floor layers inthe CH and the HA forest types on northern Vancouver Island. Means and (standard errors).fineCHcoarseHAfine^coarsep1Pfine coarseNon-woodyF 1.8 (0.15) 0.7 (0.29) 3.6 (0.51) 2.8 (1.94) 0.028 0.332H 4.4 (1.01) 25.5 (3.14) 2.7 (0.63) 10.1 (8.17) 0.218 0.154WoodyFw 0.3 (0.31) 2.4 (2.40) 0.4 (0.29) 0.3 (0.27) 0.776 0.440Hw 3.4 (2.56) 32.1 (13.40) 2.3 (1.12) 15.0 (2.40) 0.704 0.276Total 10.2 (2.16) 61.9 (14.31) 9.0 (2.02) 28.3 (9.25) 0.713 0.1201 p values for differences between forest types using Student's-t test.53Detrital pool sizesThe total mass and nutrient content of detritus in each forest type is shown in Table 4.6.Both forest types had large amounts of biomass in detrital pools: 644 Mg/ha in the old-growth CHstands, and 436 Mg/ha in the second-growth HA stands. Most of the detritus was woody (531Mg/ha in the CH type and 359 Mg/ha in the HA type). These large accumulations of wood in bothforest types may be due to the prevalence of windstorms, which result in a large number of tree bolesbeing deposited on the forest floor. Western red cedar wood is also high in extractives and secondarycompounds making it resistant to decomposition (Minore 1983). Early and late successional stagesalso generally have greater quantities of woody debris than intermediate stages (Spies et al. 1988).Both of these factors may explain the higher quantities of woody forest floor in the CH stands.The nutrient content of this detrital biomass was high: 2178 kg/ha of N in the CH and2049 kg/ha in the HA stands. These values are comparable to the reported values of 2040 kg/ha ofN for the old-growth stands in Washington state (Gessel and Balci 1965), and 2 260 kg/ha of N in a115-year-old Picea stand in Germany (Cole and Rapp 1980). Woody material in and above theforest floor contained 51-59% of the N, and 58-61% of the P. Arthur and Fahey (1990) suggestedthat the role of dead wood in forest nutrition may be less significant than expected simply on thebasis of its mass. However, when nutrients accumulated in woody forest floor material are included,decaying wood is a major component of the nutrients stored in detritus in the ecosystems in thisstudy.This detritus is also a considerable sink for carbon, with 357 Mg/ha in the CH and 246Mg/ha in the HA stands. Biomass accumulation and storage of carbon in forest floor pools generallyincreases from the tropics to the boreal regions, with intermediate values from intermediatelatitudes, in a pattern largely determined by global temperature variations (Vogt et al. 1986,Schlesinger 1991). The larger than expected accumulation in this area may result from thecombination of windstorms depositing large amounts of wood to decay on the forest floor, the cool-wet climate resulting in relatively slow decomposition, and the large size and decay resistance ofwestern red cedar boles.Table 4.6 Mean mass and nutrient content of four detrital pools in old-growth, cedar-hemlock (CH) and second-growth, hemlock-fir (HA)forests on northern Vancouver Island.CategoryCH HAMass(Mg/ha)C(Mg/ha)N(kg/ha)P(kg/ha)K(kg/ha)Mass(Mg/ha)C(Mg/ha)N(kg/ha)P(kg/ha)K(kg/ha)Dead standingand stumps80.1 45.2 78 8.3 20.0 20.9 11.7 23.4 2.2 3.2Logs andwoody debrisforest floor(woody)forest floor(non-woody)282.5166.1113.7161.191.459.6254.0769.41 077.327.946.459.143.946.870.0204.6133.777.2116.676.740.5271.3915.3839.420.749.545.928.927.934.2TOTAL 642.4 357.3 2 178.6 141.7 180.7 436.3 245.5 2049.4 118.3 94.255ConclusionsThis study indicated that (i) there was generally a greater amount of woody and non-woodybiomass in the CH than the HA type, but that the detrital biomass accumulated in both forest typeswas relatively high compared to other forest ecosystems; and (ii) the N capital in detrital pools in thetwo forest types was similar. Because there is a relatively large amount of forest floor biomass in bothtypes, it is unlikely that the physical effects of organic matter accumulation - that is , lower soiltemperature and higher moisture, can explain differences in decomposition, rates of nutrient cyclingand nutrient availability between the two types.56Chapter 5. Mineral soil propertiesIntroductionSoil pedogenesis is a function of the geological parent material, climate, topography, thevegetation cover and time (Jenny 1981). Forest structure, functioning and productivity can also bestrongly influenced by topographic position, and variability in mineral soil properties imposed bythese characteristics.The topography in the study area is relatively subdued, and the two forest types underinvestigation occur on well-drained to somewhat-imperfectly drained sites in an intricate matrix withlow-lying, poorly drained areas dominated by wetland communities containing species such asSphagnum and Myrica spp., and swamp-forest complexes with western red cedar and lodgepole pine(Pinus contorta var. contorta). The CH and the HA type are both found on the range of geologicalsubstrates that occur within the study area, and examples of the two types can be found in similartopographic positions (Lewis 1982). There is no obvious climatic variation between the two types.Differences in mineral soil at the three sites where detailed investigations were undertaken mayreflect subtle differences in topography and drainage between the two types, differences in theincidence of windthrow, or differences in vegetation composition occupying the two types. deMontigny (1992) suggested that the presence of hardpans and compacted mineral soils couldcontribute to periods of anaerobic conditions in soil under the CH type, and Lewis (1982)considered that soils under the HA type had a more friable consistence indicative of periodicwindthrow cultivation, whereas the CH soils had a firmer consistence, which influenced air andwater movement, soil temperatures and, as a result, root distribution and organic matter turnoverrates.In this Chapter the morphology of the mineral soils, and measurements of bulk density, soiltexture and soil chemical characteristics are reported. The objectives of this part of the study were toassess differences in mineral soil properties that may be the cause of differences in forest composition57and functioning, and to assess features, such as bulk density, structure, consistence, texture or soilchemical characteristics that may reflect differences in the disturbance regime between the two foresttypes. These investigations relate to the hypothesis that differences in past disturbance regime are thecause of differences in nutrient availability.MethodsMorphological propertiesAt four randomly selected points in each of the plots described in Chapter 2, a soil pit wasexcavated to a depth of 1 m, or to an impenetrable layer. Principal horizons, reflecting obviouspedological discontinuities such as density, particle size or organic matter content, were identifiedand the following morphological properties described: layer thickness, coarse fragment content, soiltexture, structure, consistence, colour, mottles, and roots (Ballard 1989). Bulk density samples weretaken at intervals of 30 cm from the soil surface to the pit bottom, using a metal cylinder (approx.330 cm3) hammered into the side of the pit.Samples of each layer and bulk density samples were taken to the laboratory and moistcolours determined. The samples were then air dried for 7 days and dry colours determined. Bulkdensity samples were separated into coarse and fine fractions using a 2 mm sieve and weighed. Asub-sample of the fine fraction was oven-dried at 105°C for 24 h to determine the moisture content,and the weight of the fine fraction was adjusted accordingly. Bulk density was calculated bysumming the weight of the coarse fraction and the adjusted weight of the fine fraction and dividingby the volume of the sampling cylinder. The fine fraction from about 3 bulk density samples fromeach depth in each plot were analysed for silt and clay content using the hydrometer method (Geeand Bauder 1982), after first removing organic matter by digestion in sodium hypo-chlorite, andcentrifuging. The sand fraction was determined after sieving the soil solution through a 53 micronscreen and drying the residue at 105°C for 24 h.58Soil chemical propertiesApproximately 500 g samples of the 0-15 cm layer of the mineral soil were taken from 10systematically-located points in each plot, before the soil pits were dug. A small hole was dugthrough the thick humus accumulation, and the mineral soil removed. Samples from deeper in theprofile were taken from the fine fraction of those bulk density samples analysed for particle sizedistribution (see above). Samples were air dried for 14 days, total N was determined colorimetricallywith a Technicon autoanalyser following a Kjeldahl digestion (Bremner and Mulvaney 1982),available P was determined by spectrophotometer following acid ammonium fluoride extraction(Olsen and Summers 1982), and K, Ca, and cation exchange capacity, by the ammonium acetatemethod (Bremner and Mulvaney 1982). Fe and Al were determined after extraction with sodiumpyrophosphate (Bascomb 1968). Samples from the deeper layers of the profile were analysed for C,N and P only.Statistical analysesDifferences in soil physical and chemical properties between the two forest types were testedby comparing mean values for each type from the three sites using Student's t-test (Sokal and Rohlf1981).Results and DiscussionMineral soil morphologyGeneralised descriptions of the soil pit information for the two forest types at the three sitesare shown in Tables 5.1-5.3. The Beaver Lake and the SCHIRP sites were within a few kilometresof each other, and occurred on similar gently undulating landforms characteristic of the SuquashBasin, were underlain by a compacted layer of till, of peri-glacial or fluvio-glacial origin, derivedfrom andesitic parent material of the Bonanza group. This compacted layer formed a strongly-cemented duric horizon at the base of the podzolic B, 20-120 cm below the top of the mineral soil.The Rupert site bordered the Suquash Basin, and soils formed directly above harder basaltic materialof the Karmutsen formation (Lewis 1982).Table 5.1 Morphology of mineral soils under the CH and the HA type at the Beaver Lake site. Values are the range from 3 or 4 soil pits ineach type at each site.^Type Horizon Mean^Range Colour moist(dry)^Structure^Consistence^Roots^Rooting^Comments^Width^(cm)^ Depth(cm) (cm) CH^LFH^34-0 20-56Bhf^0-8^0-10^5YR 3/3^weak, fine-^ few fine,few^50(5YR 4/6)^sub-angular mediumblockyBf^8-38^10-50^7.5YR 5/6^n MF2^plentiful fine,few(7.5YR 6/8) WS1,WP1 med.BCc^38+ 10YR 4/4 strong,coarse^ none(10YR 7/4)^platyimpermeable durichorizonGently sloping mid-lower slope (5%), soils in pits further up the slope were moderately well-drained and moderately pervious, the one further down wasimperfectly-drainedand moderately pervious, duric Bc horizon derived from glacial tillHA^LFH^18-0^2-40^ -Bhf 0-8^0-10 lOR 2.5/2^massive^MF1-2^few fine,few^100^most roots in top 20(2.5YR2.5/2) medium cmBE^8-78^15-^upper: 7.5YR 5/8^med. fine^WSO-1, WP1 few fine,few med.130 (10YR 6/8)^angular plent. coarselower: 10YR 5/8^blocky(10YR 6/8)BCc^78+^10YR 4/4 strong,coarse^-^none^impermeable duric(2.5Y 7/4)^platy horizonGently sloping crest (<5%), moderately well-drained and moderately pervious soils. Duric Bc horizon derived from glacial till. Thin (1-2 cm) Ae presentin some cases, but generally absent. 5 cm weathering layer above duric horizonTable 5.2 Morphology of mineral soils under the CH and the HA type at the Rupert site. Values are the range from 3 or 4 soil pits in eachtype at each site.Type Horizon Mean Range Colour moist(dry)^Structure^Consistence^Roots^Rooting^CommentsWidth (cm)^Depth(cm)^(cm)CH LFHBf^20-0^0-60^0-70^0-120 upper:5YR 4/6(7.5YR 6/6)lower: 7.5YR 5/6(10YR 6/8)moderate, finesub-angular,blockyMF2^few fine, very fewWS1, WP IMF2^mediumWS1, WP1100 C 70+ volcanic bedrockFlat to hummocky topography, shedding position. Microtopography largely caused by varying surface of volcanic bedrock. Soil development quite variable, onepit was a follisol with organic matter over bedrock, another had a well-developed B horizon over 1 m deep. Soils generally well-drained and moderately pervious.Despite the shedding situation drainage was variable because of the impermeable bedrock with water accumulating in lower situations.HA^LFH^13-0^4-24Bhf 0-6^0-11 5YR 3/2 (5YR 4/2)Bf^6-60^0-120^upper:7.5 YR 5/6(7. 5YR6/8)lower:5YR 4/6(7.5YR6/6)mod. fine sub-angular blockyweak, fineangular,blockyMF3,^few fine, plent. med.WS I , WP IMF2^few fine, plent. med.WS1,WP1120C volcanic bedrockModerate mid-slope (10%). Microtopography largely caused by varying surface of volcanic bedrock. Soil development quite variable, Soils generally well-drainedand moderately pervious. Some mottling, particularly at the border with bedrock.HA^LFH^28-0^5-47^Ae 0-4^0-5^5YR 6/4(10YR 3/2)BE'^4-41^20-50^upper: 7.5YR4/4(10YR 4/4)lower: 10YR 5/4(2.5Y 6/4)BCc^41+^2.5Y3/2 (2.5Y6/2)weak,massivemod., fine,sub-ang.blockystrong,coarse platyplent. fineMF2, few fine, no med.WS 1-2,WP1-240 none Duric horizonTable 5.3 Morphology of mineral soils under the CH and the HA type at the SCHIRP site. Values are the range from 3 or 4 soil pits ineach type at each siteSite Horizon^Mean^Range Colour moist(dry)^Structure^Consistence^Roots^Rooting^CommentsWidth^(cm)^Depth(cm) (cm)MF2,WS1, WP1MF1-2,WS2, WP2MF3,WS2, WP2CH^LFH^30-0 10-40Bhf^0-10^0-17Bf^10-50^13-60Bfg^35-50^13-17BCc^50+5YR 3/3^weak,(10YR 6/2)^massiveupper: 7.5YR4/4^weak,(10YR 5/4)^massivelower: 10YR 5/4(10YR 5/8)10YR 3/2 It(7.5 YR 4/2)10YR 5/4^strong,(2.5Y 5/4)^coarse platyfew-plent. fine& med.few fine, nomed.nonenone40^some mixing ofhorizonspresent only inone pitDuric horizonGently-sloping mid-slope (<5%). Moderately pervious, imperfectly-drained. Soil greyish yellow, yellow-red at depth. Water table fluctuating,present throughout the growing season in one pit. Thin weathering layer, of similar colour, just above the duric horizon.Gently-sloping mid-slope (<5%). Moderately pervious, imperfectly-drained. Soil greyish yellow and mottled. Water table fluctuating. Thinweathering layer, of similar colour, just above the duric horizon.62The soils in both forest types fell into the same classification. Those at Beaver Lake andSCHIRP were duric Humo-Ferric Podzols; and the Rupert soils were orthic Humo-Ferric Podzols(Agriculture Canada 1987). An Ae horizon was generally absent. The B horizon was generally auniform Bf that varied little in colour or texture through the depth of the profile. In some pits therewas a relatively thin (up to 15 cm), darker Bhf indicative of organic accumulation near the top ofthe profile. The structure of the Bf was generally weak; sub-angular blocky at the Beaver Lake andRupert sites, and more massive under both forest types at the SCHIRP site. Consistence wasgenerally friable, slightly sticky and plastic when wet.There were some general differences in soil depth, colour and drainage between the threesites, but there were no consistent differences in morphology between the CH and HA forest types.At the Beaver Lake site the CH was in a water-receiving position, whereas the HA was slightly upslope in a more shedding situation, which caused some differences in drainage between the types,however the colours in the profile of the CH soils did not indicate any gleying due to prolongedsaturation or reducing conditions. At the SCHIRP site both types were in water-receiving positions,and both had soil colours indicative of extensive periods of waterlogging. At the Rupert site bothtypes were in shedding positions, although the HA was on a steeper slope than the CH type. Becauseof the impermeability of the bedrock, water accumulated throughout the year in one pit at a lowlying point in the CH type.Soil depth was quite variable, particularly at the Beaver Lake and Rupert sites. At the Rupertsite this was probably due to the variability in the surface of the underlying bedrock, which has ledto soil accumulation in depressions. At both these sites, and in both types, tree windthrow hascontributed to the variability by creating mounds and pits and redistributing soil within the area.Soils were generally thinner and less variable at the SCHIRP site, perhaps because of its lower-lyingposition. There was little rooting in the mineral soil at this site, and therefore windthrow would nothave the same impact on the mineral soil at the SCHIRP site as it has at the other two sites wherethere was more rooting through the profile.63Lewis (1982) suggested that a higher incidence of windthrow in the HA type may lead todifferences in structure and consistence between the two types, but there was little obviousdifference in these attributes in this investigation. There was also no evidence of the formation ofhardpans within the B horizon that de Montigny (1992) suggested might lead to greaterwaterlogging in the CH type. Fine roots (<2 mm diameter) were observed throughout the profile inboth types at all sites, although they were generally few in number in the mineral soil, most rootsbeing located within the LFH layer (Table 4.5).A BCc horizon was present at the base of the B horizon in both forest types at the BeaverLake and SCHIRP sites. This horizon had a strong, coarse-platy structure, and an olive-grey colourwith some darker, reddish mottles. At these sites there was generally a thin (up to 5 cm) zone ofactive weathering immediately above the duric horizon, that was of similar colour to the BCc buthad a different structure.Classical concepts of the genesis of podzolic soils are based on the mobilisation andmovement within the solum of sesquioxides. This mobilisation is associated with the dissolution ofiron and the formation of soluble organo-metallic complexes in the upper mineral layers. Thesecomplexes are thought to be immobilised in the B horizon through the decomposition of organicligands and the precipitation of metallo-organic complexes by overloading with sesquioxides(Pritchett and Fisher 1987). However, the general absence of the Ae in podzols on Vancouver Islandled Lewis (1976) to postulate an extension of the podzolisation concept to include in-situ weatheringof the parent material. This weathering results in losses to drainage water of bases and silica, and theresidual enrichment of sesquioxides in the B horizon. 'Anti-horizonation processes' resulting fromturbation of the solum by tree windthrow and slope processes, further mask and limit the processesof eluviation, transfer and precipitation. These factors explain the general absence of an eluviated Aehorizon in the profiles of the sampled soils.Soil physical propertiesBulk density, coarse fragment content and percent sand, silt and clay are shown in Tables 5.4and 5.5. Mean bulk densities in the 0-30 cm depth ranged from 0.58 to 0.94 g/cm 3 . Those fromTable 5.4 Bulk density coarse fragment content, and particle size distribution (in the fraction smaller than 2 mm in diameter) of mineralsoils under the CH forest type at three sites on northern Vancouver Island. Values are the means and standard errors (in parentheses) from3 or 4 soil pits at each site.Site Depth(cm) nBulkdensity• /cm)Coarsefragment(% mass)sand(%)silt(%)clay(%)texture classBL 0-30 7 0.69(0.12) 40(2.6) 2 40(2.6) 39(1.9) 22(0.7) loam0-60 3 0.80(0.11) 44(6.2) 1 42 36 23 loamRU 0-30 6 0.62(0.13) 41(5.9) 3 37(3.6) 39(2.6) 24(1.1) loam30-60 3 0.58(0.29) 35(3.0) 1 47 34 18 loam60-90 1 0.82SC 0-30 5 0.83(0.04) 21(6.2) 4 47(5.9) 33(3.7) 21(2.3) loam/SL30-60 1 0.68 49 1 37 35 27 loam/CLMEAN 0-30 3 0.71(0.06) 34(6.6) 3 41(2.8) 37(2.0) 22(1.0)30-60 3 0.69(0.12) 43(4.2) 3 42(2.9) 35(0.4) 23(2.7)Table 5.5 Bulk density coarse fragment content, and particle size distribution (in the fraction smaller than 2 mm in diameter) of mineralsoils under the HA forest type at three sites on northern Vancouver Island. Values are the means and standard errors (in parentheses) from3 or 4 soil pits at each site.Site Depth(cm) nBulkdensity(g/cm3)Coarsefragment(% mass)sand(%)silt(%)clay(%)texture classBL 0-30 6 0.61(0.11) 44(2.8) 4 38(2.2) 42(1.7) 21(2.1) loam30-60 5 0.63(0.03) 44(1.3) 4 39(0.8) 38(2.6) 22(2.4) loam60-90 1 0.61RU 0-30 10 0.58(0.07) 40(3.5) 4 37(2.8) 39(0.6) 24(2.2) loam30-60 4 0.74(0.07) 42(9.4) 2 35(3.2) 41(0.8) 23(4.0) loam/CL60-90 1 0.88SC 0-30 5 0.94(0.12) 45(9.2) 4 42(1.7) 33(1.8) 24(0.8) loam30-60 0 1MEAN 0-30 3 0.71(0.12) 43(1.7) 3 39(1.7) 38(2.4) 23(1.1)30-60 3 0.68 43(0.8) 3 37 40 230-30 0.809 0.986 0.514 0.653 0.60830-60 0.978 0.952*p values for differences between the CH and the HA type using student's t test. Test was performed on log-transformed values for the 30-60 cm depth. n is the number of observations.66the 30-60 cm depth ranged from 0.58 to 0.82 g/cm 2 . The SCHIRP site had higher bulk densities inthe 0-30 layer than the other two sites, perhaps because of its somewhat restricted drainage andlower lying position, which has led to a slower rate of soil development than the other two sites.There did not appear to be a difference in bulk density with depth, and there was no significantdifference between forest types for either depth. The relatively low bulk densities are typical ofnorthern Pacific coastal forest soils (Alexander et al. 1993).Particle size distribution of the fraction below 2 mm diameter was relatively uniform acrossall sites and between the two forest types. Sand content was 40-45%, silt 33-42% and clay 21-24%.Given these values, the texture of this fine fraction was fairly consistently classified as loam, although2 samples fell into the clay-loam and 1 into the silty-loam category. These values are within therange for till-derived podzols on Vancouver Island reported by Lewis (1976), although the claycontent was in the high end of the range for all sites.Soil chemical propertiesSoil chemical properties are shown in Tables 5.6 and 5.7. Organic C, total N and available Pgenerally decreased down the profile. For the samples from the 0-15 cm depth, organic C (mean 14-15%), total N (0.27-0.38%) and available P (2.71-4.34 ppm) were considerably higher than thevalues for deeper in the profile, and higher than those reported for a similar soil by Lewis (1976).This was probably due to difficulties involved in taking samples from the upper mineral horizons,that may have resulted in some of the lower portion of the H layer material being included. Thesamples from deeper in the horizon were collected from the sides of the soil pits and are probablymore representative of the mineral soil itself. There were was a higher concentration of total N inthe 0-15 cm layer in the HA than the CH type (0.38 vs. 0.27%, p=0.027) and in the 15-30 cmlayer (0.30 vs. 0.13%, p=0.127). This was not the case for organic carbon concentrations andconsequently the C/N ratio was generally lower in the HA type (40 vs. 52 in the 0-15 cm, and 12vs. 38.5 for the 15-30 cm layer).Table 5.6 Chemical properties of mineral soils under CH forest type at three sites on northern Vancouver Island. Values for the 0-15 cmdepth are the mean of about 10 samples from each site, those for the lower horizons are the mean of samples taken from 3 or 4 soil pits. nis the number of observations.Site HorizonDepthnOrganicC (%)TotalN (To)C/N Avail P(ppm)CEC(me/100g)K(me/ 100g)Ca(me/100g)Fe(%)Al(%)BL 0-15 10 11.1 0.28 40 3.21 54.4 0.17 1.48 1.0 0.9415-30 2 2.3 0.08 28.8 0.7630-60 1 2.7 0.08 33.8 1.45RU 0-15 9 15.2 0.24 63.3 4.28 52.3 0.28 4.29 0.97 0.3815-30 3 9.0 0.18 50.0 3.5230-60 1 1.2 0.05 24.0 1.60SC 0-15 9 16.1 0.29 55.5 5.54 58.8 0.22 2.67 0.73 0.5315-30 3 3.7 0.14 26.4 1.6130-60 1 6.4 0.22 29.1 3.13MEAN 0-15 3 14.1 0.27 52.2 4.34 55.2 0.22 2.81 0.90 0.6115-30 3 5.0 0.13 38.5 1.9630-60 3 3.4 0.12 28.3 2.06Table 5.7 Chemical properties of mineral soils under HA forest type at three sites on northern Vancouver Island. Values for the 0-15 cmdepth are the mean of about 10 samples from each site, those for the lower horizons are the mean of samples taken from 3 or 4 soil pits. nis the number of observations.Site HorizonDepthnOrganicC (%)TotalN (%)C/N Avail P(ppm)CEC(me/100g)K(me/100g)Ca(me/100g)Fe(%)Al(%)BL 0-15 10 18.2 0.38 47.9 3.20 77.7 0.27 2.77 1.19 0.7515-30 4 3.3 0.47 7.0 0.8030-60 4 3.4 0.11 30.9 0.97RU 0-15 10 12.9 0.33 39.0 1.35 57.2 0.15 1.05 1.21 0.9315-30 4 3.0 0.21 14.2 1.2330-60 2 3.1 0.10 31.0 1.16SC 0-15 10 14.4 0.43 33.5 3.60 66.9 0.18 2.70 1.21 0.9415-30 4 4.6 0.23 20.0 1.7130-60 1 4.6 0.19 24.2 3.18MEAN 0-15 3 15.2 0.38 40.0 2.71 67.2 0.20 2.17 1.21 0.8715-30 3.6 0.30 12.0 1.2530-60 3.7 0.13 28.4 1.770-15 0.665 0.027 0.168 0.124 0.671 0.553 0.026 0.22015-30 0.550 0.127 0.45030-60 0.877 0.794 0.760* p values for the differences between the CH and HA type.69C and N in soil are derived from the decomposition of detritus from the vegetation cover. Thesedifferences between the types are consistent with differences in the chemical characteristics of thehumus under the two types that have been reported previously (Chapter 4, Prescott et al. 19936, deMontigny et al 1993). Thus, it appears that these between-type differences have persisted longenough to be expressed in the organic matter contained within the mineral soil.Pyrophosphate-extractable Fe was higher in the HA (1.21%) than in the CH type (0.9%,p=0.026); this could indicate that more active weathering is occurring in the HA type, or differencesin functional groups in organic ligands may mean that more Fe is being brought into solution in theHA type. For example, de Montigny et al. (1993) found a higher ratio of vanillic acid to vanillin inforest floor material from the HA type, and the acid would be a more effective ligand than thevanillin. Mean available P in the 0-15 cm layer was higher in the CH type (4.34 ppm) than the HAtype (2.71 ppm), although this difference was not significant (p=0.168) and was lower for deeperlayers.ConclusionsThe results in this Chapter indicated that there was considerable variability in the depth ofmineral soil developed within each site, and some variability between the sites in mineral soilcharacteristics. However, there was little consistent difference between the two forest types thateither (i) indicated that differences in topographic position and soil drainage could explain thedifferences in nutrient availability and productivity, or (ii) indicated that there had been majordifferences in anti-horizonation processes such as disturbance due to windthrow. N concentrationwas higher, and the C/N ratio was lower, in the upper mineral soil was higher in the HA type,suggesting that differences in humus chemistry between the two types has persisted long enough tobe expressed in soil organic matter.70Chapter 6. Effects of soil mixing on soil properties and understorey vegetationIntroductionPedoturbation, or soil disturbance and mixing, is an important pedogenic process that occursnaturally in forest soils through the activities of animals or plants, or through physical mechanisms,such as freezing and thawing, or swelling and shrinking of clays (Johnson et al. 1987). Disturbanceby ploughing or some other form of cultivation is also a common site preparation practice prior toestablishment of trees in plantations. This disturbance aims to increase early growth of plantedseedlings by creating more favourable conditions of moisture, aeration and nutrient availability, anddecreasing competition from non-crop vegetation (Burger and Pritchett 1988).Pedoturbation due to tree-falls is a widespread phenomenon in forests (Lyford and MacLean1966, Dunn et al. 1983, Cremeans and Kalisz 1988), that has a significant impact on many aspectsof forest ecosystem functioning (Schaetzl et al. 1989, Nakashizuka 1989, Peterson et aL 1990).Tree falls create a variety of soil microsites within a stand, depending on whether the trees snap off,the roots are uplifted and form a hinge, or the root ball rotates within the soil profile (Beatty andStone 1986). Uplifting or rotation of the roots of larger trees results in a major disturbance to thesoil profile, with an effect similar to mechanical ploughing (Armson 1977). Tree-falls due to wind(windthrow) has been suggested as an important mechanism for maintaining forest productivity insome areas. For example, in the coastal forests of south east Alaska it is considered to be the onlyform of disturbance occurring often enough to keep soils in a juvenile, or semi-mature stage ofdevelopment. Mature soils in this region are not as favorable for tree growth, because of the negativeeffects of podzolisation on nutrient immobilisation, and the development of impermeable horizons(Ugolini et al. 1990).Differences in nutritional conditions between the CH and HA types have been attributed tothe history of windthrow in the HA stands. It has been hypothesised that windthrow improves thephysical condition of the soil and increases nutrient availability by periodically mixing organic71horizons with mineral soil (Lewis 1982, de Montigny 1992), in a similar way to the changesobserved in coastal Alaska .In 1988 the staff of Western Forest Products Ltd. established an experiment that intended tosimulate the effect of extensive windthrow by mixing (cultivating) the mineral and organic soilsusing a large rake attached to an excavator. The experiment was conducted in an area that had beenclearcut and slashburnt on a site that previously contained examples of both forest types. Thischapter reports a study that compared a range of soil properties, and the biomass of understoreyplants, in clearcut, clearcut and mixed, and uncut sites of the two forest types, and provided a test ofthe hypothesis that the higher nutrient availability in the HA type is due to the effects of broadscalewindthrow.MethodsExperimental design and treatmentsThe experiment was established in a 97 ha area containing a mosaic of the two forest types.The area was clearcut during 1986 and slashburned in the spring of 1987. The mixing treatmentwas part of a larger experiment aimed at investigating the growth of cedar and hemlock seedlingsplanted at three densities, with and without fertilization, and with and without soil mixing, on thetwo forest types. In total, 32 plots (16 in each forest type) had the soil mixed, and half of these wereplanted with cedar and half with hemlock at 2500 stems per hectare. The mixing treatment wascarried out in January 1988 with a 3-tyned rake attached to a Caterpillar 215 excavator. The forestfloor and mineral soil were thoroughly mixed, salal rhizomes were removed by hand from thedisturbed soil, and the slash was redistributed to facilitate planting.Soil sampling and other field measurements in this study were carried out on 22 plots, 16 inthe clearcut and burnt experimental area: 4 unmixed (the CC treatment) and 4 mixed (the CCMtreatment) in each of the two forest types; and on the three surrounding sites containing uncutstands of the CH and HA forest types described in Chapter 2 (the UC treatment).72Soil propertiesSoil samples were collected in July 1992, 4.5 years after treatment. Five samples werecollected from each plot. Near each sample point, the vegetation cover and litter were removed and2 samples of the 0-20 cm soil layer were combined. Soils from the unmixed and uncut plots werelargely organic, while those from the mixing treatment had a higher mineral content. Samples werekept in coolers in the field and transported to the laboratory within 4 days. In the laboratory largerroots and stones were removed and organic samples were passed through a 5 mm sieve;predominantly mineral samples were passed through a 2 mm sieve. After sieving, samples werestored at 4°C until they were analysed.Extractable nitrate, ammonium and phosphate were determined colorimetrically using aTechnicon Autoanalyzer, after shaking 5 g of fresh mass of each sample in 100 ml of 2 M KC1 forone hour. Mineralisable nitrate and ammonium were determined by subtracting the initialextractable amounts from the amounts extracted after an anaerobic incubation of 5 g of fresh massin 25 ml of distilled water for 7 days at 30°C. Soil moisture content was estimated after oven dryinga subsample at 105°C for 24 h, and, organic matter content by the loss on ignition after 4 h at500°C. Soil pH was measured on one bulked and homogenised sample of the 5 samples from eachplot using a glass electrode in a 1:4, soil:water mixture.Nutrient availability was measured using ion-exchange resin bags. In April 1992, bagscontaining cation and anion exchange resins (21 g Amberlite IRC-50 C.P. RCOO-H, and 29 gAmberlite IRC-45 C.P. RNH3-OH), were buried at 20 cm depth at 9 points in each plot. Thesewere removed in August 1992, air dried, mixed with 200 ml of 2 M KC1 and shaken for 1 h. Theextract was analysed for NH4-N, NO3-N and PO4-P using the methods described above.Decomposition rate of cellulose is also an index of nutrient availability. At the same points asthe resin bags a 4.25 cm disc of cellulose (Whatman #1 filter paper, enclosed in a nylon mesh bagwith a pore size of 1 mm) was buried at 20 cm depth for the four month period. After collection,soil particles were washed off and the paper remaining was oven dried for 24 h at 70°C andweighed.73The rate CO2 evolution from soil is an indicator of microbial activity. The way thattreatments influence CO2 evolution is therefore an indication of the effect of treatment onconditions for microbial decomposers. CO2 evolution was assessed in the laboratory. About 80 g(wet weight) of each soil sample was removed from refrigeration and placed in a 500 ml mason jar atlaboratory room temperature (about 25°C). The jars were left open for two days at roomtemperature, then sealed. Twenty four hours after sealing, a 0.2 ml sample of gas was withdrawnthrough a rubber seal in the lid of each jar using a syringe. The CO2 concentration of this gassample was measured using an infra red gas analyser. 5 mason jars were filled with marbles to aboutthe same level as those with soil. The average value from these blank samples was subtracted fromthe measured concentration, and this adjusted value was converted to tg C released per dry g of soilover the 24 h period.Soil temperatures at 10 and 20 cm depths were measured in the field at 4 to 5 points in eachplot using digital thermometers in August 1992. Measurements were taken on clear days between 11am and 2 p.m.Understorey vegetationAbove-ground biomass of understorey plants was estimated by clipping all vegetation withinfive systematically-located 1 m 2 quadrats in each plot. Each sample was separated into species,divided into stem, leaf and flowers, and weighed. Moisture content was determined by drying a sub-sample of the biomass of each species at 70°C for 24 h, and then used to convert each sample tooven-dry mass.Statistical analysisThe results were analysed as a 2x3 factorial using a randomised design. The factors were:forest types, CH and HA; and treatment: UC, CC, and CCM. There were 3 plot replicates for theuncut forest, and 4 for the clearcut and clearcut and mixed treatments, with 5 to 9 samples taken ineach plot. The mean value was calculated for each plot, and these were used as replicates in ananalysis of variance. Treatment means within each forest type were compared using orthogonalcontrasts after the ANOVA. Bartlett's test (Sokal and Rohlf 1981) was used to compare the74homogeneity of variances of the means. If they varied significantly log or square-roottransformations were used in the ANOVA, but only the untransformed means and standard errorsare shown in the results. No analyses were carried out on the temperature measurements, becausethe number of replicates was small and variances were not homogeneous, even for transformed data.All analyses were performed using SYSTAT (Wilkinson 1990).Results and DiscussionSoil moisture, temperature, and organic matter content.Soil physical properties are shown in Table 6.1. Gravimetric moisture content was slightlyhigher in the UC-CH (78%) than the UC-HA (70%), which may be indicative of highertranspiration from the denser canopy of the second-growth HA stands. Moisture content was lowerin the CC plots on the CH type, but not on the HA type. Soil moisture has generally been observedto increase after clearcutting, because of the greater transpiration and rainfall interception by the treecanopy (McColl et al. 1977, Adams et al. 1991), although reductions in moisture content havebeen observed in other studies (Edwards and Ross-Todd 1983, Hendrickson et al. 1985). Moisturecontent on the CCM plots was significantly lower (a <= 0.05) in both forest types (about 60%).This decrease could be due to: (i) the mixing of organic and mineral soil layers during cultivation,which increased the mineral content at the soil surface, leading to increased thermal conductivityand heat admittance, and increased temperature below the soil surface, and therefore increasedevaporation; (ii) the lower moisture holding capacity of mineral soil compared with organic matter,and/or (iii) higher rates of infiltration due to the mixing treatment (Ross and Malcolm 1982).Organic matter content in both forest types was significantly lower (a <= 0.05) in the CCM(52-53%) than the other two treatments (82-86 and 77-86% for the UC and CC treatments,respectively). This reduction in organic matter content was primarily due to the mixing of themineral and organic horizons in the CCM plots.75Table 6.1 Soil physical properties in uncut, clearcut, and clearcut and mixed old-growth western redcedar and western hemlock (CH) and second-growth western hemlock and amabilis fir (HA) standson northern Vancouver Island, British Columbia. Mean and (standard error of the mean). Withineach forest type means followed by different letters were significantly different (a <= 0.05), usingorthogonal contrasts following the ANOVA. No statistical analyses were undertaken on soiltemperature data.CH HAUC CC CCM UC CC CCMMoisturecontent 78 (1.3)a 71 (1.8)a 60 (5.0)b 70 (2.0)x 72 (1.5)x 59 (2.8)Y(%)Organicmattercontent86 (2.1)a 77 (5.2)a 52 (10.3)b 82 (6.7)x 86 (6.1)x 53 (6.0)Y(%)Temp.10 cm 11.4 (0.33) 12.2 13.6 11.4 13.2 13.8(°C) (0.34) (0.98) (0.03) (0.22) (0.31)Temp.20 cm 10.1 (0.10) 11.1 12.5 10.1 11.1 13.1(°C) (0.40) (0.79) (0.10) (0.49) (0.46)76Temperature measurements at 10 and 20 cm depths were higher in CCM plots than in theCC treatment in both forest types, and higher in the CC than in the uncut forest. However, thenumber of replicates for temperature were small and variances were high, so no statistical analyseswere performed.Soil nutrients and microbial activityEstimates of pH, nutrient status (by KCI extraction, anaerobic mineralisation, and buriedion-exchange resin bags), and rate of microbial activity (through evolved CO2 and decomposition ofcellulose) are shown in Table 6.2. The pH was similar in the UC and CC treatments in both theCH and HA types (3.8-4.1), and significantly higher (a <= 0.05) in the CCM plots on the twoforest types (4.6). This increase was probably due to the mixing of the more acidic organic matterwith the less acidic mineral soil during cultivation.Extractable ammonium-N and P were significantly higher (p=0.006, 0.022) in soil from theHA than the CH type. Mineralisable ammonium was also higher in the HA type across the threetreatments, but this difference was not significant (p=0.171). The order of magnitude of thesedifferences is similar to those previously reported by Prescott et al. (1993b), although they reportedlarger differences in ammonium mineralised in an aerobic incubation.In the HA type extractable P was a significantly higher (a <= 0.05) in the CC than the UCtreatments but, except for this clearcutting alone generally had no significant effect on extractable ormineralisable nutrients in either forest type.Mineralisable ammonium-N was significantly lower (a <= 0.05) on the CCM plots than theother two treatments in the CH type, but not in the HA type. Extractable P was significantly loweron the CCM-CH plots than on the CC treatment. In the HA type, mean extractable P wassignificantly higher (a <= 0.05) in the CCM than the UC treatment. The mean of the sample valuesfor mineralisable N was lower in the CCM-HA than the other two treatments, but this differencewas not statistically significant.The ions held on buried exchange resins provide an estimate of the labile pool of nutrientsthat have been mineralised in the soil but not yet taken up by plant roots (Binkley et al. 1986).Table 6.2 Soil chemical properties and microbial activity in uncut, clearcut, and clearcut and mixed treatments in the CH and HA foresttypes. Mean and standard error (in parentheses). Within forest type values followed by different letters were significantly different (alpha<=0.05) using orthogonal contrasts following the ANOVA. Log transformed values were compared for resin ammonium, but theUCCHCC CCM UCHACC CCMExtractable (ppm)nitrate 12 (0.6)a 12 (0.3)a 11 (0.5)a 13 (0.4)x 12 (0.3)x 12 (0.2)x 0.175ammonium 34 (2.3)a 39 (4.7)a 30 (3.0)a 44 (6.4)x 41 (1.8)x 46 (0.9)x 0.006phosphate 6 (1.0)ab 14 (5.3)a 4 (1.1)b 9 (2.0)x 19 (1.2)Y 13 (2.1)Y 0.022Mineralisableammonium 48 (9.1)a 48 (7.5)a 20 (5.6)b 54 (3.8)x 49 (8.2)x 37 (4.3)x 0.171Resin (ppm)nitrate 36 (1.6)a 36 (3.5)a 45 (8.8)a 53 (19.9)x 49 (5.5)x 48 (4.9)x 0.124ammonium 270(37.8)a 262 (3.1)a 435(67.0)a 536(117.7)x 1370(68.5)x 1010 (22.1)x 0.003phosphate 312(19.7)a 114(27.3)ab 39 (17.8)b 89 (48.1)x 117 (19.6)x 157 (62.0)" 0.596Evolved C(ug/g organicmatter/day)270 (24.3)a 197(29.9)ab 98 (20.7)b 249 (14.8)x 196 (57.7)x 187 (17.2)" 0.430Cellulosedecomposition 45 (5.2)ab 58 (5.9)b 26 (5.9)a 31 (7.2)x 33 (8.0)x 59 (5.5)Y z(% wt loss)pH 3.8 (0.12)a 4.1 (0.18)ab 4.6 (0.19)b 4.0 (0.23)x 3.9 (0.22)x 4.6 (0.18)Y 0.7951 p values for differences between forest types using a two-way ANOVA.z Indicates a significant interaction(alpha <=0.001) between site and treatment.untransformed mean and standard error are_ shown.78Ammonium-N was significantly higher (p=0.003) across all treatments in the HA type compared tothe CH, but the other ions were not significantly different. Clearcutting alone had no significanteffect on this labile pool in the CH type. This suggests either that nutrient mineralisation was lowerin the CC treatment, or that nutrient uptake by salal is of the same order of magnitude as the uptakeby the old-growth CH stands. Above-ground salal biomass in the CC was similar to the UCtreatment (Fig. 6.1), but because below-ground biomass reaches a maximum 5 years afterclearcutting and burning (Messier and Kimmins 1991), total biomass is likely to be higher in theCC plots. Thus, the nutrient demand by salal on the clearcuts may equal the total demand in theold-growth stands.In the HA type, the mean of the samples for resin ammonium in the CC treatment wasmore than double that in the uncut forest, and the mean for the CCM plots was also considerablyhigher than the UC plots. However, these differences were not significant (a <= 0.05).Rates of CO2 evolved per g of soil organic matter measured in the laboratory is an index ofsize and activity of soil microbes. Rates were similar in the UC-CH and UC-HA stands, and about20-25% lower on the CC plots in both forest types, although these differences were not significant.There was no significant difference in CO2 evolved on the CCM plots compared to the CC onlytreatment in the HA type. In the CH type CO2 evolution was significantly lower (a <= 0.05) in theCCM treatment compared to the UC forest.Cellulose is a form of readily-available carbon for decomposers, and its decomposition rate insoil is an indication of demand for available carbon. This depends on microbial population size,environmental conditions for decomposers (eg. water, temperature), and the extent to which othernutrients are limiting microbial activity. The effect of the treatments on cellulose decomposition wasdifferent in each forest type. In the CH type decomposition was significantly slower (a <= 0.05) inCCM treatment (26% mass loss) than in the CC only treatment (58%). This lower rate ofdecomposition is in general agreement with the estimates of nutrient availability from extraction,mineralisation, and resin bags in the CH type. Therefore the availability of nutrients may be limitingdecomposition in the CCM treatment in this type.79In the HA type, mass loss was significantly faster (a <= 0.05) in the CCM (59%) than in the othertwo treatments (31-33%). but given that other indices of nutrient availability were generally nohigher in this treatment, the reason for the increased rate of cellulose decomposition is unclear.Understorey biomassAbove-ground understorey biomass in each site/treatment combination is shown in Fig. 6.1.Understorey biomass was much greater in the UC-CH than the UC-HA stands, due the more openstructure of the CH stands, and the high quantity of salal in the understorey. In the HA stands therewas a minor component of salal and the 'others' component was largely coniferous advanced growth.Estimates from the CC plots on both the CH (414 g/m 2) and the HA (362 g/m 2) types indicatedthat, almost 5 years after cutting and burning, above-ground salal biomass had recovered to levelssimilar to that in the UC-CH stands (431 g/m 2), but not yet at the maximum for CC sites in thisarea: 557 g/m2 reported 8 years after clearcutting and slash burning by Messier and Kimmins(1991). The allocation to different plant parts in salal was quite different between the twotreatments. In the CC area there was a much greater proportion in leaf compared to stemcomponents. Below-ground biomass is also likely to be much higher than in the UC CH stands.Salal allocates a large proportion of biomass below ground following clearcutting (Messier andKimmins 1991), whereas allocation to fine roots in the UC stands is much lower (Vogt et al. 1987,Messier 1991). This large increase in leaf and fine root biomass following clearcutting and burningmay explain the possible equivalence of nutrient uptake between the UC and CC treatmentssuggested above. On the HA sites, the biomass of salal increased substantially following clearcutting.This originates from small clumps surviving under the dense canopy of the HA stands, salalresprouts rapidly from rhizomes, and the burning treatment probably gives it an advantage overregenerating conifers. Unburnt HA stands are rapidly re-occupied by the dense stocking of coniferadvanced growth following clearcutting.The above-ground biomass of salal in both forest types was five times lower on the mixedplots than the CC and UC treatments. Other species in the CH type were not greatly affected bymixing, but the biomass of fireweed and other species in the CCM plots in the HA type was more80Fig. 6.1 Above-ground biomass of understorey species in the uncut forest, and after clearcutting andclearcut and soil mixing in two adjacent forest types on northern Vancouver Island: an old-growth,cedar-hemlock (CH) type, and a second growth hemlock-amabilis fir (HA) type. Values are themeans from three plots in the uncut forest, and four plots in the treated areas; five 1 m 2 samples perplot. Bars indicate standard errors of each mean.Uncut500400900E20010005004003002001000 eel&^othersClearcutsalad fireweed othersUncut500400300EO2001000set&^othersClearcut5004009002001000aseal fireweed other*Clearcut and cultivated5004003002001000Easaid^fireweed othersClearcut and cultivated5004003002001000Ei FlowersLeaf■ Stemsalad fireweed others81than twice that on the CC plots. These species may be benefiting from the increased availability ofnutrient-rich sites brought about by the reduction in salal by the mixing treatment in the HA type.ConclusionsIn general, the soil mixing treatment had no major effect on soil nutrients in the HA typeand a detrimental impact in the CH type. These results, almost five years after the initial treatment,suggested that the anticipated benefits of mixing and soil disturbance (i.e. increased aeration and soilnutrient availability through the mixing of mineral and organic horizons, and the bringing to thesurface of buried organic material where it can decompose more rapidly) have not occurred. Otherstudies indicated that the podzols in this area have a B horizon with low nutrient and organic mattercontent (Chapter 5). This contrasts with the higher organic matter content of this horizon in coastalAlaskan forest soils where the benefits of soil mixing and windthrow have been documented (Ugoliniet al. 1990). Organic material from deeper in the forest floor in the CH type have lower extractableand mineralisable N and P concentrations, compared to similar material in the HA type (Prescott etal. 1993b). The chemical structure of this deeper humus material also differs, possibly due totannins leaching from salal (de Montigny et al. 1993). Therefore, it appears that the nutrients inthese layers are being held in complexes that a relatively resistant to decomposition andmineralisation. Mixing these layers with the L and F layers, that have more available nutrients, hasresulted in decreased nutrient availability in the CH type.This suggests that short-term effects of disturbance and mixing of the organic and mineralhorizons by extensive windthrow is not the main cause of differences in nutrient availability andforest productivity between the two forest types. However, over the longer term, changes broughtabout by exposing deeper organic layers and the mineral soil, may have positive influence onnutrient availability and on productivity. Continued remeasurement of nutrient availability indicesin this experiment would allow these potential developments to be monitored.82Chapter 7. Litter production, nutrient resorption and decompositionIntroductionLitterfall is a major pathway for the transfer of energy and nutrients from above-groundcomponents to the soil surface in most forest ecosystems, and annual litterfall is an indicator ofprimary productivity (Bray and Gorham 1964, Vogt et al. 1986). The decomposition of foliar androot litter is the major mechanism for release of the nutrients contained in plant remains, and therate at which it occurs is a major determinant of nutrient availability for plant uptake and growth. Inmature forests, the majority of nutrients required for plant production is made available during thedecomposition process (Waring and Schlesinger 1985). The factors determining the rate ofdecomposition and nutrient release in forest litter have therefore been the subject of a considerableamount of investigation.Meentmeyer (1978) established the importance of climatic conditions in determiningdecomposition rate, but also found that where climate is less limiting, the 'quality' of the substratefor decomposers is an important determinant of the rate of mass loss. More recent studies haveconfirmed the importance of litter quality, in determining both the rate of decomposition and thenutrient dynamics in decomposing litter (e.g. Aber and Melillo 1982, Melillo et al. 1982, Flanaganand Van Cleve 1983, Berg and McClaugherty 1989, Aber et al. 1990, Harmon et al. 1990b).Quality is a general term, and is a combination of the availability to decomposers of the energycontained in C compounds and the amount and availability of N in nitrogenous compounds in thelitter material. The quality indicators that most often correlate with mass loss are the concentrationsof recalcitrant, secondary carbon compounds, generally measured as the acid-insoluble fraction andconsidered functionally equivalent to lignin (Aber et al. 1990), and of N. There is a general trendduring decomposition towards accumulation of the more recalcitrant components in relativelyundecomposable, humic substances (Bosatta and Agren 1991), and the complexing of N withinthese substances, where it is resistant to further microbial breakdown. In temperate and boreal83forests there is general pattern of early immobilisation (net accumulation) of N in decomposing litterfollowed by a net release. The presence of this initial immobilisation phase may depend on substratequality, especially initial lignin concentration (Upadhyay et al. 1989).Secondary C compounds in foliar and woody material, such as lignin, polyphenols, andtannins, have no obvious physiological role, but they are important components of the cell wallstructure, and there is an increasing recognition of their role in defence against herbivores andpathogens (Harborne 1982). There is general evidence for greater concentrations of secondarycompounds in the foliage of plant species growing on sites where nutrients are more limiting forgrowth (Waterman and Mole 1989), and foliar N concentrations are usually lower on these sites(Chapin 1980). Janzen (1974) suggested that the higher concentration of secondary C compounds isan adaptive mechanism, brought about by the higher replacement cost of photosynthetic materiallost to herbivory in nutrient-poor environments. Generally, the composition of vegetation isdifferent on nutrient-rich and nutrient-poor sites, and therefore this adaptation could be a speciescharacteristic. However, there is some indication that individuals of the same species may vary theirallocation of carbon to secondary compounds when growing on sites of differing nutritional status(Waterman and Mole 1989).Vitousek (1982) puts forward an argument for the stronger retention of nutrients by speciesadapted to growing on nutrient-limited sites. This could lead to higher retranslocation of nutrients atthe time of leaf senescence, and lower nutrient concentration in litter on nutrient-poor sites. It hasbeen suggested that among individuals within a coniferous species, the percentage of nutrientsresorbed prior to senescence is greater on nutrient poor sites, and that the concentration of ligninand other secondary compounds is higher, leading to a positive feedback between nutrientlimitation, litter quality, decomposition, nutrient availability and forest productivity (Gosz 1981,Edmonds et al. 1990). Others have argued that resorption is primarily a function of growth rate,and that high N availability will produce higher growth rates, a higher pool of resorbable N infoliage, and higher rates of resorption (Nambiar and Fife 1991).84The evidence for increased N resorption from senescing foliage on sites of lower nutrientavailability is variable. A number of studies have reported no increase (Staaf 1982, Birk and Vitousek1986, del Arco et al 1991). Where higher rates of resorption have been observed, it has generallybeen in experiments where fertiliser has been added to increase nutrient availability (eg. Turner1977, Flanagan and Van Cleve 1983).The two forest types in this investigation occur in similar topographic situations, and areunderlain by similar mineral soils. Differences in the productivity of seedlings planted on clearcutsites of the two types have been observed, but there has been no investigation of whether these arereflected in differences in the productivity of the existing stands. The availability of N in the forestfloors of the HA is higher than in those of the CH type (Prescott et al. 1993b). The amount of Ncontained in detritus in the two types is similar (Chapter 3), and therefore the lower N availability isa product of the lower rate at which N is mineralised in the forest floor of the CH type. However,the mechanisms responsible for this lower mineralisation rate have not been identified. It may bedue to higher rates of resorption, and therefore poorer quality of foliar litter, in the speciesdominating the two forest types. This could lead to slower foliar decomposition in the CH type, anddifferences in the pattern of nutrient immobilisation and release in decomposing foliage.The objectives of this study were to: (i) determine the annual mass of above-ground litterfallin the two forest types; (ii) determine if there were differences in the proportion of N and P resorbedto living tissues at the time of foliage senescence among species growing on the same sites andbetween hemlock individuals growing in the CH and HA types; and (iii) to measure rates of foliardecomposition, and the dynamics of N in decomposing foliage, in relation to the lignin/N ratio ofthe different species.MethodsLitterfall mass and nutrient content.Twenty litterfall traps were systematically located in each of the six plots described in theChapter 2. The traps were 50 cm x 50 cm in size, with a base of nylon mesh with a pore size of 185mm surrounded by a 10 cm high wooden frame, and were raised about 10 cm off the ground. Theywere cleared at approximately monthly intervals for 10 months from the end of January 1990 untilthe end of November 1990, then at the beginning of February 1991. Material smaller than 1 cmdiameter was retained and dried for 24 h at 70°C. Dried samples were sorted into the followingcategories: woody (eg. twigs, bark, cones), conifer foliage, understorey foliage, and other (eg. finebark, flowers), and weighed.Internal cycling.Samples of live foliage from two branches in the upper third of the crown of about ten treesper plot (about five cedar and five hemlock in the CH type, and five hemlock and five fir in the HAtype) were collected in winter (February 1992) by helicopter. Two small branches were clippedabout 1 m back from the tip of a main branch. Foliage from each branch were composited to make asingle sample for each tree, and dried at 70 0C for 24 h. Five of the litter traps (described above)were randomly selected from each plot. The foliar litterfall from each month for each trap wasbulked to form an annual sample, and these were sorted into species (cedar and hemlock for the CHtype, and hemlock and fir for the HA type). Approximately 0.2 g of each green foliage sample andeach foliar litter sample was digested using Kjeldahl methods with Se as a catalyst (100 g K2SO4 and1 g of Se in 11 of concentrated H2SO4). The resulting solution was analysed for N and P using theTechnicon Autoanalyser in the Forest Ecology laboratory at the University of British Columbia.It is more appropriate to express resorption on a mass of nutrient per unit of leaf area ratherthan a weight basis, because the density of foliage can change considerably during senescence, areadoes not (Fahey and Birk 1991). A small portion (0.3 to 1.0 g) of each sample was weighed, and theleaf area (in cm 2) determined using a LICOR 3100 leaf area meter, and foliage density calculated ing/cm2 . Nutrient concentration was then calculated on an area basis in mg/cm 2 . Percent resorptionwas calculated on the basis of area using the mean values of green and senesced foliage for eachspecies on each plot using the following formula (Fahey and Birk 1991):To resorbed = 100 x (green nutrient conc. - senesced nutrient conc.) / green nutrient conc.86Litter decomposition and nutrient dynamicsTwo sets of litterbags were used to investigate decomposition. A common substrate(lodgepole pine needles) was used to test between-type and between-site difference in the climaticconditions for decomposers. In February 1990, 35, nylon mesh bags, 10 x 10 cm in size with a poresize of 1 mm, containing 1 g of dried needles, were systematically located in each plot. Five bagswere brought back from each site at six monthly intervals for three years. The contents of each bagwas dried for 24 h, and weighed.To investigate the rate of mass loss and nutrient dynamics in the litter from the major plantspecies within each forest type, fresh litterfall from each tree species, and salal, was collected on meshscreen laid out in each plot. This litter was dried at 70 0C for 24 h and placed in mesh bags, asabove. In February 1991, 35 bags containing 1 g of cedar and 1 g of mixed hemlock/fir foliage, and35 bags containing a single salal leaf of known dry weight were placed at systematically-locatedpoints in each CH plot. For the HA sites 35 bags containing 2 g of mixed hemlock/fir foliage weresimilarly situated. Five bags of each group were retrieved at six monthly intervals for 2 years, and thecontents of each bag were dried and weighed as above. Litter in the bags from the CH sites,containing a mixture of cedar and hemlock and fir foliage, were sorted into these two categories andreweighed.Dried samples from each time period were ground in a Wiley mill and bulked to form asingle sample for each litter category from each site. Freshly fallen litter samples from five of thelitter traps at each site, described above, were ground and bulked to form an annual sample of eachlitter type from each site. These fresh litter samples, and the samples of litter bag material from eachtime period were analyzed at the Natural Resources Research Institute laboratory in Duluth,Minnesota. Percent N was estimated using a LECO CHN 800 analyzer.Statistical analysisThere was substantial damage and disturbance of the litterfall traps by wildlife, and the fullcomplement of 20 samples were not available for each plot in every collection. Consequently, themean monthly value of each litterfall category was calculated for each of the six plots, and these87values were summed to calculate an annual total. Forest types were compared using the annualvalues for each type across the three sites with Student's-t test (Sokal and Rohlf 1981).Percent resorption for each species, and for hemlock in both forest types, were comparedusing the mean values from each of the three sites. The variances of the means of all thespecies/forest type combinations were not homogeneous. However, variances between thecombinations of the greatest interest were homogeneous, and the following pairwise comparisonswere made on untransformed data using Student's-t test: cedar vs. hemlock in the CH type, hemlockvs. amabilis fir in the HA type, and hemlock in the CH vs. hemlock in the HA type.Mean values of mass remaining at each site were calculated for the four decomposing littertypes at each 6-monthly measurement period. These three site means were used to comparedifferences between litter types. For the twelve month collection the effect of site, and theinteraction between site and litter type for each litter type was tested using a two-way ANOVA inthe CH type, and the effect of site was tested in the HA using a one-way ANOVA. All analyses werecarried out using SYSTAT (Wilkinson 1989).ResultsLitterfall massThe annual mass of litterfall by category in each forest type is shown in Table 7.1. Therewere significant differences (alpha<=0.05) between the forest types in the annual amounts of alllitterfall categories except twigs (where the probability was 0.054). Coniferous foliar litterfall wassignificantly greater in the HA (about 2100 kg/ha) than the CH type (about 1500 kg/ha). Smallwoody litterfall (twigs and cones) was also greater in the HA (about 1700 kg/ha) than the CH type(965 kg/ha). Understorey leaves composed about 8% of the foliar litterfall in the CH type but only avery small component of the HA type (0.2%). This was to be expected, given the greater amount ofunderstorey biomass in the CH type. Total litterfall was about one-third greater in the HA type.These differences were consistent across all sites, and variability in the mean values between the siteswas quite low (coefficient of variation was 2.1% for the total in the CH, and 8.6% in the HA type),88Table 7.1 Annual above-ground litterfall (kg/ha) from the beginning February 1990 to the end ofJanuary 1991 for old-growth cedar-hemlock (CH) and regrowth hemlock-amabilis fir (HA) forestson northern Vancouver Island. Mean of three sites (material from about 20 traps at each site wascollected and weighed monthly, and monthly mean values were summed to produce an annualtotal). Figures in brackets are the standard error of the mean across the three sites.litter category CH HA p*conifer foliage 1516 (36.0) 2093 (138.2) 0.016understorey leaves 135 (22.3) 5 (1.6) 0.004twigs 806 (49.6) 1108 (100.6) 0.054cones 160 (16.3) 604 (58.0) 0.002others 478 (18.1) 362 (26.0) 0.022TOTAL 3094 (38.0) 4173 (206.2) 0.007p values indicate the level of significance of the difference between the two forest types based on aStudent's-t test.89which suggested that the 0.25 ha plot encompassed most of the within-stand variability in bothforest types.The pattern of litterfall through the year differed between the two types (Fig. 7.1). In theHA type, litterfall peaked in the late summer months of August and September, which is a normalpattern for evergreen species adapted to summer water limitation. In the CH type, peak litterfalloccurred later in the year (in October) which is more typical of a deciduous species adapted to awinter dry period.Nutrient resorptionThe concentration of N and P after a Kjeldahl digest, percentage resorption, and the relativedensity of litter vs. green foliage for the major species growing in each forest type are shown in Table7.2. Nutrient concentrations in both green foliage and foliar litter were within the ranges reportedfor these species in other studies (Tarrant and Chandler 1951, Daubenmire and Prusso 1963,Beaton et al 1965, Ovington 1965, Harmon et al. 1990b). N concentrations in cedar foliage andlitter was consistently lower than those in hemlock on the CH sites; this is also consistent with theresults of studies in which these two species are compared. Hemlock on the HA sites had a higher Nconcentration in both green foliage and litterfall than hemlock on the CH sites. The concentrationof N in green hemlock foliage on the HA sites was higher than that of amabilis fir, but because of agreater rate of resorption the concentration of N in hemlock litterfall was lower than fir.In pairwise comparisons between species in the CH type, cedar had a significantly higherrate of resorption of both N and P than hemlock (p=0.008 for N, 0.037 for P), and in the HA typeresorption in hemlock was significantly greater than amabilis fir (p=0.013 for N, and 0.044 for P).Hemlock growing in the CH type resorbed a significantly higher percentage of N (64%), thanhemlock growing in the HA type (51%, p=0.007), but P resorption was not significantly different(p=0.228).90Fig. 7.1 Mean conifer foliage litterfall (in kg/ha/day) from February 1990 to January 1991 from CHand HA stands at three sites on northern Vancouver Island. Average values for each stand wereestimated by collecting material from about 20 litter traps. Error bars indicate one standard error ofthe mean of the average values from each of the three sites.T.ava 15co.le06 102oN3ooZ 50O2, HA0 CHMONTHTable 7.2 Percent N and P (using a Kjeldahl digestion), percentage resorption (calculated per unit of leaf area), and the relative leaf density(RLD, litter density/green density in percent) for four types of foliage in western red cedar and western hemlock forests on northernVancouver Island. Mean and (standard error) across three sites.Green foliage Litter Percent resorptionN(%) P(%) N(%) P(%) N P RLD0.93 0.10 0.43 0.04 76.4 82.1 51(0.038) (0.005) (0.015) (0.002) (0.72) (0.75) (0.7)1.26 0.09 0.64 0.04 64.0 69.8 70(0.058) (0.007) (0.017) (0.001) (2.42) (3.91) (4.2)1.40 0.12 0.83 0.05 50.9 64.1 82(0.034) (0.003) (0.008) (0.001) (0.94) (0.75) (1.0)1.29 0.12 1.11 0.06 17.9 47.1 95(0.039) (0.005) (0.022) (0.003) (7.80) (5.84) (4.9)Species/typecedar,CHhemlock,CHhemlock,HAamabilisfir, HA92DecompositionMass remaining during 3 years of decomposition for the standard substrate, lodgepole pineneedles, is shown in Fig. 7.2. The rate of mass loss is almost identical in the two types, with about30% lost after 1, 45% after 2, and 70% after 3 years.Mass loss from local species' litter is shown in Fig. 7.3. There were no significant differencesamong the three study sites in the rate of mass loss tested using ANOVA (p=0.284 in the CH, and0.161 in the HA type), and the observed differences between the litter types were consistent at allthree sites. Salal leaves decomposed significantly more rapidly than conifer needles, with 57% massloss after 12 months, although rate of loss for salal declined over time with a further decline of only15% in the second year. Cedar litter decomposed most slowly, with 22% mass loss after 12 monthsand about 50% after 24 months, significantly slower (alpha <= 0.05) than hemlock in either typeduring the first 18 months, but the difference was not significant after 24 months. This may havebeen due to the approach of mixing the cedar and hemlock needles in the same bag in the CH type,and sorting them after the bags were brought back to the laboratory and dried. As decompositionproceeded there was an increasingly higher proportion of fragments that were difficult to allocate tospecies.The rate of mass loss of hemlock and fir needles was slower in the HA type than in the CHtype. This difference was significant (alpha < 0.05) at 12 months, and the trend continued throughto 24 months but the differences were not significant for the last two measurement periods.The decomposition parameter k, was calculated using the negative exponential function x/xo= e -kt (Olsen 1963), where x/xo is the proportion of mass remaining, after time t. k values after 1year were 0.25 for cedar, 0.36 for hemlock and fir in HA, 0.54 for hemlock in CH and 0.84 forsalal. k was higher in all cases after 2 years of decomposition: 0.35 for cedar, 0.41 for hemlock andfir in HA, 0.51 for hemlock in CH and 0.63 for salal.93Fig. 7.2 Mass remaining versus time for lodgepole pine needles decomposing in old-growth westernred cedar and western hemlock (CH) forests, and second-growth hemlock and amabilis fir (HA)forests on northern Vancouver Island.CDc 0.70 0.6Ea)L._ 0.5u)cn 0.40 6^12^18 24 30 3694Fig. 7.3 Mass remaining versus time for western red cedar, western hemlock and salal foliar litterdecomposing in the CH type, and mixed western hemlock and amabilis fir foliar litter decomposingin the HA type at three sites on northern Vancouver Island. Error bars are one standard error of themean of the three site values (site values are the average of 5 observations at each site). Differentletters beside each point at each six-monthly interval indicate they are significantly different (alpha<=0.05) using Tukey's test (Wilkinson 1990).10080Xcoc 60•_,cCo'EE4ca 0caco2205^10^15^20^25^30Months95N dynamics in decomposing litterThe accumulation of N in local species' litter during the 2 years of decomposition is shownin Fig. 7.4. The initial percentages of N obtained from the LECO CHN analyzer were higher forcedar and hemlock needles from the HA than those obtained from the Kjeldahl digest andcolorimetric analysis (Table 7.2). This is because the combustion and gas analysis process of theLECO analyzer recovers N from acid insoluble fractions, whereas the Kjeldahl digest does not(Bremner and Mulvaney 1982).The N concentration in all litter types increased during the two years. Salal did so mostrapidly, and to the highest level, but this increase occurred almost entirely during the first 12months of decomposition, which corresponded to the period in which salal leaves were rapidly losingmass. Aber et al. (1990) suggested that litter decay has two phases. The first phase shows relativelyconstant fractional mass loss until the mass remaining reaches about 20%, whereupon the secondphase is reached in which mass loss becomes very small. For salal, it appears that this second phasewas reached during the two years of this study, at a percentage mass remaining of about 30%. Theconifer foliar litter types exhibited slower, but more constant, increases in N over the time period,and mass loss was also consistent during the 2 years of observation, which suggests that they had notentered the second phase. Aber et al. (1990) estimated that the first phase for relatively recalcitrantlitter with high lignin/N ratio, such as these conifers, could last from 3 to 5 years. The increase inthe N concentration of cedar from 18 to 24 months, which has a steeper slope than the other timeperiods and converges with the hemlock litter from the CH type, may be an artifact of the methodused for these litter types (see above).Considerable evidence has accumulated to indicate that the relationship between massremaining and N concentration during the first phase of decomposition (described above) has aninverse-linear form (Aber and Melillo 1980, 1982, Melillo et al. 1984, Aber et al. 1990). Fig. 7.4indicated that this relationship holds up very well for the litter types in this study, with r 2 values of0.95 for cedar, 0.98 for hemlock on CH, 0.99 for hemlock on HA, and 0.94 for salal. The lower96Fig. 7.4 Relationship between percentage mass remaining and percent nitrogen for western redcedar, western hemlock and salal foliage decomposing in the CH type, and mixed western hemlockand amabilis fir decomposing in the HA type at three sites on northern Vancouver Island.r2 for salal was due to it entering the second phase of decomposition some time in the first year,while the lower value for cedar is probably due to methodological complications.DiscussionThe annual litterfall values fell within the general range of values for cold temperateevergreen forests given by Vogt et al. (1986). Total and foliar litterfall were significantly higher inthe HA than the CH type which suggested that productivity was higher in the HA stands. This isprobably due to a lower growth potential in the CH type, rather than a difference in the stage ofdevelopment of the two types, because the HA stands (now around 80 years old) should havereached a relatively steady state amount of foliage biomass. The difference in litterfall (about 3100vs. 4200 kg/ha/yr) between two coniferous forests in close proximity probably helps to explain whyVogt et al. (1986) found the global relationship between above-ground litterfall and latitude to bequite poor.The rates of resorption of N (76%) and P (82%) in western red cedar were high.Comparable studies of this species were not found. Higher rates have been reported for Larix spp.(Cole 1981, Tyrrell and Boerner 1984, Gower et al. 1989), and similar rates in Pinus silvestris(Stachurski and Zimka 1975). Western hemlock in the CH had lower resorption rates than cedar,although the values (64% N, and 70%P) are still relatively high, and comparable with many Pinusspp. (Miller 1984, Birk and Vitousek 1986, Gower et al. 1989). Amabilis fir growing in the HAtype had relatively low rates of N resorption (18%), and higher resorption of P (47%). Thesignificantly higher resorption of N by hemlock individuals growing in the CH (64%) than those inthe HA type (51%) supported the contention that coniferous species can have higher rates ofresorption on sites with lower nutrient availability (Gosz 1981, Edmonds et al. 1990), and contraryto evidence from studies which indicated that conifer species do not generally alter their rates ofresorption across gradients of N-availability (eg. Birk and Vitousek 1986). However, the consistencyin resorption rates reported by researchers such as Birk and Vitousek (1986) may be restricted tosituations where N continues to be limiting biological activity. In situations where N is not limiting,9798N concentrations in litterfall may be elevated and the rate of resorption may be reduced (Turner1977).It is possible that N is no longer limiting growth of hemlock (or amabilis fir, given its lowrate of N resorption) on the HA sites, as there is little response to N-fertilisation on these sites(Weetman et al. 1990). There was also an indication from the litter decomposition study that N wasnot limiting biological activity in the HA system, which suggests that, despite the higher Nconcentration in the hemlock and fir foliage in the HA type, it does not decompose any faster thanhemlock litter in the CH type.Other studies comparing resorption rates among species from this region were not found.The apparent greater ability of hemlock to resorb N compared to amabilis fir may explain whyhemlock survives in the cedar and salal-dominated CH type, whereas amabilis fir is a relativelyminor component in this type.The rate of mass loss for the lodgepole pine needles in this study was relatively rapid, abouttwice that observed for needles of the same origin in the dry continental climate of the RockyMountains (Taylor et al. 1991). The similarity in the rate of mass loss of these needles in the twoforest types suggested that the climatic conditions for decomposers, at least at the surface of theforest floor, are equivalent in the two forest types.The relatively rapid decay of salal leaves indicated that it is not immobilising significantquantities of nutrients in leaf litter during the course of decomposition. The k values for cedar andhemlock in the two types are within the reported ranges for conifers (Gosz 1981, de Catanzaro andKimmins 1985). Harmon et al. (1990b) compared decomposition of cedar and hemlock foliage inthe Olympic National Park in Washington State, an area with a similar temperature regime buthigher precipitation. They also found more rapid decomposition of hemlock than cedar, with k =0.724 after 12 months, higher than the values of 0.54 in CH and 0.36 in HA obtained in this study.They studied the decomposition of cedar in two different years, and the results (k = 0.386, and0.291) were also higher than the value of 0.25 obtained in this study.99Cedar and hemlock have similar lignin concentrations (23.1% and 24%, Harmon et al.1990b), and therefore lignin alone cannot explain the difference between them in the rate of massloss as it has in other studies (Meentemeyer 1978, Upadhyay et al 1989, Taylor et al 1991). Thetwo species did vary significantly in initial N concentration (Table 7.2), a difference also reported byHarmon et al (1990b). Therefore lignin/N ratio differs considerably between the two species, andit is possible that this can explain the difference in decay rate.Pastor and Post (1985) developed a model that predicted annual mass loss from annualevapotranspiration (AET) and lignin/N ratio. Using this model with an AET of 550 mm (an averagevalue for the study area), the lignin values from Harmon et al (1990b), and the Kjeldahl nitrogenvalues in Table 7.2, gave a predicted annual mass loss of 19.7% for cedar and 34.3% for hemlock.This compares with the measured values of 22% and 36% (the average for hemlock across the twotypes).The more rapid decomposition of hemlock needles in the CH type compared with that ofthe mixed hemlock and fir in the HA type had no obvious explanation. The N concentration of theneedles in the CH type was lower, and Pastor and Post's (1985) model would therefore predict thatthey decay more slowly. Any climatic differences at the surface of the forest floor between the twotypes should also be reflected in the standard substrate (lodgepole pine), but none were observed(Fig. 1).Differences in the mass (Table 7.1), and differences in the N concentration of foliar litterfall(Table 7.2) and (Fig. 7.4) that come about because of higher resorption of N by cedar and byhemlock on the CH site, indicate that there are significantly smaller amounts of N cycling in theCH type. This can partially explain the differences in N availability between the two types. Thisdifference can be further explained by the slower rate of decomposition of the cedar foliage, andpotentially higher short-term immobilisation of N in decomposing foliage. For example, using theannual foliar litterfall values in Table 7.1 (1516 kg/ha in the CH type, and 2093 kg/ha in the HAtype), and multiplying by the initial N concentrations in Fig. 7.4 (assuming CH litter is 80% cedar,100the proportion of cedar by basal area), gives an annual N input to the forest floor of 10.46 kg/ha inthe CH type and 17.58 kg/ha in the HA type.Multiplying the initial mass by the percent remaining after 12 months of each litter type(Fig. 7.3), gives 1123 kg/ha remaining in the CH type and 1457 kg/ha remaining in the HA type.Multiplying this mass by the 12 month nutrient concentrations (Fig. 7.4) gives an N content of10.00 kg/ha in the CH and 16.17 kg/ha in the HA type. Therefore, after 12 months ofdecomposition there has been a release of 0.46 kg/ha (4.4% of the initial) of N from the foliar litterin the CH type, and 1.41 kg/ha (8% of the initial) from the HA litter. Consequently, there is aconsiderably higher rate, and total amount, of N released from the litter in the HA type. Theimplications of these differences in nutrient dynamics are explored more fully using a model ofnutrient cycling and organic matter dynamics in Chapter 8.ConclusionsThe results of this study indicated that there was a smaller mass of above-ground litter fall inthe CH than in the HA type. Because of higher rates of nutrient resorption at the time of foliagesenescence, particularly by cedar, there was only 60% as much N returned to the forest floor inlitterfall in the CH as there was in the HA type. Largely because of this greater ability to resorb N,and consequently, the lower concentration of N in foliar litterfall, cedar foliage decomposed moreslowly than hemlock in the CH type. Hemlock, by resorbing a greater proportion of foliar N atsenescence in the CH type also contributed to the lower amount of N cycling in above-ground litter.Resorption by salal was not measured, but its foliage has relatively high N concentration (comparedto the conifers) and this suggests that its growth is not generally limited by N in this area. Salal leavesalso decomposed quite rapidly and therefore the N contained in its foliage is not immobilised forlengthy periods.The lower rate of litterfall and the greater rate of nutrient resorption in the CH type areindicative of species adapted to nutrient-poor conditions (Vitousek 1982). The results of this studysuggest that these adaptive mechanisms to conserve nutrients can contribute to positive feedbacks101that lead to a lower rate of N cycling and availability on N-poor sites (Hobbie 1992). However, thismay also be considered as a competitive mechanism, giving a species that can effect a reduction inthe soil nutrient supply, and exploit this lower nutrient supply more effectively, an advantage overspecies that cannot do so to the same degree.These results provide evidence to support the hypothesis that differences in thecharacteristics of the dominant tree species can explain much of the difference in N availabilitybetween the CH forest types. This hypothesis is explored further in Chapter 8.Chapter 8 Modelling organic matter and nutrient dynamicsIntroductionThe storage and flux of carbon are important components in the functioning of forestecosystems (Waring and Schlesinger 1985), and these are often intimately coupled with the cyclingof nutrients (Attiwill 1986). For example, in temperate forests N generally limits productivity whenwater is not limiting, and its availability in the soil depends largely on the rate at which N ismineralised during decomposition of organic matter returned to the forest floor in above- andbelow-ground litterfall. When climatic conditions are constant, this is mainly determined by thechemistry of the carbon associated with the N in litterfall (Meentmeyer 1978). Thus, there arestrong feedbacks between the way in which different species store and utilise the carbon fixed inphotosynthesis, and the rate at which N is released in the forest floor during decomposition and,ultimately, to the level of production that the forest sustains (Hobbie 1992). These feedbacks canlead to alternative stable states in the composition of vegetation and N supply in the soil, even whenabiotic conditions such as climate, soil parent material and topography are quite similar, in bothforest (Flanagan and Van Cleve 1983, Nadelhoffer et at 1983, Pastor et at 1984, Zak et al. 1986),and grassland ecosystems (Wedin and Tilman 1990).The landscape of the Pacific coast of North America is dominated by coniferous species.Individual trees grow to large sizes and large amounts of biomass accumulate in forest stands(Waring and Franklin 1979). However, there is considerable variation between these conifer speciesin maximum size and age, and in their susceptibility to attack by insects and diseases. This variationis indicative of differences in growth patterns and carbon allocation to secondary chemicals in theproduction process. There are also differences in the growth response of each species to varyinglevels of N availability, although this has been less intensively investigated. Despite the economicimportance of, and scientific interest in, these forests, there has been little consideration of the wayin which these differences in chemistry and response to N availability might express themselves in102103ecosystem functioning, and affect community structure, through the kinds of feedback mechanismsdescribed above.Western red cedar and western hemlock differ in their maximum life span and maximumsize, and in their resistance to insects and pathogens (Minore 1983). Western red cedar can live upto 1000 years, and can attain a maximum diameter of 250-300 cm, while hemlock rarely survivesmore than 400 years, and reaches a maximum diameter of about 120 cm (Franklin and Dyrness1973, Chapter 3). These differences in size and life history are associated with different growth ratesand regeneration dynamics in this environment, and they indicate differences in wood and foliarchemistry between the two species.The objective of this chapter was to investigate the hypothesis that the lower nutrientavailability in the forest floors of the CH stands could be explained by these differences in specieschemistry, and the way in which they express themselves in this environment. Because of the timescales involved in ecosystem development, it is difficult to test such a hypothesis using traditionalempirical approaches, such as a controlled experiment, as has been done in more rapidly developingvegetation types like grasses (Wedin and Tilman 1990). The stand history of these long-lived speciesis also difficult to reconstruct, and no study sites were available that could be used to investigateforest floor properties with varying proportions of each species in different stages of development.Computer modelling provides a way of overcoming some of these investigative difficulties. Indeveloping a model, current knowledge and ideas about the way ecosystems develop and functionare organised in a logical framework, and the implications of this empirical and theoreticalunderstanding can be investigated. Hypotheses can be tested that are difficult or impossible toexplore through empirical means (Yarie 1990). The model, in effect, is a kind of experimentalmicrocosm in which different scenarios can be tested using assumptions based on current (albeitimperfect) knowledge of ecosystem functioning. However, models are not reality, and should not beconsidered a replacement for empirical approaches to investigation and hypothesis testing.Models can depict the functioning of ecosystems at various scales of space and time(Kimmins 1988). In this study, the ecosystem properties of interest were those that develop over104decades or millennia, in a space of the order of a single forest stand. Because forests in this localityhave been the subject of relatively little scientific investigation, it was important to develop a simplemodel that described the functioning of the ecosystem using the dynamics of a few key componentsfor which calibration data were available. In longer-time step models, the dynamics of larger, slower-moving components can integrate the dynamics of smaller components that turn over more rapidly(Bunnell 1989). For example, tree diameter growth integrates the kinetics of water and nutrientuptake, photosynthesis, and carbon allocation; and the nutrient dynamics in decomposing litterintegrate the activities of various fungal and microbial decomposers. By carefully choosing thecomponents included in the simulation, a relatively simple model structure can be developed thathas a considerable degree of descriptive realism.Model structure and calibrationThe model used in this study was LINKAGES (Pastor and Post 1985,1986), a simulationmodel of forest growth and nutrient cycling based on the JABOWA model (Botkin et al. 1972,Botkin 1993). Both LINKAGES and JABOWA were developed to simulate forest growth andcommunity dynamics in north-eastern USA. LINKAGES includes more detailed representations ofclimate and soil nitrogen and water availability, and the way they affect the growth of different treespecies, than JABOWA. It explicitly represents the feedback between species chemistry, nitrogenavailability and forest production that may control species composition in these forests. It requires arelatively simple set of calibration data, and can simulate the development of both even-aged, singlespecies, and mixed-age, mixed-species stands found in the study area.LINKAGES consists of three basic components (Fig. 8.1): production, climate anddecomposition. The production component is the single-tree, non-spatial model construct ofJABOWA (Shugart 1984) which has been used extensively to simulate tree growth and communitydynamics in many different forest types around the world (for a full description see Botkin 1993). Ithas previously been applied, with some success, to the Douglas-fir ecosystems of the PacificNorthwest (Dale et al. 1986). In this component, tree establishment, growth and mortality are105Fig. 8.1 LINKAGES structure, rectangles represent subroutines, and arrows indicate important flowsbetween subroutines.TEMPEMONTHLYTEMPERATURESMOISTAETDEGREE-DAYSDECOMPAVAILABLENITROGENGMULTiDEGREE-DAY, WATER , NITROGEN MULTIPLIERSI 4•DRYDAYSLITTERILIGHT MULTIPLIER4^BIRTHDEMOGRAPHYILIGHT MULTIPLIER4^ GROWDEMOGRAPHYKILL106simulated on a small-area plot. Individual trees establish (as saplings with a dbh chosen stochasticallyfrom between 1 and 3 cm) at a user-specified rate, if light and moisture conditions are suitable forthe species. These established individuals increment in diameter on an annual time-step at a ratedetermined by the potential maximum diameter increment under optimal conditions (a function ofthe maximum age and maximum diameter of each species), and modified according to the simulatedavailability of light, water and nutrients, whichever is most limiting, for each individual. Thesemodifiers are calculated from species-specific coefficients of response functions that are entered inthe input file. Tree height is a parabolic function of diameter. Mortality is simulated in two ways: (i)exogenous mortality is simulated by killing a small proportion of trees each year, so that 1% of treesreach the potential maximum age for their species; and (ii) within stand competition is simulatedusing a flag for slow growth. In this study a diameter increment less than 10% of the potentialmaximum was considered slow growth, and the probability of mortality in a given year increased to0.2 for trees with more than 10 consecutive years of slow growth.Light at any level in the canopy is a function of the foliage biomass above that level, and isdetermined from allometric relationships between diameter and foliar biomass, and assuming that allfoliage for an individual is situated at the top of the tree, and spread across the entire plot. Availablemoisture is calculated in the climate component. The mean and the standard deviation of monthlytemperature and precipitation for the study area are read into the model as input, and normally-distributed, random values are selected to simulate an annual climate. Thornthwaite and Mather'smonthly actual evapotranspiration (AET) is calculated according to an approximation function(Pastor and Post 1984), and combined with soil moisture-holding capacity (from soil texture) todetermine the proportion of the growing season that soil moisture falls below field capacity. Thisvalue is used to reduce diameter growth.The availability of N, the nutrient assumed to be limiting tree growth, is calculated in thedecomposition component. Foliar, root and twig litterfall are calculated for each year in theproduction component from foliar biomass and foliage retention time. Woody litterfall isdetermined by tree mortality. The decomposition and N dynamics of these annual inputs of each107type, and species, of litter are monitored each year as separate cohorts. Foliar and fine root litter losemass as a function of the simulated climate (annual AET) and their lignin:N ratio, according to arelationship developed by Pastor and Post (1985). Woody litter cohorts lose mass at user-specifiedannual rates. Once a cohort reaches a certain percentage mass remaining (a function of the initiallignin concentration, DeHaan 1977), it is transferred to a combined pool of humus, which losesmass, and N, at a constant rate (1% per year in the original model, changed to 2% in this study) orwell-decayed wood which loses mass at 1.5% per year.The immobilisation or mineralisation of N is simulated for each cohort using a linearrelationship between the mass remaining and the N concentration in the remaining material (Aberand Melillo 1980, Aber et al. 1990). The coefficients of this relationship are specified as input foreach species. A similar relationship is used to model the dynamics of lignin. Lignin and Nconcentrations of each cohort change during the course of decomposition, and therefore these aredynamic components that are recalculated annually in the model. These changes affect the mass lossin the following year. The net mobilisation of N is calculated from the sum of immobilisation ormobilisation in each cohort, and the mobilisation from humus and decaying wood. A proportion ofimmobilisation is satisfied by throughfall (16% of litterfall N) and external inputs (1 kg/ha/yr in thisregion).Parameters used in the growth component of the model are shown in Table 8.1. Maximumage, dbh, and height were either obtained from the literature (Franklin and Dyrness 1973), or frommeasurements of stand structure in the study area (Chapter 3). The coefficients B2 and B3 werecalculated following the selection of the value for G, the growth scaling parameter in JABOWA. Gfor the two species was determined in an iterative process, by applying various values and using theone that gave the best fit to stand growth from permanent plots with a site index of 25 m (at breastheight age 50, John Barker, pers. comm.). Foliage and stemwood biomass relationships were takenfrom Gholz et al (1979), Dale and Hemstrom (1984), and Feller (1992). In simulating the forestsof northern Vancouver Island there were a number of simplifying factors. Firstly, the object was tosimulate nutrient availability under constant climate and mineral soil conditions for forestsTable 8.1 Parameters used for each species in the growth component of the model.108Parameter western red cedar western hemlockMaximum age (years) 750 400Maximum dbh (cm) 250 122Maximum height (m) 50 50B2 0.08 0.27B3 38.90 81.21G 100 190Stemwood biomass slope 2.081 2.257Stemwood biomass intercept-2.0927 -2.172Foliage biomass, trees < 30cm dbh - slope 1.922 2.218Foliage biomass, trees < 30cm dbh - interceptm2 leaf area/kg foliage-3.0070.0044-4.1300.0044Shade tolerance, a 10.59 10.20Shade tolerance, b 0.062 0.038Shade tolerance, c 9.31 10.15N tolerance, a 0.998 1.258N tolerance, b 0.174 -8.808N tolerance, c -0.028 -0.01Foliage retention time(years) 4 4109dominated by two species, western red cedar, and western hemlock. Amabilis fir was a minorcomponent in the stands where detailed investigations were undertaken, and fir generally has similarlife history and foliar characteristics to hemlock. Initial model runs using climatic data from thelocality with a relatively fine-textured soil indicated that the soil moisture content remained abovefield capacity throughout the growing season. This was consistent with field observations, andtherefore it was assumed that there was no moisture limitation on tree growth. The two species areclose to the center of their geographic range in this locality, so no temperature limitations wereimposed.The limiting factors on tree growth were therefore assumed to be light and nitrogenavailability. In the model, tree species are assigned shade tolerance categories based on the extent towhich growth is reduced in relation to the proportion of above canopy light reaching the tree. Inapplying the JABOWA model to the Pacific Northwest Dale and Hemstrom (1984) developedrelationships between foliage biomass and leaf area, and leaf area and percent of above-canopy light.These relationships conformed fairly well to recent data from Vancouver Island (Smith 1993), andwere used to determine the proportion of above canopy light (PACL) at various heights in thecanopy. A recent study of shade tolerance of cedar and hemlock in coastal BC by Carter and Klinka(1992) provided the following relationship between PACL and relative growth rate, and theparameters for each species:RG = [a (1-exp- b(PACL))] cwhere RG = relative growth rate, and a, b, and c are parameters. Parameters for the two species onfresh sites were used in the model.The response to N availability function of the two species was more problematical. InLINKAGES, the Mitterlich relationships developed by Aber et al. (1979) for hardwood and coniferspecies in northeastern USA were used to describe three different classes of response to varying levelsof annual net N mineralisation (not including tree uptake). However, there have been few studies of110annual N-mineralisation in the Pacific Northwest, and those have indicated considerably lowerannual rates than in the temperate, hardwood/conifer forests where Aber et al: (1979) developedtheir relationships. In addition, there has been little broadscale investigation of the response of thesetwo species to fertilisation, or to varying levels of available N. For western hemlock there was ageneral pattern of site index increasing across an N-status (measured as total and anaerobically-mineralised N) gradient from poor to medium, with no further increase on sites classified as N rich(Kayahara et al: 1993). The response of cedar is more variable, and may depend to some degree onthe extent of nitrification (Krajina et al. 1973, Minore 1983).There have been a number of studies in the study area of response to fertilisation, and of thegrowth of seedlings planted on the two site types. These indicated that hemlock responds strongly tofertilisation on the CH sites but not on the HA sites, and that hemlock grows much more rapidly onthe HA than the CH sites. Growth response to fertilisation on the HA sites was generally small. Theresponse of cedar planted on the HA sites, and to fertilisation, is much lower (Weetman et al. 1990,Thompson et al 1993). Thus it appears that N-availability in the two forest types may lie across thepoor-medium threshold for hemlock described by Kayahara et al. (1993). A similar pattern maywell apply for cedar, but the limit below which N availability has to fall before growth is significantlyreduced is below the levels found in these two forest types.Thus, although it appeared that the Mitterlich function was an appropriate descriptor of thegrowth response to N availability for the two species, the problem was to determine appropriateparameters. This was done using height growth data from an experiment where the two species wereplanted on clearcuts of the two forest types, and assuming annual N-availability was 20 kg/ha/yr onthe CH sites (Weetman et al. 1990) and 50 kg/ha/yr on the HA sites. With no measured estimateor best guess available, the latter figure was somewhat arbitrary. The difference, compared to CHsites, is greater than that reported for KC1-extractable N, but less than that reported for Nmineralised in 40 day aerobic incubations (Prescott et al. 1993b). Parameters were developed basedon the assumption that cedar growth ceased at zero net mineralisation, while hemlock growth did so111at 10 kg/ha/yr. Parameters for the relationship are shown in Table 8.1, and the response functions inFig. 8.2.The parameters used in the decomposition component of the model are shown in Table 8.2.Initial N contents were measured during this study (Chapter 7), and initial lignin values were fromHarmon et al. (1990b). The parameters for the relationships between N concentration versuspercent organic matter remaining are from unpublished studies (J. Pastor, R. Edmonds pers.comm.). The slope of the line was kept the same between the two species, but the intercept wasadjusted to reflect the differences in initial N concentration. Parameters for the lignin versus organicmatter remaining relationship were derived in a similar way.Woody debris is a major component of these systems and, to more realistically reflect their Ndynamics, the model was modified to include a larger number of woody litter types which havedifferent rates of decay according to the species. Rates of decay have been reported for logs of thesespecies in similar ecosystems (Graham and Cromack 1981, Sollins et al. 1987). These values weredoubled to obtain the numbers reported in Table 8.2 to take into account Sollins' (1982) argumentthat those estimates, based on changes in density, do not take into account fragmentation losses. Theparameters for N and lignin dynamics in decaying wood were derived from measurements of woodin various stages of decay in the study area (Chapter 4).The model simulated forest growth and nutrient dynamics on a 0.2 ha plot. The model hasstochastic elements for calculating climate and tree mortality, and allows for the simulation of up to100 plots with different random effects. Results are then reported as a mean value with 95%confidence limits. Climate is relatively consistent in this locality, and simulation of 10 plotsproduced output with confidence limits within 10% of the mean for most variables. Model runswere done on an IBM 386/25 Mhz PC, and a 3000 year simulation with 10 plots took about 45minutes. The model output was varied from that described by Pastor and Post (1985) to get a morecomprehensive picture of the simulated forest floor, and to obtain an indication of the diameter-classdistribution of the simulated stands. A component was included to simulate a windstormdisturbance, in which the probability of death was high for big trees and low for small trees.40 50 60Available N (kg/ha/yr)1.00.8^ cedar2 0.4ecc^ hemlock0.20.00 10 20 30 8070112Fig. 8.2 Growth response functions for western red cedar and western hemlock to available N.Calculated according to the equation Relative growth = a[1 - 10 -c(N+b)], where N is available N,and a, b and c are the N response parameters specified in Table 8.1.Table 8.2 Parameters used in decomposition component of the model.Litter type^Initial N (%)^N parameters:^Initial lignin (%) lignin parameters^Percent litter^Percent annualintercept, slope intercept, slope^becomes humus^mass lossor decayed woodCedar foliageHemlock foliageFine rootsCedar wood > 60cm dbhcedar wood < 60cm dbhhemlock wood >30 cm dbhhemlock wood <30 cm dbhtwigswell decayedwood0.4 0.0191, 0.0186 23.1 0.664, 0.673 500.75 0.0261, 0.186 24.0 0.673, 0.433 490.46 0.0163, 0.0117 28.3 n/a 580.06 0.0033, 0.0027 n/a n/a 500.06 0.0033, 0.0027 n/a n/a 500.06 0.0033, 0.0027 n/a n/a 500.06 0.0033, 0.0027 n/a n/a 500.3 0.0195, 0.0157 n/a n/a 610.3 0.005, 0.002 n/a n/a 25n/an/a100.61.21.62.3101.5114The model was initialised with a mass of humus and N specified in the input file. Thisrepresents a starting humus pool that loses mass and mineralises N at a constant, relatively slow, rateof 2% per year. Thus, initial N can only be varied by varying the size and N concentration of thishumus pool. Once a forest becomes established and a forest floor of decomposing cohorts is builtup, most of the mineralised N comes from these decomposing cohorts. Initialising the model with alow amount of soil N required a considerable period for cedar (the more low-N tolerant species) tobecome established and develop a forest floor. Consequently, the first 2,000 years or so of allsimulations should be considered an initialisation stage and largely ignored for the evaluation ofmodel performance. All simulations were run with a fine-textured mineral soil (field capacity of 38.3cm, wilting point of 20 cm).SimulationsFour scenarios were investigated with the model:1. CH type. Cedar and hemlock were grown together on soil with an initial soil N of 2Mg/ha. Cedar was established at one seedling per plot every four years and hemlock at one seedlingevery year.2. HA type. Hemlock established at five seedlings per plot per year, with an initial soil N of4 Mg/ha. Simulated windthrow occurred every 300 years (stems and foliage of dead trees were addedto the forest floor).3. Hemlock only, no disturbance. Five seedlings were established per plot per year, with aninitial soil N of 4 Mg/ha.4. Cedar only, no disturbance. Five seedlings were established per plot per year, with aninitial soil N of 4 Mg/ha.The first two scenarios were designed to simulate the two forest types found in the region. Ahigher rate of seedling establishment in the HA than the CH type largely reflects the presence ofsalal in CH, which was thought to limit the rate of recruitment in the CH type, while seedlingestablishment in HA stands without salal is quite prolific (Chapter 3). It has been postulated that theHA forest type has been subject to repeated windthrow (Terry Lewis, pers. comm.), so this kind of115event was included in the HA simulation. The model only simulates the dynamics of tree species,understorey species such as salal were not included. The latter two scenarios were more hypotheticaland were designed to demonstrate the effect of the different species with other things (establishment,mortality and initial N) being held equal.In order to determine the management consequences of the differences in nutrientavailability which result from stands being dominated by the different species, four further scenarioswere investigated. These involved running the CH and the hemlock-only scenarios for 3000 years,clearcutting the simulated stands (i.e. all trees were killed, stemwood was not added to the forestfloor, but foliage was) and planting cedar or hemlock plantations at 3000 stems per ha. Theseplantations were then grown for two 100-year rotations.ResultsFig. 8.3 shows an example of the diameter distributions for the two species after about 4,000years in the simulated CH stands, compared to the average values from the three stands in Chapter3. The shape of the simulated and actual distributions are quite similar. There were some gaps in themeasured cedar distribution compared with simulated one, that may indicate that disturbance notaccounted for in the simulations has disrupted the regular patterns of establishment and mortality.The hemlock distributions suggest that recruitment may be higher than the 1 stem/plot/year in thesimulation, and that the potential maximum diameter used may be a little high for the species in thistype.Above-ground live biomass simulated in the four scenarios, and 'measured' biomass amountsfor the two forest types (calculated using measured diameters and the same allometric relationshipsused in the model) are shown in Fig. 8.4. Available N is shown in Fig. 8.5, and forest floor biomassin Fig. 8.6. Values for all variables stabilised fairly early in the simulation for the single speciesscenarios, but it took 2000 to 3000 years to reach a stable equilibrium in the simulation of the CHtype. This is because of the low initial N and the low recruitment in this scenario, which meant thatit took some time before a closed canopy nutrient cycle was established.300200100Cedar------e"--\if:1 I lilt\ `s- --I—.I11■1111 1116Fig. 8.3. Actual (bars) and simulated (dotted lines) diameter distributions for western hemlock andwestern red cedar in CH stands after running the model for 4000 years.HemlockI1I7II11 1 —rl—t—N_IIIIIIII10. 90. 50. 70. 90. 110.130.150.170.190.210.230.250.270.10080to 60.c\aEa 402010. 30. 50. 70. 90. 110.130.150.170.190.210.230.250.270.mid-point dbh class (cm)TO.c>........^800ea0E 600o171c4002coel 2000.120010001000 2000 3000 40007.c......^800aaaE 600oac•co400o2cmel 2000A.1t^0010001000^2000^3000^4000^cH^HAFig. 8.4 (a) and (b) Simulated above-ground biomass for four different scenarios, and (c) meanabove-ground biomass calculated from 3 stands in each of the CH and HA forest types.(a)117Year(b)^(c)Year Measured valuesCedar onlyHemlock only _^ 1.----1 1000^2000 3000 4000(b)100 ...Initialisation period for CH —0-6080HACH1000^2000^3000^40004020118Fig. 8.5 (a) and (b) Simulated nitrogen availability (annual net N mineralisation, not including treeuptake) in the forest floor of the four scenarios.119Above-ground, live biomass in the simulated CH stand (Fig. 8.4b) varied between 550 and 630Mg/ha, slightly higher than the mean 'measured' value from three stands of about 560 Mg/ha. Thedivision of the biomass between the two species is not shown in the figure, but it was also relativelystable over the last 1000 years of the 4000 year simulation with hemlock comprising about 25% ofthe biomass. In the stands measured about 22% of the basal area was hemlock and fir.Above-ground live biomass in the simulated HA stands varied widely because of the periodicdisturbance, peak values were between 500 and 600 Mg/ha at a stand age of 300 years, and thebiomass of simulated 100 year-old stands varied between 300 and 400 Mg/ha. This compared quitewell with the 320 Mg/ha measured in the stands disturbed by the major windstorm about 85 yearsago. Biomass in the hemlock-only scenario (Fig. 8.4a), with no disturbance, stabilised between 500and 550 Mg/ha, and in the cedar-only scenario at around 800 Mg/ha. Cedar biomass was greaterbecause it can attain larger sizes, and therefore it can accumulate a considerably greater mass in eachstem than hemlock.Simulated available N (Fig. 8.5) stabilised at about 27 kg/ha/yr in the CH scenario, andvaried between 37 and 43 kg/ha/yr in the HA scenario. The variation in the HA scenario was largelybrought about by the flush of N in the litter deposited on the forest floor in the simulatedwindthrow. Although they both began with similar high levels of N, the single species simulationsshowed similar differences in nutrient availability after about 500 years of simulation. Available N inthe cedar-only scenario was almost identical to the equilibrium value in the CH scenario - 27kg/ha/yr. In the hemlock-only simulation available N stabilised at about 44 kg/ha/yr. As statedpreviously, no field estimates of annual N mineralisation are available with which to compare thesesimulated estimates. However, the simulated values for the CH stands are near the value of 20-30kg/ha/year estimated by Weetman et al. (1990), and all the values are within the range of valuesreported for the coastal forests in the Pacific Northwest (Edmonds et al. 1990). The differentialbetween the two forest types is similar to the differences in N availability measured in seedlingbioassays, and KCI extractions by Prescott et al (1993b).Simulated forest floor accumulation under the different scenarios is shown in Fig. 8.6. The120total forest floor mass in the CH scenario is about 500 Mg/ha, about 140 Mg/ha less than the 640Mg/ha measured in the CH stands (Chapter 4). Interestingly, this underestimate was about the sameamount that the model overestimates live biomass. This could have been due to the effects of the1906 windstorm, which resulted in higher mortality among larger trees than the constant exogenousmortality simulated in the model, with less live, and more detrital, biomass. It is therefore possiblethat the model relationships and parameters are simulating organic matter dynamics more accuratelythan might be suggested by these discrepancies. The proportion of the forest floor contained in thewoody and non-woody L, and F & H layers corresponds fairly well to the measured values, a furtherindication that the input rates and decay parameters for these variables in this forest type are set atappropriate levels. Similar conclusions can be drawn from the figures from the HA simulation.Forest floor biomass fluctuates considerably because of the simulated disturbance, but the forestfloor biomass at 100 years after each windstorm (marked with arrows) is a similar magnitude to the440 Mg/ha measured in the 85 year old stands. In the CH type, simulated F and H layer was largerin the simulation than measured in the field.Impact of varying N availability, caused by differences in characteristics of speciescomprising the previous stand, on simulated growth of plantations of both cedar and hemlock isshown in Fig. 8.7. Cedar growth was only slightly affected by lower N availability on ex-CH sites,and predicted biomass accumulation over 100 years is about 310 Mg/ha on ex-CH sites, and 330Mg/ha on ex-hemlock sites. Current growth projections developed by forest managers in this area(of stemwood volume, which we multiplied by the density of cedar to calculate mass, and by 1.2 toget gross bole biomass and foliage), indicate that cedar stands contain 275 Mg/ha on poorer sites and350 Mg/ha on richer ones (John Barker, Western Forest Products Ltd., pers. comm.). Simulatedgrowth of hemlock, on the other hand, was markedly lower on ex-CH sites. Biomass after 100 yearswas 230 Mg/ha, compared with 380 Mg/ha on the ex-hemlock sites. These values are lower thanthose projected by managers for site indices of 20 and 30 (440 and 540 Mg/ha, respectively).SOOTOOSOOSOO030SOO0101000SOOTOOMOSOO500SOO100100IIIIllemwred rare■ LEl F and HL and F (woody)Ei H (woody)800700a 600.c\ 500la2 400: 300a2 2001000HA800700a 600.c\ 500co)2 400: 300a2 2001000800700a 800.c\ 500al2 400re 300a2 2001000Cedar only 800700a 600.c\ 500co2 400: 300a2 2001000Hemlock only0^1000^2000^300001'Swarmed were121Fig. 8.6 Simulated forest floor accumulation by forest floor layers for the CH and HA scenarios, andmean measured forest floor accumulation in three stands of the two forest types. Simulated forestfloor accumulation for the cedar-only and hemlock-only scenarios. L and F (woody) includes woodydebris and Fw in the forest floor, H (woody) is well decomposed wood.CHYear3000 3050^3100^3150^3200YearHemlock plantations after 3000 years— — hemlock siteCH site ////./'/i//Fig. 8.7 Simulated growth of cedar and hemlock planted at 3000 stems per ha on sites previouslydominated by cedar (CH), and hemlock.Cedar plantations after 3000 yearse 500.cco2 400aaE 300o. 31:2 20000thil 100eoxi47 500.co)2 400aaE 300oB2 200006, 100eo.o^043100^3150^320012203000^3050— — hemlock site^ CH siteYear123It is interesting to note that the simulated growth of hemlock improves in the second rotation, as itbegins to influence N availability.DiscussionThese results suggest that the modified LINKAGES model, using a relatively simplecalibration data set, provided reasonable simulations of population dynamics, organic matterturnover and nutrient dynamics in this environment. Simulated values of live and detrital biomassaccumulation were generally within +/- 20% of measured mean values, and the predicted values forannual N-mineralisation (available N) were within the range of field estimates of average Navailability.Growing the two species in single species stands, with high initial N,resulted in lower Navailability in the forest floor of cedar-dominated stands than those dominated by hemlock. Thisdifference is largely due to the lower N concentration of cedar foliage, which influenced the rate ofN mineralisation in two ways. There was a lower initial pool of N for the same quantity of foliage,and the simulated decay rate of litter was slower, because the lignin:N ratio of cedar foliage washigher. N concentrations used in the model did not differ greatly (0.40% for cedar, and 0.75% forhemlock), however, N feedback mechanisms caused the relatively small variation to have a markedeffect on decomposition rate and, consequently, on N mineralisation. AET in this cool, moistenvironment was comparatively non-limiting, and the lignin:N ratio had a large effect, especiallywhen compared with drier and warmer, or colder environments (Meentmeyer 1978, Pastor and Post1985). Also, both species had comparatively high lignin concentrations (23% for cedar, 24% forhemlock), and small variations in N concentration had a considerable impact on lignin:N ratio.These simulated differences in decay rate were supported by litter bag studies using foliage of thetwo species (Chapter 7).Lower N-availability in the forest floors under cedar compared to hemlock is contrary topreviously reported studies of the two species. Cedar has previously been considered a 'calciumpump', capable of raising the forest floor pH (Krajina 1969), and allowing the development of124microbially-mediated decomposition, compared with fungal-dominated turnover under hemlock(Turner and Franz 1986). However, these latter results were from forests with a drier, morecontinental climate than the one in this study, and Stone (1975) argues that effects of cedar on pHand calcium levels largely disappear at higher levels of precipitation. It has been argued that cedarprefers N available in the form of nitrate rather than ammonium, which may explain the generallack of response to fertilisation and to planting on the HA sites. However, there is no evidence inthis locality to indicate that the presence of cedar can lead to greater rates of nitrification, as found inother areas (Turner and Franz 1985b).Because cedar has a different response function to the availability of N than hemlock, thesedifferences in N availability affect the growth of planted seedlings of each species differently, withcedar growing more slowly than hemlock on the ex-hemlock sites, but more rapidly on the ex-CHsites. There is a small decline in production once cedar becomes established on a site with a higheravailable N; however this does not lead to an ongoing feedback between declining foliar Nconcentration and productivity, and ultimately to zero production, because of the way litter N isincluded in the model. Initial litter N and lignin concentrations are fixed attributes of the species,and do not fluctuate with varying levels of N availability. Many ecologists and foresters wouldsuggest that the N concentration in foliar litter can vary considerably, for example when trees aresupplied with additional N in fertiliser. However, under N-limited conditions the concentration ofN in litterfall will be a relatively constant attribute of the species, determined by the extent to whichtrees can retranslocate nutrients prior to senescence. This 'base-level' of N is a function of thechemical composition of leaf structural components (maybe 'lignin-bound' N, Berg and Theander1984), and this will vary between species, and in turn determine the effect of species on Navailability. Because the N concentration can vary, it is therefore important to determine the species'N concentrations used in the model using foliar litter from N-limited sites.If foliar litter contains more N than this base structural amount, then this extra N consists ofmore labile proteins and amino acids, and it is logical to assume that if the species is not activelyretranslocating these compounds prior to senescence then its growth is not limited by N. If this is125the case it is also likely that the decomposers are not N-limited, and the foliage may also notdecompose any faster; in fact, decomposition can be inhibited by high N concentrations (Berg andTamm 1991). If the N-response function in the model is calibrated correctly, increases in litterfall Nfor the same species on more N rich sites should not result in any growth increase because they willoccur at, or above, the N 'saturation' level for the species.The idea that species characteristics can influence nutrient availability, and consequently,forest composition, is not new. Much of the early work investigating differences between mull andmor humus forms arrived at this conclusion (Rommell 1936, Handley 1956). However, theextension of these findings to the kind of logical conclusion described by Gosz (1981), wherefeedbacks between litter quality, nutrient availability, and production can lead to alternative positiveor negative spirals in ecosystem productivity may not be entirely appropriate. The assumptionsinherent in LINKAGES indicate that these feedbacks are bounded at the lower limit by the degreeto which a species can retranslocate N, which varies with the species and the chemical structure of itssupporting tissues, and at the upper limit by the level at which growth no longer responds toincreasing N availability. There is considerable variation among conifers in this upper limit, but it isrelatively low compared to deciduous species. Instead, what may occur are shifts in rates of N-cycling and forest productivity as forest composition changes from one set of species to another.These shifts occur in the LINKAGES model, given the interaction of life history characteristics,environment, and disturbance events, and lead to alternative stable equilibria in the absence ofdisturbance or climatic shifts. These are probably more common ecological states than thecontinuously declining or aggrading conditions that Gosz's (1981) conceptual model implies.Climatic conditions are also important in determining the extent to which species influence Navailability, because of the varying influence of the lignin:N ratio in different climates.A further factor, not varied in the model simulations, that could increase the simulateddifferences in N availability between the two types was the rate of humus turnover. This is specifiedas input to the model and was set at 2% per year for both types. However, the results in Fig. 8.6suggest that the while this figure is appropriate for the CH type, it may be low for the hemlock126dominated stands. This is further supported by the results of standard chemical and NMR imaginganalyses of the forest floors in the two types (de Montigny et al: 1993), which indicate that there aresome important differences in the chemical composition of the humus between the two types,perhaps due to tannins originating in salal, which is leading to N being immobilised in stable tannin-protein complexes.Two other factors have the potential to further reduce the nutrients available for the growthof planted seedlings, and were not included in the management scenarios used to produce Fig. 8.7.The first is the shrub salal. It is the dominant component of the understorey in the CH stands, but arelatively minor component of the HA stands, and it rapidly forms a continuous cover on the CHsites following clearcutting. Measurements in clearcuts indicate that salal can attain pre-harvest levelsof above-ground biomass (about 4 Mg/ha) within 5 years after cutting (Messier and Kimmins 1991,Chapter 6). Allocation above-ground to leaves rather than stem is about 75% in the clearcuts, butonly 25% in the uncut stands. Salal also allocates a considerably greater proportion below-ground inthe clearcut areas. Taking just the above-ground leaf component of 3 Mg/ha at five years, andmultiplying by the leaf N concentration of 1% (Chapter 7), indicates that salal can take up at least 6kg/ha/yr of N during the early years after cutting. This would decrease the N available for conifergrowth to about 20 kg/ha/yr, further reducing seedling growth.Secondly, the management practice of slash burning could have a significant impact onnutrient availability in the short-term. During the modelling, the importance of maintaining theactively decomposing layer of forest floor became apparent. The majority of available N was releasedduring the first few years of foliar litter decomposition. Following this, the rate of N release wasdetermined by humus turnover, which was quite slow. If tree establishment was not rapid enoughfor them to take advantage of the nutrient flush created by a disturbance and rapidly recreate a forestnutrient cycle, then growth was very slow and the establishment of a fully stocked stand took a longtime. This was part of the reason the CH simulation took so long to reach a steady state. Burning oflogging slash and the actively decomposing L layer following clearcutting can produce a small, short-term increase in N-mineralisation, but it can also remove a considerable portion of the labile pool of127N in these ecosystems, and may substantially reduce long-term N availability. Thus, while the largeamounts of humus in these ecosystems gives the impression that they are well buffered againstdisturbance impacts such as fire, these considerable stores of N are not readily mineralised. Thecontrary conclusion that ecosystems containing large accumulations of detrital biomass have lowamounts of available nutrients, may well be a more general one, and managers should take everyeffort to conserve the available nutrient pool in these forests. High detrital accumulations,particularly in temperate environments such as the one in this study, probably indicate that much ofthe organic C and N in the system either begins in, or can quickly become part of, chemicalcomplexes that are resistant to decomposition. This phenomenon is probably also a majorcontribution to the differences in N-availability between mull and mor humus forms (Handley1954).The forest floor conditions created by cedar may lead to the development of conditions inwhich cedar is able to grow more rapidly than hemlock despite lower N availability. This suggeststhat, in the absence of a catastrophic disturbance, the CH type is a self-sustaining system in whichhemlock will remain a smaller component. However, this will also depend on the relative abilities ofthe two species to establish under a canopy. If hemlock is more successful at this stage of its lifehistory, then it may eventually achieve a greater dominance. It has been speculated that the CH typeis the climatic 'climax' for this locality, eventually replacing hemlock-dominated stands (Terry Lewispers. comm.). However, the mechanism for the transition has not previously been identified. Bothspecies similarly high shade tolerance (Carter and Klinka 1992), and both are capable of establishingon organic substrates, such as decaying logs (Chapter 3). However, the mechanism could be thedifference in growth response to N-availability. Salal may mediate this development by invadinglater stages of the hemlock stands once gaps begin to form in their canopy, and sufficient light isavailable at the forest floor. Salal would lessen the dominance of hemlock by competing with thehemlock advanced growth for light and nutrients, while cedar, once established, is more tolerant oflower nutritional conditions. However, once the cedar-salal association has developed it is likely to128take a particularly severe disturbance to allow hemlock to re-establish dominance, because of thelasting effect that the association has on forest floor nutrient mineralisation.There is considerable potential to improve various aspects of this investigation. The responsefunctions for each species to varying rates of annual N mineralisation need to be developed in amore thorough way, using growth information for the two species planted on a wider range of sites,perhaps using data from fertilisation trials. The rates of turnover of the various forest floorcomponents shouls be investigated further, and there was no field information available on the decayrates and nutrient dynamics in decomposing fine roots. Differences in root nutrient content anddecay rates could make a further substantial contribution to the observed differences in Navailability. To more accurately depict these ecosystems it would be valuable to include the dynamicsof understorey species (particularly salal), because the biomass of the understorey and its effect onlight available for seedling establishment is probably an important determinant of the dynamics ofthese ecosystems. To further understand the mechanisms involved in the efficiency of nutrientuptake and nutrient use by different species, and to simulate how these might affect the competitivedynamics among the different species, a more mechanistic model of nutrient cycling would berequired.ConclusionsThe results of this study tend to support the hypothesis that the characteristics of thedominant coniferous species, particularly lower foliar N concentration, can explain much of thedifference in N availability between the two forest types. The interaction of coniferous litter andsalal exudates may be contributing further to lower rates of N availability in the CH type, byforming stable protein-tannin complexes deeper in the humus profile. The management practice ofslash burning may further lower nutrient availability by burning off much of the foliar litter fractioncontaining most of the labile N. The small amount of N mineralised during the burn is rapidlytaken up by salal. The lower N availability caused by these four factors has a greater impact on theproductivity of more nutrient demanding species such as hemlock and spruce than it does on thegrowth of western red cedar.129130Chapter 9. ConclusionsIntroductionThe aims of this study were to (i) describe the differences in structure and functioningbetween the two dominant forest types growing on well-drained to somewhat-imperfectly drainedsites on northern Vancouver Island; and (ii) to determine the mechanisms associated with differencesin functioning, in particular differences in nitrogen availability in the forest floors between the twoforest types. A second objective was to explore the potential mechanisms and patterns of successionaldevelopment in the two types. As stated in Chapter 1, there had been little previous investigation ofstructure and functioning in unmanaged stands of the two forest types and much of the workdescribed in this thesis is therefore exploratory and descriptive. As a consequence, many of theconclusions drawn below are speculative, based on the evidence currently available and, in mostcases, they present a basis for further investigation through more detailed description, analysis,experimentation and synthesis.Results of the investigations described in the body of the text indicated that the two foresttypes had significant structural differences, largely due to differences in response to the catastrophicwindstorm that struck the area in 1906. These structural differences are also indicative of differentrates and patterns of regeneration of the dominant species. The evidence that remains of thestructure of forests in the HA type prior to the windstorm suggests that the current differences inspecies composition between the two types has persisted for some time. At two of the three sitesinvestigated, the HA type, now dominated by relatively even-aged, second growth stands of westernhemlock and amabilis fir, had an older, more uneven-aged structure prior to the storm. While theymay have been more open stands with a greater biomass of understorey plants, these 'old-growth'HA stands probably contained a large bank of hemlock and fir seedlings as advanced growth in theunderstorey, similar to that currently present in the HA stands, that enabled these species to respondrapidly to the windstorm and to re-establish their dominance on the site.131On the other hand, the CH type was dominated (in terms of basal area) by western red cedarindividuals up to 1000 years old, with a relatively large number of hemlock in smaller size classes.Cedar is present in low numbers in all size classes, but there appears to be sufficient regenerationpresent for it to continue to maintain its current level of dominance. The hemlock in these standshave a reverse-J diameter distribution, with a large number of trees in smaller size classes and fewstems in the upper part of the canopy. This canopy is considerably more open than in the HA type,and the biomass of understorey plants (largely salal) is much greater (Chapter 6). The 1906 stormhad a variable impact in the CH type. Tree boles were deposited on the forest floor, and canopyopenings created, but the effect was much more sporadic than in the HA type. This is likely to bedue to the greater wind firmness and bole strength of the larger cedar, and the heterogeneity of theCH canopy. These large downed boles and root mounds, that are the product of windstorms overcenturies, now form the majority of sites for establishment of young trees of both species.The study identified differences in functioning between the two types. The mass of litterfall,and the differences in the amount of N cycling in the litterfall was lower in the CH type. The rate ofnutrient resorption at the time of leaf senescence by western red cedar was higher than that ofhemlock on the CH type, and the rate of resorption by hemlock in the CH type was greater thanthat of either hemlock or amabilis fir in the HA type. The rate of decomposition of the litter of thedominant species (cedar) was slower on the CH type, whereas hemlock in both the CH and the HAtype decomposed more rapidly than cedar. These differences were due to differences in substratequality, rather than climatic effects, because a standard substrate, lodgepole pine needles,decomposed at almost the same rate in each type.In the beginning of the study, three broad hypotheses were put forward that could havepotentially explained the differences in productivity and nutrient availability identified by others,and the differences in functioning identified during this study. These hypotheses are discussed inturn below.132Causes of differences in functioningSiteLewis's (1982) conclusion, that differences in site characteristics did not explain differencesbetween the two types in structure, productivity or nutrient availability, was supported by this study.Mineral soil properties varied among three sites where detailed investigations were undertaken, butthere was little difference between the two forest types at each site. HA stands occur on well-drainedsites, as well as on soils with a high water table for most of the year (the SCHIRP site), and CHstands were found on upland, water-shedding situations (the Rupert site). Studies of decompositionshowed that the same substrate decomposed at similar rates in each type, indicating that conditionsfor decomposers at the surface of the forest floor were similar.There is one pattern that was fairly consistent in observations across the study area, andsurrounding areas with subdued topography. As mentioned in Chapter 2, the well-drained, tosomewhat imperfectly-drained situations in this area form a mosaic with lower lying poorly-drainedareas of bog, and bog-woodland, fen and swamp forest that Lewis (1982) classified as the S6-S11ecosystem associations. In general, it is the CH type that is contiguous with these lower lyingsystems, and the HA type is not commonly adjacent to them. This observation has some significancein the consideration of the dynamics of the forests in the area, and it will be discussed below.DisturbanceThe disturbance hypothesis had two components. The first was the influence of windthrowon soil physical properties. There has been considerable attention paid locally to the potentialinfluence of windthrow in maintaining higher stand productivity. This influence was thought to bedue to the physical effects of mixing mineral and organic horizons and the improved aerationconditions that such mixing creates for microbial or fungal decomposers, based on the beneficialeffects of windthrow (Armson 1977, Ugolini et al 1990), and mechanical cultivation (Burger andPritchett 1988) that have been observed in other areas.However, in a test of the effects of a mechanical mixing treatment, that was partly intendedto simulate the effects of windthrow, the treatment had a negative effect on most indices of soil133nutrient availability and microbial activity, particularly in the CH type (Chapter 6). Measurementsof the growth of seedlings planted in these treated areas indicated that both cedar and hemlock havea small growth response to the treatment (Messier et al. 1993, Thompson et al. 1993). This couldbe attributed to the reduction in the cover of salal, due to the removal of salal rhizomes from themixed soil during the treatment. Therefore, the physical effect of disturbance by wind on soilproperties did not appear to be the major cause of observed differences in ecosystem functioning andproductivity. Wind may have an important influence on forest functioning, however, by creatingconditions that favour the regeneration strategy of hemlock and amabilis fir at the expense of cedar.The second component of the disturbance hypothesis was that different periods of time sincea major disturbance has led to differences in the accumulation of organic matter in the two types.Deeper organic matter accumulation may provide an explanation for differences in functioningbetween the two systems, because it may lead to changes in the physical and chemical characteristicsof the forest floor, such as lower temperatures and therefore lower rates of nutrient mineralisation.Organic matter accumulates over time since disturbance in most temperate and boreal forestecosystems. This comes about because disturbances such as fire consume organic matter, andtemperature and moisture conditions are more favorable for decomposition following disturbances.A large disturbance causes a major reduction in inputs of litter from the canopy and, with fasterdecomposition and lower inputs, the amount of forest floor organic matter declines (Sprugel 1986).After the canopy closes, litter inputs return to pre-disturbance levels, and particularly in coniferoussystems, forest floor temperature and moisture tend to decrease and decomposition slows down.Consequently, organic matter accumulates over time since disturbance. This phenomenon is largelyindependent of species, and may be accelerated by the generally higher proportion of woodycompared to foliar litter entering the forest floor later in stand development (Vitousek et aL 1988).Steady state accumulations of organic matter, where inputs are balanced by decomposition, arereached in most forest types. These levels of accumulation are generally highest in coniferous forestsat high latitudes and altitudes, where decomposition is slowest (Vogt et al. 1986), and in some highlatitude situations positive feedbacks between organic matter accumulation, changes in soil physical134properties (temperature and depth of permafrost), slower decomposition, lower nutrient availabilityand declining productivity may lead to the replacement of forests by bogs or bog-woodlands if adisturbance does not occur (Van Cleve et al. 1991).Estimates obtained during this study indicated that there were large accumulations of detritalbiomass in and on the forest floor in both forest types (Chapter 4). Differences in functioning andnutrient availability cannot be ascribed to the kinds of physical changes brought about by theaccumulation of organic matter that have occurred in biomes like the taiga or boreal forest (Heilman1966, 1968, Van Cleve and Yarie 1986). However, the deep organic accumulations observed in theHA type may be a transient condition, because there was such a large mass of material deposited onthe forest floor in these stands during the 1906 windstorm. The current high level in HA forestsmay decline over time to a lower steady state if no further disturbance occurs. Nonetheless, thecurrently high quantities appear to preclude the organic matter accumulation hypothesis.It is possible that differences in the stage of development of the two stands, caused bydifferent patterns of disturbance, could have led to differences in functioning. Vitousek et al. (1988)put forward the argument that allometric changes later in stand development produce higher inputsof woody compared with foliar material in above ground litter, and that this leads to a greater forestfloor C/N ratio in stands in later stages of development that consequently influences nutrientmineralisation. However, the mass of woody material deposited during the windstorm tends to thenegate this argument for this particular study, and where windstorms are the dominant disturbance,the amount of wood in the forest floor may be high during all stages of stand development. Woodydebris can also be high during the early stages of stand development in situations where wildfire ismore frequent (Spies et al. 1988).Vitousek et al. (1988) argued that a high overall ratio of C to N in the forest floor can leadto lower nutrient availability, but this assumes that woody and non-woody material becomeintimately associated in the forest floor, and that nutrients are readily transferred from low C/N tohigh C/N material. This view of the forest floor is probably too simplistic, and observation suggeststhat woody and non-woody fractions generally remain more spatially separated until very late stages135of decay. Concentration of nutrients in decaying wood does increase with decay (Chapter 4), andwhile this suggests immobilisation, the extent to which wood decay organisms (mainly fungi)actively transfer these nutrients from the labile nutrient pool is uncertain. Labile N is being added tothe forest floor in throughfall and atmospheric deposition, and this may satisfy much of theimmobilisation potential of woody substrates. There is also considerable evidence that N-fixationoccurs in decaying wood (Sollins et al. 1987, Hendrickson 1991). Roots are more intimatelyassociated with wood in the forest floor, particularly in later stages of decay (Chapter 4, Table 6),and there may be a greater transfer of nutrients from root litter to wood.Perhaps the most striking evidence for the lack of association between the amount of forestfloor wood, nutrient availability, and consequently forest productivity, is the pattern of developmentof the HA stands following the windstorm. It appears that there was high nutrient availability andhigh productivity following windthrow, despite the very large quantity of wood deposited on theforest floor. While N-fixation may have occurred and could still be occurring in this decaying wood,the rate of fixation and rate of subsequent release to the labile pool would not be fast enough tosupply the demand of a rapidly growing forest. The windstorm also deposited large amounts of greenfoliage on the forest floor. The decomposition of this nutrient rich foliage, and subsequent rapidcycling once the HA canopy was established, probably sustained the growth of these stands.Modelling of the organic matter dynamics indicates that wood will be a major component of theforest floor in both forest types, even if no disturbance occurs (Chapter 8, Fig. 6), but thedifferences in the nutrient availability and productivity will persist. However, differences in thedecomposability of the wood of the different dominant species will lead to differences in the steadystate value of accumulated wood.SpeciesThe above discussion indicates that site differences (topography and geology) were not themajor cause of the observed differences in functioning and productivity, and that these differencescould not be explained by type and frequency of disturbance, or stage of development of the two136forest types. The only other predominant difference that had the potential to explain the differencesin functioning was the difference in species composition.The characteristics of the species in the two forest types that appear to be having the greatestinfluence on ecosystem functioning, are: (i) the poorer litter quality of western red cedar foliagecompared with western hemlock and amabilis fir, in particular differences in the lignin/N ratio, thatled to slower rates of decay and lower rates of forest floor nutrient mineralisation; (ii) the larger bolesize and higher concentrations of biotoxic substances in cedar wood, leading to a slower decay rate ofwestern red cedar wood compared with the other two species (Sollins 1982, Graham and Cromack1981, Sollins et al. 1987); and (iii) the influence of tannins and phenolics leaching from salal (deMontigny et al. 1993).The accumulation of woody material contributes to the substantial forest floors found inboth types, and a considerable store of nutrients gradually accumulates in this material (Chapter 4).However, the input of nutrients to the forest floor in woody litter is only a small fraction of the totalnutrient input in litter, and it is spatially and temporally heterogeneous. The rate of decay of woodymaterial is relatively slow, and therefore the rate of release of these nutrients is slow compared to thatof foliage. In the modelling exercise in Chapter 8 it was found that N availability in the model wasvery insensitive to differences the rate of wood decay (unreported data). Thus, while differences inwoody litter quality does have a small affect on the rate of nutrient mineralisation, the functioningof these ecosystems is probably relatively insensitive to differences between species in the rate ofwood decay.Salal dominates the understorey of the CH stands and rapidly reoccupies this type followingclearcutting and burning by resprouting from underground rhizomes, but is largely absent from theHA type. The competitive influence of salal on conifer regeneration has been well-documented(Weetman et al. 1990, Messier 1991, Messier and Kimmins 1993), and there has been someattention paid to its potential influence on forest floor chemistry in the CH type. de Montigny et aL(1993) used 13CPMAS NMR and cupric oxide oxidation to compare the chemical composition ofdifferent layers of the forest floor in the two forest types. They found no major differences in the137chemical composition of decomposing woody horizons in the two types, but minor differences werefound in the non-woody horizons. Comparison of the o-alkyl C content and ratio of carbohydrate tolignin monomer units indicated that carbohydrates may be more effectively decomposed in the HAtype. The ratio of vanillic acid to vanillin (acid to aldehyde) was higher in the HA type, indicatingmore extensive decomposition of lignin, and there was higher tannin content in the non-woodyhorizons of the CH type, that may have originated in salal.Thus, there is an indication of more extensive decomposition of carbohydrates and lignin innon-woody forest floor material from the HA type, and a retention of tannins in the CH. Tanninsare potentially significant because they can reduce the biodegradability and humification of organicmatter by precipitating proteins, coating non-proteins such as cellulose and hemi-cellulose and byinactivating enzymes. The slower and less complete breakdown of carbohydrate and lignin C infoliar and root litter could explain some of the differences in N availability between the two types.However, the breakdown of these substances, and the coating of proteins and non-proteins, arelikely to more significant during the second stage of decomposition (Aber and Melillo 1990), whenmaterial is deeper in the forest floor. Thus, it probably explains the greater accumulation of biomassin the H layer of the CH type (Chapter 4, Table 5). However, it may not have a great impact onnutrient availability, because the majority of nutrients are mineralised during the first stage, which islargely controlled by intrinsic properties of the decomposing litter. The rate of nutrient supply fromthe deeper layers of the forest floor may be an important consideration if the L layer is removed, andthere is a lack of litter input for a number of years. This situation occurs when the forests areclearcut and the slash is burned.Thus, while all three factors influence nutrient cycling and availability, the analysis presentedusing a computer model of nutrient dynamics in decaying litter (Chapter 8) indicated that foliarlitter quality has the greatest potential to influence nutrient availability, and the subsequentproductivity of seedlings planted on clearcut sites. This is consistent with other studies that havefound a close connection between foliar litter quality and N mineralisation in the forest floor (eg.Stump and Binkley 1993).138The general evidence indicating that species characteristics may affect nutrient cycling andproductivity has been reviewed previously (Chapters 1 and 8). The evidence from this study suggeststhat the two dominant coniferous species establish alternative nutrient cycling regimes in theirrespective environments, a phenomenon that has also been observed for the same species in anotherlocality (Turner 1984). The principle components of these regimes are: (i) the chemical properties ofthe leaf leachate, litter and soil; (ii) microbially mediated transformation rates of organic matter tomineral nutrients; and (iii) the nutrient uptake and assimilation characteristics of the tree species.While the quality of foliar litter has been a major object of investigation in this study, a considerableproportion of forest production is also directed below-ground, and differences in the quality of rootlitter may be having as great, or greater, impact on the turnover rate of organic matter and nutrientavailability as foliar litter does.If cedar has advanced in the area by spreading out from poorly-drained sites or poor, drierrocky ridgetop sites, then it has originated in conditions with relatively low N availability. In N-limited conditions carbon is more available to plants through photosynthetic activity, relative to N,than in N-rich conditions. It appears that cedar, through being able to tolerate the poorer nutritionalconditions, and the poorer drainage associated with these sites, has an advantage over most of theother conifers found in this area. Cedar has a relatively high tolerance to disease and insect attack(Minore 1983), and therefore appears to have the capability of directing more of this excess C todefensive, secondary compounds than other conifers. This may be one of the reasons why it cantolerate the more stressful situation. In these N-limited situations (and the two phenomena may wellbe physiologically connected - Chapin et al. 1987) cedar resorbs a greater proportion of N and Pfrom foliage prior to senescence, and its litter quality is reduced, and therefore N transformationrates in the forest floor under cedar are lower than under hemlock. This lower mineralisation rateslows the growth of hemlock to a greater degree than cedar, enabling cedar to expand slowly acrossthe landscape.There may be some symmetry in this competitive relationship. There is some evidence toindicate that cedar can grow rapidly with added nitrate-N, but shows no significant response to139added ammonium-N (Krajina et al 1973). While hemlock litter creates a forest floor with higherrates of N-mineralisation, virtually all of the N is produced as ammonium (Prescott et al. 1993b).This enables young hemlock to grow rapidly because they respond to additional ammonium, but iteffectively prevents cedar from taking advantage of any increased resource that might occur if thehemlock canopy is disturbed.Competition theoryGiven that the differences in ecosystem functioning cannot be attributed to sitecharacteristics, and that disturbance has a mainly indirect effect by influencing species composition,the predominating influence of species composition on functioning and nutrient availability can beconsidered in terms of plant competition, and the way the two dominant species, western red cedarand western hemlock interact and influence the availability of resources used for plant growth. Twotheories that attempt to explain the mechanisms of plant competition are the plant strategy model ofGrime (1979), and the resource ratio hypothesis of Tilman (1988). A common theme of theseconceptual models is that the competitive abilities of plant species interact with resource availabilityand disturbance to determine vegetation composition, with one of the key resources being light(Austin 1986, Keddy and MacLellan 1990).In simple summary, the plant strategy model of Grime (1979) predicts that those plants thatare the best competitors for light (i.e. that grow the tallest, fastest) will also have the most resourcesavailable to put below ground and extract water and nutrients, and therefore will continue to growthe fastest. Under Grime's theory, differences among competing species in resource acquisition rates,once established, are maintained and magnified during competition, because of this positivefeedback between growth and resource capture. Grace (1990) called the mechanism involved in thiskind of competition 'resource pre-emption'.On the other hand, the resource-ratio hypothesis (Tilman 1988, Wedin and Tilman 1993) isthat the species that can reduce the concentration of the limiting resource to the lowest level and stillmaintain its population (i.e. the one with the lowest resource requirement, or R*) will be the140superior competitor, rather than the species that can grow to the largest size, the fastest growing, orthe first to colonise. Tilman argues that the level to which a species competes for, and consequentlyreduces light or nutrients depends on its allocation pattern, and that there is an inherent trade-offbetween competitive ability for above- and below-ground resources. Grime's theory predicts thatcompetitive rankings of species should be constant along productivity gradients, whereas Tilman'stheory suggests they should change (Wedin and Tilman 1993).In this study, both major species have relatively low light requirements (i.e. high relativegrowth rates at low light levels) although hemlock has a higher absolute rate of growth in fullsunlight than cedar (Carter and Klinka 1992). Water is not generally limiting (Lewis 1982), andtherefore the primary resource in short supply is soil nutrients, in particular N. The presence ofcedar on more nutrient-limited bogs, low-lying situations and rocky outcrops, higher rates of Nresorption in foliage prior to senescence, and higher nitrogen use efficiency of cedar-dominatedstands in N-limited conditions, all suggest that it is a superior competitor under low-N conditions,and has a lower R* for N. However, it is generally absent from the HA stands that have higher Nmineralisation. As a result, there does appear to be a trade-off between acquiring N more efficiently,and competing effectively for light under conditions of high N availability, which tends to supportTilman's theory.A further justification of Tilman's resource reduction argument comes about through theway cedar utilises N and the way that this feeds back into soil N availability. This argument suggeststhat a more competitive species reduces the level of available resource (i.e. it intercepts light becauseit grows taller, or it obtains more soil nutrients through a more efficient root structure, mycorrhizalassociation, or by accessing nutrients in organic forms). This leaves less resources available for itsown progeny and for other species, but, as long as the resource is not reduced to below the R* ofthat species, the progeny will be more competitive than those with a higher R*. However, by havinga higher potential rate of N resorption at the time of leaf senescence, cedar also reduces forest floorN-availability by having a lower quality of litter for decomposers. This kind of resource-reductionmechanism has been previously identified in studies of grassland species Wedin and Tilman (1990),141and is the basis of some classical explanations of competitive interactions between coniferous species(Gordon Weetman pers. comm.).A number of other factors, such as disturbance or herbivory, determine whether suchmechanisms will lead to the long-term exclusion of one species at the expense of another. Whilethere are differences in the R* of cedar and hemlock for N (as suggested by Fig. 3 in Chapter 8),they are relatively small in comparison to a wider range of species, and therefore it is unlikely thathemlock will ever be completely excluded from the CH type on the basis of competition for Nalone.SuccessionLewis (1982) hypothesised that there was a successional pathway from the HA to the CHtype, and that in this environment western red cedar is the more important species later in thesuccessional sequence. However, the evidence for a successional linkage, given the longevity of thetree species, the time spans involved in developmental change and the frequency of disturbancessuch as windstorms is largely inferential, as is the case for other forest types in this region (Franklinand Hemstrom 1981).Succession can generally be defined as the process whereby one assemblage of plants replacesanother on the same piece of land. It has often been divided into two distinct categories, primaryand secondary succession. Primary succession is the development of vegetation on areas that arecompletely bare of all plants, propagules, or organic material, and with little soil development: forexample, the conditions occurring after glacial retreat, major earth movement or open-cut mining.Secondary succession begins after disturbances such as wildfire or windstorms, which kill most of thecurrent plant population but leave intact more-or-less mature soils often containing a sizable bank ofseeds or vegetative propagules (Macintosh 1981, Crawley 1986, Pickett et al. 1987). Secondarysuccession is the phenomenon that can potentially lead to the development of the CH type from theHA type in the study area.142In considering succession it is important to distinguish between the following concepts(Pickett et al. 1987):1. Pathways; the temporal patterns of vegetation change;2. Causes, an agent, circumstance or action responsible for succession patterns. A cause may bean interaction of mechanisms;3. Mechanisms, i.e. interactions that contribute to successional change; and4. Models, conceptual constructs to explain successional pathways.The biogeoclimatic ecosystem classification system developed for British Columbia uses thevegetation considered to be the 'climatic climax' for a region as a basis for defining broad geographiczones (Pojar et al. 1991). This system gives some consideration to successional pathways, butmechanisms, and causes of succession are not extensively considered. Because of its assumed greatershade tolerance and its ability to germinate under the canopy of the overstorey in humus or decayingwood, western hemlock is considered the climax species for this locality. More recent study (Carterand Klinka 1992, John Karakatsoulis, per. comm.) indicates that cedar and hemlock both have ahigh degree of shade tolerance. Hemlock also dominates the seral stands establishing after majorwindthrow, indicating that it can be relatively opportunistic in its resource utilisation (Chapter 3).Hebda (1983) suggested that the facilitation model of Connell and Slatyer (1977) may be applicablein this area, with the expansion of western red cedar requiring a higher degree of organicaccumulation and soil maturity facilitated by the occupation of the site by other species. But thisstudy identified no consistent differences in mineral soil properties between cedar and hemlockdominated types.The mechanism for the hypothesised successional transition from one type to the other hastherefore been unclear. Competition for resources has been considered one of the major mechanismsinvolved in successional change, and most attention has been focused on the role of light and theway its availability changes during the succession. Less attention has been given to varying levels ofnutrients and water, which are generally considered as fixed attributes of the site, that do not143fluctuate with shifts in vegetation (Botkin 1993). Because of consistent rainfall throughout the yearand cool cloudy summers, water is generally not limiting to tree growth in the study area. Light inthe lower canopy also needs to be at very low levels before the growth of any of the dominant speciesis substantially reduced (Carter and Klinka 1992). Therefore, competition for nutrients is probablyan important part of any successional mechanism. However, species longevity, pattern and mode ofregeneration, the presence of salal, and the incidence of disturbance are also important parts of theprocess of succession in this environment.The discussion above has described how the dominant tree species create alternative nutrientcycling regimes that favour the growth of their own progeny, and how this can lead to competitiveinteractions that may lead to shifts in species dominance. In Chapter 3 a scenario was described inwhich cedar could only expand within the narrow margin near established cedar trees because of themode of reproduction of its mycorrhizal associates. Once cedar becomes established it can graduallyexert an influence on N availability and become more dominant in the forest via the mechanismsdescribed above. If catastrophic windstorms occur, they give other species, such as hemlock andamabilis fir, that maintain a large bank of seedlings as advanced growth in their understorey, anadvantage and the advancement of cedar can be halted. In other areas, geomorphic processes, such ascolluvial or fluvial earth movement, or fire, may allow other species such as Douglas-fir to regenerateand restrict cedar dominance. Although cedar can also establish in these more disturbedenvironments, its dominance is not so pronounced. This very gradual expansion of cedar possiblyexplains why it is a relatively late arrival as a dominant in this locality, compared to other coniferousspecies (Hebda 1983).However, in the modelling exercise described in Chapter 8, it became apparent that, as longas the recruitment of hemlock into the stand as advanced growth remained relatively high, then itwould continue to dominate the stands and maintain a higher rate of nutrient availability andexclude cedar establishment. Thus, to explain the transition from HA to CH, a mechanism wasrequired that caused a reduction in the number of hemlock advanced growth. This could comeabout as a result of the gradually increasing influence of cedar litter on forest floor nutrient144availability at the ecotone between the two types, as suggested above. Invasion by salal may alsocontribute to such a reduction and facilitate the expansion of cedar into the HA stands.Salal has been demonstrated to have an important impact in the early growth and dynamicsof conifers by competing with seedlings for nutrients (Weetman et al. 1989a and b, Weetman 1990,Messier 1991), possibly because its ericoid mycorrhizae can access organic forms of N and P (XiaoGoh Ping pers. comm.). Cedar is more tolerant of this competition for nutrients, perhaps because ofits ability to utilise N more efficiently. However, salal needs a certain amount of light to becomeestablished and maintain itself under a forest canopy (Messier et al 1989), and conditions under the1906 HA stands in this study are generally too dark and it is largely absent. Although it can befound under HA stands along the transitions with the CH type. If a HA stand continues to developundisturbed, then the canopy will become more open, and patches of salal can become established ingaps. Salal seeds are formed in fruits consumed by birds and bears and are more widely dispersedthan cedar seed.Once salal is established it casts a heavy shade on the forest floor, and its leaf leachate mayinhibit conifer seed germination (de Montigny 1992), lowering the number of hemlock seeds thatgerminate and become advanced regeneration. Salal competes strongly with the conifers fornutrients, but cedar is more tolerant than hemlock of this competition and consequently cedar islikely to have an advantage in salal-dominated understorey. Because salal forms a dense cover undermore open forests, conifer seedlings become more dependent for their establishment on the largedecaying boles and root mounds that are provided by the death of large trees.This kind of 'shrub mediated' successional development has been reported previously. Forexample, Rhus typhina increased the survivorship of trees invading a Michigan old-field by thinningthe dense herb cover that formerly inhibited the trees (Werner and Harbeck 1983). However, theplacement of this effect in one of the three pathways described by Connell and Slatyer (1977) isproblematical. The interaction between Rhus and the trees, or in our case salal and cedar, would belabeled facilitation. However, as Pickett et al. (1987) point out such relationships are asymmetrical,and from the point of view of the hemlock, the situation is inhibitory. Furthermore, if the situation145was tested with an experiment carried out late in the interaction, for example by removing salal oncecedar was established, cedar would probably grow faster, and the situation would be labelledinhibitory to cedar. This is the case with current experimentation with salal removal in clearcut areas(Weetman et al. 1990, Messier 1991).Sufficient opening of the HA stands for salal to establish may take a considerable time,perhaps another 200 years beyond the current age of the HA stands in this study. However, olderhemlock stands with an understorey of salal have been observed in and around the study area (TerryLewis, pers. comm.). If another catastrophic windstorm occurs before then, expansion of the cedar-salal association will be thwarted. If a disturbance does not occur, development of a mature cedarforest with a range of size classes and many large trees will take considerably longer, perhaps 1000 to2000 years. Analysis of the pollen record indicates that cedar has only been a dominant species in theforests of the study area for about 3000 years (Hebda 1983). Therefore, cedar may only be in thebeginning of an expansionary phase in this area. The continuation of this expansion will depend onfuture climatic conditions, and the type and frequency of future disturbance.Once a mature cedar forest has established, a catastrophic windstorm is unlikely to bringabout a reversion to a hemlock-dominated forest. Observation of the effects of clearcutting in theCH type indicates that following such catastrophic disturbance the slower N supply from the cedar-dominated forest floor persists, and that if present prior to cutting, salal can respond rapidly to theincreased light availability. Both these factors prevent the rapid re-establishment of dense hemlockregeneration, and contribute to the maintenance of the cedar/salal complex as a relatively stablevegetational condition in this environment, under the current natural disturbance regime.Manipulation by forest managers, through dense planting and fertilisation, presents somepotential to alter this situation, and this is discussed under Management Implications. A warmerclimate with prolonged summer drought, conditions that currently prevail in the xeric subzone ofthe CWH and the CDF zone south of the study area on the east coast of Vancouver Island (Pojar etal. 1991), would lead to an increase in the incidence and intensity of wildfire, and this would changespecies composition, organic matter dynamics and nutrient cycling considerably.146The succession model presented is one in which cedar, probably with the help of salal,expands by gradually exerting its influence within a relatively narrow band of contact with the HAtype. But how does cedar get into the landscape in the first place? The answer here appears to lie inits tolerance to lower nutritional conditions, and to waterlogging (Minore 1983). These conditionsallow cedar to develop in the mosaic of bog and bog-woodlands that occur in low-lying situationwithin the study area. Hemlock and amabilis fir are unable to tolerate such conditions and are absentfrom these environments. From these low-lying 'refuges', or from other microsites in more uplandsituations where impermeable bedrock, high rainfall and low-evapotranspiration have created morewaterlogged situations, cedar can gradually expand into more upland areas.However, the CH forest type appears to be a stable condition for well-drained to somewhatimperfectly drained situations, maintaining a slower rate of nutrient cycling that is consistent withthe nutritional requirements of the dominant species, and there is no evidence to indicate that bogsor bog-woodlands may eventually predominate on upland situations as they can do further north(Banner et aL 1983).Recent research regarding succession has focused attention on the degree of fluctuation inthe global climate and how this may affect the relative ability of plant species to compete for andoccupy the same portion of land. By integrating the results of paleobotanical studies and comparingthem with climatic analyses, Davis (1986) found that even in the short-term, communities do notrespond as units to directional climatic change; instead, the variations in rates of change amongcomponent species cause floral and faunal disequilibrium and change the patterns of speciesabundances that result from competition and predation.The above discussion indicates the mechanisms that could lead to the kind of vegetationalshift hypothesised by Lewis (1982). However, given the longevity of the species in the area,particularly that of cedar, and the factors that potentially make its rate of expansion very low, thetime scales involved in such transitions are long. Thus it is difficult to observe such shifts by directobservation. Recent research has indicated that climate is fluctuating on the order of centuries, asimilar order of magnitude to the life span of coniferous tree species, and therefore it is very difficult147to attribute any perceived shifts in vegetation to the traditional autogenic changes associated withsuccession, as can be done for organisms (even trees) with much shorter life spans. Vegetationalshifts that are considered to be due to autogenic changes associated with stand development andsuccession, could in fact largely be due to climatic variation. This potential is true, not only for thestudy area, but for most areas dominated by long-lived tree species (eg. Lertzman 1991).Management ImplicationsProblems associated with the productivity of planted conifers provided the impetus for thisstudy, and for a number of other investigations into different aspects of the nutrition andfunctioning of these forests. The poor growth of regeneration in the CH type following clearcuttingand slashburning has been identified as being due to nutritional limitations, and these limitationscome about for two reasons: (i) the low rate of nutrient supply from the forest floor, and (ii) intensecompetition for this low labile pool from salal. The reasons for salal being able to compete moreeffectively have been considered by others (Messier 1991, Xiao Goh Ping pers. comm.). The majorcontributors to the low nutrient supply from the forest floor have been identified above. During thediscussion of the factors that contribute to low nutrient supply, the importance of lower foliar litterquality (and root litter quality, although this was not studied) in determining nutrient availabilitywas identified. While this is leading to the nutritional limitations experienced by conifer seedlings,hemlock and Sitka spruce in particular, the management practices being used in the area alsopotentially contribute to the problems that planted seedlings experience, by further reducing thelabile nutrient pool, and by enhancing salal's competitive effectiveness.The current practice on areas managed by Western Forest Products Ltd. is to clearcut theforest, burn the logging slash in the spring, and plant the treated areas with a relatively densestocking (1200 spha) of western red cedar seedlings. Burning was switched from fall to spring toreduce the intensity, and some of the older cutovers where the nutritional limitations were mostapparent may have been subject to more intense fall burns. Burning is carried out for soundoperational reasons. The cutover areas are covered in heavy slash, and sufficient planting spaces are148not available on unburnt sites. However, while the burning is not intense, it does consume all of thefoliage and fine twigs that were contained in the standing biomass of the pre-harvest forest. Some ofthe nutrients in this material are mineralised, but a considerable proportion are lost throughvolatilisation. The effect of burning on nutrient mineralisation has not been measured directly.Some of the mineralised nutrients are taken up by conifers, and the general experience is thatconifers do grow reasonably well for the first 3-5 years, the period in which salal biomass is relativelylow. However, as salal expands above- and below-ground in the high light environment, and it takesup a substantial quantity of nutrients during the first 8 years following clearcutting and burning,when it reaches peak biomass (Messier 1991). Compared to some other shrub and herb species thatrapidly invade and dominate clearcut areas, salal is relatively slow to begin recycling these nutrientsin litterfall. Beyond 5 years after burning, any foliar or root litter present before clearcutting, thatwas not consumed in the slashburn, would have moved into the second stage of decomposition(Chapter 7) and, in the absence of litter inputs from the understorey, nutrient supply from the forestfloor will remain low until tree and shrub litterfall reaches a peak following tree canopy closure.Consequently, the combined effects of (i) burning off much of the fresh or recent fallenlitter, (ii) the immobilisation of most of the nutrients mineralised in salal, and (iii) the lack of freshlitter input from the overstorey, or the understorey, is leading to a condition of severe nutritionalstress for planted seedlings. These are sufficient to cause the death of some planted hemlock andspruce, and while cedar is more tolerant of such deficiencies and generally survives, it also exhibitssymptoms of nutritional deficiencies (Messier 1991).The persistence of these nutritional deficiencies depends on the species. Because cedar cantolerate the deficiencies it can survive and form a relatively closed canopy above the salal. Thisgradual closure can cause salal to allocate more biomass to stem and less to leaves (Chapter 6, Fig. 1),and as a consequence much of the nutrients immobilised in the salal biomass present on theclearcuts may become available to the trees. This could explain the acceleration in growth of cedarplanted on sites dominated by salal once crown closure has been achieved that has been observed insome areas (Paul Bavis pers. comm.). Fertilisation, particularly with N and P, can lead to an149improvement in cedar growth, but whether this is of a sufficient magnitude to justify its costrequires further investigation (Bill Thompson pers. comm.).While hemlock is unlikely to grow well on clearcut CH sites without additional fertilisation,if a dense stocking of hemlock can be established and if a reasonable growth rate can be achievedwith fertilisation, then it may be possible, over time, to create a hemlock nutrient cycling regime inthe CH type. Because nutrient availability is largely a function of the rate of nutrient mineralisationof a relatively small pool of decomposing foliage and fine roots, changing the quality of this pool bychanging the species may not take as long as one might expect. Whether this is desirable from amanagement point of view will depend on whether the objective is to grow cedar or hemlock.lithe objective is to grow cedar, then re-planting the clearcut sites with a relatively densespacing of cedar will be sufficient, although managers should anticipate slow early growth because ofthe period of salal dominance. Planting in a mixture with an N fixing species such as red alder (Alnusrubra) may be of some benefit, but there are likely to be difficulties associated with establishing alderin these thick organic soils.If the objective is to grow hemlock, then more intensive management and repeatedfertilisation of planted hemlock is justified. Fertilisation will be necessary to overcome the nutrientlimitations created by the current forest floor and losses in burning. However, burning appears to benecessary to provide sufficient planting spaces to achieve a dense hemlock stocking. Overcoming theeffect of salal is more problematical, but heavy fertilisation may also have some effect on salaldominance (Prescott et al 1993a).Other practices that merit attention are selection cutting of the larger cedar and regenerationin small gaps. While there are considerable operational difficulties associated with removing the largecedar without damaging the residual stems, there may be some potential advantages associated withapplying a selection system. There are a considerable number of cedar, and hemlock, in smaller sizeclasses in these stands that are currently of little value, but these would increase in value, and attainsawlog size more quickly than planted seedlings if released from overstorey competition. By creatingrelatively small openings, salal may not be able to shift its allocation from above- to below-ground,150and from stem to leaf, as extensively as it does after clearcutting. If cedar were planted in the gaps,they would have little below-ground competition, and being relatively tolerant of competition forlight, it is possible that they could fill the gap and create a denser patch. If this was done on acontinuing basis in the stand, then a more balanced, reverse-J structure could be achieved.Future researchStudies of various other aspects of the functioning of these ecosystems have been under wayfor some time. These include: fertilisation and salal removal experiments, mycorrhizal functioningand soil faunal interactions, and investigations of microbial populations and microbial functioning.Apart from the silvicultural experimentation suggested above, other research projects that would beof value based on the findings of this study include the following.N-mineralisation and availabilityA detailed, long-term study to determine annual rates of N-mineralisation in situ, usingeither the buried-bag (Pastor et al. 1984) or the contained-core (Raison et aL 1987) method wouldprovide a relatively precise determination of rates of N supply in the forest floor. The study couldinclude a comparison of the two forest types, and measurement across a chronosequence of clearcutsites. The contribution of the different woody and non-woody layers of forest floor to N-mineralisation would test the results from the modelling exercise which indicated that most of theforest floor nutrition comes from relatively early stages of decomposition of foliage and fine roots.The 15N dilution studies currently in progress by other investigators will also contribute tounderstanding of N dynamics in these forests.A study of root litter quality and dynamics and the effect of root decay on N availabilitywould fill a major gap in our understanding of the overall pattern of N mineralisation in the forestfloor.N-fixation by free-living bacteria in forest floor and decaying wood may be making asubstantial contribution to N inputs in both forest types, and may explain some of the difference in151nutrient availability in the two forest types. A study of the rates of N-fixation using acetylenereduction would quantify this effect.Long-term monitoring and population biologyContinued monitoring of the SCHIRP experiment (described in Chapter 6) will provide animportant test of the potential to alter nutrient cycling regimes by changing species and/or fertilising.The rates of these potential changes have received little previous investigation, and have importantscientific and management implications.The factors important in seedling establishment, and the natural dynamics of both foresttypes received some consideration during this study. However, longer-term and more detailedinvestigations are required to fully test the hypotheses put forward. This includes a test of factorspreventing the establishment of cedar in the HA stands, and a test of factors restricting seedlingestablishment to logs in the CH type. Such an investigation would include manipulation of lightlevels, seed bed and the requirements for mycorrhizal inoculation by different species.ModellingThe modelling exercise described in Chapter 8 was a first attempt at integrating our currentunderstanding of the different components of these ecosystems and how they interact to affectecosystem functioning. Many aspects of this exercise could be improved, in particular: betterestimation of foliar and stemwood biomass, for the species in this area, particularly that of largecedar; better estimates of the rate of production of new foliage and foliage retention time;information on root decay and nutrient dynamics; better estimates of species growth responsefunctions to differing levels of N availability; improved information on the rate of wood decay, andnutrient dynamics in wood; the inclusion of understorey species growth and nutrient responses; andsome information on the rate of turnover and nutrient mineralisation in decomposing material onceit becomes more humified and moves into the second stage of decay.The aim of Chapter 8 was to investigate the long-term impact on functioning and nutrientavailability of the dominance by different species. From a managerial point of view, and to furtherdevelop an understanding of these systems modelling of the shorter term (up to 100 years) dynamics152is also important. This would involve the simulation of effects such as conifer seedling-understoreycompetitive interactions, short-term growth responses and nutrient fluxes, and quantification of theeffect of different management practices such as slash burning on nutrient dynamics and forestgrowth.153ReferencesAber, J.D., Botkin, D.B. and Melillo, J.M. 1979. Predicting the effects of different harvestingregimes on productivity and yield in northern hardwoods. Can. J. For. Res. 9:10-14.Aber, J.D., and Melillo, J.M. 1980. Litter decomposition: measuring relative contribution of organicmatter and nitrogen to forest soils. Can. J. Bot. 58:416-421.Aber, J.D., and Melillo, J.M. 1982. Nitrogen immobilisation in decaying hardwood leaf litter as afunction of initial nitrogen and lignin content. Can. J. Bot. 60:2263-2269.Aber, J.D., Melillo, J.M., and McClaugherty, C.A. 1990. Predicting long-term patterns of mass loss,nitrogen dynamics and soil organic matter formation from initial fine litter chemistry intemperate forest ecosystems. Can. J. Bot. 68:2201-2208.Adams, D.L., and Mahoney, R.L. 1991. Effects of shade and competing vegetation on growth ofThujap/icata regeneration. West. J. Appl. For. 6: 21-22.Adams, P.W., Flint, A.L., and Fredriksen, R.L. 1991. Long-term patterns in soil moisture andrevegetation after a clearcut of a Douglas-fir forest in Oregon. For. Ecol. and Manage.41:249-263.Agee, J.K., and Huff, M.H. 1987. Fuel succession in a western hemlock/Douglas-fir forest. Can. J.For. Res. 17:697-704.Agriculture Canada Expert Committee on Soil Survey 1987. The Canadian System of SoilClassification. 2nd ed. Agric. Can. Publ. 1646. 164p.Alaback, P. 1982. A comparison of old-growth forest structure in the western hemlock-Sitka spruceforests of SE Alaska. In: Fish and Wildlife Relationships in Old-growth Forests. Edited by W.Meehan, T.R. Merrell and A. Hanley, Symp. Proc. Am. Inst. Fish. Res. Biol., Juneau, Alaska.pp. 219-225.Alexander, E.B., Shoji, S., and West, R. 1993. Andic soil properties of spodosols in nonvolcanicmaterials of Southeast Alaska. Soil Sci. Soc. Am. J. 57:472-475.Allen, T.F.H., and Hoekstra, T.W. 1989. Further comment on Carney's article. Functional Ecology3:642-643.American Society for the Testing of Materials, 1978. Standard test methods for specific gravity ofwood and wood-base materials. Annual Book of ASTM Standards. Section 4. Construction.ASTM, Philadelphia.Armson, K.A. 1977. Forest Soils: Properties and Process. Univ. Toronto Press, Toronto.154Arthur, M.A., and Fahey, T.J. 1990. Mass and nutrient content of decaying boles in an Engelmannspruce-subalpine fir forest, Rocky Mountain National Park, Colorado. Can. J. For. Res.20:730-737.Attiwill, P.M. 1986. Interactions between carbon and nutrients in forest ecosystems. Tree Physiol.2:389-399.Austin, M. 1986. The theoretical basis of vegetation science. Trends Ecol. Evol. 1:161-164.Ballard, T.M. 1989. Forest soils manual, Forestry 312. Dept. Soil Science, U.B.C., Vancouver, B.C.67p.Banner, A., Pojar, J., and Rouse, G. E. 1983. Post-glacial paleoecology and successional relationshipsof a bog woodland near Prince Rupert BC. Can. J. For. Res. 13:938-947.Bascomb, C.L. 1968. Distribution of pyrophosphate extractable iron and organic carbon in soils ofvarious groups. J. Soil Sci. 19:251-268.Beaton, J.D., Moss, A., MacRae, I., Konkin, J.W., McGhee, W.P.T. and Kosick, R. 1965.Observations on foliage nutrient content of several coniferous tree species in BritishColumbia. For. Chron. 41:222-236.Beatty, S.W., and Stone, E.L. 1986. The variety of soil microsites created by tree falls. Can. J. For.Res. 16:539-548.Berg, B., and McClaugherty, C.A. 1989. Nitrogen and phosphorus release from decomposing litterin relation to the disappearance of lignin. Can. J. Bot. 67:1148-1156.Berg, B., and Tamm, C.O. 1991. Decomposition and nutrient dynamics of litter in long-termoptimum nutrition experiments. I. Organic matter decomposition of Norway spruce needlelitter. Scan. J. For. Res. 6:305-321.Berg, B., and Theander, 0. 1984. The dynamics of some nitrogen fractions in decomposing Scotspine needles. Pedobiologia, 27:161-167.Bingham, B.B., and Sawyer Jr., J.O. 1988. Volume and mass of decaying logs in an upland old-growth redwood forest. Can. J. For. Res. 18:1649-1651.Binkley, D. 1984. Ion exchange resin bags: factors affecting estimates of nitrogen availability. SoilSci. Soc. Am. J. 48:1181-1184.Binkley, D., Aber, J., Pastor, J. and Nadelhoffer, K. 1986. Nitrogen availability in some Wisconsinforests: comparisons of resin bags and on-site incubations. Biol. Fert. Soils. 2:77-82.Birk, E.M., and Vitousek, P.M. 1986. Nitrogen availability and nitrogen use efficiency in loblollypine stands. Ecology 67:69-79.155Bosatta, E. and Agren, G.I. 1991. Dynamics of carbon and nitrogen in the organic matter of soil: ageneric theory. Am. Nat. 138:227-245.Botkin, D.B., Janak, J.F., and Wallis, J.R. 1972. Some ecological consequences of a computer modelof forest growth. J. Ecol. 60:849-872.Botkin, D.B. 1993. Forest Dynamics. Oxford Univ. Press, Oxford. 309p.British Columbia Forest Service, 1976. Whole stem cubic metre volume equations and tables. ForestInventory Zones A, B and C. Ministry of Forests, Forest Inventory Branch, Victoria, B.C.Bray, J.R., and Gorham, E. 1964. Litter production in forests of the world. Adv. Ecol. Res. 2:101-158.Bremner, J.M. and Mulvaney, C.S. 1982. Nitrogen - total. In Methods of Soil Analysis. Part 2Chemical and Microbiological Properties. Edited by A.L. Page , R.H. Miller and D.R. KeeneyAgronomy 9:595-624.Brown, J.K. 1974. Handbook for Inventorying Downed Woody Material. USDA-For. Serv. Gen.Tech. Rep. INT-16, Missoula, Montana..Bunnell, F.L. 1989. Alchemy and uncertainty: what good are models? USDA For. Serv. Gen. Tech.Rep. PNW-232, Portland. 27p.Burger, J.A., and Pritchett, W.L. 1988. Site preparation effects on soil moisture and availablenutrients in a pine plantation in the Florida flatwoods. For. Sci. 34:77-87.Carney, H.J. 1989. On competition and the integration of population, community and ecosystemstudies. Funct. Ecol. 3:637-641.Carter, R.E., and Klinka, K. 1992. Variation in shade tolerance of Douglas fir, western hemlock andwestern red cedar in coastal British Columbia. For. Ecol. and Manage. 55:87-105.Chapin, F.S. III 1980. the mineral nutrition of wild plants. Ann. Rev. Ecol. Syst. 11:233-260.Chapin, F.S. III, Bloom, A.J., Field, C.B., and Waring, R.H. 1987. Plant responses to multipleenvironmental factors. BioScience, 37:49-57.Chilton, R.R.H. 1981. A summary of climatic regimes of British Columbia. B.C. Ministry ofEnvironment, Victoria, B.C. 46p.Christy, E.J., and Mack, R.N. 1984. Variation in demography of juvenile Tsuga heterophylla acrossthe substratum mosaic. J. Ecol. 72: 75-91.Cole, D.W. 1981. Nitrogen uptake and translocation by forest ecosystems. In Terrestrial NitrogenCycles. Edited by F.E. Clark and T. Rosswell. Ecol. Bull. (Stockholm) 33:219-232.156Cole, D.W., and Rapp, M. 1980. Elemental cycling in forest ecosystems. In Dynamic Properties ofForest Ecosystems. Edited by R.E. Reichle. International Biological Programme No. 23. pp.341-409.Connell, J.H. and R.O. Slatyer (1977) Mechanisms of succession in natural communities and theirrole in community stability and organisation. Am. Nat. 111: 1119-1144.Crawley, M.G. 1986. Plant Ecology. Blackwell Publ. London, Engl. 496p.Cremeans, D.W. and Kalisz, P.J. 1988. Distribution and characteristics of windthrowmicrotopography on the Cumberland Plateau of Kentucky. Soil Sci. Soc. Am. J. 52:816-821.Dale, V.H., and Hemstrom, M. 1984. Climacs: a computer model of forest stand development forwestern Oregon and Washington. USDA For. Serv. Res. Pap. PNW-327 Portland, Oreg. 60p.Dale, V.H., Hemstrom, M., and Franklin, J. 1986. Modeling the long-term effects of disturbanceson forest succession, Olympic Peninsula, Washington. Can. J. For. Res. 16:56-67.Daubenmire, R. and Prusso, D.C. 1963. Studies of the decomposition rates of tree litter. Ecology44:589-592.Davis, M.B. 1986. Climatic instability, time lags and community disequilibrium. In CommunityEcology. Edited by J. Diamond and T.J. Case, Harper and Row, New York. pp. 269-284.de Catanzaro, J. and Kimmins, J.P. 1985. Changes in the weight and nutrient composition oflitterfall in three forest ecosystem types on coastal British Columbia. Can. J. Bot. 63:1046-1056.DeHaan, S. 1977. Soil Organic Matter Studies, Vol. 1. Int. Atomic Energy Agency, Vienna, pp. 21-30.del Arco, J.M., Escudero, A., and Garrido, M.V. 1991. Effects of site characteristics on nitrogenretranslocation from senescing leaves. Ecology 72:701-708.de Montigny, L. 1992. An investigation into the factors contributing to the growth check of coniferregeneration on northern Vancouver Island. Ph.D Thesis, Dept. Forest Sciences, U.B.C.,Vancouver, B.C. 191p.de Montigny, L.E., Preston, C.M., and Hatcher, P.G., and Kogel-Knabner, I. (in press)Comparison of humus horizons from two ecosystem phases on northern Vancouver Islandusing 13C CPMAS NMR spectroscopy and CuO oxidation. Can. J. Soil Sci. .De Vries, P.G. 1986. Sampling Theory For Forest Inventory. Springer-Verlag, New York. 399p.Demaerschalk, J.P. and Omule, S.A.Y. 1982. Estimating breast height diameters from stumpmeasurement in British Columbia. For. Chron. 58: 143-145.157Dunn, C.P., Guntenspergen, G.R. and Dorney, J.R. 1983. Catastrophic wind disturbance in anold-growth hemlock hardwood forest, Wisconsin. Can. J. Bot. 61:211-217.Edmonds, R.L., Binkley, D., Feller, M.C. et al. 1990. Nutrient cycling: effects on productivity ofNorthwest forest ecosystems. In Maintaining the Long-term Productivity of Pacific NorthwestForest Ecosystems. Edited by D.A. Perry, R. Meurise, B. Thomas et al. Timber Press,Portland, Ore. pp.17-35.Edwards, N.T., and Ross-Todd, B.M. 1983. Soil carbon dynamics in a mixed deciduous forestfollowing clear-cutting with and without residue removal. Soil Sci. Soc. Am. J. 47:1014-1021.Egler, F.E. 1954. Vegetation science concepts. I. Initial floristic composition - a factor in old-fieldvegetation development. Vegetatio 4:412-417.Fahey, T.J. and Birk, E. 1991. Measuring internal distribution and resorption. In Techniques andApproaches in Forest Tree Ecophysiology. Edited byJ.P. Lassoie and T.M. Hinckley, CRCPress, Boca Raton, pp.225-245.Feller, M.C. 1992. Generalized versus site-specific biomass regression equations for Pseudotsugamenziesii var. menziesii and Thuja plicata in coastal British Columbia. Bioresource Technology39:9-16.Flanagan, P.W. and Van Cleve. K. 1983. Nutrient cycling in relation to decomposition and organicmatter quality in taiga ecosystems. Can. J. For. Res. 13:795-817.Forcier, L.K. 1975. Reproductive strategies and the co-occurrence of climax tree species. Science189: 808-810.Franklin, J.F., Cromack K., Denison, W., McKee, A., Maser, C., Sedell, J., Swanson, J. and Juday,G. 1981. Ecological characteristics of old-growth Douglas fir forests. USDA, For. Serv. Gen.Tech. Rep. PNW-118, Pacific Northwest Forest and Range Exp. Stat. Portland, Oregon. 48p.Franklin, J.F. and Dyrness, C.T. 1973. Natural Vegetation of Washington and Oregon. USDA,For. Serv. Gen. Tech. Rep. PNW-8, Pacific Northwest Forest and Range Exp. Stat., Portland,Oregon, USA. 417p.Franklin, J.F., and Hemstrom, M.A. 1981. Aspects of succession in the coniferous forests of thePacific Northwest. In Forest Succession: Concepts and Application. Edited by D.C. West,H.H. Shugart, and D.B. Botkin, Springer, Berlin. pp. 212-229.Gee, G.W. and J.W. Bauder 1982. Particle-size analysis. In Methods of soil analysis. Part 1. Physicaland mineralogical methods. 2nd. ed. Edited by A. Klute, Agronomy 9:383-411.Germain, A.Y. 1985. Fertilisation of stagnated Sitka spruce plantations on Northern VancouverIsland. M.F. thesis, Dept. Forest Sciences, U.B.C., Vancouver, B.C. 185p.158Gessel, SP. and Balci, A.N. 1965. Amount and composition of forest floors under Washingtonconiferous forests. In Forest Soil Relationships in North America. Edited by C.T. Youngberg,Oregon State Univ. Press, Corvallis, pp. 11-24.Gholz, H.L., Grier, C.C., Campbell, A.G., and Brown, A.T. 1979. Equations for estimatingbiomass and leaf area of plants in the Pacific Northwest. Res. Pap. 41, For. Res. Lab., OregonState Univ., Corvallis, 37p.Gibson. N., and Brown, M.J. 1991. The ecology of Lagarostrobus franklinii (Hook.f.) Quinn inTasmania. 2. Population structure and spatial pattern. Aust. J. Ecol. 6: 223-229.Gosz, J.R. 1981. Nitrogen cycling in coniferous ecosystems. In Terrestrial Nitrogen Cycles. Editedby F.E. Clark and T. Rosswell. Ecol. Bull. (Stockholm) 33:405-426.Gower, S.T., Grier, C.C. and Vogt, K.A. 1989. Above-ground production and N and P use byLarix occidentalis and Pinus contorta in the Washington Cascades, USA. Tree Physiol. 5:1-11.Grace, J.B. 1990. On the relationship between plant traits and competitive ability. In Perspectiveson Plant Competition. Edited byJ.B. Grace and D. Tilman, Academic Press, New York, pp.51-65.Graham, R.L., and Cromack, K. Jr. 1982. Mass, nutrient content and decay rate of dead boles inrain forest of Olympic National park. Can. J. For. Res. 12:511-521.Gregory, R.A. 1957. Some silvicultural characteristics of western red cedar in southeastern Alaska.Ecology 38: 646-649.Grier, C.C. 1978. A Tsuga heterophylk-Picea sitchensis ecosystem of coastal Oregon: decompositionand nutrient balance of fallen logs. Can. J. For. Res. 8: 198-206.Grime, J.P. 1979. Plant Strategies and Vegetation Processes. Wiley, Chichester, 222p.Handley, W.R.C. 1954. Mull and Mor Formation in Relation to Forest Soils. Brit. For. Comm.Bull. No. 23, HMSO, London, Engl. 115p.Harborne, J.B. 1982. Introduction to Ecological Biochemistry. 2nd Ed., Academic Press, London.Harcombe, P.A., and Marks, P.L. 1978. Tree diameter distributions and replacement processes insoutheast Texas forests. For. Sci. 24: 153-166.Harmon, M.E., Cromack Jr., K., and Smith, B.G. 1987. Coarse woody debris in mixed-coniferforests, Sequoia National Park, California. Can. J. For. Res. 17:1265-1272.Harmon, M.E., Ferrell, W.K., and Franklin, J.F. 1990a. Effects on carbon storage of conversion ofold-growth forests to young forests. Science 247:699-702.Harmon, M.E. and Franklin, J.F. 1989. Tree seedlings on logs in Picea-Tsuga forests of Oregon andWashington. Ecology 70:48-59.159Harmon, M.E., Franklin, J.F., Swanson, et al. 1986. Ecology of coarse woody debris in temperateecosystems. Adv. Ecol. Res. 15:133-302.Harmon, M.E. and Hua, C. 1991. Coarse woody debris dynamics in two old-growth ecosystems.BioScience 41:604-610.Harmon, M.E., Baker, G.A., Spycher, G. and Greene, S.E. 1990b. Leaf-litter decomposition in thePicea/Tsuga forest of Olympic National Park, Washington, U.S.A. For. Ecol. and Manage.31:55-66.Harper, J.L. 1977. The Population Biology of Plants. Academic Press, New York, 892p.Haynes R.J. 1986. The decomposition process: Mineralisation, immobilisation, humus formation,and degradation. In R.J. Haynes (ed.) Mineral Nitrogen in the Plant-Soil System. AcademicPress, Orlando FL. 483p.Haywood, J.D. 1983. Small topographic differences affect slash pine response to site preparation andfertilization. South J. Appl. For. 7:145-148.Hebda, R.J. 1983. Late-glacial and post-glacial vegetation history at Bear Cove Bog, northeastVancouver Island, British Columbia. Can. J. Bot. 61:3172-3192.Hebda, R.J. and Mathewes, R. 1984. Holocene history of cedar and native Indian cultures of theNorth American Pacific Coast. Science 225:711-713.Heilman, P.E. 1966. Changes in distribution and availability of nitrogen with forest succession onnorth slopes of interior Alaska. Ecology 47: 825-831.Heilman, P.E. 1968. Relationship of availability of phosphorus and cations to forest succession andbog formation in interior Alaska. Ecology 49:331-336.Hendrickson, O.Q. 1991. Abundance and activity of N2-fixing bacteria in decaying wood. Can. J.For. Res. 21:1299-1304.Hendrickson, O.Q., Chatarpaul, L., and Robinson, J.B. 1985. Effects of two methods of timberharvesting on microbial processes in forest soil. Soil Sci. Soc. Am. J. 49:739-746.Henry, J.D., and Swan. J.M.A. 1974. Reconstructing forest history from live and dead plantmaterial - an approach to the study of forest succession in South-west New Hampshire.Ecology 55: 772-783.Hilborn, R. and Stearns, S.C. 1982. On inference in ecology and evolutionary biology: the problemof multiple causes. Acta Biotheoretica 31:145-164.Hobbie, S.E. 1992. Effects of plant species on nutrient cycling. Trends Ecol Evol. 7:336-339.Janzen, D.H. 1974. Tropical blackwater rivers, animals and mast fruiting by the Dipterocarpaceae.Biotropica 6:69-103.160Jenny, H. 1981. The Soil Resource. Springer-Verlag, New York. 377p.Jessome, A.P. 1977. Strength and related properties of woods grown in Canada. For. Tech. Rep.No. 21. FORINTEK Canada Corp., Ottawa. 37p.Johnson, D.L., Watson-Stegner, D., Johnson, D.N., and Schaetzl, R.J. 1987. Proisotropic andproanisotropic processes of pedoturbation. Soil Sci. 143:278-292.Kayahara, G.J., Klinka, K., and Carter, R.E. 1993. Site index of western hemlock ( Tsugaheterophylla (Raf.) Sarg.) in relation to soil moisture, nutrient and foliar chemical measures.For. Sci. (in review).Keddy, P.A., and MacLellan, P. 1990. Centrifugal organization in forests. Oikos 59:75-84.Keenan, R.J., Prescott, C.E., and Kimmins, J.P. 1993. Mass and nutrient content of forest floor andwoody debris in western red cedar and western hemlock forests on northern Vancouver Island.Can. J. For. Res. (in press)Kimmins, J.P. 1988. Community organization: methods of study and prediction of the productivityand yield of forest ecosystems. Can. J. Bot. 66:2654-2672.Krajina, V.J. 1969. Ecology of Forest Trees in British Columbia. Ecology of Western NorthAmerica 2:1-146.Krajina, V.J., Madoc-Jones, S., and Mellor, G. 1973. Ammonium and nitrate in the nitrogeneconomy of some conifers growing in Douglas-fir communities of the Pacific North-West ofAmerica. Soil Biol. Biochem. 5:143-147.Lambert, R.L., Lang, G.E., and Reiners, W.A. 1980. Loss of mass and chemical change in decayingboles of a subalpine balsam fir forest. Ecology 61:1460-1473.Lertzman, K.P. 1992. Patterns of gap-phase replacement in a subalpine, old-growth forest. Ecology73: 657-669.Lewis T. 1976. The till-derived podzols of Vancouver Island. Ph.D Thesis, Dept. of Soil Science,U.B.C. Vancouver, 158p.Lewis, T. 1982. Ecosystems of Block 4, TFL 25. Internal Report. Available from Western ForestProducts Ltd. 1140 West Pender St. Vancouver. BC, Canada. 75p.Little, S.N., and Ohmann, J.L. 1988. Estimating nitrogen lost from forest floor during prescribedfires in Douglas-fir/western hemlock clearcuts. For. Sci. 34:152-164.Lorimer, C.G. 1985. Methodological considerations in the analysis of forest disturbance history.Can. J. For. Res. 15: 200-213.Lyford, W.H. and MacLean, D.W. 1966. Mound and pit microrelief in relation to soil disturbanceand tree distribution in New Brunswick, Canada. Harvard For. Paper 15.161Mac Arthur, R.H., and Wilson, E.O. 1967. The Theory of Island Biogeography. PrincetonUniversity Press, Princeton, New Jersey. 203p.Macintosh, R.P. 1981. Succession and ecological theory. In Forest Succession: Concepts andApplications. Edited by D.C. West, .H.H. Shugart and D.B. Botkin, Springer-Verlag, NewYork p. 10-23.Maser, C. and Trappe, J.M. (eds.) 1984. The seen and unseen world of the fallen tree. USDA For.Serv. Gen. Tech. Rep. PNW-164, Portland, Oreg. 56p.Mathewes, R.W. 1973. A palynological study of post glacial vegetation in the University ResearchForest, Southwestern B.C. Can. J. Bot. 51:2085-2103.McColl, J.G. 1977. Retention of soil water following forest cutting. Soil Sci. Soc. Am. J. 41:984-988.McKee, A., LaRoi, G., and Franklin, J.F. 1982. Structure, composition and reproductive behaviourof terrace forests, South Fork Hoh River, Olympic National Park. In Ecological Research inNational Parks of the Pacific Northwest. Edited by E.E. Starkey, J.F. Franklin and J.W.Matthews, Oregon State University, Forest Research Laboratory, Corvallis, pp. 22-29.Meentemeyer, V. 1978. macroclimate and lignin control of litter decomposition rates. Ecology59:465-472.Melillo, J.M., Aber, J.D., and Muratore, J.M. 1982. Nitrogen and lignin control of hardwood leaflitter decomposition dynamics. Ecology 63:621-626.Melillo, J.M., Naiman, R.J., Aber, J.D. and Linkins, A.E. 1984. Factors controlling mass loss andnitrogen dynamics of plant litter decaying in northern streams. Bull. Mar. Sci. 3:341-356.Messier, C. 1991. Factors limiting early growth of Thuja plicata, Tsuga heterophylla and Piceasitchensis seedlings on Gaultheria shallon dominated cutovers in coastal British Columbia,Ph.D thesis, Dept. Forest Sciences, U.B.C., Vancouver, B.C. 160p.Messier, C. and Kimmins, J.P. 1991. Above- and below-ground vegetation recovery in recentlyclearcut and burned sites dominated by Gaultheria shallon in coastal British Columbia. For.Ecol. and Manage. 46:275-294.Messier, C., Keenan, R.J. and Kimmins, J.P. 1993. The effects of ploughing on soil nutrient status,recovery of competing vegetation and conifer growth on cutovers originating from old-growthforests in British Columbia. Newforests (in review).Miller, H.G. 1984. Dynamics of nutrient cycling in plantation ecosystems. In Nutrition ofPlantation Forests. Edited by G.D. Bowen and E.KS. Nambiar. Academic Press pp. 53-76.Minore, D. 1983. Western red cedar: A literature review. USDA For. Serv. Gen. Tech. Rep. PNW-150, Portland, Oreg. 70p.162Mueller-Dombois, D. and J. Ellenberg 1974. Aims and Methods of Vegetation Ecology. JohnWiley and Sons, New York. 547p.Muller, R.N. and Lui, Y. 1991. Coarse woody debris in an old-growth deciduous forest on theCumberland Plateau, southeastern Kentucky. Can. J. For. Res. 21:1567-1572.Nadelhoffer, K.J., Aber, J.D., and Melillo, J.M. 1983. Leaf-litter production and soil organic matterdynamics along a nitrogen availability gradient in Southern Wisconsin. Can. J. For. Res.13:12-21.Nakashizuka, T. 1987. Regeneration dynamics of beech forests in Japan. Vegetatio 69: 169-175.Nakashizuka, T. 1989. Role of uprooting in composition and dynamics of an old-growth forest inJapan. Ecology 70:1273-1278.Nambiar, E.K.S., and Fife, D.N. 1991. Nutrient retranslocation in temperate conifers. Tree Physiol.9:185-208.Nelson, D.W., and Sommers, L.E. 1982. Total carbon, organic carbon, and organic matter. InMethods of Soil Analysis. Part 2 Chemical and Microbiological Properties. Edited by A.L. PageAgronomy, 9(2): 539-580.Ohlsen, S.R. and L.W. Sommers 1982. Phosphorus. In Methods of Soil Analysis. Part 2 Chemicaland Microbiological Properties. Edited by A.L. Page Agronomy, 9(2):403-48.Oliver, C.D., Nystrom, M.N., and DeBell, D.S. 1988. Coastal stand silvicultural potential forwestern red cedar. In Western Red Cedar - Does it Have a Future? Edited by N.J. Smith,Faculty of Forestry, U.B.C., Vancouver, B.C. pp. 39-46.Oliver, C.D, and Larson, B.C. 1990. Forest Stand Dynamics. McGraw-Hill, New York, USA.467p.Olson, J.S. 1963. Energy storage and the balance of producers and decomposers in ecologicalsystems. Ecology 44:322-331.Ovington, J.D. 1965. The composition of tree leaves. Forestry, 41: 22-28.Parkinson, J.A. and Allen, S.E. 1975. A wet oxidation procedure for the determination of nitrogenand mineral nutrients in biological material. Comm. Soil Sci. Plant Analysis 6:1-11.Pastor, J., Aber, J.D., McClaugherty, C.A., and Melillo, J.M. 1984. Above-ground production andN and P cycling along a nitrogen-availability gradient on Blackhawk Island., Wisconsin.Ecology 65:256-268.Pastor, J. and Post, W.M. 1984. Calculating Thornthwaite and Mather's actual evapotranspirationusing an approximating function. Can. J. For. Res. 14:466-467.163Pastor, J. and Post, W.M. 1985. Development of a linked forest productivity-soil-process model.Env. Sci. Div. Publ. No. 2455, Oak Ridge Nat. Lab., Oak Ridge, Env. Sci. Div. Publ. No.2455. 161p.Pastor, J. and Post, W.M. 1986. Influence of climate, soil moisture, and succession on forest carbonand nitrogen cycles. Biogeochemistry 2:3-27.Peterson, C.J., Carson, W.P., McCarthy, B.C. and Pickett, S.T.A. 1990. Microsite variation andsoil dynamics within newly created treefall pits and mounds. Oikos 58:39-46.Pickett, S.T.A.. and White P.S. 1985. The Ecology of Natural Disturbance and Patch Dynamics.Academic Press, New York. 472p.Pickett, S.T.A., Collins, S.L., and Armesto, J.J. 1987. A hierarchial consideration of the causes andmechanisms of succession. Vegetatio 69: 109-114.Pojar, J., Klinka, K., and Demarchi, D.A. 1991. Coastal western hemlock zone. In Ecosystems ofBritish Columbia. Edited by D. Meidinger and J. Pojar , Ministry of Forests, Victoria, B.C.330p.Prescott, C.E., Coward, C.P., Weetman, G.F. and Gessel, S.P. 1993a. Effects of repeated nitrogenfertilisation on the ericaceous shrub salal (Gaultheria shallon) in two coastal Douglas-firforests. For. Ecol. and Manage. (in press)Prescott, C.E., McDonald, M.A., and Weetman, G.F. 1993b. Differences in availability of N and Pin the forest floors of adjacent stands of western red cedar-western hemlock and westernhemlock-amabilis fir on northern Vancouver Island. Can. J. For. Res. 23:605-610Pritchett, W.L. and Fisher, R.F. 1987. Properties and Management of Forest Soils. John Wiley andSons, New York, 493p.Raison, R.J., Connell, M,J., and Khanna, P.J. 1987. Methodology for studying fluxes of soil N-mineralisation in situ. Soil Biol. Bio.19:521-530.Rommell, L.G. 1935. Ecological problems of the humus layer in the forest. Memoir 170, CornellUniversity Agric. Exp. Stat., New York, 28p.Ross, S.M., and Malcolm, D.C. 1982. Effects of intensive forestry ploughing practices on an uplandheath soil in south east Scotland. Forestry, 55:155-171.Runkle, J.R. 1981. Gap regeneration in some old-growth forests of the eastern United States.Ecology 62: 1041-1051.Scagel, R.F., Bandoni, R.J., and Rouse, G.E. et aL 1965. An Evolutionary Survey of the PlantKingdom. Wadsworth Publishing Co. Belmont, Ca 658p.Schaetzl, R.J., Burns, S.F., Johnson, D.L., and Small, T.W. 1989. Tree uprooting: review ofimpacts on forest ecology. Vegetatio 79:165-76.164Schlesinger, W.H. 1991. Biogeochemistry: An Analysis of Global Change. Academic Press, SanDiego, 443p.Schmelz, D.V., and Lindsey, A.A. 1965. Size class structure of old-growth forests in Indiana. For.Sci. 11: 258-264.Schmidt, R.L. 1955. Some aspects of western red cedar regeneration in the coastal forests of BritishColumbia. Research Note No. 29, British Columbia Forest Service, Victoria, B.C. Canada,10p.Scott, M.L. and Murphy, P.G. 1987. Regeneration patterns of northern white cedar, an old-growthforest dominant. Am. Midl. Nat. 117:10-16.Shugart, H.H. 1984. A Theory of Forest Dynamics. Springer-Verlag, New York, 278p.Smith, N.J. 1993. Estimating leaf area index and light extinction coefficients in stands of Douglas-fir (Pseudotsuga menziesa Can. J. For. Res. 23:317-321.Smith, N.J., and Clark, D.R. 1990. Estimating salal leaf area index and leaf biomass from diffuselight attenuation. Can. J. For. Res. 20: 1265-1270.Sokal, R.R., and Rohlf, F.J. 1981. Biometry 2nd. Ed. W.H. Freeman and Co. New York, 859p.Sollins, P. 1982. Input and decay of coarse woody debris in coniferous stands in western Oregonand Washington. Can. J. For. Res. 12:18-28.Sollins, P. Cline, S.P., Verhoeven, P., Sachs, D., and Spycher, G. 1987. Patterns of log decay in old-growth Douglas-fir forests. Can. J. For. Res. 17:1585-1595.Spies, T.A., Franklin, J.F., and Thomas, T.B. 1988. Coarse woody debris in Douglas-fir forests ofwestern Oregon and Washington. Ecology 69:1689-1702.Sprugel, D.S. 1985. Natural disturbance and ecosystem energetics. In The Ecology of NaturalDisturbance and Patch Dynamics Edited by S.T.A. Pickett and P.S. White. Academic Press,New York. pp. 335-352.Staaf, H. 1982. Plant nutrient changes in beech leaves during senescence as influenced by sitecharacteristics. Acta Oecologia 3:161-170.Stachurski, A., and Zimka, J.R. 1975. Methods of studying forest ecosystems: leaf area, leafproduction and withdrawal of nutrients from leaves of trees. Ekol. Pol. 23:637-638.Stewart, G.H. 1986a. Population dynamics of a montane conifer forest, western Cascade Range,Oregon U.S.A. Ecology 67: 534-544.Stewart, G.H. 1986b. Forest dynamics and disturbance in a beech/hardwood forest, Fiordland, NewZealand. Vegetatio 68: 115-126.165Stone, E.L. 1975. Effects of species on nutrient cycles and soil change. Phil. Trans. R. Soc. Lond. B271:149-162.Stump, L. and Binkley, D. 1993. Relationships between litter quality and nitrogen availability inRocky Mountain forests. Can. J. For. Res. 23:492-502.Tarrant, R.F., and Chandler, R.F. 1951. Observations on litter fall and foliage nutrient content ofsome Pacific Northwest tree species. J. For. )0(:914-915.Taylor, B.R., Prescott, C.E., Parsons, W.F.J., and Parkinson, D. 1991. Substrate control of litterdecomposition in four Rocky Mountain coniferous forests. Can. J. Bot. 69:2242-2250.Thompson, W.A., Barker, J., Weetman, G.F. 1993. Growth and foliar nutrition of western redcedar and western hemlock on northern Vancouver Island in rsponse to fertilisation and sitescarification. Can. J. For. Res. (in review).Tilman, D. 1988. Plant Strategies and the Dynamics and Structure of Plant Communities.Princeton Univ. Press, Princeton, 360p.Turner, D.P. 1984. Interactions among forest overstory, understory, and soils: mechanisms of patchdynamics in a Tsuga heterophylla/Thuja plicata ecosystem. Ph.D thesis, Washington StateUniv., Pullman.Turner, D.P., and Franz, E.H. 1985a. Size class structure and tree dispersion patterns in old-growthcedar-hemlock forests of the northern Rocky Mtns. (USA). Oecologia 68: 52-56.Turner, D.P., and Franz, E.H. 1985b. The influence of western hemlock and western redcedar onmicrobial numbers, nitrogen mineralisation and nitrification. Plant and Soil 88:259-267.Turner, J. 1977. Effects of nitrogen availability on nitrogen cycling in a Douglas-fir stand. For. Sci.23:307-316.Tyrrell, L.E., and Boerner, R.E.J. 1987. Larix laricina and Picea mariana: relationships among leaflife span, foliar nutrient patterns, nutrient conservation, and growth efficiency. Can. J. Bot.65:1570-1577.Ugolini, F.C., Bormann, B.T., and Bowers, F.H. 1990. The effect of treethrow on forest soildevelopment in southeast Alaska. unpublished manuscript, USDA For. Serv. Pac. North WestRes. Stn., Portland, Oreg. 27p.Ulanowicz, R.E. 1990. Aristotelean causalities in ecosystem development. Oikos 57:42-48.Upadhyay, V.P., Singh, J.S., and Meentemeyer, V. 1989. Dynamics and weight loss of leaf litter incentral Himalayan forests: abiotic versus litter quality influences. J. Ecol. 77:147-161.Van Cleve, K., F.S. Chapin III, L.T. Dyrness and L.A. Viereck 1991. Element cycling in taigaforests: state factor control. BioScience 41:78-88.166Van Cleve, K. and J. Yarie 1986. Interaction of temperature, moisture and soil chemistry incontrolling nutrient cycling and ecosystem development in the taiga of Alaska. In ForestEcosystems of the Alaskan Taiga. Edited by K. Van Cleve, F.S. Chapin, P.W. Flanagan, L.A.Viereck and L.T. Dyrness (eds.) Springer-Verlag New York. 230p.Veblen, T.T., Schlegel, F.M., and Escobar, B. 1980. Structure and dynamics of old-growthNothofagus forests in the Valdivian Andes, Chile. J. Ecol. 68: 1-31.Vitousek, P. 1982. Nutrient cycling and nutrient use efficiency. Am. Nat. 4:553-571.Vitousek, P.M., T. Fahey, D.W. Johnson, and M.J. Swift 1988. Element interactions in forestecosystems: succession allometry and input-output budgets. Biogeochemistry 5:7-34.Vogt, K.A., Grier, C.C., Meier, C.G., and Keyes, M.R. 1983. Organic matter and nutrientdynamics in forest floors of young and mature Abies amabilis stands in western Washington, asaffected by fine root input. Ecol. Monogr. 53:139-157.Vogt, K.A., Grier, C.C., and Vogt, D.J. 1986. Production, turnover, and nutrient dynamics ofabove- and belowground detritus of world forests. Adv. Ecol. Res. 15:303-377.Waring, R.H., and Franklin, J.F. 1979. Evergreen coniferous forests of the Pacific Northwest.Science 204:1380-1386.Waring, R.H., and Schlesinger, W.H. 1985. Forest Ecosystems: Concepts and Management.Academic Press, San Diego, 340p.Waterman, P.G., and Mole, S. 1989. Soil nutrients and plant secondary compounds. In MineralNutrients in Tropical and Savanna Ecosystems. Edited by J. Proctor, Brit. Ecol. Soc. Spec.Publ. No. 9. Blackwell Scientific, Oxford, pp.241-253.Wedin, D.A., and Tilman, D. 1990. Species effects on nitrogen cycling: a test with perennialgrasses. Oecologia 84:433-441.Wedin, D.A., and Tilman, D. 1993. Competition among grassed along a nitrogen gradient: initialconditions and mechanisms of competition. Ecol. Monogr. 63:199-229.Weetman, G.F., Fournier, R., Barker, J., Schnorbus-Panozzo, E., and Germain, A. 1989a. Foliaranalysis and response of fertilized chlorotic Sitka spruce plantations on salal dominated cedar-hemlock cutovers on Vancouver Island. Can. J. For. Res. 19: 1501-1511.Weetman, G.F., Fournier, R., Barker, J., Schnorbus-Panozzo, E., and Germain, A. 1989b. Foliaranalysis and response of fertilized chlorotic western hemlock and western red cedarreproduction on salal dominated cedar-hemlock cutovers on Vancouver Island. Can. J. For.Res. 19: 1512-1520.Weetman, G.F., Fournier, R., Schnorbus-Panozzo, E., and Barker, J.1990. Post-burn nitrogen andphosphorus availability of deep humus soils in coastal British Columbia cedar/hemlock forestsand the use of fertilisation and salal eradication to restore productivity. In Sustained167Productivity of Forest Soils. Edited by Gessel, S.P. et al. Faculty of Forestry, UBC, Vancouver,451-459.Wenger, K.F. 1984. (Editor) Forestry Handbook. John Wiley and Sons., New York. 1335p.White, P.S., MacKenzie, M.D., and Busing, R.T. 1985. A critique on overstory/understorycomparisons based on transition probability analysis of an old-growth spruce fir stand in theAppalachians. Vegetatio 64: 37-45.Wilkinson, L. 1990. SYSTAT: The System for Statistics. SYSTAT, Inc. Evanston, IL. 677p.Yarie, J. 1990. Role of computer models in predicting the consequences of management on forestproductivity. In Impact of Intensive Harvesting on Forest Site Productivity. Edited by Dyck,W.J. and Mees, C.A. For. Res. Inst. Bull. No. 159. Rotorua. pp. 3-18.Zak, D.R., Pretzinger, K.S. and Host, G.E. 1986. Landscape variation in nitrogen mineralisationand nitrification. Can. J. For. Res. 16:1258-1263.

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:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0075208/manifest

Comment

Related Items