UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Impacts of partial harvesting on stand structure and wildlife habitat in the Prince George Forest Region Nishio, Grant Richard 2006

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

Item Metadata

Download

Media
831-ubc_2006-0279.pdf [ 6MB ]
Metadata
JSON: 831-1.0074996.json
JSON-LD: 831-1.0074996-ld.json
RDF/XML (Pretty): 831-1.0074996-rdf.xml
RDF/JSON: 831-1.0074996-rdf.json
Turtle: 831-1.0074996-turtle.txt
N-Triples: 831-1.0074996-rdf-ntriples.txt
Original Record: 831-1.0074996-source.json
Full Text
831-1.0074996-fulltext.txt
Citation
831-1.0074996.ris

Full Text

IMPACTS OF PARTIAL HARVESTING ON STAND STRUCTURE A N D WILDLIFE HABITAT IN THE PRINCE GEORGE FOREST REGION by G R A N T RICHARD NISHIO B.Sc, University of British Columbia, 2002  A THESIS SUBMITTED IN PARTIAL F U L F I L M E N T OF THE REQUIREMENTS FOR THE D E G R E E OF M A S T E R OF SCIENCE in  THE F A C U L T Y OF G R A D U A T E STUDIES (Forest Science)  THE UNIVERSITY OF BRITISH C O L U M B I A March 2006  © Grant Richard Nishio, 2006  Abstract The present mountain pine beetle (MPB) epidemic is the largest forest insect infestation in Canada's history. The most common M P B management strategy involves clearcut harvesting large areas. However, i f all of the beetle-killed pine is removed with large clearcuts, the landscape will suffer dramatic reductions in wildlife habitat for many years. Alternatively, sometimes different types of partial harvesting are used that conserve some of the habitat values that would otherwise be removed with clearcutting. This study, done in the Prince George Forest Region, examines two types of partial harvesting I refer to as 'aggregate treatment' and 'dispersed treatment'. These treatments were not originally prescribed as true retention systems, but they are similar in design to aggregated and dispersed retention systems. The aggregate treatments retain small unharvested areas and the dispersed treatments retain individual dispersed trees (usually Douglas-fir veterans) throughout the cutblock. Two related studies were completed at the same site using the same data. The purpose of Study 1 was to determine how the abundance of several wildlife habitat variables differed between dispersed, aggregate, and unharvested control treatments. The purpose of Study 2 was to estimate the probabilities of windthrow and mortality according to tree, site, and neighbourhood factors. Tree size and species influenced windthrow occurrence, but as expected, windthrow was chiefly related to post-harvest exposure. Recommendations for future partial harvesting strategies must include consideration of the potential risk of windthrow. Large Douglas-fir trees have the lowest risk of windthrow and should be retained in dispersed treatments. Two or three time the desired number of trees should be retained in the dispersed treatments to compensate for the inevitable occurrence of windthrow. Aggregate patches should be at least one hectare in size to minimize windthrow damage. Lodgepole pine is at a high risk of M P B attack and non-pine species should therefore be retained in aggregates. It is also recommended to retain deciduous species in aggregates whenever possible since they have high wildlife value, but were relatively scarce in the study area.  Key words: Partial harvesting, dispersed retention, aggregate retention, mountain pine beetle.  11  TABLE OF CONTENTS ABSTRACT  ii  T A B L E OF CONTENTS  iii  LIST OF T A B L E S  v  LIST OF FIGURES  vi  ACKNOWLEDGEMENTS  vii  1 INTRODUCTION 1.1 P R O B L E M S T A T E M E N T 1.2 R E S E A R C H OBJECTIVES 1.3 R E S E A R C H QUESTIONS 1.4 SCOPE A N D A P P R O A C H OF STUDY 1.5 S A M P L I N G P R O C E D U R E R A T I O N A L E  1 4 4 4 5 7  2 LITERATURE REVIEW 2.1 W H A T IS WILDLIFE HABITAT? 2.2 W H A T A R E ATTRIBUTES OF WILDLIFE HABITAT? 2.3 STAND STRUCTURE C H A N G E S WITH SUCCESSIONAL S T A G E OF STAND 2.4 D I S T U R B A N C E EFFECTS O N STAND STRUCTURE A N D WILDLIFE H A B I T A T 2.5 EFFECTS OF S T E M DENSITY O N WILDLIFE 2.6 SHRUB H A B I T A T 2.7 S N A G SIZE 2.8 STAND STRUCTURE 2.9 L I T E R A T U R E R E V I E W S U M M A R Y 2.10 DESCRIPTION OF A T A R G E T STAND WITH THE M O S T WILDLIFE V A L U E  9 12 13 15 .16 18 19 19 20 22 24  3 S T U D Y 1: E F F E C T O F P A R T I A L H A R V E S T I N G T R E A T M E N T T Y P E O N SUPPLY OF WILDLIFE HABITAT ATTRIBUTES 3.1 INTRODUCTION 3.2 METHODS 3.2.1 STUDY SITE 3.2.2 SAMPLING P L A N 3.2.3 STATISTICAL A N A L Y S E S 3.2.4 RECONSTRUCTION OF P R E - T R E A T M E N T S T A N D 3.2.5 C R O W N / D B H RATIOS 3.3 RESULTS '. 3.3.1 STANDING L I V E CONIFERS (StLvCon) 3.3.2 STANDING L I V E DECIDUOUS (StLvDec) 3.3.3 STANDING D E A D CONIFERS (StDdCon) 3.3.4 STANDING D E A D DECIDUOUS (StDdDec) 3.3.5 D O W N CONIFERS (DnCon) 3.3.6 D O W N DECIDUOUS (DnDec) 3.3.7 B A S A L A R E A PER H E C T A R E (BATree) 3.3.8 B A S A L A R E A OF STANDING TREES (UpBATree)  28 28 30 30 31 35 37 37 38 41 41 42 43 43 44 44 45  in  3.3.9 C A N O P Y C O V E R (CovA) 46 3.3.10 SHRUB C O V E R V O L U M E (shrubvol) 46 3.3.11 C O A R S E W O O D Y DEBRIS (CWD) 49 3.3.12 C W D INPUT F R O M POST-HARVEST D O W N E D TREES 49 3.3.13 DISTRIBUTION OF STANDING TREE SIZES IN THE T R E A T M E N T TYPES.50 3.3.14 C R O W N / D B H RATIO DISTRIBUTIONS 52 3.4 DISCUSSION 52 3.5 C O N C L U S I O N 57 4 STUDY 2: PROBABILITY OF INDIVIDUAL TREE MORTALITY.AS A FUNCTION OF TREE AND NEIGHBOURHOOD VARIABLES 4.1 INTRODUCTION 4.2 METHODS 4.2.1 S T U D Y SITE 4.2.2 S A M P L T N G P L A N 4.2.3 TREE L O C A T I O N COORDINATES A N D E X P O S U R E V A R I A B L E S 4.3 STATISTICAL A N A L Y S E S 4.3.1 STEPWISE LOGISTIC REGRESSION 4.3.2 V A R I A B L E S U S E D IN STATISTICAL A N A L Y S E S 4.4 RESULTS 4.5 DISCUSSION 4.6 C O N C L U S I O N  61 61 63 63 63 63 64 64 65 67 ...80 85  5 CONCLUSIONS AND RECOMMENDATIONS 5.1 S U M M A R Y OF RESULTS OF S T U D Y 1 A N D S T U D Y 2... 5.2 R E C O M M E N D A T I O N S 5.3 LIMITATIONS OF THIS S T U D Y 5.4 R E C O M M E N D A T I O N S FOR F U R T H E R STUDIES  87 87 92 95 97  REFERENCES  99  APPENDICES 107 A P P E N D I X 1: SPECIES IN MPB-INFESTED A R E A S TN B C 107 APPENDTX 2: BRITISH C O L U M B I A MINISTRY OF FOREST'S BIODIVERSITY OBJECTIVES : 110 A P P E N D I X 3: TESTS FOR N O R M A L I T Y A N D H O M O G E N E I T Y OF V A R I A N C E FOR 10 H A B I T A T V A R I A B L E S WITHIN T R E A T M E N T S 112  iv  LIST OF TABLES Table 1. Habitat variables used in the analyses 34 Table 2. Critical Values (CV) of Kruskal-Wallis test for comparing mean scores between treatments (alpha = 0.05) 37 Table 3. Kruskal-Wallis one-way analysis of variance by ranks test 39 Table 4. Kruskal-Wallis mean scores 40 Table 5. Absolute differences in Kruskal-Wallis mean scores 40 Table 6. Tree and neighbourhood variables used in logistic regression modelling 66 Table 7. Spearman's correlations of variables used in logistic regression models 68 Table 8. Logistic regression models 72 Table 9. Model 1 .Logistic regression odds ratio estimates for likelihood of being dead from the total population of trees 72 Table 10. Model 2:Logistic regression odds ratio estimates for likelihood of being dead from population of standing trees 74 Table 11. Model 3:Logistic regression odds ratio estimates for likelihood of being down from population of trees in aggregate and dispersed 76 Table 12. Model 4:Logistic regression odds ratio estimates for likelihood of being down from total population of trees excluding controls 79 Table 13. Deciduous trees in treatments (WD = Wind Damaged) 79 Table 14. Model 5: Logistic regression odds ratio estimates for likelihood of being down from population of trees in aggregate 80 Table 15. Abundance of wildlife habitat variables in retention treatments compared to unharvested control areas 87 Table 16. Light and wind exposure and impacts on stand structure by treatment 88  v  \  LIST OF FIGURES Figure 1. Pre-harvest stems per hectare (SPFf) and basal areas (BA) Figure 2. Proportions of tree species in pre-harvested stands Figure 3. Standing live conifers stems per ha (sph) by treatment Figure 4. Standing live deciduous stems per ha (sph) by treatment Figure 5. Standing dead conifers stems per ha (sph) by treatment Figure 6. Standing dead deciduous stems per ha (sph) by treatment Figure 7. Down conifers stems per ha (sph) by treatment Figure 8. Down deciduous stems per ha (sph) by treatment Figure 9. Post-harvest basal area per hectare by treatment Figure 10. Standing basal area Figure 11. Basal area of post-harvest windthrown trees Figure 12. Canopy cover Figure 13. Shrub cover volumes (m ) by treatment. Figure 14. Relationship of shrub cover and canopy cover. : Figure 15. Grass cover (%) and shrub volume (m /ha) for dispersed and aggregate treatments Figure 16. Shrub cover volumes versus grass cover Figure 17. Grass cover in years since harvest (YSH) classes Figure 18. Shrub volumes (m /ha) in years since harvest (YSH) classes Figure 19. Distribution of C W D Figure 20. Coarse woody debris inputs from windthrown trees Figure 21. Proportion of total trees that were windthrown post-harvest in aggregate and dispersed treatments Figure 22. Distribution of standing post-windthrow tree sizes Figure 23. Crown/dbh ratios for conifer and deciduous species. Figure 24. Proportions of windthrown trees in dbh size classes Figure 25. Proportions trees in height classes that are windthrown Figure 26. Proportions trees in height/diameter ratio classes that are windthrown Figure 27. Proportions trees in VRfetch classes that are windthrown Figure 28. Proportions of trees in direx30 classes 1-8 that are windthrown Figure 29. Proportions of species in treatments Figure 30. Percent mortality of lodgepole pine in treatments Figure 31. Proportion of down trees in Y S H classes Figure 32. Windthrow and non-windthrow mortality of trees in dispersed and aggregate treatments Figure 33. Proportions of species of windthrown trees from the population of dead trees in aggregate and dispersed treatments 3  3  3  vi  ACKNOWLEDGEMENTS The writing of this thesis has been a challenging and very fulfilling experience and I am grateful to many people for their help. I would like to give special thanks to Dr. Steve Mitchell. Steve's expertise, generous support, and his instinctual ability to teach meant I had the best graduate supervisor I could possibly hope for. I would also like to thank my other committee members Dr. Peter Marshall and Dr. Ann Chan-McLeod. Peter gave the gift of earthly logic to what first appeared as statistical fuzziness. Ann's keen scientific scrutiny always kept the edge on my personal accountability. M y thanks go to Naa Lanquaye for her assistance with GIS procedures and to Robyn Scott for creation of the exposure variables (VRfetch, Direx30). M y thanks also go to Canadian Forest Products Ltd. for their support of this project, particularly to Kerry Deschamps and the supervisory staff at the Prince George office. I would like to thank Geoff Tomlins and Stephanie Ortlepp at Pacific Geomatics Ltd. for their contribution of the SPOT 5 imagery. I would also like to thank all my family including Geri, Stephen, Mya, Michiko, Rosco, Stanley, and Cathy for their enthusiastic support and kind patience. Finally, I would like to express my sincere gratitude to the Mountain Pine Beetle Initiative for the funding that made this study possible.  vii  1  INTRODUCTION The mountain pine beetle (scientific names of all species in this report are given in  Appendix 1) is native to the forests of northwestern North America and exists at low levels in virtually all stands containing lodgepole pine. Recently the population growth of mountain pine beetle (MPB) in British Columbia has expanded to epidemic levels. A mountain pine beetle mortality model (BCMPBv2) by Eng et al. (2005), based on aerial surveys of the province, projects the peak of the outbreak will be in 2006 with infected tree volumes greater than 90 million m per year. The worst case scenario estimates infested tree volumes exceeding 25 3  million m per year will continue until 2015. The model estimates 14 million hectares will be 3  M P B infected by 2009. This represents over 615 million m or 60% of the pine on the timber 3  harvesting landbase. Mortality rates are not expected to drop to pre-outbreak levels until after 2020 (Eng et al. 2005). On Sept. 14, 2004, the chief forester, Larry Pedersen, announced an increase in the annual allowable cut (AAC) by 27 per cent to 23.4 million m in the three northcentral B.C. timber supply areas most affected by the M P B infestation. Pedersen stated "the A A C in the Prince George, Quesnel and Lakes timber supply areas will go up in total by about 4.9 million m , effective Oct. 1, and the increase will focus on salvaging pine forests with 3  moderate to high levels of mortality as a result of the beetle". There is a valid concern by many that the dramatic increase in the rate of harvesting will increase risk of habitat degradation from the widespread use of clearcut logging. Provincial biodiversity objectives are often met by retaining portions of unharvested forest outside of harvested areas. The economic, silvicultural, and operational advantages of clearcut harvesting methods make it the most commonly employed strategy for dealing with large M P B infested stands in B C . Depending on the time since initial beetle attack, the harvesting strategy is to harvest before the beetles leave the infected trees and re-infect other live trees; to harvest  1  susceptible pine trees before they become infected; or to salvage harvest before decay sets in and timber values are lost. Alternate M P B management options include small patch cuts, selection harvesting, fall and burn, controlled fires, and thinning (sometimes termed "beetle-proofing"). However, these methods are used less frequently than the clearcutting option, particularly during outbreak conditions. The problem is wide scale clearcut harvesting could potentially result in an ecologically degraded landscape of stands that are missing many of the wildlife habitat features found in natural stand structures. Chan-McLeod and Bunnell (2003) identified 195 vertebrate species (Appendix 1) that could be impacted by conventional M P B management in the interior of British Columbia. Forest elements (eg., large trees, shrubs) result in stand features (eg., age-distributions, vegetation complexes) that determine habitat values (eg., nesting structures, cover, foraging substrates). Management of these stand features is achieved through retention of the various elements. However, retained features change after harvesting as a result of their newly exposed circumstances and the treatments can only be considered a success if the desired stand features are sustained or produced over the long-term. The constraints of meeting the provincial biodiversity guidelines and minimizing impacts to wildlife imposes challenges to effective M P B management. Different types of harvesting methods and levels of harvesting intensities will interact with the post-harvest stand dynamics to produce a variety of post-harvest stand structures. It is the intention of this research to determine how the abundance of wildlife habitat values as provided by individual trees, coarse woody debris (CWD), and shrub cover, etc. varies with four different partial harvesting treatments (three levels of dispersed treatment and one aggregate treatment) and unharvested control areas. When designing the cutblocks examined in this study, forest managers selected trees for retention according to their individual characteristics (eg., windthrow resistance, non2  pine species, size, wildlife value) and for the most part their selection is related to availability and operational feasibility. Since trees are selected for retention according to certain criteria, I will refer to this as 'retention selection bias'. For example, when it is possible, it is common to retain individual dispersed Douglas-fir veterans (dispersed treatment) throughout the cutblock. It is also common to retain groups of trees (aggregate treatment) that are different (eg. mixed species, immature trees, trees with special wildlife value) from the rest of the cutblock. The features that are retained are unique to the particular area where the partial harvesting treatment is implemented. In this study treatments retaining groups or aggregates of trees such as wildlife tree patches (WTP) will be referred to as aggregate treatments and treatments retaining dispersed individual trees will be referred to as dispersed treatments. This study measured the abundance of habitat variables by using simple random sampling (SRS) methods to collect data from aggregate treatments, dispersed treatments, and unharvested control areas. There were not enough suitable treatments available to randomly select treatments from a larger population of treatments so this study was designed as a casestudy. As a case study, the treatment blocks were selected to represent the four partial harvesting treatments of Dispersed-H (high retention), Dispersed-M (medium retention), Dispersed-L (low retention), Aggregate, and unharvested Controls. In fact, all three of the dispersed treatment classes retained very low stem densities, but they were subjectively grouped relative to each other into high, medium, and low retention classes according to their postharvest stem densities. Treatment blocks were further grouped into two drainage classes (drainage, non-drainage) as to whether or not they had the presence of some type of drainage or wet ground. This was done to include any variability that may result from the presence or absence of water. The use of fixed area plots allowed variables to be easily converted to area based per hectare (ha) units for statistical analyses. Randomly locating plots in the treatment blocks 3  allowed standard statistical analyses to be used. In addition to the fixed area plots, line intersect sampling (LIS) was used to measure C W D volumes. Post-harvest C W D input volumes were estimated by calculating the volumes of the trees in the fixed area plots that became windthrown after harvesting (down trees). Dispersed treatments contained very low residual stem densities so large rectangular (1/3 ha) fixed area plots were used in these blocks to measure tree data. This allowed a sufficient number of trees to be captured that would not have been possible with the smaller (1/100 ha) fixed area plots used in the aggregate and control treatments.  1.1 P R O B L E M STATEMENT How does partial harvesting for mountain pine beetle control or salvage affect stand structure and supply of the forest elements that provide wildlife habitat? 1.2 R E S E A R C H OBJECTIVES The broad objective of this study is to evaluate whether different harvesting treatments will produce a corresponding variety of the forest elements that provide wildlife habitat. If detectable relationships can be established between the persistence of habitat features such as standing live and standing dead trees, and the site and or neighbourhood variables, these models can be used to provide potential habitat abundance according to the type and intensity of the partial harvesting treatment. Ultimately, it is hoped that this knowledge can be incorporated into the planning of M P B management strategies that are operationally and economically feasible, but also effectively address wildlife habitat concerns. 1.3 R E S E A R C H QUESTIONS There are three main research questions addressed by this research study: 1. Does the aggregate treatment result in a different abundance of the stand structure elements that provide wildlife habitat than the three dispersed treatments and how do these treatments differ from the unharvested forest?  4  2. Do the aggregate and dispersed treatments affect the probabilities of tree mortality or windthrow? 3. How do tree, neighbourhood, and site factors affect the probabilities of tree mortality or windthrow?  1.4 SCOPE A N D A P P R O A C H OF STUDY The harvesting methods studied in this research are from the two broad categories of aggregate and dispersed treatments. The aggregate treatment category includes the retained patches of unharvested forest within cut-blocks that have most if not all stems removed except for reserve areas or patches of non-harvested ground. The dispersed treatment category includes blocks that have residual stems that are dispersed at various densities throughout the area of the cut-block. A set of dependent tree status variables and a set of independent site and neighbourhood variables were measured in each of the sample plots. The tree status variables represent attributes specific to the individual trees. The site variables represent the treatments and site conditions of the residual stand. The exposure variables represent neighbourhood conditions resulting from the locations of individual trees. The field data were collected in the spring and summer of 2004. The data were imported into the SAS (SAS Institute Inc. 2004) statistical software computer program for various analyses including the calculation of treatment means, Kruskal Wallis non-parametric analyses, and the creation of five logistic regression models. Analyses were completed in 2005 to indicate the relationships between the tree status, site, and neighbourhood variables that result from the different types of harvesting treatments. This study uses retrospective sampling rather than an experimental design to directly infer treatment effects. One of the most important contributions of this study is the determination of whether the present partial harvesting practices can actually be successful at meeting the intended objectives of wildlife habitat supply at the stand level. O f course, "biodiversity objectives" must be interpreted and managed at the landscape level, but that also 5  requires an understanding of how stand-level wildlife habitat values are specifically affected by the different harvesting treatments (eg., partial harvesting). The lodgepole pine dominated stands in this study are relatively common in the forests of the interior plateau region of British Columbia. For this reason, even though the statistical rigour of this study is reduced due to the nature of a "case study" analysis the findings in this study should provide some knowledge that is generally applicable at the regional scale. The three research questions in this thesis were addressed in two separate studies. Study 1 estimated the differences in the forest elements that provide wildlife habitat between the treatments. Since retention selection bias can significantly influence the final outcome of trees in the different treatments, Study 1 cannot be considered a true measure of treatment effect. Rather, it is an indirect measure of habitat values. Habitat values were inferred by measurements of the abundance of 11 variables that represent the basic elements of wildlife habitat. Analysis of the data was accomplished by using various frequency distribution graphs and the Kruskal-Wallis non-parametric one-way analysis of variance by ranks to estimate significant differences in habitat variable abundance between the treatments. Study 2 estimated the probabilities of tree mortality and windthrow according to tree, site and neighbourhood factors. This was accomplished through the creation of five logistic regression models estimating the significant relationships of tree mortality and windthrow. Treatment effect was characterized using exposure indices. The effect of tree attributes variables (eg., species, height, dbh) on windthrow and mortality was also investigated. Only the most parsimonious logistic regression models with significant relationships (alpha = 0.05) are presented in the thesis. Understanding the relationships among these variables can provide managers insight into the specific harvest design and tree characteristics that will produce the desired results in the target stand.  6  1.5 S A M P L I N G P R O C E D U R E R A T I O N A L E Sampling and analytical procedures must be designed according to the specific conditions and objectives of the research project. A study on the varying effects of different partial harvesting treatments needs to group the sampling into treatment classes. This will allow the groups of field measurements to be linked to their respective classes and analyses to be done within and between treatments. If a regression analysis is desired, the treatment classes should capture as broad a range of effects as possible. This can be accomplished by measuring areas within treatment classes that are representative of the full spectrum of site conditions (eg., soil conditions, site series, exposure). Grouping treatments into different age classes (eg., years since harvest) will allow an estimation of change over time. Measuring dispersed treatments with differing levels of retention can provide potential impacts according to varying harvesting intensities. If the measured attributes can be separated into independent variables (eg., type of treatment) and dependent variables (eg., the forest elements that provide habitat), and relationships established between them, it may be possible to develop models that can help link treatments to habitat objectives. The sampling design should also include a range of treatment controls that are relatively similar to the treatment areas before they were harvested. The selection of measurement variables should reflect the features or attributes that are important to the study. For instance, if wildlife habitat is being studied, features with expected significant wildlife values such as foraging resource, cover attributes, breeding requirements, etc. need to be measured. Statistical procedures can objectively decide which variables have significant relationships between or within the groups of measured variables. Even in the event that no significant relationships are indicated with the measured variables, there may be considerable value in reevaluating previously held assumptions that are not supported by the study. A well designed sampling plan and analysis can also indicate which treatments are  7  meeting their objectives and producing effective implementation of long term management strategies. One of the most common sampling methods used in forestry is simple random sampling (SRS). Simple random sampling randomly selects sampling points from a population of possible points such as locations or coordinates. Various types of plots can be employed at the selected sampling points. Fixed area plots are a type of cluster sampling commonly used in forest inventories. Fixed area plots are specific areas of land considered to be representative of the stand the plot is located within. Any variable of interest that can be captured in a fixed area plot can be measured. Trees and their attributes can be measured in fixed area plots and calculated on a per hectare basis by using a plot multiplier. Line intersect sampling (LIS) is a technique where lines of a specific length are located randomly in the sampling area. Each line is a cluster of the elements that intersect that line. LIS is commonly used to estimate volumes of coarse woody debris. The probability of an element crossing the line is a function of the length of the element (eg., length of C W D piece). The direction of the line can have an important influence (eg., bias) on the probability of intersecting pieces of CWD. Harvesting operations will tend to align C W D in particular directions associated with yarding procedures. If the LIS line is parallel with the CWD alignment, the probability of intersecting pieces is greatly reduced. Using randomly selected LIS line directions along with an accompanying second line oriented 90 degrees to the first line will minimize CWD alignment bias. Diameters and tilt of intersected CWD pieces are recorded and formulas are used to estimate CWD volumes per hectare (Lemay and Marshall 1990).  8  2  LITERATURE REVIEW Since alternative non-clearcut M P B partial harvesting methods can be used to conserve  forest elements that provide wildlife habitat, it should potentially be able to mitigate some of the habitat losses normally associated with standard clearcut harvesting methods. Partial harvesting can be used to retain dispersed individual trees or groups of trees in aggregates. Implementing habitat management strategies at the stand-level requires conserving the specific elements of stand structure that support habitat values important to wildlife. The six key structural elements of dead and dying trees, downed wood, shrubs, hardwoods, riparian habitat, and serai stage distribution are commonly altered by forest practices (Bunnell et all999a). Consequently the provincial biodiversity guidelines provide retention targets for these elements. Appropriately designed partial harvesting methods can retain these particular elements of stand structure. In order for a partial harvesting treatment to be considered successful, the long-term surviving stand should sufficiently resemble the target stand. The wildlife objectives can only be achieved if the desired habitat values are sustained by the stand elements that provide them. Habitat values may be supplied by individual live or dead standing trees (snags), groups of standing trees, down trees (CWD), or shrub cover. The effects of windthrow can be one of the most important factors influencing the long-term success or failure of a partial harvesting treatment (Halpern and Spies 1995). Windthrown trees will supply new C W D , but the survival of standing stems is necessary to sustain many other habitat values ( M l et al. 2003). Survival of aggregates or patches of standing stems may be required for survival of organisms contained within aggregates^ old snags, or structural cover. The spatial arrangement of individual standing stems can be important for providing connectivity, or meeting species territorial requirements. Marcot (1983) found dispersed retention actually increased both abundance and richness of secondary cavity nesters compared to mature and old-growth forests, but abundance of primary nesting species was reduced. Windthrow events can effectively remove many of the habitat 9  values that were previously supplied by the standing stems. The probabilities of windthrow are affected by factors of species, exposure, and individual tree characteristics (eg., size, slenderness). Some level of windthrow is expected along the downwind edges of harvested areas (Jull et al. 2003) so it should always be considered in any long-term planning strategy. Long-term input of downed wood or coarse woody debris is important for sustaining a healthy forest ecosystem (Keisker 2000). There are numerous habitat values provided by CWD resulting from the tree species, decay patterns, and structural arrangement. CWD input is the result of tree mortality and disturbance events such as insects, wind, root rot, fire, etc. (Jull et al. 2003). However, clearcut harvesting can potentially reduce the abundance of large C W D unless specific steps are taken to leave downed wood on site after logging activities have been completed. Also, retention of some trees is required to sustain a long-term input of CWD. Trees age and die and become coarse woody debris (CWD) in various decay classes and configurations. The volume, condition and rate of C W D recruitment is an important consideration since CWD provides a variety of wildlife habitat during its progressively changing structural arrangements and stages of decay (Kimmins 1996). British Columbia Ministry of Forest's (BCMoF) best management practice for managing CWD is to 'leave a range of species, decay classes and size classes of C W D spread out on the site, including large enough pieces that will last for a long time' (Densmore 2002). However, due to safety concerns as declared by the Worker's Compensation Board guidelines, dead trees are most often felled in harvested areas. This means future C W D recruitment must come from residual immature trees that mature and eventually die. The only other source of CWD will be from the regenerated stand, which will normally require even longer periods of time to mature and die. Much of the value of CWD for wildlife is directly related to piece size, and state of decay. Managing CWD for wildlife requires long-term planning since the process of tree decay cannot be achieved in a short period (Stevens 1997). 10  The wildlife value of a tree will vary according to the wildlife species, and the condition and species of the tree (Harestad and Keisker, 1989). Impacts on wildlife will vary according to which kinds of trees are retained and how they are retained (eg., individual or groups of trees, density/ha, size class, species, location, stand structure complexity). The habitat value of a tree varies with changes in mortality status, species, and whether the tree is standing or down. In order to simplify and group these characteristics, the following 11 habitat variables were used in this study: standing live conifers (StLvCon), standing live deciduous (StLvDec), standing dead conifers (StDdCon), standing dead deciduous (StDdDec), down conifers (DnCon), down deciduous (DnDec), basal area of trees left after harvesting both standing and down at time of sampling (BATree), basal area of trees standing at time of sampling (UpBATree), % canopy cover (CovA), shrub cover volume (ShrubVol), and coarse woody debris (CWD). Coarse woody debris input from post-harvest windthrown trees was also measured. The distribution of tree sizes, especially the presence of large trees, is important to wildlife so tree size distributions in treatments are indicated in a graph of stem densities/ha (Figure 22). Different partial harvesting methods can be used to address specific requirements of wildlife. Dispersed treatments may be appropriate when structures need to be well distributed (eg., territorial habitat requirements). However, the closed conditions in aggregate treatments can include features that would not survive as well in open conditions (eg., shade-tolerant species, windthrow-prone snags, late seral-vegetation). Aggregate shape, orientation, size, and distribution can be adjusted to meet specific objectives (eg., ecological, microclimate, windthrow resistance). Aggregate treatments have operational and biological advantages compared to dispersed treatments. However, dispersed treatments can provide connectivity and structural diversity in a block. A mixture of treatments is generally considered better than a single type of treatment pattern throughout the block.  11  Provincial area-based biodiversity guidelines (Appendix 2) specify retention levels for non-harvested areas (eg., aggregate reserves) within cut-blocks. The total aggregate reserve areas represent a proportion of the total harvested area. Another approach is to retain some residual stems (eg., non-harvested individual trees) as dispersed treatments. A calculation is done where each of the individual dispersed stems represents an area and the sum of these individual areas is considered to represent a dispersed reserve area. The sum total of aggregate reserve areas and dispersed reserve areas contribute towards meeting the area requirements of the biodiversity guidelines. 2.1  W H A T IS WILDLIFE HABITAT? According to Bunnell et al. (1999a), the major elements of the forest that are "relevant to  sustaining vertebrate richness are dead and dying wood, downed wood, shrubs, hardwoods, riparian habitat, and early and late successional stages". Environment Canada (2005) uses the term habitat to describe "the environment where wild species live. It includes all the conditions that species need to thrive such as climate, water, and the availability of food and shelter. Habitat requirements can vary in size and location, depending on the time of year and the life cycle of the organism". Habitat shelter or cover, is described in physical terms and can be considered a structural resource. Cover can be provided by vegetation, topography, water, or various elements of stand structure. The function of cover (eg., nesting, feeding, resting, escape) is linked to how wildlife populations respond to their environments and to changes caused by forest practices (Watts 1983). The structural features of a forest can be divided into habitat elements, habitat structures, and landscape features (Center for Applied Conservation Research 2004). Habitat elements provide the habitat required by wildlife, and depending on the particular species of wildlife, is defined at the stand or microhabitat level. Habitat elements include dead/dying trees, hardwoods, shrubs, large live trees, canopy, downed wood, and microhabitat elements. Habitat 12  structure is the combination of habitat elements that together produce habitat (eg., structural heterogeneity, stand age). Tree size, percent canopy cover, and number of canopy layers are measures used to classify structure. A landscape feature results from the combination of two or more stands including forested and non-forested areas. The broader scale effect is more than the sum of the individual stand effects. Landscape feature identification utilizes a landscape-level concept and requires more than just stand-level information. Some examples of landscape features include forest-age distribution, habitat interior/ edges/ aggregate size distribution, riparian habitat, and roads. Habitat niches are a product of plant communities, successional stage, and environmental factors including soil types, moisture regimes, microclimate, slope, aspect, elevation, and temperature. The feeding and/or reproduction needs of wildlife species are individually adapted to the different combinations of plant communities and successional stages (Thomas 1979). Management activities alter succession in plant communities. Consequently, the change in succession alters the habitat niches associated with the successional stages. Habitat supply refers to the "supply of a wide range of habitats that are used by all species" (BC MoF 2005). Habitat use identifies the "types of habitats used by wildlife; the way an animal uses or consumes a collection of physical and biological components (resources) in a habitat" (Colostate 2005).  2.2  W H A T A R E ATTRIBUTES OF WILDLIFE HABITAT? Wildlife tends to respond more to forest structure rather than to individual elements of  habitat. Therefore, combinations of elements required for reproduction, forage, and survival cover, influence species richness more than single elements (Bunnell et al. 2003). However, some specific elements can be individually significant. Individual dead or dying trees can provide nesting and foraging sites for raptors (Zemlak et al. 1995) and cavity nesters (Robinson and Mark 2001). A study done in the north central interior of British Columbia 13  (Zemlak et al. 1995) found dead snags were the most preferred nesting site for raptors (58%) followed by cottonwood (22%) and then aspen (17%). Cavity nesting birds prefer larger trees (Bunnell et al. 1999a), but selection is dictated by suitable decay patterns. The patterns of decay can vary greatly for different species of conifers. Decay commonly found in dead Douglas-fir trees results in the sapwood decaying before the heartwood. Bunnell et al. (1999a) found the softer exterior sapwood of decaying Douglas-fir was preferred by weaker excavators such as chickadees. B y the time the Douglas-fir heartwood had decayed enough to be excavated, the outer sapwood was no longer firm enough to provide a sound cavity. Heart-rot fungi dies soon after the tree falls, so large hollow trees must be created by decay fungi in living trees (Rayner and Boddy 1988). Hardwoods are commonly preferred by birds and mammals (Harestad and Keisker 1989). Approximately 40 species of birds in British Columbia are largely restricted to nesting in deciduous trees and another 45-50 species prefer deciduous stands (Bunnell et al. 1999a). Hardwoods also provide a nutrient rich, easily digestible litter that encourages arthropod fauna and benefits ground dwelling insectivores. Many small litter dwelling mammals, including Pacific water shrew, Trowbridge's shrew, shrew-moles, white-footed vole, long-tailed vole, Pacific jumping mouse, pinion mouse, and chipmunks, are more abundant in deciduous stands than conifer stands (Bunnell et al. 1999a). A study in north-central British Columbia found aspen was most preferred for nesting in proportion to its occurrence, but conifers were preferred for foraging by woodpeckers (Martin and Eadie 1999). The B C Forest Practices Code addressed biodiversity objectives at the stand level by incorporating wildlife tree and C W D retention guidelines to provide desired stand structure and species composition. Serai stage can be a significant factor for providing habitat for species. Some species such as the three-toed woodpecker are associated with late serai elements (eg. nesting and foraging snags) while other species such as the neotropical migrant, orange-crowned warbler are 14  normally associated with elements of early serai stages (Wells et al. 1998). Older stands contain large diameter trees, snags, downed wood, and structural complexity in both canopy and nearground layers. However, some of the features of older stands can also be found in younger stands. The implications for management are that habitat associated with old-growth can be conserved in much younger managed stands i f indices of 'old-growthness' are adequately defined and features are deliberately maintained (Holt et al. 1999). Davis et al. (1999) found basal area and live crown volume had the strongest correlation with bird abundance in the ESSF zone in east central British Columbia. The level of basal area where crown closure occurred resulted in the strongest shift in abundance. Davis et al. also found volume of coarse woody debris was positively correlated with Winter wren and Wilson's warbler abundance.  2.3 STAND STRUCTURE C H A N G E S WITH SUCCESSIONAL S T A G E OF STAND Old forests are more structurally complex and contain more shade-intolerant species than young forests (McCleary and Mowat 2002). Lodgepole pine tends to become the effective climax species in areas where fires are frequent such as the west Chilcotin. Old growth lodgepole pine forests can only occur i f there is a sufficiently long absence of fire, but infrequent fires can allow successional replacement by longer-lived non-pine species. This makes old growth lodgepole pine forests scarce in the interior of British Columbia. Old growth lodgepole pine forests produce large very old pine trees with sweeping smooth boles and are rare, indicating their existence may be ecologically significant within the landscape (Silva Ecosystems 1992). Climax lodgepole pine forests can also occur on sites too dry or cool to sustain other tree species (Pojar 1984). Pojar (1995) found the older stages of both pure aspen and mixed aspen-conifer stands in the Prince Rupert Forest Region contained the highest diversity of vegetative structure and supported the greatest diversity of bird communities. The complexity of vegetation increases with the increasing age of the serai stage and results in a higher abundance and diversity of 15  breeding species (Pojar 1995). The older stages of both pure aspen and mixed aspen-conifer stands contained the highest diversity of vegetative structure and supported the greatest diversity of bird communities. Loss of the younger developmental stages of aspen impacted species associated with sapling thickets such as ruffed grouse and alder flycatcher (Pojar 1995). Low basal areas and live crown volumes along with higher volumes of grass and forbs are found in the early serai stages and result in bird communities that are different from the older serai stages in the Cariboo region of British Columbia (Davis et al. 1999). Davis et al. (1999) found warbling vireo, orange-crowned warbler, MacGillivary's warbler, and Lincoln's sparrow more abundant in early serai stands. Structural attributes such as high basal area and live crown volumes were similar in the mid and late serai stages and Davis et al. felt this resulted in similarities in the bird communities. A few species such as winter wren and boreal chickadee were more abundant in late serai stands, but generally the composition of bird communities was similar for mid serai and late serai stands (Davis et al. 1999). 2.4 D I S T U R B A N C E EFFECTS O N STAND STRUCTURE A N D WILDLIFE H A B I T A T The present forests of the central interior of British Columbia are the result of large stand-replacing disturbances such as fire or insect outbreaks and are characterised by a mosaic of stand types that vary in species composition, age class, and structure (McCullough et al. 1998). The forests in the Sub-Boreal Spruce zone (SBS) are largely composed of older stands resulting from large scale disturbances approximately 140 years ago (Steventon 1997). The present M P B outbreak is an example of this type of large scale disturbance (Steventon 1997). The impact on stand structure and habitat are different depending on whether the disturbance event is from an insect infestation such as mountain pine beetle or i f it is from fire. Disturbance from both fire and beetle infestations can leave an abundance of snags, but the vegetation and the type of snags will be different. A mountain pine beetle infestation will leave large areas of standing dead trees with undisturbed understory vegetation. A n MPB-attacked 16  stand will leave standing dead pine snags, and the regenerating stand will be composed of the shade-tolerant tree species (eg., spruce, fir) surviving in the understory. The vegetation surviving a beetle infestation can provide cover for species such as snowshoe hares and marten (Stadt 2001). In contrast, fire can kill both standing trees and understory vegetation. Regeneration following severe fires can result in large areas of a single pioneer species such as lodgepole pine initially with an early serai stage understory. Fires can also leave patchy areas of unburned ground, creating a mosaic of different conditions. Also, fire-killed trees are casehardened and are not as suitable for cavity nesting birds as beetle-killed trees (Bull et al. 1997). Salvage harvesting a stand alters the natural successional pathway of a beetle infected stand by producing a replacement stand with early serai vegetation and shade-intolerant trees that is not characteristic of a beetle outbreak. Current forest management practices have caused an increase of these even-aged early serai stand types accompanied with a decrease in older serai stands types (Stadt 2001). However, any disturbance event can provide benefits for some species and negative effects for others, and post-fire and post-harvest stands with their early serai vegetation can supply habitat for many species including forage for ungulates (Schuerholz et al. 1988). Bark beetle caused mortality is also an important source of C W D in the central interior of British Columbia (Clark et al. 1998). Managed forests typically have reduced levels of C W D since harvesting activities remove large diameter logs. Windfirmness is influenced by several factors including level of exposure, site conditions, stand history, tree health, and species (Stathers et al. 1994). Windthrow and other types of mortality (eg., dessication) will respond to varying levels of exposure from different types of harvesting treatments. When soil conditions do not have root restrictions, Douglas-fir, which has a taproot, is generally considered the most windfirm of the species measured in this study (Burns and Honkala 1990a). Scott (2005) found species (including Douglas-fir) was not a factor influencing proportions of windthrow at Clayoquot Sound on the west side of Vancouver 17  Island, but found species was a factor on a site on the eastern side of the island. Nevertheless, all trees will likely be more susceptible to windthrow with the relatively sudden increase in the level of exposure in dispersed treatments with very low retention levels. Aggregate treatments have less protection than unharvested areas (eg., controls), but have more protection than dispersed treatments. Until roots decay and fail, MPB-killed trees without foliage will initially have a reduced risk of windthrow due to a reduced sail area (Stathers et al. 1994). Varying levels of exposure and disturbance should also affect shrub cover volumes.  2.5 EFFECTS OF S T E M DENSITY O N WILDLIFE Boyland and Bunnell (2002) indicated that cavity-nesters would decline i f snag densities fell below a threshold density. Bunnell et al. (1999c) suggested a minimum of two large snags/ha and 10-20 smaller snags/ha be retained to sustain cavity nesters. Boyland and Bunnell (2002) felt there would be little to gain by retaining more than four large snags/ha. The value of a snag or wildlife tree will also increase when it is located in areas that contain a diversity of structural elements such as rocky outcrops, riparian areas, etc. (BCMoF 1995). Another important factor influencing wildlife habitat value related to the stem density of the residual stand is the windchill factor. The windchill factor affecting animals is influenced by the thermal cover provided by surrounding trees. A 2003 study by Natural Resources Canada estimated a body would lose heat 1.25 - 1.5 times faster in stands with 4 meter and 5 meter spacings respectively, and 3.0 times faster in clear-cut stands compared to unharvested stands. In addition, security cover for ungulates was found to be higher for the first five years in stands that were spaced compared to unspaced stands. However, five years after spacing, the security cover was found to be lower in the stands that were spaced. This may be explained by the growth of vegetation immediately after spacing (from 0-5 years) initially providing additional cover until canopy closure after five years when the loss of vegetation combined with the  18  reduced stem density would result in less overall cover than before spacing (Natural Resources Canada 2003).  2.6 SHRUB H A B I T A T Shrubs provide a variety of habitat values including berries for many wildlife species, browse for ungulates, cover for small and large mammals, and breeding sites for shrub nesting species. Shrub use can vary according to species. Orange-crowned warblers and song sparrows usually nest on the ground but may nest in low shrubs (Bunnell et al. 1999a). Pojar (1995) found a strong relationship with tall shrubs and warbling vireo, American redstart, brownheaded cowbird, yellow-rumped warbler, Swainson's thrush and black-capped chickadee. Low shrub cover and openings were associated with MacGillivray's warbler and orange-crowned warbler (Pojar 1995). Shrub nesters typically comprise 10-12% of the forest avifauna, but proportions are lower in stands where tall robust shrubs are uncommon (Bunnell et al. 1999d). Tall dense shrub cover can provide many benefits to wildlife including providing nesting and rearing habitat for songbirds and grouse, providing cover for small mammals, maintaining diversity of arthropod communities, and providing a moist stable microclimate that can promote sensitive plant species (Bunnell et al. 2003). However, Vega (1993) found dispersed retention of individual trees increased rates of predation on shrub nesters because trees provided perch sites for predators. These studies suggest retaining a variety of shrub volumes, species, and spatial arrangements is the most conservative strategy for managing the habitat of shrub-using species. 2.7 S N A G SIZE Since hardwoods decay faster than conifers, larger birds can find sufficient rot to create large cavities in smaller hardwood trees, whereas conifer trees need to be older and larger to provide the same size of decay pocket (Martin 2003). Densities of cavity-nesting birds are correlated with densities of large snags, but not smaller snags (Raphael and White 1984). 19  However, smaller snags are used for foraging and higher snag densities generally tend to increase species richness other than cavity nesters (Dickson et al. 1983). The data from the study by Bunnell et al. (1999c) suggested that the abundance of Redbreasted, Red-naped, and Williamson's sapsuckers, White-headed, Lewis's, and Pileated woodpeckers, and Northern flickers would decline where conifer snags were below a threshold size of 50cm. However, Bunnell et al. found Black-capped chickadee, Downy woodpecker, Hairy woodpecker, and Three-toed woodpecker selected nests in trees with dbh's less than 50cm, in inland forests. Smaller cavity nesters do not appear to be limited by current forestry practices, but Lewis's woodpecker, White-headed woodpecker, and William's sapsucker require larger trees and are listed at risk in the Pacific Northwest. Most cavity nesters currently listed at risk in the Pacific Northwest use large snags. Large trees are also used by squirrels, tree voles, bats, and some terrestrial salamanders (Bunnell et al. 1999a). The denning requirements for marten, fisher, lynx, and wolverine are found in stands characterized by large trees, hollow trees, down trees, snags, and stumps. Black bear dens are often found in large hollow trees, stumps and logs (Bunnell et al. 1999a). Forestry planning that focuses on harvesting the optimal milling diameters of 45-50cm will reduce the availability of tree sizes used by woodpeckers and other species that depend on them.  2.8 STAND STRUCTURE There are many definitions of stand structure including the following: •  the "vertical and horizontal make up or appearance of the stand" (BCMoF 2002),  •  "the physical and temporal distribution of trees and other plants in a stand" (Oliver and Larson 1990),  •  "the vertical and horizontal organization of plants" (Kimmins 1996), and  20  •  "an attribute of forest stand condition that describes the horizontal and vertical arrangement of trees; stand structure affects the light penetration and development, crown length and limbiness, stem H D R and the internal arrangement of wood fibres" (Farnden et al. 2003).  Horizontal structure in a stand will be the result of a combination of successional stages creating a mosaic of habitats. This structural mosaic is influenced by variations in soil depth, moisture, productivity, the presence of natural openings (eg., rock outcrops, rock talus), adaptations to disturbance events (disease patches, windthrow), and variations in the microtopography. Different wildlife species use different structural features to meet their habitat needs (eg., cover, forage, reproduction). The horizontal patchiness provides a diversity of structural features that can support more species over time. Vertical structure is the structure from top to bottom (eg., the layers of tree/vegetation/CWD) of the forest stand. A single species even-aged stand will have trees with similar growth rates and heights, and produce a minimum diversity of vertical structure. A n uneven-aged stand of varied species composition and growth rates will result in a much higher diversity of vertical structure. Patches of disturbance (eg., root rot, insects) can slow growth rates or cause mortality and result in variations in canopy height. Multi-layered canopies typical of lower density stands provide a diversity of structural habitat features that are suitable for foraging, nesting, resting, and snow/rain interception. A stand containing a mixture of tree species (eg., deciduous and coniferous) will provide a variety of structures that are important to many species (Alberta Centre for Boreal Studies 2000). Large fires will typically produce even aged stands containing patches of older forest that the fire missed and patches of younger more recently regenerated forest. Fires can also allow deciduous species to replace conifers. Of course, this situation may only persist until shade tolerant conifers (eg., climax species) overtake the shade-intolerant deciduous species. 21  Riparian areas can also influence stand diversity as a result of reduced fire effects and unique moisture regimes that produce different vegetation communities (Alberta Centre for Boreal Studies 2000). Some structural diversity can be maintained using different partial harvesting treatments including reserve patches in various arrangements and retention of dispersed individual trees. The appropriate selection and spatial arrangement of specific attributes of stand structure could presumably provide a host of associated wildlife values. However, success of a partial harvesting treatment depends on whether or not the retained elements providing habitat remain intact (or properly develop) over the long-term.  2.9 L I T E R A T U R E R E V I E W S U M M A R Y Several specific forest elements have been identified as providing significant habitat values to wildlife. In the Pacific Northwest, 69 vertebrate species commonly use cavities, 47 species respond positively to down wood, and 25-30% of the terrestrial vertebrate fauna use cavities. The stage of decay, type of snag, species of tree, and size of tree are attributes important to wildlife. It is apparent from many studies that the attributes of stand structure used most commonly by the 195 species identified by Chan-McLeod and Bunnell (2003) can be classified into the general groups of large dead/dying trees, hardwood trees, conifer trees, downed wood, shrubs, variations in stem densities, "old growth" stand conditions, early serai vegetation, and riparian habitat. Among these forest elements, large dead/dying trees, hardwoods, downed wood, shrubs, and old-growth conditions appear to be used by the most species from the list of 195 identified species. With the exception of old growth conditions, these elements can be conserved at the stand-level through the implementation of appropriately designed partial harvesting methods. These elements can also be easily measured in different partial harvesting treatments to infer variations in the level of potential habitat values present. The decay stage of snags is an important factor and was originally intended to be included in 22  this study. Unfortunately, very few snags with obvious signs of decay were recorded in any of the treatments or controls and were therefore not included in the final analyses. The scarcity of snags with suitable level of decay indicates an important habitat value is not available to primary cavity excavators. The long term supply of habitat at the landscape level relies on the collective input from sources at the stand level. Maintaining an adequate supply of habitat while harvesting large areas of land requires comprehensive planning. Strategies can include conserving areas of unharvested forest and retaining specific forest elements in partial harvesting treatments. If much of the trees are removed, but specific forest elements are retained, presumably some habitat values can be preserved. Further local studies need to be done to confirm whether or not retention of these variables in the partial harvesting treatments actually provide the actual wildlife values they were intended for. Partial harvesting treatments can be designed to retain many of these key elements, but the long term planning success of habitat supply depends on whether or not these elements respond to the treatments in a predictable manner. If standing trees with particular characteristics (eg., live, dead, large, old age) are required they must not only be left after harvesting, but they must also remain standing over the long term. Also, post-harvest levels of C W D volumes can vary according to the type of harvesting treatment and/or pre-harvest levels of CWD. However, C W D input can be planned from expectations of downed trees i f reasonably accurate predictions of windthrow can be made. Rates of windthrow will vary over time in response to peak wind events, but the level of risk for an individual tree varies with tree, site and neighbourhood factors. Criteria for selecting which trees to retain in a M P B management plan can include retaining non-pine species and immature pine. Classification of tree species, size, mortality status, windthrow status, C W D , and shrub volumes, along with abundance measurements of 23  these variables will allow indirect estimations of available habitat. Calculations of mortality and windthrow probabilities will allow predictions to be made on the future status of individual tree elements (eg., standing, live or dead, C W D input). Comparing data from the different treatments will provide estimations of how post harvest habitat values differ between treatment types. Mortality and windthrow probabilities can be used to improve the effectiveness of long term habitat management planning by predicting what the future stand will look like.  2.10 DESCRIPTION OF A T A R G E T STAND WITH THE MOST WILDLIFE V A L U E The current literature recognizes the main forest elements that provide important wildlife habitat include dead and dying wood, downed wood, shrubs, hardwoods, riparian habitat, and early and late successional stage vegetation. These different elements provide structures for various types of cover (eg., reproduction, hunting, escape) and foraging opportunities. Factors influencing tree habitat values include tree size, species, and age (eg., decay patterns), structural complexity, and spatial distribution. Maintaining a sustainable supply of wildlife habitat requires strategies that incorporate the dynamic processes (eg., windthrow, insects) that provide for the long-term input of the required forest elements (eg., old snags, CWD, late serai vegetation). General threshold levels for retention should also be incorporated into wildlife habitat planning strategies. Results from a study of coastal ecosystems by Chan-McLeod and Bunnell (2003) suggested species diversity would remain constant at retention levels 20-100%, but decline rapidly when retention levels were below 20%. Further, Chan-McLeod and Bunnell found individual species abundance would shift significantly with changes in retention levels below 20%. A partial retention treatment that attempts to meet the widest range of habitat needs presumably should retain an accompanying wide range and variety of the different elements that provide that habitat. Retaining both large and small trees will provide nesting and foraging opportunities. Retention of both live and dead trees is important to meet the immediate and 24  future needs of wildlife. Aggregate treatments will allow the survival of features that would not endure in the exposed conditions of a dispersed treatment. Chan-McLeod and Bunnell (2003) found some songbirds are more affected by dispersion pattern than retention level. Aggregated patches retained wildlife communities resembling those in old growth forests and dispersed retention patterns supported avian communities associated with clearcuts. Dispersed retention of individual trees can provide a distribution of habitat elements important for connectivity or territorial requirements. Dispersed treatments mimic the frequent stand-initiating wildfire disturbance regime (NDT3) in this study area, in that they leave even-aged lodgepole pine stands with a component of scattered Douglas-fir veterans (BCMoF 1995). Deciduous trees are relatively rare in the mature, lodgepole pine dominated stands examined in this study. Whenever possible, a mixture of deciduous and coniferous species should be retained to provide nesting and foraging habitat for strong and weak cavity excavators. Down wood is important to many species including birds, small mammals, amphibians, and invertebrates. Retention of shrubs is important to provide forage (eg., berries, insects) and cover (eg., nesting, security) for some species. The unique conditions of riparian areas are important to many species both aquatic and terrestrial. Retaining unharvested strips along streams (eg., riparian buffer zones) provides protection of riparian habitat and contributes to the area-based biodiversity objective. Riparian habitat also provides an inter-connecting network of corridors allowing migration and re-colonization of plants and animals. This study indicated dispersed treatments can supply early serai stage shrub and grass species while aggregates can retain late serai stage shrub species. At the landscape level, a mixture of early and late serai stages should be retained. Meeting the goals of ecosystem management at the landscape level requires the incorporation of effective stand level habitat management. Partial harvesting treatments can be implemented to retain a specific set of habitat values at the stand level. Ecosystem management at the landscape level requires providing a 25  long-term sustainable supply of all required habitat values and is beyond the scope of this study. Nevertheless, it can be assumed that retaining the correct mixture of unharvested forest, standlevel treatments and corridors with access to mature forest is necessary for biodiversity objectives to be met at the landscape level. At the stand level, a particular target stand design will depend on which specific set of habitat values are desired to be retained. One target stand may contain a relative abundance of large conifer trees, CWD, and late serai understory vegetation. Another target stand may contain an abundance of deciduous trees, shrubs, and early serai vegetation. Issues concerning connectivity may be addressed with a focus on riparian habitat, retention of wildlife corridors, and for certain purposes, dispersed individual tree retention. The design of the target stand will be dictated by the long term landscape level objectives and the potential habitat values available in the pre-harvested stand. The set of objectives associated with the desired target stand will dictate which partial harvesting methods will be the most effective at providing the required values. Retaining MPB-attacked pine trees individually or in groups can provide important habitat value. Beetle-attacked trees provide a source of food for woodpeckers in the first year (while beetle are present) and subsequently produce a standing dead snag in the years following. A study in the Nelson Forest Region by Steeger and Dulisse (1997) found mature beetleattacked lodgepole pine trees were important for nest trees and was the primary tree species for feeding by three-toed woodpeckers (Picoides tridactylus). A California study of ponderosa pine by Shea et al. (2002) found bark beetle-killed trees offered a "biologically rich snag that is both suitable and acceptable to cavity dependent species". Steeger and Dulisse (1997) recommend retaining wildlife tree patches containing M P B attacked trees within cutblocks in lodgepole pine stands.  26  The high risk of M P B attack for lodgepole pine must be considered in any long-term planning in this region since the inevitable result is the standing mortality of most if not all mature pine trees. M P B killed trees will eventually decay and become valuable wildlife trees and/or a source of CWD. If large areas of ground are not disturbed from harvesting, the understory will remain intact and produce the natural vegetation communities resulting from an M P B infestation. Large scale beetle infestations are a type of natural disturbance associated in this region and it is best to include some large undisturbed M P B attacked areas to mimic nature. Provincial guidelines recommend that"... during tree pest and disease treatments, some areas should be left untreated" (BCMoF 1995). The current management practices already benefit the species that are adapted to young managed forests, so retaining unharvested MPB-attacked areas will allow the natural ecosystem processes to continue and provide habitat for species dependent on those processes (Stadt 2001). Since forest companies will not be able harvest all infested stands before the commercial value of the trees are lost it is a sound strategy to retain unharvested areas in strategic locations that will maximize benefits to wildlife (Chan-McLeod and Bunnell 2003).  27  3 STUDY 1: EFFECT OF PARTIAL HARVESTING TREATMENT TYPE ON SUPPLY OF WILDLIFE HABITAT ATTRIBUTES 3.1  INTRODUCTION Two partial harvesting treatments 'dispersed' and 'aggregate' were studied to estimate  how the resulting stands differ in abundance of the forest elements that provide wildlife habitat The treatments were also compared to unharvested control areas. Canadian Forest Products Limited (Canfor) is an industrial collaborator for this project and the study site was done on their licensed area near Prince George. These types of partial harvesting treatments are used by Canfor to help meet the British Columbia Ministry of Forest's area based biodiversity objectives. Canfor's goals are met by retaining small unharvested areas of forest (aggregate treatment) and dispersed individual trees (dispersed treatment) that collectively amount to a target percentage of the overall harvested area. There are several factors that can be considered when selecting which trees should be left in aggregate and dispersed treatments. Leave trees in dispersed treatments should be relatively windfirm to maximize the probability that the trees will remain standing. It is expected that the aggregate treatments will provide more protection from wind than dispersed treatments that leave residual trees exposed (Scott, 2005). Factors considered for selection in aggregate retention include resistance to beetle-attack (non-pine species), specific wildlife values (eg., unique tree or site characteristics, riparian areas, vegetation), commercial values, and operational and silvicultural feasibility. Since aggregates retain areas of unharvested ground, they are better able to sustain most features than dispersed treatments. Whether an aggregate or dispersed treatment is successful depends on whether it meets its intended objective. Canfor's biodiversity objectives, as stated in the silviculture prescriptions for the study blocks, are derived from the area-based guidelines set in the British Columbia Ministry of Forest (BCMoF) Biodiversity Guidebook (1995). These guidelines calculate the percentage of a cutblock required for retention (eg., wildlife tree patches) according to the 28  "percentage of the biogeoclimatic subzone within the landscape unit available for harvest" (BCMoF 1995). Provincial guidelines further state "stand level recommendations are designed to maintain or restore, in managed stands, important structural attributes such as wildlife trees (including standing dead and dying trees), coarse woody debris, tree species diversity, and understory vegetation diversity". Also, "wildlife tree patches (WTP's) should be well distributed across the landscape. The maximum distance between WTP's (500 m) is based on territory size and dispersal requirements of wildlife" (BCMoF 1995). Meeting and sustaining these broad objectives requires that an adequate number of trees remain live and standing. One measure of treatment success can therefore given by calculations of leave tree distribution and windthrow/mortality status. Estimations of the abundance of certain stand features (eg., distributions of tree species and tree sizes, shrub cover volumes, CWD) can provide an indirect assessment of the wildlife values retained by the treatments. This study estimates the abundance of 11 wildlife habitat variables considered important to the species in the harvested region and tests for differences in abundance between aggregate treatments, three levels of dispersed treatments (High, Medium, and Low), and unharvested control areas. This study also includes an estimation of C W D input from post-harvest downed trees. The conversion of a standing tree to a downed tree (windthrow) contributes directly to an increase in C W D volume. Some of the hypotheses tested in this study include:  HQ\ There is no difference in the abundance of habitat variables between the five treatments (Control; Aggregate, DispH, DispM, and DispL). H\: The abundance of habitat variables is different in at least one of the treatments. HQ\ There is no difference in the volume of shrub cover between the five treatments (Control, Aggregate, DispH, DispM, and DispL). 29  H\: The volume of shrub cover is different in at least one of the treatments. Ho: There is no change in the volume of shrub cover over the span of seven years since harvest. H\: The volume of shrub cover changes over the span of seven years since harvest. A particular partial harvesting treatment may initially retain the desired set of stand characteristics, but these characteristics and their accompanying habitat values may not remain intact for long. Assessment of success must be based on what remains over the longer term. The outcomes in this study were measured in cutblocks that had been harvested over the previous seven years. This timeframe is considered adequate for assessing the longer term likelihood that individual trees remain standing since many trees that are vulnerable to windthrow are likely blown down in less than seven years following harvest. Burton et al. (2001) found windthrow risk in northern British Columbia appeared to increase until about three years after harvest. The question of long-term non-windthrow mortality is more difficult to assess since there are many additional factors involved (eg., dessication, insects, infection from wounds, etc,) that may take several years to become apparent.  3.2 METHODS 3.2.1 S T U D Y SITE The study site is located in the Prince George Forest Region just southwest of Prince George in the Gregg and Pelican units in Canadian Forest Products Limited's (Canfor) Timber Supply Area 15. This area is in the Biogeoclimatic Ecosystem Classification (BEC) Sub-Boreal Spruce (SBS) zone subzone/variant dw2 (Delong et al. 1993). The SBSdw2 is dry and warm relative to other biogeoclimatic zones in the region. Winter precipitation is relatively low for the region, with snow packs accumulating up to 2 metres. Climatic factors limiting growth are drought on drier sites and frost on frost-prone sites. Predominant soils are Luvisols with 30  gravelly loam and clay loam textures. There are also components of Dystric Brunisols with gravelly sand and loamy sand textures. The forests of the SBSdw2 are diverse with mixtures of lodgepole pine (PI), Douglas-fir (Fd), and hybrid white spruce (Sx). Lodgepole^pine and Douglas-fir dominate on drier sites, and hybrid white spruce dominates wetter sites. Black spruce (Sb) grows in wetlands and poorer upland sites, and presence of subalpine-fir increases at higher elevations. Trembling aspen (At) is abundant in upland deciduous forests and black cottonwood (Ac) grows along rivers and streams (Delong et al. 1993). The natural disturbance regime of the SBSdw2 in the Prince George Forest Region is classified as NDT3 and describes a landscape with frequent stand initiating events. The mean fire return interval is about 125 years. Douglas-fir is a more fire-resistant species than pine so the abundance of Douglas-fir will determine the number, size and aggregation of mature remnant trees that survive extensive crown fires to provide structural diversity. The frequent stand-initiating wildfires in these dry forests have created a landscape of a mosaic of even-aged stands of different ages (BCMoF 1995). Uniform lodgepole pine stands often contain a component of scattered Douglas-fir veterans.  '  3.2.2 S A M P L I N G P L A N Canfor's rationale for using the partial harvesting treatments used in this study was to meet an area-based biodiversity objective. Aggregates were selected for retention when small areas of within a cutblock had certain characteristics such as non-pine species, immature trees, or specific wildlife values (eg., gullies, riparian areas, swampy ground). The treatments in this study that retained aggregates or groups of trees (eg., WTP's) within harvested blocks will be referred to as aggregate treatments. When pre-harvest stands contained a component of large old Douglas-fir veterans, they were often retained as individual leave trees within the cutblocks.  31  These treatments will be referred to as dispersed treatments in this study. Occasionally, i  dispersed treatments retained individual trees of species other than Douglas-fir. In order to capture a wide range of variability, cutblocks within each treatment were i  grouped to include post-harvest ages from 1 to 7 years since harvest (YSH 1-7), and blocks with and without the presence of water (eg., streams, swamps, seepage or wetter ground). It was initially expected that periodically wet ground conditions could influence windthrow potential. Including all of these parameters in the selection process meant there were a limited number of cutblocks available from the study site that could be classified into the appropriate categories of aggregate and dispersed treatments. Consequently, there were insufficient blocks to randomly select treatments from a larger population of candidate blocks. However, there were sufficient blocks to represent all of the treatments required for the sampling matrix. The data were obtained from 35 total cutblocks comprised of 10 dispersed treatment blocks, 11 aggregate treatment blocks, and 14 blocks containing both aggregate and dispersed treatments. The sampled cutblocks ranged in size from 20-70 ha and only included stands where the main species was originally pine. Nine unharvested areas were also sampled and used as controls. The dispersed treatment blocks were further grouped into the three different relative dispersed retention levels of low, medium, and high (DispL, DispM, and DispH). The method for classification of retention levels for dispersed treatments was subjectively based on I  information from the silviculture prescription documents for the harvested cutblocks and from air photos where available. It was found during sampling that all but one of the dispersed treatment blocks actually contained very low stem densities. Nevertheless, representation from all three retention classes provided a range of retention values that could be used in the regression analyses. A l l five treatments were sampled using three circular (5.65m radius) fixed area (l/100ha) plots per treatment block that measured site and tree data. Each dispersed treatment 32  block was also sampled with three additional large rectangular fixed area (40m x 83.3m =l/3ha) plots in which only tree data were recorded. The circular l/100ha plots were too small to capture enough tree data in the low stem densities of the dispersed treatments so the larger rectangular plots were used in these three treatments to gather only tree data. The one exception i  was the dispersed treatment (Block 565-1) that had a much higher stem density (407sph). In this block we used three mid-sized circular (12.61m radius) fixed area (l/20ha) plots to capture tree data.  :  The total number of plots in which tree data were recorded was 174 (24 dispersed blocks x 3 plots + 25 aggregate blocks x 3 plots + 9 control blocks x 3 plots = 174). A n additional 72 dispersed site data plots (24 x 3 - 72) made the total number of plots measuring site and tree data 246 (174 + 72 = 246). Plots were randomly located by placing a metric dot grid (4 dots per square cm) in a North/South orientation over a 1:10000 map of the treatment cutblock and using a random number generator to identify the three dot numbers to be used for the three plot locations. During field sampling, i f a portion of a plot landed outside the treatment, it was moved just sufficiently in the reverse direction of travel to allow the total plot area to remain in the treatment. Variables representing various tree attributes and site features (Table 1) were measured at each plot. These variables included species, dbh, height, and tree status (live or dead, standing or down). These data were used to create six tree variables (StLvCon, StLvDec, StDdCon, StDdDec, DnCon, and DnDec) and the tree basal area variable (BATree). The plots also recorded the percent ground cover and average height of shrubs in the plot. The ground area of the shrub cover (dripline of foliage) was multiplied by the average height of the shrubs to produce an estimate of the 3-dimensional volumes of shrub cover (shrubvol). A mirror densiometer was used to measure percent canopy cover (CovA) at each plot. Line Intersect Sampling (LIS) was used to measure C W D at each plot. The LIS method used two 25 metre 33  lines emanating from the plot centre to a sum of 50 metres of sampling line per plot. The first 25m line was placed in a random direction and the second 25m line was placed at a 90 degree angle clockwise to the first line. If a sampling line traveled outside of the treatment it was continued in the reverse direction from the opposite end to produce a total line length of 25m. Post-harvest C W D input was estimated by calculating the volumes of trees in the l/100ha plots that were windthrown after harvesting. Table 1. Habitat variables used in the analyses., Label StLvCon  Name Standing Live Conifers  Unit Measurement Stems/Ha  Description Standing live conifer species include lodgepole pine, Douglas-fir, interior spruce, black spruce, and sub-alpine-fir.  StLvDec  Standing Live Deciduous Standing Dead Conifers Standing Dead Deciduous Down Conifer  Stems/Ha  Standing live deciduous species include trembling aspen, paper birch, cottonwood, and alder. Standing dead conifer species same as standing live conifer species.  DnDec  Down Deciduous  Stems/Ha  BATree  Basal Area  m /Ha  UpBATree  Standing Basal Area  m /Ha  Basal area of live and dead trees standing at time of sampling.  Shrubplotvol  3-D Shrub Volume  m /Ha  3-dimensional projection of shrub cover.  Name % Canopy Cover  Unit Measurement % Canopy  StDdCon StDdDec DnCon  Table 1 cont. Label CovA CWD  Coarse Woody Debris  i  Stems/Ha Stems/Ha  ; ]  Down conifer are conifers that were left standing after harvest, but were down at time of sampling.  Stems/Ha  Down deciduous conifers that were left standing after harvest, but were down at time of sampling. Basal area per ha of live and dead trees left standing immediately after harvest.  2  2  3  m /Ha 3  Standing dead deciduous species same as standing live deciduous species.  ;  34  Description Percent canopy closure Post-Harvest CWD is coarse woody debris that was present immediately after harvesting. *CWD input is post-harvest down trees.  *CWD input is considered as trees that were left standing after harvest, but subsequently became down trees (eg., windthrow). The basal area of trees was calculated from the diameter at breast height (dbh). The basal area of pre-harvest trees (dispersed treatments) was estimated from stump diameter classes recorded in 5cm increments. Since trees were cut at stump height (30cm), the diameters of stumps were larger than they would be at the height of dbh (1.3m). To adjust this difference, recorded stump diameters were reduced by 13%. This was the mean difference between dbh and stump height found in a sample of 13 lodgepole pine trees in a different study (Byrne 2005) from the same area. Basal area calculations using stump diameter classes instead of dbh measurements only produced an approximate pre-harvest basal area. Nevertheless, it did provide a rough estimate of basal area before harvesting took place. Basal areas of trees in aggregate treatments and controls were measured at dbh allowing relatively more accurate comparisons between these two treatments. Volumes of windthrown trees were conservatively estimated as the product of basal area times 1/3 of the height (BA x 1/3 ht). Data from variables were converted to per hectare units using the inverse of the plot size as a plot multiplier. Plot level statistical analyses were conducted after this conversion.  3.2.3 STATISTICAL A N A L Y S E S Tests for normality and homogeneity of variance on the habitat variables in the different treatments indicated that most of the variables were not normally distributed nor did they have homogeneous variances (Appendix 3). Since these assumptions were not met, the parametric statistical analysis of variance ( A N O V A ) test was not used. Rather, the non-parametric, Kruskal-Wallis: one-way analysis of variance by ranks, was used to test whether the differences between treatments were significant. The Kruskal-Wallis test ranks all observations sequentially and a formula approximates the variance of the ranks. Tied observations are I  assigned an average value of the tied ranks. When the samples are from different populations, 35  the sums of the ranks are different, a large H-value is produced, and the null hypothesis is i  rejected.' If the samples are from the same population, the sums of the ranks are approximately ! I  the same, the H-value is small, and the null hypothesis is not rejected. Using this non-parametric method is appropriate because the variables do not need to be i i  normally distributed and the hypotheses can be tested without knowledge of population parameters (Siegel and Castellan 1988). One disadvantage of the Kruskal-Wallis test is that the hierarchal ranking of observations does not provide the sensitivity to discern the level of i difference between the values of individual observations. Non-parametric methods use less information and are less sensitive than their parametric counterparts. Given the large i differences in abundance of most of the habitat variables, the Kruskal-Wallis method was considered adequate to indicate i f differences between treatments were significant (Siegel and Castellan 1988). However, it is possible some important differences could be missed due to the inherent insensitivity of the test.  !  A significant Kruskal-Wallis test indicates that at least one of the treatments is different than the others, but it does not provide information about which treatments are different. This distinction can be made by comparing the differences between the treatment mean scores to i  i  critical values. Table 2 shows calculations of critical values used to compare mean scores between treatments using a significance level of 0.05. The critical values (CV) are calculated with the following formula: C V = Z i ha/k(-i) x ;Square Root [(N(N+1)/12) (l/n +l/n )]. a P  •  u  v  Z i ha/k(-i) is the 'abscissa value from the normal distribution above which lies alpha/k(ka p  1) percent of the distribution' and can be obtained from a 'critical Z values for #c multiple comparisons table' (Siegel and (Castellan 1988). •  The number of comparisons (#c) uses the formula #c = k(k-l)/2.  •  In this case, there are five (k = 5) different treatments so #c = 5(5-l)/2 = 10.  i  36 I  •  With #c= 10 and alpha = 0.05, the Z-value Z i ha/k(-i) = 2.807. a P  i  • •  ni = number of plots in treatment.  ;  ricontrol 27, noispH 24, noispM 24, nQispL = 24, nAggregate 75 i N = total number of plots. N = (27+24+24+24+75) = 174. =  =  =  =  Table 2. Critical Values (CV) of Kruskal-Wallis test for comparing mean scores between treatments (alpha = 0.05). i  Treatment Comparison  Critical Value 'Calculation  Critical Value  Control vs. Disp (H, M , L)  2.807 Sqrt [(174(174+l)/12)(l/27+l/24)  39.6682  2.807 Sqrt [(174(174+l)/12)(l/27+l/75)  31.7346  Aggregate vs. Disp (H, M , L)  2.807 Sqrt [(174(174+l)/12)(l/75+l/24)  33.1609  Disp(H, M , L) vs. Disp(H,M, L)  2.807 Sqrt [(174( 174+1)/12)( 1724+1/24)  40.8183  Control vs. Aggregate  '  3.2.4 RECONSTRUCTION OF P R E - T R E A T M E N T S T A N D Stumps were counted and recorded for species and diameter classes in all plots sampled in the dispersed treatments. The stump counts were used to estimate pre-harvest species stems per hectare (sph) and approximate basal areas. Using pre-harvest tree densities and basal area i  approximations allowed the pre-harvest stand to be "reconstructed". Relatively low sph/basal i  area ratios were assumed to indicate larger mean tree sizes. The species distributions, stem densities, and basal areas from the reconstructed dispersed treatments were then compared to aggregate and control treatments to contrast pre-treatment differences. i  3.2.5 C R O W N / D B H RATIOS  j  One of the factors influencing windthrow is the ratio of the crown sail area to the dbh of the tree (Stathers et al. 1994). It seemed possible that newly exposed trees in the dispersed treatments might adapt open-grown crown forms that could affect windthrow susceptibility. In j order to gain some insight into whether or not crown/dbh ratios were changed as a result of  37  exposure in dispersed treatments, crown/dbh ratios of the six major tree species were compared between the control, aggregate, and dispersed treatments. 3.3 RESULTS  ,  Pre-harvest stem densities were generally lower in the harvested areas (dispersed treatments) than in the aggregates or controls, but basal areas were relatively higher (Figure 1  1). This meant pre-harvest dispersed stands had larger mean dbh than aggregates and controls. i Mean dbh was also larger in aggregates than controls. There were some differences in preharvest species composition between treatments (Figure 2 ) . The aggregate treatments i  contained the lowest proportion of lodgepole pine (PI) and the highest proportions of non-pine 1 and 2) indicate a biasinfor harvesting larger pine in dispersed treatments and for retaining conifers. Differences pre-harvest stem density, basal area, and species proportions (Figures smaller pine and non-pine species in aggregate treatments. This trend is the result of focusing i  harvesting efforts on mature pine that was either killed by M P B or at the highest risk of future attack. Pre-harvest SPH and Basal Area 1400  j  1200  -•  1000  -•  x  800 -•  ft  600 ••  •• 60 -•50  4 40  Pre-harvest SPH and Basal Area 1400  nSPH _  B  ••  T  1200 -•  A  6  0  -•50  OSPH _  1000 -•  „ x  800 -•  ft 600 -•  400 ••  400 -• 200 -•  4 10 m  0  DispL DispM DispH Aggr Control  0 DispL DispM DispH  Figure 1. Pre-harvest stems per hectare (SPH) and basal areas (BA).  38  Aggr Control  B  A  Proportion of PI in Pre-harvest Stand  Proportion of Fd in Pre-harvest Stand  .» 100% -i Q.  V)  o 20%  M  80% -  115%  "ID55 60% -  40% 20% 0% se > l_  !  OS  SZ Q.  to <D O OJ Q.  30%  -i  j  5% -  -rm  OJ  DispL  DispM  DispH  Aggr  arl  Proportion of Sx in Pre-harvest Stand -i  n  r—i  ;£  o% I  j  M  , I'- •! , W , I- I , r~i  DispL  !  »V is  •-  DispM  DispH  Aggr  ,  Ctrl  Proportion of Sb in Pre-harvest Stand  I I 2% -, • u  i — i  1  il  VI  •5<D 5 2p% >  , >  10% 0J I CL  o%  |  DispL  DispM  DispH  Aggr  Ctrl  i  OT  V, 2%  5?  'I , DispL  1  DispM  1  DispH  '  Proportion of Ac in Pre-harvest Stand  r-— V  ' l_  Si  0% DispL  DispM  DispH  Aggr  .  Ctrl  g 0.8% -I  ! £ 0.4% -  1%  i  Aggr  !! "55I 0.6% -  Q.  I <D L_ Q.  | I [  ou 4% < 3% 0D  SZ  /•  i  Proportion of Bl in Pre-harvest Stand  ID > OS  1% -  £ I # 0% -I  CO  -  0.2% Ctrl  !^ i  0  i  0  %  _|  1  DispL  , N*! , I M ,  (  DispM  DispH  Aggr  Ctrl  Figure 2. Proportions of tree species in pre-haryested stands. The results of the Kruskal-Wallis test (Table 3) indicate that the abundance of all habitat variables except the two standing deciduous variables (StLvDecha, StDdDecha) differ significantly among the five treatments (Control, Aggregate, Disp-H, Disp-M, and Disp-L). The Kruskal-Wallis number of observations (n) and mean scores are shown in Table 4. i  Table 3. Kruskal-Wallis one-way analysis of variance by ranks test. Variable Chi-Square i Pr > Chi-Square* StLvConha 98.7 j .<.0001 StLvDecha 8.3 : 0.0827 StDdConha 91.0 <.0001 StDdDecha 2.1 0.7125 DnConha 32.6 j <.0001 DnDecha 10.1 0.0391 BATreeha 115.8 <.0001 UpBATree 88.1 ! <.0001 shrubvol 19.5 i 0.0006 CWD 32.6 j <.0001 CovA 112.2 | <.0001 * p-values less than 0.05 are considered significant and are indicated in bold !  :  39  Table 4. Kruskal-Wallis mean scores. N  StLvConha StLvDecha StDdConha StDdDecha DnConha DnDecha BATreeha UpBATreeha ShrubVol CWD CovA  Control  DispH  27  24  DispM  DispL  Aggregate  24  24  75  121.63 86.19  49.58 104.50  40.48 89.08  37.92  118.26  88.42  152.37  57.85  81.73  49.17  85.31  92.69  50.67  97.69  88.96  85.48  86.81  i  79.21  91.71  99.07  95.67 51.00  99.10  85.19  '  89.02 37.67  52.31  !  53.27  115.37 38.67  115.37 106.09  49.81  84.146  i  91.00  94.81  98.68  49.33  106.31  :  104.75  59.75  98.58  121.87  45.96  !  36.25  39.37  120.22  na na  106.27  137.26 140.96  1  i ;  |  The absolute value differences in Kruskal-Wallis mean scores are shown in Table 5. i  Significant differences (significance level <0.05) greater than critical values between treatments are indicated in bold. Down trees (DnCon, DnDec) were not recorded in controls so these j differences in mean scores are not applicable (na). Table 5. Absolute differences in Kruskal-Wallis mean scores.  StLvConha  Ctrl DispH 72.05  Ctrl DispM 81.15  Ctrl DispL 83.71  Ctrl Aggr 3.37  Aggr DispH 68.68  Aggr DispM 77.78  Aggr DispL 80.34  DispH DispM 9.10  DispH DispL 11.67  DispM DispL 2.56  StLvDecha  18.31  2.90  2.23  4.45  22.77  7.35  6.68  15.42  16.08  0.67  StDdConha  94.52  103.20  101.70  54.68  39.83  48.52  47.02  8.69  7.19  1.50  [  StDdDecha  7.37  3.64  0.16  1.49  5.88  2.15  1.33  3.73  7.21  3.48  DnConha  na  na  na  na  7.20  19.86  7.36  27.06  14.56  12.50  DnDecha  na  na  na  na  10.48  3.83  13.92  6.65  3.44  10.08  BATreeha  86.26  99.59  106.49  21.89  1 6437  77.70  84.60  13.33  20.23  6.90  UpBATreeha  88.65  87.69  102.30  34.87  53.78  52.82  67.43  0.96  13.65  14.60  shrubvol  34.33  41.19  45.00  48.87  14.53  7.68  3.87  6.85  10.67  3.81  CWD  56.98  55.42  10.42  49.25  7.73  6.17  38.83  1.56  46.56  45.00  CovA  75.91  85.62  82.49  1.65  I  83.97  80.84  9.71  6.58  3.12  74.26  Significant differences greater than the critical values are indicated in bold \  i  i  40  Graphs of mean densities (sph), volumes (m ), or areas (m ) (including standard error 3  2  bars) of the habitat variables are given in Figures 3- 1'8. Some of the graphs in these figures are accompanied by a table of the significant differences in K W mean scores. The graphs indicate i j  the observed mean abundances in the treatments and the K W mean scores table estimates i f these differences are significant. If the K W mean score differences are not significant they are i I  j  not listed in the K W mean scores table.  i  3.3.1 STANDING L I V E CONIFERS (StLvCon) StLvCon densities in the control and aggregate treatment did not differ significantly from each other, but both were significantly higher than densities in the dispersed treatment (H, t  M , and L). Total standing conifer densities including both live (Figure 3) and dead (Figure 5) trees were higher in controls than in aggregates; but the proportion of non-pine conifers (Fd, Sx, Bl) was higher in aggregates (Figure 2). In spite of higher conifer densities in controls, higher i i  pine mortality from M P B in controls resulted in similar abundance of standing live conifers in controls and aggregates (Figure 3.  ! Significant differences in KW mean scores  StLvCon  700 600 500 i 400 300 200 100 H 0  Treatments  M  DispL Figure 3.  DispM  DispH  Aggr  KW mean score difference  Control-DispH  72.0463  Control-DispM  81.15046  Control-DispL  83.71296  Aggregate-DispH  68.67667  Aggregate-DispM  77.78083  Aggregate-DispL  80.34333  Control  Standing live conifers stems per ha (sph) by treatment.  3.3.2 STANDING L I V E DECIDUOUS (StLvDec) Although the graph of standing live deciduous (Figure 4) indicated variations in the abundance of standing live deciduous between the control and the dispersed treatments, and  41  between the aggregate and dispersed treatments, the Kruskal-Wallis test found no significant differences. This may be due to insensitivities in the Kruskal-Wallis test and low abundance of standing live deciduous in all treatments, especially dispersed treatments, and high variability between aggregate and control.  StLvDec 60 -i 50 40 X Q- 30 01 20 10 0• DispL  DispM  DispH  Aggr  Control  Figure 4. Standing live deciduous stems per ha (sph) by treatment.  3.3.3 STANDING D E A D CONIFERS (StDdCbn) For StDdCon, the Kruskal-Wallis test indicated that control, aggregate and dispersed i  treatments are significantly different from each 'other but the high, moderate and low dispersed treatments are not significantly different from each other (Figure 5). Abundance of StDdCon is relatively high in control, moderate in aggregate, and very low in dispersed treatments. A n important consideration when comparing standing dead conifers in controls and aggregates is that there are twice as many MPB-susceptible lodgepole pine in controls than in the aggregates. '  Significant differences in K-W mean scores  StDdCon 800 i 600 -  fe  X Q. 400 -  w  Treatments Control-DispH Control-DispM Control-DispL Control-Aggregate Aggregate-DispH Aggregate-DispM Aggregate-DispL  200 0 • DispL  DispM  DispH  Aggr  Control  Figure 5. Standing dead conifers stems per ha (sph) by treatment.  42 i  KW mean score difference 94.516 103.204 101.704 54.684 39.832 48.520 47.020  3.3.4 STANDING D E A D DECIDUOUS (StDdDec) The graph of standing dead deciduous (Figure 6) suggested a treatment effect, but the Kruskal-Wallis test found no significant differences between treatments. This may be due to i i  insensitivities in the Kruskal-Wallis test and the overall low abundance and high variability of i  standing dead deciduous. Densities of standing dead deciduous are very low in all treatments.  16 14 12 10 X Q_ 8 i/i 6 4 2 0  StDdDec  -, DispL  DispM  DispH  Aggr  Control  I I  I  Figure 6. Standing dead deciduous stems per ha (sph) by treatment. i i  3.3.5 D O W N CONIFERS (DnCon)  j  The graph of down conifers (Figure 7) suggests a variation in the abundance of down conifers between treatments, but the Kruskal-Wallis test found no significant differences. This may be due to the lack of sensitivity of the Kruskal-Wallis test. The aggregate treatments contained much higher densities of trees than dispersed treatments and commonly suffered extensive windthrow, especially in small aggregates and near aggregate edges. Post-harvest downed conifers were not recorded in the control plots because of the inability to assign a time since harvest. However, the relatively low number of firm solid logs lacking any signs of decay detected on the line transects suggested few trees were windthrown in the control plots over the period of interest.  j  43  DnCon  140 120  n  100 H  x ft  80 60 40 -I 20 -I 0 DispL  DispM  DispH  Aggr  j  Figure 7. Down conifers stems per ha (sph) by treatment. !  3.3.6 D O W N DECIDUOUS (DnDec)  j  The graph of down deciduous trees (Figure 8) suggests a difference in the abundance of down deciduous between dispersed and aggregate treatments, but the Kruskal-Wallis test found i  no significant differences. This was probably due to the lack of sensitivity of the Kruskal-Wallis test and the overall low abundance of down deciduous trees in all treatments. The densities of | down deciduous trees are low in all treatments.! Down trees were not recorded in controls.  DnDec  20 15 ft  1 0  5 I  0  DispL  DispM  I|  DispH  Aggr  Figure 8. Down deciduous stems per ha (sph) by'treatment.  3.3.7 B A S A L A R E A PER H E C T A R E (BATree) Significant Kruskal-Wallis differences in basal area were indicated between control and dispersed treatments, and between aggregate and dispersed treatments, but not between aggregate and control treatments. Since this basal area variable is an estimate of all trees (live and dead) left standing immediately after harvesting it was expected that the dispersed treatments would contain considerably less basal area than aggregate and control treatments (Figure 9). 44  Significant differences in KW mean scores  BATree  Treatments  60 50 40 30 20 10 0  DispL  Figure 9.  DispM  DispH  Aggr  Control  Control-DispH Control-DispM Control-DispL Aggregate-DispH Aggregate-DispM Aggregate-DispL  KW mean score difference 86.259 99.593 106.488 64.3667 77.700 84.596  Post-harvest basal area per hectare by treatment.  3.3.8 B A S A L A R E A OF STANDING TREES (UpBATree) There were significant Kruskal-Wallis differences between control and dispersed treatments, and between aggregate and dispersed treatments in the basal area of standing trees at the time of sampling (UpBATree). However, UpBATree also indicated a difference between control and aggregate treatments (Figure 10). This difference is likely due to aggregates experiencing more windthrow than controls. Presumably, the majority of basal left after harvesting (Figure 9) was intended to remain standing in both treatments, but more basal area (eg., more trees) was windthrown in aggregates than dispersed treatments (Figure 11) simply because there was more basal are available to be windthrown in the aggregate treatments (Figure 9). Significant differences in KW mean scores  UpBATree  50 i 40 -  Treatments Control-DispH Control-DispM Control-DispL Control-Aggregate Aggregate-DispH Aggregate-DispM Aggregate-DispL  J= 30 • E 20 •  10 • _r*L  0• DispL  DispM  DispH  Aggr  Control  Figure 10. Standing basal area.  45  KW mean score difference 88.650 87.692 102.296 34.870 53.781 52.822 67.427  Basal Area of Windthrown Trees  II*  0  DispL  i * 1  1  DispM  •  1  DispH  1  Aggr  Figure 11. Basal area of post-harvest windthrown trees.  3.3.9 C A N O P Y C O V E R (CovA) The Kruskal-Wallis test indicated significantly different canopy cover in control and aggregate treatments compared to the dispersed treatments. These differences are simply the result of many more trees in the control and aggregate treatments compared to the scarce residual trees retained in the dispersed treatments. The overall low densities of standing trees in the three Dispersed treatments meant there were no significant differences in canopy cover between them (Figure 12).  40%  3 o o  Significant differences in KW mean scores  CovA  i  30%  Treatments  Control-DispH Control-DispM Control-DispL Aggregate-DispH Aggregate-DispM Aggregate-DispL  I 20% c  0% DispL  DispM  DispH  Aggr  KW mean score difference 75.912 85.620 82.495 74.262 83.970 80.845  Control  Figure 12. Canopy cover.  3.3.10 SHRUB C O V E R V O L U M E (shrubvol) The Kruskal-Wallis test indicated significant differences in shrub volumes between control and dispersed treatments and between control and aggregate treatments. The Kruskal-  46  Wallis test did not indicate significant differences between aggregate and dispersed treatments or among the dispersed treatments (Figure 13). Figure 14 indicates the relationship of shrub cover and canopy cover. Canopy cover (CovA) was similar between control and aggregate treatments, but shrub cover was highest in aggregate and lowest in control (Figure 14).  Significant Differences in KW Mean Scores KW mean score Treatments difference Control-DispM 41.185 Control-DispL 44.998 Control-DispH 34.331 Control-Aggregate 48.865  shrubvol 8000 i 6000 4000 • 2000 0 DispL  DispM  DispH  Aggr  Control  Figure 13. Shrub cover volumes (m ) by treatment. 3  6000 -I 5000 • jp 4000 -  j 40% -• 30%  ^ 3000 •  >. o o  -• 20% Q.  ;,;3  shrub volume -CovA  c  •S 2000 • 1000 • | i 0 •  • 3-D •  .''*•  —h-  fe"  DispL DispM DispH Aggr  -•10%  o  JZL -• 0% Ctrl  Figure 14. Relationship of shrub cover and canopy cover. Figure 15 indicates the relationship of shrub cover and grass. This graph suggests an inverse relationship of grass to shrub volumes in the dispersed treatment group. The data infers grass is favoured over shrubs in the highly exposed and highly disturbed dispersed treatments. Shrubs are favoured over grass in the undisturbed, and moderately exposed aggregate treatments. However, both grass cover and shrub volumes are low in the relatively low-light undisturbed controls (Figure 15). Figure 16 illustrates no relationship exists between shrub cover volume and grass cover indicating competition is not a factor. Figure 17 indicates large 47  increases in grass cover in the dispersed treatments for the first two years (YSH 1-2). Grass cover increases in the aggregate treatments are less dramatic. Shrub volume showed an increasing trend for the first three Y S H classes, suggesting a change over time, but did not continue to increase with additional time (Figure 18). Shrub V o l u m e s and % G r a s s Cover 6000  T  4000  -•  -  '-  50 ,_  fm. -  9  HP  -•  —t—  —t—  —1—  DispL DispM DispH Aggr  30  S3  20  o.  10  *  CD  8* .  [fl Ctrl  a m3/ha shrub volume — % grass| cover  0  Figure 15. Grass cover (%) and shrub volume (mVha) for dispersed and aggregate treatments. m3/ha Shrub Volume vs % Grass Cover 7 G % >  5  3  4  I  3 2 1 20  40  60  80  100  % grass cover  Figure 16. Shrub cover volumes versus grass cover Grass Cover in Y S H Classes  80 -| 70 60 (U  • grass (disp)  50 -  m grass (aggr)  A  o 40 o S« 30 • 20 10 0 YSH1  YSH2  YSH3  YSH4  YSH5  YSH6  YSH7  Figure 17. Grass cover in years since harvest (YSH) classes. No data was collected from the dispersed treatments for YSH 6.  48  Shrub Volumes in Y S H Classes  18000 15000  a Disp D Aggr  12000 9000 6000 H 3000 0  YSH1  YSH2  YSH3 YSH4 YSH5  YSH6 YSH7  Figure 18. Shrub volumes (m /ha) in years since harvest (YSH) classes. 3  3.3.11 C O A R S E W O O D Y DEBRIS (CWD) The ' C W D ' variable only recorded the C W D present immediately after harvesting and did not include any additional input from post-harvest downed trees. C W D was lowest in the controls and highest in the high and medium dispersed treatments (Figure 19). The KruskalWallis test indicated significant differences between control and Disp H - M , between control and aggregate, between aggregate and DispL, between DispH and DispL, and between DispM and DispL. Significant Differences in KW Mean Scores  CWD 200  -i  150  Treatments Control-DispH Control-DispM Control-Aggr Aggr-DispL DispH-DspL DispM-DspL  i  £ 100 -I E 50 H  i i Ijjiw  0 DispL  DispM  DispH  Aggr  Control  KW mean score difference 56.979 55.417 49.247 38.830 46.562 45.000  Figure 19. Distribution of CWD.  3.3.12 C W D INPUT F R O M POST-HARVEST D O W N E D TREES Aggregates supplied more total trees to be windthrown than dispersed treatments and therefore produced the most new C W D input from down trees (Figure 20). In fact, aggregates often resulted in a high occurrence of windthrow, especially smaller aggregate sizes. However, a higher proportion of retained trees in the dispersed treatments were windthrown compared to 49  aggregate treatments (Figure 21). It was expected that the proportions of windthrown trees would be similar in all dispersed treatments or have a slight inverse relationship to retention levels. It is unclear why the moderate retention had a lower proportion of downed stems than the other dispersed treatments; it may be due to sampling error. Proportions of windthrown trees are based on both live and dead standing trees, but presumably most if not all of the retained trees in the dispersed treatments were live immediately following harvest. C W D Input f r o m W i n d t h r o w n T r e e s  70 60 50  EJ C W D Input from Wlnctthrown Trees  H  40 30 20 1 0 -\ 0 DispL  DispM  DispH  Aggr  Figure 20. Coarse woody debris inputs from windthrown trees.  Proportions of Down Trees in Treatments 70% -i 60% CO 50% cu C D -t= 40% CZ  o  30% 20% 10% 0% -  DispL  DispM  DispH  Aggr  Figure 21. Proportion of total trees that were windthrown post-harvest in aggregate and dispersed treatments.  3.3.13 DISTRIBUTION OF STANDING TREE SIZES IN THE T R E A T M E N T TYPES There are dramatic differences in the distribution of post-windthrow standing tree sizes between the treatments (Figure 22). The most obvious difference is between the dispersed treatments and the two other treatments (control and aggregate). This was expected since relatively few trees were left after harvesting in the dispersed treatments and a high proportion 50  of those leave-trees were windthrown. Most trees selected for retention in the dispersed treatments were large Douglas-fir. This is the most appropriate choice since Douglas-fir is not at risk to M P B attack and is considered more windfirm than spruce. The actual stem densities in dispersed treatments were quite low. The densities of all species combined in the dispersed treatments only ranged from 7-79 sph with the exception of one dispersed block (Block 565-1) with an unusually high density of 407 sph. The mean densities of large (>40cm dbh) trees remaining standing in the dispersed treatments were very low with a range of only 1.08 to 6.38 sph (Figure 22). Mean densities of large trees were 18.52 sph in controls and 21.62 sph in aggregate treatments.  Densities of Standing Tree Size Classes 700 600 500 x 400 & 300 200 100 0  — large trees • medium trees A small trees  DispL  DispM  DispH  Aggr  Control  Densities of Standing Tree Size Classes 700 600 500 x 400 & 300 200 100 0  — large trees • medium trees A small trees  A  DispL  DispM  DispH  Aggr  Control  Figure 22. Distribution of standing post-windthrow tree sizes. Large trees >40cm dbh, medium trees 20-40cm dbh, small trees < 20cm dbh.  51  3.3.14 C R O W N / D B H RATIO DISTRIBUTIONS There was little difference in crown/dbh ratios of pine (PI) between the dispersed and aggregate treatments, but ratios were relatively smaller in controls (Figure 23). Spruce trees (Sx) appeared to have higher crown/dbh ratios than Douglas-fir (Fd) in dispersed treatments, but lower crown/dbh ratios than Douglas-fir in the other treatments.  If this trend actually exists, it  may mean spruce crowns (and possibly pine crowns) are being released and becoming larger and/or Douglas-fir is developing smaller crown profiles in the newly exposed conditions in dispersed treatments. Crown/dbh ratios of aspen (At) and birch (Ep) appear to be slightly higher in aggregate treatments compared to dispersed and control treatments. Conifer Species Crown DBH Ratio Distributions 25 0 20  A •  1 15 n  T3 | 10 o o 5  •  •  API • Fd  A  ASx  A  Dispersed  Aggregate  Control  Deciduous Species Crown DBH Ratio Distributions 20  I  15H  •  •  •  O  i_  • At • Ep Actwd  I 1C §  5H  Dispersed  Aggregate  Control  Figure 23. Crown/dbh ratios for conifer and deciduous species.  3.4 DISCUSSION There were significant differences in the abundance of many habitat variables between treatments. However, the deciduous variables (StLvDec, StDdDec, DnDec) were not significantly different among any of the treatments. Some of these differences may be real, but 52  due to the inherent insensitivity in the Kruskal-Wailis test, differences were not considered significant. A N O V A did indicate a significant difference in StLvDec, StDdDec, and shrubvol between block drainage classes indicating the presence of water is a factor for abundance of deciduous trees and shrub cover. This is not surprising since trees and understory vegetation in riparian areas can often be different from surrounding forest areas (BCMoF 1995). It has been well documented that deciduous trees are preferred habitat for primary excavators (Harestad and Keisker 1989, Martin and Eadie 1999, Bunnell et al. 1999a, and others), but can be replaced by conifers when required. A study in the southern interior of British Columbia found all of the 243 nests of primary excavators were in aspen, even though aspen represented only 5% of the species (Harestad and Keisker 1989). The same species of primary excavators used Douglas-fir and western hemlock in coastal forests (Bunnell and Allaye-Chan 1984). The inference is that primary excavators can adapt to coniferous species in the absence of adequate suitable deciduous trees. In spite of the literature supporting cavity excavator preference for deciduous trees, there was little evidence of cavity excavation detected in deciduous trees in any of the treatments, including the controls. However, there was also little evidence of cavity excavation in conifer trees. There was evidence of woodpecker feeding on trees indicating their presence in the study area so the lack of cavities may indicate low occurrence of trees with suitable decay patterns in the sampled area. It is unlikely that treatment effect alone influenced the lack of cavities since all treatments including controls contained little evidence of new or old cavity excavation. A study of artificial snags in west Kootenay, B C , found the proximity of adjacent cover was not important for nest selection in 'stubbed' trees by woodpeckers and many nests were even found in clearcuts (Harris 1995) suggesting treatment had little effect on location of cavity nesting selection. The density of standing live conifers in aggregates was similar to that in controls. The high density of lodgepole pine in the controls, coupled with mortality from M P B , contributed to 53  high relative abundance of standing dead conifers (StDdCon) in controls. Basal area left immediately after harvesting (BATree) was not significantly different between control and aggregate. However, the basal area of trees that remained standing at time of sampling (UpBATree) was significantly higher in control than in aggregate because there was a higher occurrence of windthrow in aggregates. Standing dead trees were probably not retained in dispersed treatments because of safety concerns so it can be assumed that downed trees likely originated from standing live trees. The retention level for standing live trees has important implications for wildlife since variations of basal area and shrub vegetation provide different types of habitat. A study in the Cariboo by Davis et al. (1999) found the low basal areas and live crown volumes along with the higher volumes of grass and forbs found in the early serai stages resulted in bird communities that were different from the older serai stages. Davis et al. (1999) found warbling vireo, orangecrowned warbler, MacGillivary's warbler, and Lincoln's sparrow more abundant in early serai stands. Older forest species such as winter wren were more abundant in stands with high basal area. Early forest species such as the alder flycatcher were more abundant in stands with low basal area. Early forest species such as Wilson's warbler, Lincoln's sparrow, and chipping sparrow were positively correlated with forb and grass cover. The response of Lincoln's sparrow to basal area suggested that a lack of cover may also be important (Davis et al. 1999). C W D was highest in the dispersed treatments and lowest in the controls. The sum of pre-harvest C W D plus the additional post-harvest logging waste may account for the higher volumes of C W D in the Dispersed treatments compared to the unharvested controls. The difference in C W D between control and aggregate may be related to site factors associated with the selection of aggregate areas that result in slightly higher volumes of C W D (eg., sites with a history of older mortality). However, new C W D input from windthrow was highest in the aggregates due to a high occurrence of windthrow along aggregate edges and in small aggregate 54  patch sizes. The high proportional occurrence of windthrow in the dispersed treatments will provide long-term large C W D from downed trees. Down wood is important in a variety of ways to many forest species. In the Pacific Northwest, 47 vertebrate species respond positively to down wood (Boyland and Bunnell 2002). Down wood modifies and stabilizes the microclimate and provides shelter from temperature and moisture extremes. C W D provides habitat for small mammals and affects their populations, which in turn affects their predator populations (Stevens 1997). The canopy cover variable (CovA) is simply the result of the abundance of standing live trees. Standing live trees were abundant in control and aggregate treatments but, low stem densities in the dispersed treatments resulted in a significantly lower CovA. Overhead canopy cover was similar for control and aggregate treatments (Figure 12), but light from aggregate edges increased available light in aggregate treatments. Aggregates likely have the highest shrub volumes because of a combination of undisturbed ground and moderate light conditions. Grass response is higher than shrub response in dispersed treatments, but is not due to competition between the two. In fact, grass cover did not show any direct association to shrub cover (Figure 16). The increase in grass cover may simply be a response to the increased light and disturbed ground in dispersed treatments. If the exposed dispersed treatments are experiencing drier conditions, grass may be growing better than shrubs due to a difference in soil moisture requirements. Dispersed areas have higher light conditions than aggregates, but also contain highly disturbed ground. It may be the increase in ground disturbance that reduced the pre-existing understory vegetation coupled with the highly exposed conditions of dispersed treatments is a combination not conducive to shrub growth. On the other hand, the moderate light and undisturbed ground in aggregates may not only allow shrub growth to survive, but perhaps even to thrive. Other factors influencing the apparent higher volume of shrub cover in aggregates may be related to certain characteristics associated 55  with aggregate selection that naturally contain higher shrub volumes (eg., non-productive brush, wet soil conditions). Before treatment, the harvested areas (dispersed treatments) contained the highest abundance of mature pine trees, but the lowest abundance of spruce, subalpine-fir, aspen, cottonwood, and immature pine. However, the newly exposed conditions of the dispersed treatments may allow some deciduous species (eg., aspen) to establish where they were not growing in the pre-harvest stand. Many birds and vertebrates prefer hardwoods, but most species can adapt to using conifers or reduced levels of hardwoods to meet their needs. Consequently, compositions of vertebrate species do not vary much between deciduous and coniferous stands, although there is a tendency for bird species richness to increase as proportions of aspen increase in inland forests (Bunnell et al. 1999a). Aspen can provide valuable habitat at all stages of growth and could become important for browsing within a few years of harvesting, and foraging and nesting in the years following (Burns and Honkala 1990b). Regardless of treatment, mature pine trees are highly susceptible to M P B attack. There has been a strong selection bias to either preemptively harvest susceptible mature pine trees or salvage harvest dead or dying M P B attacked trees. As expected, there were very few mature pine trees retained in dispersed treatments. Pre-harvest proportions of pine were over 80% in dispersed treatments, 50% in aggregate treatments, and 75% in controls. This meant aggregates were selected in part for beetle resistance since aggregates contained fewer pine trees than the surrounding pre-harvest stand. The implications for the interpretation of the results of this study are variables related to factors such as species abundance (eg., pine vs non-pine), windthrow (eg., standing dead pine), and tree size (mature vs immature pine) will be more affected by harvesting selection bias than by treatment effect. The main factors influencing the abundance of the tree habitat variables are windthrow resulting from increased wind exposure, low retained tree densities in the dispersed treatments, 56  and mortality from mountain pine beetle. Beetle-killed trees lose their foliage and are therefore expected to be initially less prone to windthrow, this will result in more pine in the controls that are dead (StDdCon), but remained standing. The graph of StDdCon (Figure 5) and the species graph of total and down trees (Figure 32) support this explanation. The factors influencing shrub volumes are less clearly apparent. Varying light conditions likely play a role in shrub volume growth, but ground disturbance could also be a factor. There was a greater abundance of fireweed, a shade-intolerant pioneer species, in the dispersed treatments and more bunchberry, a more shade-tolerant species, in the aggregate and control treatments. Dispersed treatments contained the highest abundance of grass cover and moderate shrub volumes. Control areas contained the lowest shrub volumes, but also the lowest grass cover. It is likely that the lower light conditions in controls do not allow either grass or shrubs to thrive. Aggregate treatments contained the highest shrub volumes, but grass cover was almost as low as controls. This suggests the moderate light conditions and the lack of ground disturbance in aggregates are adequate for shrub growth, but not for grass growth. Grass cover increased rapidly in the dispersed treatments for the first two years following harvest (Figure 17). The presence of grass and shrubs are important considerations for habitat management. Grass is a valuable source of food and cover for many species (Mackinnon et al. 1999). Shrub cover provides many benefits to wildlife including providing nesting and rearing habitat for songbirds and grouse, providing cover for mammals, maintaining diversity of arthropod communities, and providing a moist stable microclimate that can promote sensitive plant species (Bunnell et al. 2003).  3.5  CONCLUSION The hypothesis tested in Study 1 asked the question "does aggregate treatment result in a  different abundance of wildlife structural/habitat features than dispersed treatment and how do they both differ from unharvested forest". Results indicate aggregate treatments, dispersed 57  treatments, and controls did differ in the abundance of the 11 sampled habitat variables. Abundance of most variables in dispersed treatments was different from aggregate and control treatments. There were also some differences between aggregates and controls, likely due to differences in exposure (eg., aggregate edge influence) and aggregate retention selection bias (eg., species, size). Differences between the levels of dispersed treatments were often less apparent. The abundance of standing dead conifers resulted from the higher likelihood of mortality for lodgepole pine compared to other species. The abundance of down trees resulted from the higher likelihood of a tree being windthrown in the dispersed compared to aggregate treatment. These results were expected since lodgepole pine is highly susceptible to M P B attack and it is logical to assume increased exposure in dispersed treatments would result in greater windthrow. CWD input from windthrown trees was higher in aggregate compared to dispersed treatments simply because the greater number of trees in aggregate treatments provided more trees to be windthrown. The initial expectation was that shrub volumes would be highest in dispersed treatments due to higher light conditions. However, shrub volumes were highest in aggregate treatments, moderate in dispersed treatments and lowest in control areas. Shrub volumes increased for two years in dispersed and for three years in aggregate treatments. Ultimately, the results indicate aggregate treatments provide more shrub volumes than dispersed treatments. It is likely the different conditions created by the control, aggregate, and dispersed treatments resulted in variations in the abundance of many habitat variables. However, it is important to remember that any bias that was part of the process for selecting which individual trees or groups of trees to retain in dispersed or aggregate treatments will strongly influence what remains intact after any treatment. For example, regardless of the treatment, any individual tree or group of trees that are selected for their qualities of windfirmness (eg., species, size) will have a lower chance of experiencing windthrow than those that are randomly chosen. 58  Also, aggregates that originally contained higher or lower abundances of other variables such as C W D or even shrub cover would likely retain those levels after an aggregate treatment has been done. Therefore, variable abundances can depend on the aggregate site selection (or individual tree selection) process as much or more as the type of treatment. Additional studies are required to determine the threshold amounts of the different habitat variables that are needed to sustain the desired biodiversity. However, the available literature does provide sufficient insight to allow a general estimation of whether abundance of many of the sampled habitat variables are adequate or not. According to densities recommended by Boyland and Bunnell (2002), neither the dispersed or aggregate treatments are meeting the minimum requirement for two large snags/ha. The minimum requirements for 1020 smaller snags are being met in the aggregate treatments due to M P B mortality of pine trees, but not in dispersed treatments due to windthrow. Boyland and Bunnell (2002) consider a large snag as 50 cm dbh (30 cm dbh in less productive forest) and preferred species include hardwoods, Ponderosae pine, Douglas-fir, and larch. There are sufficient large trees available to meet minimum large snag requirements in the dispersed treatments, but they are live trees and therefore do not qualify as snags. However since these large trees are old Douglas-fir veterans they provide many valuable habitat features such as deeply furrowed bark, large branches, etc. Also, i f standing dead trees are desired, management can easily employ various methods including girdling, herbicide, and fungal inoculation to create large snags. Artificially inoculating live trees with decay fungi was found to be more effective at creating suitable wildlife trees than either herbicide or girdling (Bunnell et al. 1999d). The mean densities of deciduous species in aggregates is half of that in controls and considerably lower in dispersed treatments. It is unlikely that deciduous presence will provide significant wildlife habitat in the dispersed treatments. Any requirements for canopy cover in dispersed treatments will likely not be met due to very low stem densities and high windthrow 59  occurrence. Total long-term requirements for C W D will probably be met because of the windthrow events in both dispersed and aggregate treatments. Shrub management strategies must be linked with the specific type of partial harvesting treatment. There are more early serai understory plant species (eg., fireweed, western hardhack, sweet-scented bedstraw) found in aggregate treatments, more late serai shrub species (eg., bunchberry, false Solomon's seal, queen's cup) found in dispersed treatments and more overall shrub volumes found in aggregates than in dispersed or controls.  60  4 STUDY 2: PROBABILITY OF INDIVIDUAL TREE MORTALITY AS A FUNCTION OF TREE AND NEIGHBOURHOOD VARIABLES 4.1  INTRODUCTION The success of a partial harvesting treatment depends on whether or not the desired stand  features are maintained or produced over the longer term. High mortality of retained trees may be an undesirable outcome i f an abundance of live overstory trees in some configuration (eg., aggregate, dispersed, corridor of live trees) is a longer term stand objective. However, the value to cavity-nesters is often much higher i f a standing tree is dead or dying rather than live and healthy. If a component of standing dead (or dying) trees is desired in the target stand, mortality of some trees would be considered a successful outcome and could be incorporated into the long-term strategy. Planning the long-term success of a stand that has a particular combination of stand characteristics including live trees, standing dead or dying trees, and C W D input from downed trees requires an understanding of the stand dynamics that influence these factors and the way post-harvest exposure affects windthrow and standing mortality rates. In this study, two main questions were investigated: 1. Do the different partial harvesting methods of aggregate treatment and dispersed treatment affect probabilities of tree mortality or windthrow? 2. How do tree, neighbourhood, and site-level factors affect the probabilities of tree mortality or windthrow? To address these two questions, five logistic regression models were created to estimate the probabilities of a tree being: 1. dead from the total population of trees (includes live, dead, standing, and down trees at time of sampling), 2. dead from the population of standing trees (includes only trees standing at time of sampling),  61  3. down from the population of trees in aggregate and dispersed treatments (includes standing and down trees at time of sampling), 4. down from the population of trees in aggregate and dispersed treatments (population is the same as model 3, but model 4 includes exposure variable Direx 30). 5. down from the population of trees in the aggregate treatment (model 5 includes exposure variable VRfetch).  The population of all sampled trees is referred to as the "total population" and includes both standing and down trees. The "population of standing trees" includes only the trees that were found to be standing at time of sampling. The "population of trees in aggregate and dispersed treatments" includes the trees from the aggregate and dispersed treatments, but excludes the trees from controls. The "population of trees in the aggregate treatment" includes trees found only in the aggregate treatments. Model 1 estimated the probability of mortality from the total population of trees. Presumably, live trees left after harvesting were intended to remain live and few i f any dead trees were left in dispersed treatments. Model 1 therefore provides a comparison of mortality status in the dispersed treatments to that in the aggregates and controls. Model-2 estimates probabilities of an individual tree being dead from the population of standing trees. Unlike model 1, the intention of model 2 was not only to identify predictors of mortality, but also to provide some insight into the supply of standing dead snags. Model 2 indicates the likelihood of a tree dying and then remaining standing in the years following harvesting. In other words, model 2 estimated the probability of non-windthrow mortality in the stand for the residual trees. For the most part, model 2 provides the probability of M P B attack in the pine trees left after harvesting. There may have been some dead residual trees left after  62  harvesting in the aggregate treatments, but it is unlikely there were many dead trees left after harvesting in the dispersed treatments due to safety concerns. Models 3 and 4 estimated the probabilities of an individual tree being windthrown using only the trees in dispersed and aggregate treatments. This was done because windthrown trees in the controls were not recorded as down trees. Exposure variables were included in models 3 and 4 to estimate the treatment effects of increasing exposure associated with the different treatments. Model 4 estimated the probability of windthrow resulting from exposure by including the direx30 variable in the model. Model 5 used only the population of trees in the aggregate treatments. Model 5 included the VRfetch exposure variable and estimated the probability of windthrow in the aggregate treatments. The intention of model 5 was to provide insight into windthrow susceptibility according to aggregate size and tree proximity to edge. Stepwise regression was used to eliminate non-significant (alpha = 0.05) variables and the five best models rated according to a combination of % concordance, c-value, and HosmerLemeshow goodness of fit p-values were kept. A l l five models retained at least one tree and one neighbourhood variable. 4.2 M E T H O D S 4.2.1  STUDY SITE The study site for Study 2 is the same as that used in Study 1.  4.2.2 S A M P L I N G P L A N The data collected for Study 2 was gathered at the same time and used the same sample plots as Study 1.  4.2.3 TREE L O C A T I O N COORDINATES A N D E X P O S U R E V A R I A B L E S Tree locations in dispersed treatments were measured in U T M coordinates using a handheld GPS receiver (Garmin, Model: eTrex). It was not possible to receive coordinates under the canopy in aggregates and controls so aggregate plot coordinates were located from 1:5000 block 63  maps and control plot coordinates were located from 1:20000 maps, both supplied by Canadian Forest Products Limited (Canfor). Since all the trees sampled in the aggregates were located inside 5.65m radius (l/100ha fixed area) plots, for the purposes of exposure calculations all the trees in a given plot were considered to be at the same U T M coordinates. A SPOT5 satellite, pan-enhanced 2.5m false colour image of the study site (taken in August 2004, supplied by Pacific Geomatics Ltd.), and geo-referenced block boundary polygons (supplied by Canfor) were imported into ArcView GIS 3.2. Tree and plot coordinates and image information were used to digitize the retention polygons and tree locations that were used in VRfetch and Direx30 calculations. The retention level within each polygon was estimated to the nearest 20% crown closure from the SPOT5 image. VRfetch is a fetch variable developed by Scott (2005) to include a measure of the partially obstructed distances encountered in dispersed retention treatments. Specifically, it is the segment length multiplied by the removal level (100 minus percent cover) at points located every 30m for 300m in each of the eight cardinal directions surrounding the plot (Scott 2005). The sum of the unobstructed line segments (>95% removal) are added to the products of the partially obstructed line segments to calculate the VRfetch value. Direx30 is a measurement that sums the number of the eight cardinal directions that have an unobstructed (>95% removal) distance of 30 meters immediately adjacent to the point of interest.  4.3 STATISTICAL A N A L Y S E S 4.3.1  STEPWISE LOGISTIC REGRESSION The goal of logistic regression is to predict a discrete outcome, usually dichotomous (eg.,  live or dead, up or down) from a set of predictor variables. Predictor variables can be any combination of continuous, discrete, or dichotomous variables and do not need to be normally distributed, linearly related, or of equal variance within each group (Tabachnick and Fidell 2001). Logistic regression uses a logit transformation to predict the log odds that an indicator 64  variable is equal to 1. Stepwise logistic regression is used to eliminate non-significant variables and improve the fit of the model and is considered a suitable method to discover relationships between variables (Trexler and Travis 1993). Several techniques may be used to test the coefficients for significance for inclusion or elimination from the model. A pair of outcomes is concordant when the response with the larger value has the higher probability of occurring (Tabachnick and Fidell 2001). The fit of a model increases with an increase in its proportion of concordance (% Concordance). The cvalue measure of concordance is the probability of a correct classification of a random pair of cases from both outcome categories. The c-value can vary from 0.5 (same as chance) to 1.0 (perfect prediction). The Hosmer-Lemeshow goodness of fit (H-L) statistic evaluates goodness of fit by creating 10 groups ordered according to their estimated probability and comparing the number observed in each group to the number predicted by the model. A good model produces a non-significant chi-square statistic indicating the model prediction does not differ significantly from the actual observed values. The Wald test evaluates the contribution of an individual predictor to a model. A significant result indicates a reliable association with outcome. The 95% Wald Confidence Limit provides an upper and lower spread of the point estimate (alpha = 0.05). The odds ratio is the increase or decrease in odds of being in one outcome (eg., dead, down) when the value of the predictor increases by one unit. A point estimate greater than 1.0 indicates an increase in odds and a point estimate less than 1.0 indicates a decrease in odds.  4.3.2 V A R I A B L E S U S E D IN STATISTICAL A N A L Y S E S Seven tree variables and three neighbourhood variables (Table 6) were used in the stepwise logistic regression models to estimate factors contributing to individual tree mortality status. The variables associated with site conditions (site series, soil texture, slope, and soil drainage) did not improve model fit and were therefore not included in any of the final models. Models 1, 2, and 3 had better fit when the three levels of dispersed retention (High, Medium, 65  and Low) were combined into a single treatment. In these cases, the five treatment classes (DispL, DispM, DispH, aggregate, and control) were reduced to the three treatment groups of Control, Aggregate and Dispersed. Models 2 and 4 had better fit when the nine tree species were reduced to five tree species groups. Stepwise logistic regression eliminated other nonsignificant (alpha=0.05) variables and only the most useful and parsimonious models were kept. Spearman's correlation was used to measure the strength of the linear relationship between variables. Spearman's correlation was used because the data was not normally distributed. Correlated variables (>0.6) were not used together in any of the logistic regression models. Table 6. Tree and neighbourhood variables used in logistic regression modeling.  Label  Name  Description Tree Variables  SPEC  TREESPEC  HT  Species  This variable classifies the tree according its species (Douglas-fir = Fd, lodgepole pine = PI, interior spruce = Sx, black spruce = Sb, subalpine-fir =B1, aspen = At, cottonwood = Ac, paper birch = Ep, alder = Dm).  Tree species  This variable converted the nine tree species into five tree species groups (Douglas-fir = Fd, lodgepole pine= PI, interior spruce, black spruce and subalpine-fir = SxBl, aspen = At, and birch, cottonwood and Dm = Deed).  Height (m)  This variable used a hand-held laser ranging instrument (Laser Technology, Inc. Model: Impulse LTI) to measure tree heights in metres.  HDR  Height/dbh ratio  DBH  Diameter at breast height  This variable is the diameter measured at dbh. Diameter units are measured in centimeters.  LIVEORDEAD  Live or dead  This variable classified a tree as live or dead. Although some downed trees (eg., uprooted) were still living they were recorded as dead. Therefore, a tree could be classed as standing and dead, but not down and live.  UPORDOWN  Up or down  This variable classified a tree as standing (up) or down. Leaning trees were considered 'up' trees.  This variable is the ratio of height(m) / dbh(m) and is a measurement of 'slenderness'.  66  Table 6 cont. Label  Name  Description Neighbourhood Variables  TREATMENT TYPE  TREATMENT  This variable classified the five treatment types according level of retention. DispL=dispersed with low retention, DispM=dispersed with medium retention, DispH=dispersed with high retention, Aggregate=reserve patch of trees, and Control^nharvested forest.  TREATMENT GROUP  Treatment group  This variable reduced the five treatment types to the three treatment groups Aggregate, Disp, and Control. A l l three dispersed treatments were classed as one treatment group (Disp).  Variable retention fetch  This variable is the sum of the distances of unobstructed and partially obstructed line segments for a distance up to 300 meters in the eight cardinal directions. The VRfetch variable is a measure of the tree's access to the sum of the obstructed and unobstructed wind (see Chapter 4.2.3).  Direx 30  This variable is a measure of exposure that sums the number of cardinal directions (from a possible total of eight) that have openings at minimum distances of 30 meters. For example, a tree with a direx-30 value of three means that there are a total of three cardinal directions that have openings of at least 30 meters associated to that tree (see Chapter 4.2.3).  VRfetch  DIREX30  4.4 RESULTS Some of the variables used in the logistic regression models were positively correlated and some were negatively correlated to each other (Table 7). Height correlated positively to dbh, but correlated negatively to height diameter ratio (HDR) direx30 and VRfetch. H D R correlated negatively to dbh, direx30, and VRfetch. Dbh correlated positively to direx30 and VRfetch. Direx30 correlated positively to VRfetch. The most highly correlated pairs of variables are ht-dbh (0.851) and direx30-VRfetch (0.673). These correlated pairs of variables are therefore not paired together in any of the logistic regression models. 67  Table 7. Spearman's correlations of variables used in logistic regression models. Variable 1 ht ht ht ht HDR HDR HDR dbh(cm) dbh(cm) direx30  Variable 2 HDR dbh(cm) direx30 VRfetch dbh(cm) direx30 VRfetch direx30 VRfetch VRfetch  Correlation Coefficient -0.01343 0.85116 - 0.24429 -0.12472 -0.45100 -0.00866 -0.08242 0.01226 0.08950 0.6730  level of significance 0.5670 < .0001 < .0001 < .0001 < .0001 0.7140 0.0005 0.6039 0.0001 < .0001  n 1818 1818 2124 2124 1818 1793 1793 1793 1793 2127  The proportions of windthrown trees by dbh, height, HDR, VRfetch, and Direx30 class are given in Figures 23 to 27. Descriptions of variables and treatments are provided in Table 6. The exposure variables VRfetch and Direx30 reflect treatment (dispersed has higher exposure than aggregate treatment) and edge effect of aggregates. As expected, varying both tree and exposure variables produced changes in windthrow occurrence. Figure 24 illustrates the proportions of each dbh class that are windthrown. The larger size classes (>50cm) had a much lower proportion of down trees in the dispersed treatments. There were no large trees (>50cm) in the aggregate treatments. Generally, aggregate experienced lower proportions of windthrow in most size classes compared to dispersed treatments.  68  D i s p M (N=322)  100%  DispL (N=408)  100% -i  I  80% -I i60%  t  40%  80% 60% 40%  20%  20% H  0%  0% 10cm 20cm 30cm 40cm 50cm 60cm 70cm 80cm  10cm 20cm 30cm 40cm 50cm 60cm 70cm 80cm  dbh class  dbh class  Aggregate (N=670)  DispH (N=519)  100%  100%  80%  n  80%  i 60%  | 60%  20%  20%  EL  0%  r=n  0%  11  111  li  10cm 20cm 30cm 40cm 50cm 60cm 70cm 80cm  10cm 20cm 30cm 40cm 50cm 60cm 70cm 80cm  dbh class  dbh class  Figure 24. Proportions of windthrown trees in dbh size classes. N=number of observations.  Figure 25 illustrates the proportions of each tree height class that are windthrown. The shortest (10m) and tallest (50m) height classes have the lowest proportions of windthrow. Similar to the trend for dbh classes, aggregate experienced lower proportions of windthrow in all height classes compared to dispersed treatments. There were no 50m tree height classes in aggregate treatments.  Proportion of W i n d t h r o w n Trees in Height C l a s s e s 100% to  -1  • DispL  80% -  -  I *  60% 40% 20% 0%  x DispM  •  CD  ADispH  •  ft •  T.  —Aggregate  A x  A  10m  20m  30m  40m  50m  Figure 25. Proportions trees in height classes that are windthrown.  69  Figure 26 illustrates the proportions of trees in each height/diameter ratio (HDR) class that are windthrown. The least slender of the H D R classes (HDR50) experienced the lowest proportion of windthrow in all dispersed treatments. The most slender of the H D R classes (HDR200) experienced less windthrow in DispH and aggregate compared to DispL and DispM. Aggregate treatments consistently experienced the lowest overall proportional windthrow occurrence. Proportion of Windthrown Trees in (HDR) Classes x  100%  I 60% -I •g 40% 20%  • DispL  •  80%  x DispM  • A  ADispH  •  —Aggregate  X  0% HDR50  HDR100  HDR150  HDR200  Figure 26. Proportions trees in height/diameter ratio classes that are windthrown. Figure 27 illustrates the proportions of trees in each VRfetch class that are windthrown. Proportions of windthrown trees are lowest in the lowest exposure class VRfetch500. There is a general trend for proportions of windthrown trees to increase with an increase in VRfetch class. Figure 28 illustrates the proportions of windthrown trees in the direx30 classes (1-8). Proportion of Windthrown Trees in VRfetch Classes  100%  • • DispL  80%  1 60%  a  cz  % 40% 20%  0%  A  X  x DispM ADispH  X  X  —Aggregate  X  VRfetch500 VRfetcM 000 VRfetchi 500 VRfetch2000 VRfetch2500  Figure 27. Proportions trees in VRfetch classes that are windthrown.  70  Proportion of Windthrown Trees in DispL  J  0  0  Proportion of Windthrown Trees in DispM  %  J  g 80% ^ 60% I 40% 3 20% H 0% 1  2  3 4 5 6 . Direx30Class  0  0  %  S 80% ~ 60% 1 40% 20% 0%  7  1  Proportion of Windthrown Trees in DispH  2  3 4 5 6 Direx30Class  7  Proportion of Windthrown Trees in Aggregate  100%  100%  CO  CO  g 80% ~ 60% 1 40% -| T3 20% 0% 3 4 5 6 Direx30Class  7  8  g 80% ~ 60% I 40% o T3 20% 0%  3 4 5 6 Direx30Class  7  Figure 28. Proportions of trees in direx30 classes 1-8 that are windthrown. Table 8 gives the five most parsimonious logistic regression models, their reduced set of predictor variables and model effectiveness as indicated with % concordance, c-value, and Hosmer-Lemshow (H-L) chi square and p-value. The odds ratio estimates (Tables 10 to 14) indicate the increase or decrease in odds of being dead (models 1 and 2) and down (models 3, 4 and 5).  71  Table 8. Logistic regression models. Model #  1) Likelihood of being dead 2) Likelihood of being dead 3) Likelihood of being down 4) Likelihood of being down 5) Likelihood of being down  Data Set  Variables  %Concordance  c-value  Total population of trees Population of standing trees  treatment group, spec, ht(m), YSH  69.4  0.695  41.1335 p= <.0001  treatment group, treespec  73.6  0.802  5.4149 p=0.4918  Population of trees from Disp and Aggregate Population of treesfromDisp and Aggregate Population of trees in Aggregate  treatment group, spec  62.8  0.742  13.1511 p= 0.0407  treespec, direx30, ht(m)  72.6  0.728  24.2008 p=0.0021  Dbh(cm), VRfetch  75.0  0.753  20.0456 p=0.0102  H-L chi sq, p-value  For Model 1 the likelihood of being dead from all sources of mortality is over six times higher for lodgepole pine (PI) than for cottonwood (Ac). A tree is almost 3.6 times more likely to be dead in the dispersed (Disp) treatment group compared to the aggregate treatment, and twice more likely if it is in the control treatment than in the aggregate treatment (Table 9). The likelihood of tree mortality increased with increasing tree height (ht). Height units are recorded in meters to the first decimal so the likelihood of mortality is actually higher than the 1.031 value might suggest. There is also a small increase in the likelihood of tree mortality with an increase in years since harvest (YSH). Table 9. Model l:Logistic regression odds ratio estimates for likelihood of being dead from the total population of trees. Model 1 estimates mortality from all sources including windthrow. number of observations =1817 Treatment group Treatment group Spec Spec Spec Spec Spec Spec Spec Spec  Effect  Point Estimate  95% Wald Confidence Limits  Control vs Aggregate Disp vs Aggregate At vs Ac Bl vs Ac Ep vs Ac Fd vs Ac PI vs Ac Sb vs Ac Sx vs Ac Dm vs Ac Ht(m) YSH  2.090 3.573 2.076 0.559 1.135 1.408 6.481 1.965 1.278 1.107 1.031 1.094  1.399 - 3.124 2.611-4.890 0.185-23.258 0.039-7.949 0.103- 12.502 0.133- 14.955 0.614-68.423 0.115-33.617 0.120 - 13.634 0.042-29.331 1.017- 1.044 1.022- 1.172  Table 6 provides descriptions of variables used in all Study 2 models.  72  The difference in mortality between the control and aggregate treatments is associated with the proportion of pine. Figure 29 illustrates the proportion of pine (PI) is higher in the control than the aggregate treatments, coinciding with the relatively higher mortality from M P B in the controls. Pine species mortality has a higher probability in controls than in aggregates (Model 1) and is therefore proportionately higher in control (66%) compared to aggregate treatments (48%) (Figure 30). Figure 31 does not indicate an increasing trend for proportions of down trees with increasing Y S H , but does indicate proportionately more windthrow in the dispersed compared to the aggregate treatment.  Proportion of Species in Treatments  100% -, 80% -  co cd  CL  60% -  fall  o cd  40% -  co  o  20% 0% - J01 DispL  DispM  JJ DispH  Aggregate  Control  Figure 29. Proportions of species in treatments (from the population of all trees sampled).  Proportions Lodgepole Pine Mortality in Treatments  100% •: 80% -I CO  "° 40% -I 20% 0%  rip DispL  DispM  DispH  Aggregate  Control  Figure 30. Percent mortality of lodgepole pine in treatments.  73  Proportions of Down Trees in Y S H Classes 80%  -I  70% -  •  60% -  S  50%  I  40%  20%  -I A  10% 0%  A Aggregate  •  H  l_  30%  • Dispersed  YSH1  A A  A  YSH2  YSH3  A YSH4  YSH5  YSH6  YSH7  Figure 31. Proportion of down trees in YSH classes.  Model 2 indicates mortality for standing trees (non-windthrow mortality), PI has a 10 times greater likelihood of being dead than species group SxBl, and aspen (At) has a likelihood of being dead nearly five times greater than SxBl (Table 10). The probability of a tree being dead is more than twice as high in the control than in the aggregate treatment, coinciding with the higher densities of lodgepole pine in controls than aggregates (Figure 2). The abundance of standing mortality is associated with abundance of pine in aggregate and control treatments. However, even though there are few standing pine in the dispersed treatments, an individual standing tree is 1.5 times more likely to be dead in the dispersed compared to the aggregate treatment group. Table 10. Model 2:Logistic regression odds ratio estimates for likelihood of being dead from population of standing trees. This model estimates non-windthrow mortality in all treatments. number of observations =1817 Treatment group Treatment group Treespec Treespec Treespec Treespec  Effect  Point Estimate  95% Wald Confidence Limits  Control vs Aggregate Disp vs Aggregate At vs SxBl Deed vs SxBl Fd vs SxBl PI vs SxBl  2.176 1.44 4.761 0.807 0.733 10.351  1.591 -2.977 0.865 -2.399 2.026 - 2.026 0.285-2.282 0.384- 1.400 6.638- 16.141  Table 6 provides descriptions of variables used in all Study 2 models.  74  Model 3 estimates the likelihood of being windthrown from the population of trees in the aggregate and dispersed treatments, according to the variables of treatment group and species. Down trees were not recorded in controls, but occurrence of windthrow was observed to be very low. The probability of being down is eight times greater in the dispersed compared to the aggregate treatments (Table 11). However, during sampling, it was observed that trees in smaller sized aggregates and on the edges of aggregates were highly susceptible to windthrow damage. Model 3 indicated that all species (At, B l , Ep, Fd, PI, Sb, Sx, and Dm) had lower probabilities of being down than cottonwood (Ac), but the likelihood of Spruce (Sx) being down is the highest (0.678) compared to cottonwood. Actual occurrence of cottonwood was limited with only two live standing trees in controls, one live standing tree in a dispersed treatment, and one dead windthrown tree in an aggregate treatment. The occurrence of alder and black spruce are also minor. The type of mortality was commonly related to species. For example, pine mortality was most often from M P B and spruce mortality was most often from windthrow. The very low probability (<0.001) of black spruce being down is not considered reliable since only nine standing black spruce trees were recorded from one plot in the bottom of a small gulley in a relatively protected area inside a large aggregate in Block 544-4. Black spruce often has shallow rooting and can be susceptible to windthrow (Burns and Honkala 1990a). Similarly, most of the subalpine fir were located in a gulley inside a single aggregate in Block 592-1.  75  Table 11. Model 3:Logistic regression odds ratio estimates for likelihood of being down from population of trees in aggregate and dispersed. This model estimates the probability of windthrow in aggregate and dispersed treatments, but not in controls. number of observations =1481 Treatment group Spec Spec Spec Spec Spec Spec Spec Spec  Effect  Point Estimate  95% Wald Confidence Limits  Disp vs Aggregate AtvsAc BlvsAc EpvsAc FdvsAc PlvsAc SbvsAc SxvsAc DmvsAc  8.094 0.379 0.098 0.285 0.408 0.406 O.001 0.678 0.225  5.844- 11.210 0.015 -9.439 0.003 -3.231 0.012-6.973 0.017-9.676 0.017-9.691 <0.001 - >999.999 0.028- 16.180 0.004- 11.731  Table 6 provides descriptions of variables used in all Study 2 models.  Compared to the aggregate treatments, the dispersed treatments experienced a much higher proportion of windthrow mortality for most species, but due to the low stem densities, a lower overall total of dead trees (Figure 32). In Figure 32, the 'total trees' class includes all trees both live and dead. The 'dead trees' class includes all down trees, but dead trees may also be dead and still standing (eg., beetle-killed). There were no dead subalpine fir found in the DispM or aggregate treatments (Figure 33). There were no dead cottonwood or alder found in the dispersed treatments and all cottonwood and alder in the aggregate treatments were windthrown. The one cottonwood tree found in the dispersed treatment was live and all dead birch in the aggregate treatments remained standing. Proportional windthrow to dead trees varied between treatments, but was generally greater for most species in the dispersed compared to the aggregate treatments (Figure 33). Proportions of standing dead trees were generally low for most treatments, and the highest proportions of standing dead trees were pine (PI), and aspen (At). Proportions of down trees by species are not shown for controls, but abundance of standing dead pine was much higher in controls than other treatments.  76  Aggregate Treatment Dead and Down Trees  Dispersed Treatment Dead and Down Trees  30 25  600 500 400  El total trees • dead trees  20 H  O down trees  10 •  •total trees H dead trees • down trees  I'.  300  200 100 0  5 0 PI  Fd  S»  Sb . Bl alder  Iff I L I f e , PI  At ctwd Ep  I  Fd S K Sb Bl alder At ctwd Ep  Figure 32. Windthrow and non-windthrow mortality of trees in dispersed and aggregate treatments. % down PI  100%  % down Fd 100%  -1  « 80%  «  80% -  £= 6 0 % 4.  4=  60% -  1  I  40% 20%  40% -  3*  20% -  0%  0 % -• DispL  DispM  DispH  Aggregate  DispL  % down Bl 100%  n  5  20% 0%  DispH  Aggregate  % down Sx  S3  8 0 % 4,  |  <r.  •*  DispM  DispH  40% -  5  20% 0%  Aggregate  DispL  % down alder  DispM % down At  1 0 0 % -,  100%  <3 8 0 % -  « 80%  4=  60% -  4= 6 0 %  |  40% -  |  5  20% 0 % -• DispL  DispM  DispH  i  40% 20% 0%  Aggregate  miDispL  DispM  DispH  Aggregate  % down ctwd  % down Ep  100%  100% <3 8 0 %  <> j 80%  •fc 6 0 %  4= 6 0 %  | 40%  |  5  Aggregate  4= 6 0 %  DispL  5  DispH  100%  S i  -  -  DispM  5  20%  40% 20%  0%  0% DispL  DispM  DispH  DispL  Aggregate  DispM  DispH  Aggregate  Figure 33. Proportions of species of windthrown trees from the population of dead trees in aggregate and dispersed treatments. These graphs indicate the proportions of trees of each species that died, but did not produce a standing snag.  77  Model 4 (Table 12) estimates the likelihood of being down from the population of trees in the aggregate and dispersed treatments, but unlike Model 3, includes the neighbourhood (exposure) variable direx 30. Model 4 estimates the probability of windthrow and includes the exposure variable direx 30 to infer a treatment effect resulting from increased exposure in different treatments. This model indicated the likelihood of being down was 1.471 times higher for Douglas-fir compared to the combined spruce/fir species class (SxBl). This may be in part because the spruce/fir class includes black spruce, and as mentioned previously, the data for this species is not considered dependable. The likelihood of a tree being down increased (1.221) with increases in the exposure variable direx30. There was also an increase (1.025) in the likelihood of being down with an increase in height. These results suggest a relationship between exposure and height, and risk of windthrow. Table 13 illustrates the proportions of wind damaged deciduous trees in the aggregate and dispersed treatment groups. The dispersed treatment group contained 129 total deciduous trees and the aggregate treatment group contained 27 total deciduous trees. A l l down trees were wind damaged (WD Down Trees), but some wind damaged trees remained standing (WD UP Trees). Even though the dispersed treatments experienced more exposure, there appeared to be proportionately less wind damaged deciduous trees in the dispersed treatments (49% in dispersed compared to 66% in aggregate). In the dispersed treatment 37% of the deciduous trees were down. In the aggregate treatment 33% of the deciduous trees were down. A significantly higher proportion of wind damaged deciduous trees remained standing in the aggregate treatments (33%) than in the dispersed treatments (12%). This means there are proportionately more wind damaged standing deciduous trees in aggregate treatments with decay wound entry points (eg., broken tops and branches); these trees can become a future source of valuable standing wildlife trees.  78  Table 12. Model 4:Logistic regression odds ratio estimates for likelihood of being down from total population of trees excluding controls. The direx 30 variable characterizes the level of exposure in  different treatments. number of observations = 1456 Effect  treespec treespec treespec treespec  Point Estimate  95%Wald Confidence Limits  1.006 0.730 1.471 0.573 1.221 1.025  0.524- 1.930 0.432- 1.235 1.071-2.019 0.391 -0.840 1.177- 1.266 1.010-1.040  At vs SxBl Deed vs SxBl Fd vs SxBl PI vs SxBl direx30 Ht(m)  Table 6 provides descriptions of variables used in all Study 2 models.  Table 13.  Deciduous trees in treatments (WD = Wind Damaged).  Treatment  WD Up Trees 15 9  Dispersed Aggregate  WD Down Trees 48 9  Total trees 129 27  Total WD 63 18  %WD Total 49% 66%  % WD Up Trees % 33% 1 2  % WD Down Trees 37% 33%  Model 5 estimates the likelihood of being down from the population of trees in the aggregate treatments according to the variables dbhem and VRfetch. The probability of being down increases 1.037 times with each unit increase (cm) in diameter and increases 1.003 times with each meter increase in VRfetch (Table 14). Since the VRfetch variable increases as level of exposure increases (eg., proximity to edge.), model 5 suggests the probability of being windthrown increases with an increase in proximity to the aggregate edge. Model 5 indicates the probability of being windthrown increases slightly with increasing dbh. The graphs of down trees by size class in the dispersed treatments indicate the probability of being windthrown increases with increasing dbh up to an approximate threshold dbh of 40-50cm at which point windthrow risk decreases with increasing dbh (Figure 24). There were no trees over this threshold dbh found in the aggregate treatments, but presumably this same trend exists in these relatively lower exposure treatments. Figure 24 indicates that, the higher level of exposure in dispersed treatments results in higher proportions of windthrow compared to aggregate treatments, regardless of dbh.  79  Table 14. Model 5:Logistic regression odds ratio estimates for likelihood of being down from population of trees in aggregate. The VRfetch characterizes the level of exposure. Number of observations = 670 Effect  Dbhcm VRfetch  Point Estimate  95%Wald Confidence Limits  1.037 1.003  1.016-1.059 1.002- 1.003  Table 6 provides descriptions of variables used in all Study 2 models.  4.5 DISCUSSION Models 1 and 2 indicated lodgepole pine (PI) had a higher likelihood of being dead than other species. This is not surprising since the study site is located in a M P B infested area. The severe M P B losses also affected other results. One of the factors that made other species (eg., spruce) more prone to windthrow than pine is the lower sail area of dead pine (less foliage). Mortality from all sources (Model 1) is over twice as likely compared to mortality using only standing trees (Model 2) in the dispersed treatment group. This is because mortality from windthrow is included in Model 1. Figure 31 indicates windthrow did not follow a consistent increasing trend with increases in Y S H class. In other words, the occurrence of windthrow did not increase dramatically after the first year. These results agree with a study in lodgepole pine forests in west central British Columbia by Waterhouse and Armleder (2004) that found windthrow rates were highest in the three months following harvest. The Waterhouse and Armleder study found no significant differences in windthrow of live or dead trees over 12.4 cm dbh between treatments over a 5.3 year period which also agree with the results of this study. Model 1 indicated that the likelihood of mortality was higher in the control compared to the aggregate treatment. This is due to the fact that there was more pine at risk to beetle attack in the controls compared to the aggregates (Figure 2). As was noted in Chapter 3, it appeared that aggregates had been retained in portions of the stand with less lodgepole pine. Figure 30 indicates the proportion of pine mortality from all sources (including windthrow) was lowest in the aggregates. The higher occurrence of pine in controls resulted in more standing dead pine in  80  controls compared to aggregate treatments. It is also possible that pine trees in aggregate conditions were less susceptible to beetle attack due to increased light and wind (traits of beetleproofing) compared to the relatively dense unharvested control areas (Pacific Forestry Centre 2002). Both models 1 and 2 indicated a higher likelihood of a tree being dead in the dispersed treatment group compared to the aggregate treatment. Most of the mortality in the dispersed treatment group was from windthrow, but the newly exposed post-harvest conditions in the dispersed treatments also increased sun exposure and drying winds increasing potential risk of mortality from drought stress (Black and Oke 2000). Harvesting damage to stems was low, with less than 5% of retained trees in dispersed treatments having logging scars. Most of the scarred trees were Douglas-fir veterans. Even though 15% of the trees scarred during harvesting died, it is unclear whether the harvesting scars contributed significantly to the mortality of these trees. It is possible that all trees in the dispersed treatments were experiencing some level of increased exposure stress increasing the risk of mortality from various sources. However, there was little evidence of Douglas-fir bark beetle in any of the sampled trees indicating stress was not likely a major factor affecting Douglas-fir mortality (Leslie and Bradley 2001). Figure 30 indicates pine mortality was proportionately lowest in aggregate treatments and proportionately highest in dispersed treatments. Figure 32 illustrates most pine mortality in dispersed treatments is from windthrow., It is possible that standing dead pine trees in dispersed treatments were either previously infected, but undetected at the time of harvest (green attack) or attacked very soon after harvesting. In contrast, the fact that a group of trees were selected for aggregate retention in a beetle infested area suggests pine trees in the aggregate were not infected at time of harvest. There may be certain characteristics of those selected trees that make them less susceptible to infection (eg., immature trees, increased vigour). In fact, Figure 1 indicates mean dbh's are smaller in the aggregate areas than in the pre-harvest dispersed 81  treatments suggesting trees are likely younger and possibly more vigorous in the aggregates. Model 3 indicates windthrow risk was much higher in dispersed treatments than in aggregate treatments and higher for spruce than for other species. These results are further supported by Model 4 that indicates trees are more susceptible to windthrow with an increase in the exposure variable direx30. A study in mature lodgepole pine forests by Whitehead and Brown (1997) found a similar trend where commercial thinning to a 50-65% basal area reduction aggravated windthrow. Model 4 indicated the likelihood of being down was highest for Douglas-fir compared to all other species. This seems unusual since spruce has been reported as being less windfirm than Douglas-fir (Burns and Honkala 1990a). Burton (2001) also found spruce to be less windfirm at edges than Douglas-fir or pine. Since most of the trees in the dispersed treatments were Douglas-fir, this apparent trend is likely less related to species than to the fact that all species in the dispersed treatments have a greater risk of windthrow. In fact, Figure 32 indicates Douglas-fir suffered the least proportional windthrow of all species in both the dispersed and aggregate treatments. It was expected that deciduous species would be susceptible to wind damage resulting from increased exposure (Burns and Honkala 1990b). The study results indicate that deciduous species in dispersed treatments experienced more proportional windthrow compared to the aggregate treatments. Aggregates reduced windthrow occurrence, but wind damage still occurred to branches and tops. Wind damage in aggregates likely occurred through contact with other trees during high winds. The implication is that deciduous trees in aggregates will provide damaged live standing trees with natural wound entry points (eg., broken branches, broken tops). This situation will result in decay patterns desirable for primary cavity nesters (Boyland and Bunnell 2002). Model 5 indicated windthrow risk is greater with increases in dbh and exposure. Delong et al. (2001) also found windthrow increased in reserves as the median dbh of stems increased. 82  However, my study indicated windthrow risk increased with size up to 40-50 cm, but then decreased with further increases in size (Figure 24). Risk of windthrow was lowest for very small (<20 cm) and very large (>50 cm) trees. These results agree with the Waterhouse and Armleder (2004) study in lodgepole pine forests in west-central British Columbia that indicated smaller trees were more susceptible and larger trees less susceptible to windthrow in partial harvested cutblocks. Jull (2001), found wind damage rates of dispersed Douglas-fir leave trees in the Sub-Boreal Spruce (SBS) biogeoclimatic zone were greatest for dbh classes less than 45 cm dbh and lowest in dbh classes greater than 65 cm dbh corresponding with the larger dominant or 'emergent' trees in the pre-harvest stand canopy. The graphs of down trees in tree height classes (Figure 25) indicated windthrow risk was lowest for tree heights taller than 40m, agreeing with lull's results. The graphs of down trees in H D R classes (Figure 26) indicated windthrow risk was lowest for the least slender class. These results support the assertion that large, dominant trees have adapted more wind-resistance (eg., lower HDR's, stronger root systems) than more slender trees that have never been exposed to wind prior to harvesting. Figure 26 indicated the most slender tree class experienced less windthrow in the aggregate and DispH treatments compared to DispM and DispL, and the aggregates experienced less overall windthrow suggesting increasing exposure increases windthrow risk. Burton (2001) found slender trees with HDR's greater than 69 were more likely to be windthrown in the first few years following harvesting, while stout trees with lower HDR's were more likely to remain standing. Jull (2001) also found increasing rates of windthrow with increases in HDR. Jull found trees with HDR's less than 50 had windthrow rates of 3.9%, HDR's of 50-70 had windthrow rates of 10.7-11.4%, HDR's of 70-90 had windthrow rates of 15%, and HDR's greater than 90 had windthrow rates higher than 20%. Figure 26 indicated an increasing trend of windthrow with increasing slenderness in most  83  dispersed treatments, but windthrow in the aggregate treatment remained consistently low. This result suggests windthrow is not related to slenderness in aggregate treatments. Other factors important to keep in mind include the influence of species and level of exposure. Most of the very large, tall trees were old Douglas-fir veterans located in dispersed treatments. Dispersed treatments produce a higher exposure to wind, but due to stand history and species characteristics, large Douglas-fir veterans should be more windfirm relative to other trees. Factors related to windfirmness include tree size and rooting pattern related to species, soil conditions, and exposure history (Kimmins 1996, Smith et al. 1997). The post-fire disturbance history and dominant positions of Douglas-fir veterans would have allowed these large old trees to adapt to exposed conditions by developing more taper and stronger root systems (Smith et al. 1997). Model 5 indicated an increase in exposure, as measured by the VRfetch variable, increased the occurrence of windthrow. This suggests increases in proximity to edge and/or decreases in the size of aggregates increased the risk of windthrow. These results agree with DeLong et al. (2001) who found windthrow increased dramatically as reserve size decreased below one hectare. In general, the results also agree with the trend found in a study of retention patches in the SBSmc by Burton (2001) that found mean windthrow was 13% in the centre of small patches (<33m across), 6% in the centre of medium patches (34-60m across), and 5% in the centre of large patches (>60m across). Burton also found windthrow was 10% at stand edges and 3% in stand interior conditions. The mortality results provide arguments for the types of trees to retain for wildlife habitat. The large diameter, deeply furrowed bark, large branches, and age-related conditions of Douglas-fir provide important structural elements for wildlife. The relative windfirmness of Douglas-fir veterans provide the best opportunity for large live trees to remain standing in the regenerating stand. If a Douglas-fir veteran dies but remains standing it will provide a large 84  dead snag valuable to several wildlife species. Retaining most other species in dispersed treatments will result in a much higher occurrence of windthrow. However, retaining live pine will likely result in MPB-killed trees that initially have a relatively low risk of windthrow due to reduced foliage. It is important to note that beetle-killed pine may be relatively more wind resistant initially, but become less wind resistant over the long term as root systems decay. Lodgepole pine typically have small root systems that break during storms (Stathers et al. 1994). A study of lodgepole pine stands in west central British Columbia by Waterhouse and Armleder (2004) found 53% of windthrow was due to root breakage of pine attacked by M P B in the early 1980's. Aggregates will provide better protection than dispersed treatments for retaining other less wind-resistant species (eg., spruce, subalpine-fir). It is recommended to retain both coniferous and deciduous snags for wildlife habitat (Boyland and Bunnell 2002). Boyland and Bunnell (2002) further recommend including a component of smaller snags for foraging sites. Retaining different sizes of snags for perching, foraging, and hawking sites tends to increase richness and abundance of non-cavity-nesting bird species (Dickson et al. 1983). Hardwood snags are preferred by some species and are created quickly, but they cannot completely replace conifer snags. Pojar (1995) found winter wren were only present in mature mixed aspen-conifer stands and older conifer stands, concluding there was some type of dependence on the conifer component.  4.6 C O N C L U S I O N This study indicated probabilities of mortality and windthrow are affected by species and type of treatment. The first question asked in this study was "do the different partial harvesting methods of aggregate treatment and dispersed treatment affect probabilities of tree mortality or windthrow". The five logistic regression models indicated different partial harvesting methods have different results. The most influential treatment effect for placing a tree at risk of mortality 85  was windthrow from increased exposure. The second question asked in this study was "how do tree, neighbourhood, and site factors affect the probabilities of tree mortality or windthrow". Generally, the most influential factor causing mortality was species. Specifically, lodgepole pine was indicated to be at the greatest risk for mortality. It was expected that the current M P B epidemic would produce this result. Model 5 indicated windthrow risk increased with increases in dbh or exposure. However, very large, tall trees had a lower risk of windthrow. Windthrow susceptibility was also influenced by species. Spruce was the most susceptible and Douglas-fir was the least susceptible. M P B attacked pine also had a lower risk of windthrow. Selecting large veteran Douglas-fir leave trees would likely provide the best opportunity for retaining standing trees in dispersed treatments and aggregate treatments are better able to provide protection for less windfirm trees.  86  5 CONCLUSIONS AND RECOMMENDATIONS 5.1 S U M M A R Y OF RESULTS OF S T U D Y 1 A N D S T U D Y 2 The overall goal of this research was to characterize the effect of retention level and pattern on stand and tree-level attributes that contribute to wildlife habitat. The positive, neutral and negative effects of the harvesting treatments on 11 variables are summarized in Table 15. Generally, there is a much lower abundance of most variables in the dispersed treatments than in the controls. However, shrub volumes and C W D were more abundant in dispersed and aggregate treatments than in controls. Down trees (DnCon, DnDec) were not recorded in controls so direct comparisons cannot be made. Nevertheless, windthrow in control areas was rare and the abundance of down trees was observed to be higher in both dispersed and aggregate treatments than controls. Table 15. Abundance of wildlife habitat variables in retention treatments compared to unharvested control areas . 1  Variable StLvCon StLvDec StDdCon StDdDec DnCon DnDec BATree UpBATree CovA ShrubVol CWD  DispL  DispM —  na na  DispH —  na na  Aggregate —  —  + +  na na  na na -  + + + —  + +  +  + + + +  + +  + +  + +  The number of plus (+) or minus (-) signs indicates the approximate proportional difference between the treatments and the unharvested control areas, e.g. ++ means twice as high. The sign indicates approximately equal proportions.  Different levels of post-harvest wind and light exposure produced different impacts on stand structure (Table 16). Dispersed treatments experienced the highest exposure to wind and light and produced relatively high windthrow, moderate shrub volumes, and high grass cover. Aggregate treatments experienced moderate exposure to wind and light and produced relatively 87  moderate windthrow, high shrub volumes, low grass cover, and a lower risk of M P B attack. Control treatments experienced low exposure to wind and light and produced low windthrow, low shrub volumes, low grass cover, and a higher risk of M P B attack. Some impacts were influenced as much or more by the leave tree criteria used in the retention selection process. For example, risk of windthrow and M P B attack were affected by tree size and species. Table 16. Light and wind exposure and impacts on stand structure by treatment. Treatment Type  Main Effects  Impacts on Stand Structure  Dispersed  high exposure to wind and light  • • •  high windthrow moderate shrub volumes high grass cover  Aggregate  moderate exposure to wind and light  • • • •  moderate windthrow high shrub volumes low grass cover moderate M P B attack  low exposure to wind and light  • • • •  low windthrow low shrub volumes low grass cover high M P B attack  Control  The results of the Studyl and Study 2 indicate similar trends. The dominant tree species is lodgepole pine and the region is heavily infested by M P B . This situation results in high standing tree mortality rates for pine wherever pine are found. Even though the controls contained more overall conifer trees, both the control and aggregate treatments had a similar abundance of standing live conifers because controls contained more beetle-killed pine. Given the high populations of M P B in this area, and the lower rate of attack in aggregates, it is possible that aggregates have properties that confer more beetle-resistance than controls. There is likely some increase in tree vigour created by the higher light conditions resulting from the edge effect in aggregates (vigorous trees are more beetle resistant). Also, M P B appear to prefer the lower temperatures and light intensities, typical of dense stands to land and initiate attack (Natural 88  Resources Canada 2002). Increased winds disperse the pheromone plumes that serve to concentrate attacks (Natural Resources Canada 2002). M P B are not strong fliers and may also be negatively affected by the relatively higher winds in and around aggregates compared to controls (Pacific Forestry Centre 2002). Whether windthrow is measured in stems per hectare or basal area, the abundance of down trees is highest in aggregate treatments (Study 1), but probability of being down is higher in dispersed treatments (Study 2). Windthrow risk increased with increased exposure and spruce is the most likely species to be windthrown. The control treatment had the highest occurrence of pine and contained the highest abundance of standing dead conifers (Study 1). The probability of being windthrown in an aggregate increased with an increase in proximity to edge. Study 2 also indicated very small trees (<20cm dbh, <10m ht) and very large trees (>40cm dbh, >40m ht) were less prone to windthrow than typical main canopy (medium sized) trees. The estimated pre-harvest stand density results indicate that retained aggregates and dispersed trees were somewhat different from the stand averages. Pre-emptive harvesting focused on M P B susceptible mature pine and salvage harvesting focused on larger mature pine under M P B attack. Aggregates contained a higher abundance of non-pine species than preharvest dispersed treatments indicating a selection bias for retaining non-pine, M P B resistant species during cutblock layout. This is a good harvesting strategy in this area. In terms of the study, this 'selection bias' needed to be accounted for in the interpretation of treatment effects in Study 1. Study 1 addressed the question "does aggregate treatment result in a different abundance of wildlife structural/habitat features than dispersed treatment and how do they both differ from unharvested forest". The results indicate that different partial harvesting treatments do result in different abundances of important habitat variables. Aggregate treatments result in higher 89  abundances of most habitat variables simply because aggregates retain more trees, and fewer of these trees are subsequently windthrown. Deciduous trees in aggregates are still wind damaged, but there is sufficient protection to allow many damaged stems to remain standing. Damaged standing deciduous trees can provide valuable wildlife habitat. The conditions in aggregates also result in the highest abundance of shrub cover. However, large veteran Douglas-fir and deciduous trees occurred infrequently in the pre-harvest stands and in the aggregates. Aggregate treatment would therefore be less effective at maintaining these important structural elements unless they were specifically focused on for retention. Douglas-fir veterans can be retained in dispersed treatments wherever they occur throughout the stand with moderate levels of windthrow loss. The unharvested forest provides the lowest abundance of shrub cover, but the highest abundance of snags in the form of MPB-killed trees. Study 2 asked the two questions "do the different partial harvesting methods of aggregate treatment and dispersed treatment affect probabilities of tree mortality or windthrow" and "how do tree, neighbourhood, and site factors affect the probabilities of tree mortality or windthrow". Most differences could be attributed to the different levels of exposure between the treatments, but individual tree characteristics were also important. Most differences in nonwindthrow mortality could be attributed to species (eg., MPB-killed pine). Differences in beetle mortality for pine may be influenced by a combination of factors including individual tree characteristics (eg., size, age, vigour), site factors (eg., increased wind, exposure), and stand history (eg., pre-harvest infection). The results of this study have various implications for forest managers, biologists, and researchers. Since managers are using the partial harvesting treatments to meet biodiversity objectives that are based on conserving areas of forest cover, the level of windthrow is a crucial factor in determining long-term treatment success. If an area of forest is conserved during harvesting only to be substantially blown down by wind soon after harvest, the conserved 90  area no longer provides the same habitat values. Generally, few trees were left after harvesting in the dispersed treatments (<100 sph) and 57 % of these were windthrown by the time of sampling. The study results indicate Douglas-fir veterans have the best chance of remaining standing in the dispersed treatments and larger aggregates experience less proportional windthrow than smaller aggregates. Biologists are concerned with windthrow and mortality status since the habitat value of a tree changes i f it is live, dead, standing, or down. Standing Douglas-fir veterans in the dispersed treatments provide wildlife values including large branches for perching or nesting platforms, deeply furrowed bark as a foraging substrate for woodpeckers, and age related decay patterns desirable for cavity excavators. Standing wind-damaged deciduous trees can provide important wildlife values. Wind damage includes uprooting, broken stems, broken tops and broken branches. Wind damaged deciduous trees were more often broken in aggregates, but uprooted in the dispersed treatments. Some windthrow is inevitable in dispersed and aggregate treatments and can be an important source of future large C W D . Input from windthrown trees will supply long lengths of large C W D that can be used as travel corridors and cover for small mammals (eg., voles) as well as hunting opportunities for their predators (eg., marten). The branches on down trees can provide substitute escape cover for small animals until shrub cover and tree seedlings regenerate (Smith et al. 1997). As trees eventually decay they can provide moist localized conditions for salamanders and invertebrates and foraging substrates for insect-feeding animals (Kimmins 1966). Aggregate treatments resulted in higher volumes of shrub habitat than dispersed treatments and unharvested control areas. Aggregates supply islands of concentrated habitat in the form of high shrub cover volumes, and a mixture of species of standing live, dead, and dying trees. Dying trees can provide habitat value to many species including nesting and foraging sites for raptors (Zemlak et al. 1995) and cavity nesters (Robinson and Mark 2001).  91  Dispersed treatments will supply a component of large old wildlife trees and can actually increase both abundance and richness of secondary cavity nesters Marcot (1983). It would be valuable for researchers to explore why lodgepole pine was less susceptible to M P B attack in the aggregates than in the unharvested control areas. Increased wind, light, tree vigour, and higher proportions of non-pine species may have affected the rates of M P B attack. Further studies focusing on those factors are needed to confirm these assumptions. There may be threshold levels of exposure (eg., light, wind), or densities of pine that provide characteristics of beetle resistance. Aggregate exposure, size, location, and orientation may also be important factors. Experimental design should include measurements before and after harvesting. This will allow separation of the effects of the aggregate treatments from differences in pre-harvest stand conditions (eg., species mixture, tree size, age).  5.2 R E C O M M E N D A T I O N S Any improvements in the ability to predict how individual trees or groups of trees will respond to a particular partial harvesting treatment can assist managers in planning long-term habitat management strategies. It is essential to effective wildlife management that the target stand is clearly defined and based on the needs of the species of interest. The desire for the presence of particular elements in the target stand (eg., large conifers, standing dead trees, shrub cover) will dictate which trees or groups of trees should be retained to provide the best opportunity of sustaining the required habitat values. However, retaining all the appropriate stand features during harvesting does not necessarily mean that they will remain intact over time, short or long term. The main recommendations from this study are as follows: •  Dividing the desired level of long term retention by the proportion of trees expected to survive windthrow gives a more realistic number of standing trees to retain during harvesting.  92  Some trees will be windthrown and should therefore be included in projections of long term C W D recruitment. With the exception of large veteran Douglas-fir which are scattered throughout preharvest stands, aggregate treatments are better able to sustain most elements that provide habitat better than dispersed treatments. Long-term habitat management plans should include a component of aggregate retention. Results indicated M P B mortality of pine was slightly lower in aggregates, but it is premature at this point to recommend aggregate treatments as a way of sustaining lodgepole pine. Very small trees and very large trees are less likely to be windthrown than medium sized trees, so these are the best trees to retain in dispersed treatments. Even though there is a high windthrow risk when retaining individual dispersed trees, it is one worth taking. Since most of the area-based biodiversity objective is being met with aggregate treatments, any loss to windthrow in dispersed treatments (eg., individual trees) will not severely impact the total reserve area credited to the biodiversity objective and a few dispersed trees will still provide more potential habitat value than no retained trees. Since more than half of the trees are windthrown in the dispersed treatments, it is probably best to retain at least 4-6 sph of the available large Douglas-fir veterans to provide the best chance for an adequate supply (eg., 2-3 sph). Retaining smaller, more slender Douglas-fir in dispersed treatments is less desirable since they have fewer wildlife attributes and a higher susceptibility to windthrow. Lodgepole pine has a high probability to be attacked by M P B so alternate species should be retained to meet any long-term requirements for live tree values. It is recommended  93  to retain both pine and non-pine species in aggregates to provide a desirable mixture of live trees and dead standing snags. The likelihood of a tree to be windthrown in an aggregate increases as its level of exposure increases (eg., proximity to edge, decreasing aggregate size). DeLong et al. (2001) found that windthrow in reserves larger than five hectares in northern British Columbia was similar to that in large contiguous areas of forest. DeLong et al. recorded maximum windthrow losses ranging from 100% in reserves smaller than one hectare to 25% in reserves larger than one hectare. Burton (2001) found the zone of highest windthrow (>10%) extended 18-32m into the windward edge of retention aggregates. Recommended minimum aggregate size should be at least three tree lengths wide or approximately one hectare. Species is also an important factor affecting windthrow and mortality. Douglas-fir is generally considered to be the most windfirm conifer of those sampled in this study and the results confirm it is the best choice for retention in a dispersed treatment. Any specific habitat features considered vulnerable to exposure (eg., old snags, windthrow prone species, shade-tolerant vegetation) should be retained inside an aggregate to provide some level of protection. Since more than half the trees will likely be windthrown in a dispersed treatment, the best insurance that sufficient trees will be available over the long term is to retain two or even three times the minimum required. This study found a range of one to six stems/ha of standing large trees in the dispersed treatments. If the minimum recommended densities of 2-3 large dead snags/ha are not met, living trees can be converted to standing snags by girdling or fungal injections. Aggregates have much lower windthrow rates meaning actual retained minimums can be closer to target minimums. Nevertheless, it is inevitable that some windthrow will occur  94  along aggregate edges so levels of retention in aggregates should be at least 20% higher than what is required in the target stand. Also, stand features that require time to develop such as late serai species, wildlife trees, CWD volumes, etc. will be better protected in aggregates than dispersed treatments. •  Since there were low densities of deciduous trees found in all treatments they should be selected to be retained in aggregates whenever possible. Deciduous trees provide important wildlife values and have a better chance to remain standing in the less exposed aggregates.  •  Shrub cover volumes are most likely reduced with high levels of ground disturbance and exposure meaning most required shrub cover should be retained in aggregates rather than in the dispersed treatments. However, early serai vegetation and grass can be sustained to provide forage and grazing in dispersed treatments.  •  If post-harvest levels of C W D are expected to be insufficient to meet habitat objectives, additional down wood such as lower grade logs/trees can be left on site during harvesting. Windthrown trees will also supply C W D in dispersed treatments.  •  It may be necessary to include special management procedures to retain specific features important to any identified species of interest (red or blue-listed species, or species at risk). Maintaining viable areas of "old growth" will require retention of unharvested areas sufficiently large to sustain a complex set of conditions rather than just individual features. This means the aggregate size needs to be large enough to maintain interior forest conditions within the reserve patch.  5.3 LIMITATIONS OF THIS S T U D Y The stand reconstruction process provided an estimation of live stems per hectare, basal area, species, and dbh distribution of the pre-harvested dispersed treatments. However, it was  95  not possible to determine pre-harvest conditions such as tree status (eg., health, mortality), tree height, shrub cover, or C W D precisely. The non-random selection of treatment blocks creates the potential for bias. As has been noted several times in this thesis, post-harvest results may result more from pre-harvest differences in stand conditions rather than any treatment effects resulting from a particular retention level or pattern, For example, the pre-treatment stand conditions affect the individual tree characteristics, but are not related to a post-harvest treatment effect. Tree characteristics (eg. size, taper, species) that are used to select which individual or groups of trees to retain in dispersed or aggregate treatments may potentially influence the final status of the tree as much as or more than the effects of the treatment. If preharvest stand history allowed some trees (eg., Douglas-fir veterans) to adapt windfirmness, those trees would remain standing better than other trees. If aggregates contained lower densities of pine species, this may have lowered rates of M P B attack more than any beetle resistance inherent to aggregates. Nevertheless, differences in the abundance of habitat variables between treatments can be estimated and inferences to habitat values be made. Also, relationships between individual tree behaviour and tree, site, and neighbourhood factors can be estimated. Retention selection bias was less of an issue in Study 2 because individual tree characteristics and treatment effects, as reflected by the exposure variables, were included in the models. Treatment effects can be indirectly linked to relationships such as that between occurrence of tree windthrow and varying levels of exposure. Exposure variables describe actual differences in levels of exposure between treatments (eg., aggregate treatment versus dispersed treatment) and therefore can be considered to suggest treatment effect from an individual tree perspective. This study provided estimates of the abundance and future status of several variables representing habitat values, but did not establish any actual correlation between these variables and their significance to wildlife. It was assumed that the chosen variables had wildlife value 96  based on a large body of literature, but local habitat preferences would need to be verified with local wildlife use studies. The related 2005 study in the same area by Chan-Macleod (unpublished) found mature forest bird species did not appear to well at low retention levels. Windthrown trees were not recorded in the controls. Consequently the probability of being windthrown in treated stands could not be compared to unharvested stands. However, recent windthrow (e.g. during the 7 year period prior to sampling) was extremely rare within control areas indicating that this source of damage becomes significant only after a partial harvesting treatment has been done. The range of stand types and general soil and topographic conditions within the study area is low. Other than the small differences in pre-harvest stand density and Douglas-fir abundance, no other systematic variation with treatment was observed. The sites in this case study are typical of the region and the results should therefore be applicable to similar stand types within the region. 5.4 R E C O M M E N D A T I O N S FOR F U R T H E R STUDIES Greater statistical rigour could be obtained with an experiment specifically designed before harvesting to measure the changes resulting from different partial harvesting treatments. The experimental design could use stands that have been identified as having very similar characteristics (eg., site, tree, and neighbourhood features) prior to any harvesting activities. Randomly assigning treatments to candidate stands removes the possibility of the pre-harvest variability in the stands being mistaken for treatment effects. For example, elements such as mortality, post harvest C W D and shrub volumes could be compared to pre-harvest values to better measure the effects of treatments. Ideally the dispersed treatments would cover a greater range of retention levels, and the aggregate treatments a greater range of aggregate sizes. Permanent sample plots could be installed in stands prior to treatments and monitored for several years following harvesting activities. Understanding the actual changes that happen  97  over time would contribute to the development of predictive models which can assist managers to plan effective long-term habitat management strategies. A study of live pine trees surviving in M P B infested stands (before and after harvesting) in the Prince George Forest Region may provide information on factors that contribute to beetle resistance. Any additional understanding of the individual tree characteristics, site conditions, and neighbourhood factors that result in lower M P B infection rates would greatly enhance the effectiveness of present M P B management. Regional post harvest wildlife studies focused on the specific elements identified as having habitat value would provide empirical confirmation of local wildlife usage in treatment areas. Both stand level and landscape level studies are needed to determine how much of the various elements are actually required to meet wildlife habitat needs. These types of studies will increase the confidence of managers that their habitat retention efforts are actually accomplishing the ecosystem management goals that the area-based biodiversity objectives are designed to achieve.  98  REFERENCES Alberta Centre for Boreal Studies. 2000. Forest structure and pattern: the foundation of forest biodiversity. Fact Sheet. 2 p. http://www.borealcebtre.ca/facts/forests.html Black, T. A . and T. R. Oke. 2000. Soil Science/Geography 204. Forest and agricultural climatology. Lecture handouts. University of British Columbia. Boyland, M . and F. L . Bunnell. 2002. Vertebrate Use of Dead Wood in the Pacific Northwest. http://www.forestry.ubc.ca/atlas-simfor/webdocs/extension/Vertebrates Bradford, P. 1995. Wildlife tree Stand-level biodiversity workshop proceedings. Editors: P. Bradford, T. Manning, and B. Lanson. B.C. Environment. Canadian Cataloguing in Publication Data. Victoria B.C. Bradley, T. 1993. Silva Ecosystem Consultants Ltd. http://www.silvafor.org/publications/library/docs/Mountain%20Pine%20Beetle.pdf British Columbia Ministry of Forests. 1995. Biodiversity Guidebook. Forest Practices Code of British Columbia Act. Strategic Planning Regulations. Operational Planning Regulation. British Columbia Ministry of Forests. 1999. Introduction to Silvicultural Systems, second edition. Forest Practices Branch, Victoria, B C . http://www.for.gov.bc.ca/hfd/pubs/SSIntroworkbook/index.htm British Columbia Ministry of Forests. 2002. Stand Level Biodiversity, Web Based Training Course. ht1p://www.for.gov.bc.ca/hfp/fordev/biodiversity/index.htm. British Columbia Ministry of Forests. 2004 a. M P B Action Plan Update. http://www.for.gov.bc.ca/hfp/mountain_pine_beetle/actionplan042004.pdf). British Columbia Ministry of Forests. 2004 b. Wildlife Tree Committee Coarse Woody Debris Policies, http://www.for.gov.bc.ca/hfp/wlt/cwd-policy-01 .htm. British Columbia Ministry of Forests. 2005. Silviculture Strategy. Habitat Supply Overview. htt4)://www.for.gov.bc.ca/hfp/silstrat/habitat-overview.htm. Bull, E.L., and M . G . Henjum. 1990. Ecology of the great gray owl. U.S.D.A. Forest Service. General Technical Report. PNW-265, Portland, Oregon. Bull, E.L., C. G. Parks, and T. R. Torgersen. 1997. Trees and logs important to wildlife in the interior Columbia River Basin. United States Department of Agriculture, Forest Service, Pacific Northwest Research Station, Portland, Oregon. General Technical Report PNW-GTR-391. Bunnell, F.L., B.G. Dunsworth, D.J. Huggard, and L . L . Kremsater. 2003. Learning to sustain biological diversity on Weyerhaeuser's coastal tenure. The Forest Project, Weyerhaeuser, Nanaimo, B C .  99  Bunnell, F. L., L. L . Kremsater and M . Boyland. 1998. A n ecological rationale for changing forest management on MacMillan Bloedel's forest tenure. Contract report to MacMillan Bloedel Ltd. 281 p. Bunnel, F. L., L . L . Kremsater and E. Wind. 1999a. Managing to sustain vertebrate richness in forests of the Pacific Northwest: relationship within stands. Environmental Reviews 7: 97-146. ;  Bunnell, F.L, E. Wind, M . Boyland, and I. Houde. 1999b. Diameters and heights of cavity sites: their implications to management. Ecology and Management of Dead Wood in Western Forests. Nov. 2-3, 1999. The Wildlife Society, Reno, Nevada. Bunnell, FL, Houde, I, Johnston, B , and Wind, E. 1999c. How dead trees sustain live organisms in western forests. Ecology and Management of Dead Wood in Western Forests. Nov. 2-3, 1999. The Wildlife Society, Reno, Nevada. Bunnell, F.L., I. Houde, B. Johnston, R.W. Wells, and E. Wind. 1999d. Relations of forestdwelling vertebrates and other organisms to stand attributes in forests of the Nelson Forest Region. Centre for Applied Conservation Biology, Publication No. M-14, University of British Columbia, Vancouver, B.C. Bunnell, F.L. and A. Allaye-Chan. 1984. Potential of winter range reserves for ungulates as habitat for cavity-nesting birds. In Fish and Wildlife relationships in old-growth forests. Symposium Proceedings, Juneau, Alaska, 12-15 April, 1982. W.R. Meehan, T.R. Merrel Jr., and T.A. Hanley (editors). American Institute Fish Resource Biology. 357-366. Burns R. M . and B. H. Honkala. 1990a. Silvics of North America: Volume-1 Conifers. United States Department of Agriculture. Forest Service. Agriculture Handbook 654. 675 p. Burns R. M . and B. H. Honkala. 1990b. Silvics of North America: Volume-2 Hardwoods. United States Department of Agriculture. Forest Service. Agriculture Handbook 654. 877 p. Burton, P. J. 2001. Windthrow patterns on cutblock edges and in retention aggregates in the SBSmc. Symbios Research & Restoration. In Windthrow Assessment and Management in British Columbia. Proceedings of the Windthrow Researchers Workshop held January 31February 1, 2001 in Richmond, British Columbia Compiled by S.J. Mitchell and J. Rodney. Buskirk, S.W., S.C. Forrest, M . G . Raphael, and H.J. Harlow. 1989. Winter resting site ecology of Marten in the central Rocky Mountains. Journal of Wildlife Management. 53: 191-6. In Keisker, D. G. 2000. Types of wildlife trees and coarse woody debris required by wildlife of north-central British Columbia. B C MoF. Work. Pap. 50. Victoria, B.C. Byrne, K . 2005. Critical turning moments and drag equations for British Columbia conifers. MSc Thesis. University of British Columbia, Vancouver, Canada. Center for Applied Conservation Research 2004. Biodiversity and Forest Management in British Columbia, http://142.103.128.161/ci_ind2_l.asp?id=213).  100  Chan-McLeod, A . A. and F. Bunnell. 2003. Potential approaches to integrating silvicultural control of mountain pine beetle with wildlife and sustainable management objectives. Mountain Pine Beetle Symposium: Challenges and Solutions. Natural Resources Canada, Canadian Forest Service. Information Report BC-X-399: 267-276. Clark, D.F., D.D. Kneeshaw, P. J. Burton, and J.A. Antos. 1998. Coarse woody debris in subboreal spruce forests of west-central British Columbia. Canadian Journal of Forest Research. 28(2): 284-90. Clayoquot Scientific Panel (CSP). 1995. Scientific Panel for Sustainable Forest Practices in Clayoquot Sound. Sustainable ecosystem management in Clayoquot sound. Planning and Practices, Report 5. Victoria B.C. Colostate. 2005. Wildlife Habitat, Introduction. www.cnr.colostate.edu/class_info/fw471/lectures/Lecturel5...Wildlife Habitat Intro. Corn, J.G. and M . G . Raphael. 1992. Habitat characteristics at marten subnivean access sites. Journal of Wildlife Management. 56: 442-448. In Buskirk S.W. and L.F. Ruggerio. 1994. The scientific basis for conserving forest carnivores American Marten, Fisher, Lynx, and Wolverine. U S D A Forest Service. General Technical Report RM-246. Coxson, D., S. Stevenson, and J. Campbell. 2003. Short-term impacts of partial cutting on lichen retention and canopy microclimate in an Engelmann spruce - subalpine fir forest in northcentral British Columbia. Canadian Journal of Forestry Res. 33: 830-841. Davis, C M . 1978. A nesting study of the brown creeper. Living Bird 17: 237-263 Davis, L.R., M.J. Waterhouse, and H . M . Armleder. 1999. A comparison of the breeding bird communities in serai stages of the Engelmann Spruce-Sub-alpine Fir Zone in East Central British Columbia, Working Paper 39. B.C. Ministry of Forests Research Branch. Victoria B.C. Delong, C , D. Tanner, and M . Jull. 1993. A field guide for site identification and interpretation for the southwest portion of the Prince George Forest Region. British Columbia Ministry of Forests. Crown Publications. Victoria B C . 290 p. DeLong, S. C , P. Burton, T. Mahon, P. Ott, and D. Steventon. 2001. Factors affecting windthrow in reserves in northern British Columbia. British Columbia Forest Service. In Windthrow Assessment and Management in British Columbia. Proceedings of the Windthrow Researchers Workshop held January 31-February 1, 2001 in Richmond, British Columbia Compiled by S.J. Mitchell and J. Rodney. Densmore, N . Optimizing wildlife trees and coarse woody debris retention at the stand and landscape level. Winter workshop. 2002. Northern Interior Vegetation Management Association and Northern Silviculture Committee. Prince George B C . Dickson, J.G., R.N. Conner, and J.H. Williamson. 1983. Snag retention increases bird use of a clear cut. Journal of Wildlife Management 47:799-804.  101  Eng, M . , A. Fall, J. Hughes, T. Shore, B. Riel, P. Hall, and A . Walton. 2005. Provincial-Level Projection of the Current Mountain Pine Beetle Outbreak: A n Overview of the Model (BCMPB v2) and Results of Year 2 of the Project. Canadian Forest Service, B C Forest Service. 51p. Environment Canada. 2005. Habitat Conservation web site. http://www.cws-scf.ec.gc.ca/habitat/index_e.cfm. Farnden, C , I.S. Moss, and T. Earle. 2003. Prototype field guide for identifying stand structure classes in the Cariboo Forest Region. Prepared for Lignum Limited of Williams Lake, British Columbia. 92 p. Halpern, C. P. and T. Spies. 1995. Plant species diversity in natural and managed forests of the Pacific northwest. Ecological Applications, 5 (4). 913-934. The Ecological Society of America. Harestad, A. S. and D. G. Keisker. 1989. Nest tree use by primary cavity-nesting birds in south central British Columbia. Canadian Journal of Zoology 67:1067-1073. Harris B. 1995. The use by wild birds of three meter stubs as artificial snag habitat. In Wildlife tree Stand-level biodiversity workshop proceedings. Editors: P. Bradford, T. Manning, and B. Lanson. B.C. Environment. Canadian Cataloguing in Publication Data. Victoria B.C. Holt R.F., T.F. Braumandl, and D.J. Mackillop. 1999. A n index of old-growthness for two B E C variants in the Nelson Forest Region. Final report, prepared for B C Ministry of Environment, Lands and Parks, Nelson, B C . Hutto, RL. 1995. Composition of bird communities following stand-replacement fires in northern Rocky Mountain (USA) conifer forest. Conservation Biology 9:1041-1058. Jull, M . 2001. Wind Damage and Related Risk Factors for Interior Douglas-fir Leave Trees in Central B C . In Windthrow Assessment and Management in British Columbia. Proceedings of the Windthrow Researchers Workshop held January 31-February 1, 2001 in Richmond, British Columbia Compiled by S.J. Mitchell and J. Rodney. Jull, M . , S. Stevenson, A. Eastman, and P. Sanborn. 2003. Northern wetbelt silvicultural systems project: A study of alternative silvicultural systems in northern wetbelt ICH and ESSF forests. B.C. Ministry of Forests, Victoria. Keisker, D. G. 2000. Types of wildlife trees and coarse woody debris required by wildlife of north-central British Columbia. B C MoF. Work. Pap. 50. Victoria, B.C. Kimmins, J.P. 1996. Forest Ecology: A foundation for sustainable management. 2 Macmillan Inc., New York. 596 p.  nd  Edition.  Lemay, V . and P. Marshall. 1990. Forestry 238: Forest Mensuration. Distance Education and Technology, Continuing Studies. University of British Columbia, Vancouver, British Columbia, Canada. 214 p.  102  Leslie, E. and T. Bradley. 2001. Brief literature review of the Douglas-fir bark beetle. Silva Ecosystem Consultants Ltd. http://www.silva.org/publications/library/docs/DougFirBarkReview.pdf. Mackinnon, A., J. Pojar, and R. Coupe'. 1992. Plants of Northern British Columbia. 2 Edition. British Columbia and Lone Pine Publishing. 352 p.  nd  Marcot, B.G. 1983. Snag use by birds in Douglas-fir clearcuts. In Snag habitat management: proceedings of the symposium, June 7-9, 1983, Flagstaff, Ariz. U.S.D.A. Forest Service General Technical Report. RM-GTR-99, Fort Collins, Colorado. 34-139. Martin, K . and J. M . Eadie. 1999. Nest webs: a community-wide approach to the management and conservation of cavity nesting forest birds. Forest Ecology and Management 115: 243-257. Martin, K . 2003. Treehouse to Share. Canadian Wildlife Service, Environment Canada and University of British Columbia, http://www.pwc.bc.doe.ca/ec/frap/index.html McCleary, K . and G. Mowat. 2002. Using forest structural diversity to inventory habitat diversity of forest-dwelling wildlife in the West Kootenay region of British Columbia. Research Report. B.C. Journal of Ecosystems and Management. 2 (2). http://www.forrex.org/jem/2002/vol2/no2/artl.pdf. McCullough, D.G., R.A. Werner, and D. Neumann. 1998. Fire and insects in northern and boreal forest ecosystems of North America. Annual Review of Entomology. 43: 107-27. Natural Resources Canada, 2003. Silviculture and the Mountain Pine Beetle-Habitat and Visual Quality Objectives. Natural Resources Canada Web Site. http://wvvw.pfc.forestry.ca/entomology/mpb/management/silviculture/habitat_e.html Natural Resources Canada. 2002. Silviculture and the Mountain Pine Beetle - Prescription. Mountain Pine Beetle. Pacific Forestry Centre. 2p. Oliver, C D . and B.C. Larson. 1990. Forest Stand Dynamics. McGraw-Hill Inc., New York. Pacific Forestry Centre. 2002. Mountain pine Beetle. Prescription for Beetle Proofing. Natural Resources Canada. http://www.pfc.forestry.ca/ca/entomology/mpb/management/silviculture/prescription_e.html Pojar, J. 1984. Ecological classification of lodgepole pine in Canada. In Silva Ecosystem Consultants. 1992. Old Growth Literature Review. http://www.silvafor.org/publications/library/docs/01d%20Growth%20Ecology.pdf Pojar, R.A. 1995. Breeding Bird Communities in Aspen Forests of the Sub-boreal Spruce (dk Subzone) in the Prince Rupert Forest Region. Province of British Columbia, Ministry of Forests Research Program. Raphael, M . G . and M . W . White. 1984. Use of snags by cavity-nesting birds in the Sierra Nevada. Wildlife Monograph. No. 86. 66 p.  103  Rayner, A . D . M . , and L. Boddy. 1988. Fungal decomposition of wood, its biology and ecology. JohnWiley & Sons, New York. Robinson, J. and T. Mark. 2001. California Partners In Flight Coniferous Forest Bird Conservation Plan for the Red-breasted Nuthatch (Sitta Candensis). U S D A Forest Service, Tahoe National Forest. SAS Institute Inc. 2004. SAS 9.1.3 Cary, N C , USA. Scott, R. E. 2005. Modeling windthrow risk in coastal variable retention using tree, neighbourhood, and stand attributes. MSc. Thesis. Forest Science. University of British Columbia. Schuerholz, G., P. McNamee, and M . R. C. Massie. 1988. Estimation of the effect of intensive logging on ungulates (cervids) in the White River drainage. Canadian Forestry Service. Pacific Forestry Centre. Victoria, B C . Information Report BC-X-303. Shea, P. J., W. F. Laudenslayer Jr., G. Ferrell, and R. Borys. 2002. Girdled versus Bark Beetlecreated Ponderosa Pine Snags: Utilization by Cavity-dependent Species and Differences in Decay Rate and Insect Diversity. U S D A Forest Service Gen. Tech. Rep. PSW-GTR-181. Siegel, S. and N . J. Castellan Jr. 1988. Nonparametric Statistics for the behavioral sciences, second edition. McGraw-Hill, Inc. 399 p. Silva Ecosystem Consultants. 1992. Old Growth Literature Review. http://www.silvafor.org/publications/library/docs/01d%20Growth%20Ecology.pdf Spies, T.A. and S.P. Cline. 1988. Coarse wood debris in manipulated and unmanipulated coastal Oregon forests. In C. Maser, R.F. Tarrant, J.M. Trappe, and J.F. Franklin (tech eds). From the Forest to the Sea: A Story of Fallen Trees. Gen. Tech. Rep. PNW-GTR-229, Pacific Northwest Research Station. Spies T. A . 1998. Forest Structure: A Key to the Ecosystem. Northwest Science 72 Special Issue No. 2. U S D A , Forest Service, Pacific Northwest Research Station, Corvallis, Oregon, U S A , 9733lby the Northwest Scientific Association. Stadt, J. 2001. The Ecological Role of Beetle-Killed Trees: A review of salvage impacts. Ministry of Water, Land and Air Protection. Habitat Protection, Skeena Region. http://ww.apbbc.bc.ca/htdocs/Ecological%20Role%20of%20Beetle%20killed%20trees%20%20a%20review%20of%20salvage%20impacts.htm. Stathers, R.J., T.P. Rollerson, and S.J. Mitchell. 1994. Windthrow handbook for British Columbia forests. B.C. Min. For., Victoria, B.C. Working Paper 9401. 32 p. Steeger, C. and J. Dulisse. 1997. Ecological Interrelationships of Three-Toed Woodpeckers with Bark Beetles and Pine Trees. Nelson Forest Region, Ministry of Forests. Extension Note 035.  104  Stevens, V . 1997. The ecological role of coarse woody debris, an overview of the ecological importance of C W D in B C forests. Research Branch. B C Ministry of Forests., Victoria, B.C. Working Paper 30. 25 p. Stevenson, S. and D. Coxson. 2004. Short-term effects of partial cutting on arboreal forage lichens at Pinkerton Mountain. Forest Research Update. Issue N R 3. University of Northern British Columbia. Steventon, J. D. 1997. Historic disturbance rates for interior biogeoclimatic subzones of the Prince Rupert Forest Region. Ministry of Forests, Prince Rupert Forest Region, Research Branch Extension Note #26. Stone, J.N., and J.L. Porter 1998. What is Forest stand structure and how to measure it. Proceedings of a workshop on structure, process, and diversity in successional forests of coastal British Columbia, February 17-19, 1998, Victoria, British Columbia. Research Branch, British Columbia Ministry of Forests. Northwest Science 72, Special Issue No. 2. Tabachnick B.G., and L.S. Fidell, 2001. Using multivariate statistics. 4 edition. Allyn & Bacon, Pearson Education Company. M A . U.S. 966 p. th  Thomas, J.W. 1979. Introduction. In Wildlife habitats in managed forests: The Blue Mountains of Oregon and Washington. J.W. Thomas (technical editor). U.S. Department of Agriculture Forest Service, Washington, D.C. Handbook 553: 10-21. Trexler, J.C., and J. Travis. 1993. Nontraditional Regression Analyses. Ecology 74. 1629-1637. Vega, R.M.S. 1993. Bird communities in managed conifer stands in the Oregon Cascades: habitat associations and nest predation. M.Sc. thesis, Oregon State University, Corvallis, Oregon. Waterhouse M.J. and H . M . Armleder. 2004. Windthrow in partially cut lodgepole pine forests in west-central British Columbia. B.C. Min. For., Res. Br., Victoria, B.C. Exten. Note 70. <http://w\vw.for.gov.bc.ca/hfd/pubs/Docs/En/En70.htm> Waterhouse, M.J., and R. Dawson. 1999. Bird communities in Interior Douglas-fir forests. Pages 90-112. In: Hollstedt, C, A. Vyse, and D. Huggard, (Editors). Managing the dry Douglas-fir forests of the southern interior: workshop proceedings, 1997, April 29-30, Kamloops, B C . B C Ministry of Forests. Watts S.B. 1983. Forestry Handbook for British Columbia 4 Edition. The Forestry Undergraduate Society, University of British Columbia. D.W. Friesen and Sons Ltd. Cloverdale, B.C. th  Wells, R., I. Houde, M . Kellner, and D. Haag. 2003. Low Level Indicators for Biodiversity: A preliminary evaluation of the Prince George T S A (McGregor Model Forest). Wells, R., E. Valdai, C. Steeger, and P. Vernier. 1998. SFMFOR Habitat Analysis of Two FSSEVI Harvest Scenarios in the Rocky Mountain Trench. B C Ministry of Forests, Rocky Mountain Forest District. 105  Whitehead, R.J. and B.N. Brown. 1997. Windthrow after commercial thinning in mature lodgepole pine. Canadian Forest Service. File Report, p 58001-97- 01. Zemlak, R., M . Wood and F. Corbould. 1995. Williston and Dinosaur Reservoirs raptor survey 1995: Survey 1, Raptor nest locations. Peace/Williston Fish and Wildlife Compensation Program, Report No. 43. 8 pp.  106  APPENDICES APPENDLX 1: SPECIES IN MPB-INFESTED A R E A S IN B C in 2002 Species  Scientific Name  blackbird, Brewer's blackbird, red-winged blackbird, rusty blackbird, yellow-headed bluebird, mountain bobolink bunting, lazuli bunting, snow chickadee, black-capped chickadee, boreal chickadee, mountain cowbird, brown-headed creeper, brown crossbill, red crossbill, white-winged crow, American eagle, bald eagle, golden falcon, peregrine' finch, cassin's finch, purple flicker, northern flycatcher, alder flycatcher, dusky flycatcher, Hammond's flycatcher, least flycatcher, olive-sided flycatcher, Pacifi c-sloped flycatcher, yellow-bellied goldfinch, American goshawk, northern' grosbeak, black-headed grosbeak, evening grosbeak, pine grosbeak, rose-breasted grouse, blue grouse, ruffed  Euphagus cyanocephalus Agelaius phoeniceus Euphagus carolinus Xanthocephalus xanthocephalus Sialia currucoides Dolichonyx oryzivorus Passerina amoena Plectrophenax nivalis Poecile atricapilla Poecile hudsonica Poecile gambeli Molothrus ater Certhia americana Loxia curvirostra Loxia leucoptera Corvus brachyrhynchos Haliaeetus leucocephalus Aquila chrysaetos Falco peregrinus Carpodacus cassinii Carpodacus purpureus Colaptes auratus Empidonax alnorum Empidonax oberholseri Empidonax hammondii Empidonax minimus Contopus cooperi Empidonax diffi cilis Empidonax fl aviventris Carduelis tristis Accipiter gentilis Pheucticus melanocephalus Coccothraustes vespertinus Pinicola enucleator Pheucticus ludovicianus Dendragapus obscurus Bonasa umbellus  meadowlark, western"' merlin mockingbird, northern nighthawk, common nuthatch, red-breasted nuthatch, white-breasted osprey ovenbird owl, barred owl, great gray owl, great-homed owl, long-eared owl, northern hawk owl, northern pygmy owl, northern saw-whet phoebe, Say's pigeon, band-tailed raven, common redpoll, common redpoll, hoary redstart, American robin, American sapsucker, red-breasted sapsucker, yellow-bellied shrike, northern siskin, pine solitaire, Townsend's sparrow, American tree sparrow, Brewer's sparrow, chipping sparrow, clay-colored sparrow, fox sparrow, golden-crowned sparrow, Harris's sparrow, lark sparrow, Lincoln's sparrow, savannah  grouse, spruce harrier, northern hawk, cooper's hawk, red-tailed hawk, rough-legged hawk, sharp-shinned hummingbird, anna's hummingbird, Calliope hummingbird, Rufous  Falcipennis canadensis Circus cyaneus Accipiter cooperii Buteo jamaicensis Buteo lagopus Accipiter striatus Calypte anna Stellula calliope Selasphorus rufus  jay, gray jay. Stcller's"' Junco, dark-eyed j unco kestrel, American kingbird, eastern kingbird, western kingfisher, belted kinglet, golden-crowned kinglet, ruby-crowned lark, horned" longspur, Lapland  Perisoreus canadensis Cyanocitta stelleri hyemalis Falco sparverius Tyrannus tyrannus Tyrannus verticalis Ceryle alcyon Regulus satrapa Regulus calendula Eremophila alpestris Calcarius lapponicus  sparrow, song sparrow, swamp sparrow, vesper sparrow, white-crowned sparrow, white-throated starling, european swallow, bank swallow, barn swallow, northern rough winged swallow, tree swallow, violet-green swift, Vaux's tanager, western thrush, hermit's thrush, Swainson's thrush, varied veery vireo, Cassin's vireo, red-eyed vireo, warbling  2  -  2  1  Species  2  2  2  107  Scientific Name  Stumella neglecta Falco columbarius Mimus polyglottos Chordeiles minor Sitta canadensis Sitta carolinensis Pandion haliaetus Seiurus aurocapillus Strix varia Strix nebulosa Bubo virginianus Asio otus Surnia ulula Glaucidium gnoma Aegolius acadicus Sayornis saya Columba fasciata Corvus corax Carduelis flammea Carduelis hornemanni Setophaga ruticilla Turdus migratorius Sphyrapicus ruber Sphyrapicus varius Lanius excubitor Carduelis pinus Myadestes townsendi Spizella arborea Spizella breweri Spizella passerina Spizella pallida Passerella iliaca Zonotrichia atricapilla Zonotrichia querula Chondestes grammacus Melospiza lincolnii Passerculus sandwichensis Melospiza melodia Melospiza georgiana Pooecetes gramineus Zonotrichia leucophrys Zonotrichia albicollis Stumus vulgaris Riparia riparia Hirundo rustica Stelgidopteryx serripennis Tachycineta bicolor Tachycineta thalassina Chaetura vauxi Piranga ludoviciana Catharus guttatus Catharus ustulatus Ixoreus naevius Catharus fuscescens Vireo cassinii Vireo olivaceus Vireo gilvus  A P P E N D I X 1 continued. mammals and herptiles Species  Scientific Name  Species  warbler, black-and-white warbler, chestnut-sided warbler, blackpoll warbler, Cape May warbler, Macgillivray's warbler, magnolia warbler, Nashville warbler, orange-crowned warbler, palm warbler, Tennessee warbler, Townsend's warbler, Wilson's warbler, yellow warbler, yellow-rumped waterthrush, northern waxwing, bohemian waxwing, cedar woodpecker, black-backed woodpecker, downy woodpecker, hairy woodpecker, pileated woodpecker, three-toed wood-pewee, western wren, house wren, marsh wren, winter yellowthroat, common  Mniotilta varia Dendroica pensylvanica Dendroica striata Dendroica tigrina Oporornis tolmiei Dendroica magnolia Vermivora rufi capilla Vermivora celata Dendroica palmarum Vermivora peregrina Dendroica townsendi Wilsonia pusilla Dendroica petechia Dendroica coronata Seiurus noveboracensis Bombycilla garrulus Bombycilla cedrorum Picoides arcticus Picoides pubescens Picoides villosus Dryocopus pileatus Picoides tridactylus Contopus sordidulus Troglodytes aedon Cistothorus palustris Troglodytes troglodytes Geothlypis trichas  common shrew dusky shrew pygmy shrew little brown myotis western long-eared myotis Yuma myotis long-legged myotis silver-haired bat big brown bat hoary bat Townsend's big-eared bat'' grizzly bear black bear fisherl marten least weasel short-tailed weasel long-tailed weasel" mink river otter wolverine striped skunk coyote gray wolf red fox mountain lion bobcat lynx yellow-pine chipmunk red squirrel northern flying squirrel beaver deer mouse bushy-tailed rat northern bog lemming'' brown lemming southern red-backed vole'' heather vole meadow vole long-tailed vole western jumping mouse meadow jumping mouse"' porcupine snowshoe hare"' elk"' white-tailed deer mule deer moose woodland caribou (mountain) long-toed salamander western toad Pacific treefrog spotted frog wood frog common garter snake  1  2  J  1  2  2  from Chan-McLeod and Bunnell (2003)  Scientific Name  Sorex cinereus Sorex monticolus Sorex hoyi Myotis lucifugus Myotis evotis Myotis yumanensis Myotis volans Lasionycteris noctivagans Eptesicus fuscus Lasiurus cinereus Corynorhinus townsendii Ursus arctos Ursus americanus Martes pennanti Martes americana Mustela nivalis Mustela erminea Mustela frenata Mustela vison Lontra canadensis Gulo gulo luscus Mephitis mephitis Canis latrans Canis lupus Vulpes vulpes Puma concolor Lynx rufus Lynx canadensis Tamias amoenus Tamiasciurus hudsonicus Glaucomys sabrinus Castor canadensis Peromyscus maniculatus Woodrat Neotoma cinerea Synaptomys borealis Lemmus trimucronatus Clethrionomys gapperi Phenacomys intermedius Microtus pennsylvanicus Microtus longicaudus Zapus princeps Zapus hudsonius Erethizon dorsatum Lepus americanus Cervus canadensis Odocoileus virginianus Odocoileus hemionus Alces alces Rangifer tarandus caribou Ambystoma macrodactylum Bufo boreas Pseudacris regilla Rana pretiosa Rana sylvatica Thamnophis sirtalis  ^ed-Listed, Blue-Listed, At-risk elsewhere in B C (outside M P B infested regions) 2  3  108  A P P E N D I X 1 continued. The following is a list of trees, shrubs, and insect species used in this report, but not included the M P B species list from Chan-McLeod and Bunnell 2003. Common Name black cottonwood black spruce bunchberry Douglas-fir fireweed lodgepole pine spruce (interior) subalpine-fir trembling aspen western hemlock bunchberry false Solomon's seal fireweed sweet-scented bedstraw queen's cup western hardhack Douglas-fir bark beetle mountain pine beetle  Scientific Name Populus trichocarpa Picea mariana Cornus canadensis Pseudotsuga menziesii Epilobium angustifolium Pinus contorta Picea glauca (hybrid) Abies lasiocarpa Populus tremuloides Tsuga heterophylla Cornus canadensis Smilacina racemosa Epilobium angustifolium Galium triflorum Clintonia uniflora Spiraea Douglassi Dendroctonous pseudotsugae Dendroctonous ponderosae  109  A P P E N D I X 2: BRITISH C O L U M B I A MINISTRY OF FOREST'S BIODIVERSITY OBJECTIVES Table 20(a). Percentage of a cutblock area required as wildlife tree patches when landscape units have been designated and landscape level biodiversity objectives have been established. % of the area available f o r h a r v e s t i n g in a l a n d s c a p e unit that has already b e e n  % of the biogeoclimatic s u b z o n e within  harvested without r e c o m m e n d e d wildlife tree retention  90  70  50  10  7  5  3  1  0  30  9  7  5  3  1  50  11  9  7  5  3  70  . 13  11  9  7  5  90  15  13  11  9  7  30  10  Note: The table axes refer to the area of the landscape unit. Table 20(b). Percentage of a cutblock area required as wildlife tree patches when landscape units have not been designated. % of the area available for harvesting that has already been harvested without recommended wildlife tree retention 3  % of the biogeoclimatic subzone within the landscape unit available for harvest  90  70  50  30  10  10  10  8  6  4  3  30  12  10  8  6  4  50  14  12  10  8  6  70  16  14  12  10  8  90  18  16  14  12  10  [a] Since no landscape unit objectives have been established, the area refers to the area of an interim landscape unit or a portion of a forest development plan that forms a contiguous geographic unit. Application of table 20(a) and (b) This is a one-time calculation for each biogeoclimatic subzone within the landscape unit (or interim landscape unit or portion of a forest development plan forming a contiguous geographic unit) unless the landscape unit objectives change, a new landscape unit is designated, or operability limits change (changing the area available for harvest). A separate prescription is made for each subzone within the landscape. X-axis numbers (columns) are the proportion of the subzone within the landscape unit (or interim landscape unit or portion of a forest development plan forming a contiguous geographic unit) that is identified as available for harvest (that is, not inoperable or in some sort of reserve status, such as riparian or protected 110  area). Y-axis numbers (rows) are the proportion of the available landscape (above) that has already been harvested without application of this guidebook's recommendations or similar prescriptions. Example: For each biogeoclimatic subzone in the landscape unit (Table 20(a)), calculate the area available for harvest (the X-axis). For example, i f 30% of the SBSmc area is available for harvest, then, using the 30% column, the recommended minimum proportion of each cutblock to be managed for wildlife trees is between 1 and 9%. If 50% of the available area has already been harvested without application of these or similar guidelines (Y-axis), then 5% of each new cutblock would need to be left for wildlife tree patches. This can be distributed adjacent to cutblocks in riparian or other long-term leave areas when feasible. Where landscape units have not been designated, the same calculation can be done using Table 20(b). Appendix-1 is from the 1995 British Columbia Ministry of Forest's Biodiversity Guidebook.  Ill  A P P E N D I X 3: TESTS FOR N O R M A L I T Y A N D H O M O G E N E I T Y OF V A R I A N C E FOR 10 H A B I T A T V A R I A B L E S WITHIN T R E A T M E N T S Shapiro-Wilk Normality Test P-Values: Control Treatment Variable P-Value Standing Live Conifers per Ha 0.1403 Standing Live Deciduous per Ha <0.000 Standing Dead Conifers per Ha 0.2236 Standing Dead Deciduous per Ha O.0001 Down Conifers per Ha <0.0001 Down Deciduous per Ha no results available Post-Harvest Basal Area 0.6216 3-D Shrub Cover Volume <0.0001 Post-Harvest C W D 0.0019 % Canopy Cover 0.1575 Significant P-Values (non-normality) are indicated in bold font.  Shapiro-Wilk Normality Test P-Values: Aggregate Treatment Variable P-Value Standing Live Conifers per Ha 0.0001 Standing Live Deciduous per Ha 0.0001 Standing Dead Conifers per Ha 0.0001 Standing Dead Deciduous per Ha 0.0001 Down Conifers per Ha 0.0001 Down Deciduous per Ha no results available Post-Harvest Basal Area 0.0001 3-D Shrub Cover Volume 0.0001 Post-Harvest C W D 0.0001 % Canopy Cover 0.0021 Significant P-Values (non-normality) are indicated in bold font. Shapiro-Wilk Normality Test P-Values: DispH Treatment Variable P-Value Standing Live Conifers per Ha 0.0001 Standing Live Deciduous per Ha 0.0001 Standing Dead Conifers per Ha 0.0001 Standing Dead Deciduous per Ha 0.0001 Down Conifers per Ha 0.0001 Down Deciduous per Ha results available no Post-Harvest Basal Area 0.0001 3-D Shrub Cover Volume 0.0001 Post-Harvest C W D 0.0001 % Canopy Cover 0.0001 Significant P-Values (non-normality) are indicated in bold font. 112  Shapiro-Wilk Normality Test P-Values: DispM Treatment Variable P-Value Standing Live Conifers per Ha <0.0001 Standing Live Deciduous per Ha <0.0001 Standing Dead Conifers per Ha <0.0001 Standing Dead Deciduous per Ha <0.0001 Down Conifers per Ha <0.0001 Down Deciduous per Ha no results available Post-Harvest Basal Area <0.0001 3-D Shrub Cover Volume <0.0001 Post-Harvest C W D O.OOOT % Canopy Cover <0.0001 Significant P-Values (non-normality) are indicated in bold font.  Shapiro-Wilk Normality Test P-Values: DispL Treatment Variable P-Value Standing Live Conifers per Ha <0.0001 Standing Live Deciduous per Ha <0.0001 Standing Dead Conifers per Ha <0.0001 Standing Dead Deciduous per Ha <0.0001 Down Conifers per Ha O.OOOT Down Deciduous per Ha no results available Post-Harvest Basal Area <0.0001 3-D Shrub Cover Volume <0.0001 Post-Harvest C W D O.0001 % Canopy Cover <0.0001 Significant P-Values (non-normality) are indicated in bold font.  Homogeneity of Variance Test P-Values of 10 Habitat Variables Variable  Standing Live Conifers per Ha Standing Live Deciduous per Ha Standing Dead Conifers per Ha Standing Dead Deciduous per Ha Down Conifers per Ha Down Deciduous per Ha Post-Harvest Basal Area 3-D Shrub Cover Volume Post-Harvest CWD % Canopy Cover  Levene's  O'Brien's  <.0001 0.0064 <.0001  <.0001 0.0063 <.0001  0.7828 0.0948 0.2207  0.7864 0.0989 0.2280  0.6651  <.0001  <.0001  <.0001 0.0190 <.0001 <.0001  0.1868  0.1923  <.0001 <.0001  <.0001 <.0001  Brown and Forsythe's <.0001 0.0238 <.0001 0.0004  0.2446  Bartlett's  Welch's  <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001  <.0001  Significant P-Values (non-homogeneous variance) are indicated in bold font.  113  0.0461 <.0001  0.3364 <.0001  0.1226 <.0001 <.0001 <.0001 <.0001  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0074996/manifest

Comment

Related Items