QUANTIFYING KEY METRICS OF ECOSYSTEM BIODIVERSITY IN NATURAL AND MANAGED SUB- BOREAL FORESTS OF BRITISH COLUMBIA by Colin E . Chisholm BSc. , University of Northern British Columbia , 2000 THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE IN NATURAL RESOURCES AND ENVIRONMENTAL STUDIES ( FORESTRY) UNIVERSITY OF NORTHERN BRITISH COLUMBIA April 2021 ©Colin E . Chisholm , 2021 APPROVALS PAGE Abstract Forest management in the central interior of British Columbia has been active for over a century. Industrial forest practices in the region are based on the premise that harvest and subsequent stands regeneration is sustainable, but recent investigations raise questions about long-term ecological sustainability and impacts on biodiversity. I evaluate here , using a chronosequence of forest stands , the impacts of stand harvest on biodiversity status and recovery. Aerial laser scanning is used to enhance analysis and model impacts spatially. I provide a novel assessment of key biodiversity metrics of diversity, richness , abundance, and modeling using linear discriminant analysis and random forest frameworks. Results show that vegetation community composition and coarse woody debris ( CWD ) , a key habitat for numerous taxa , are both impacted by harvest history. Predictive mapping of CWD provides insights and a further tool for decision makers to manage and ensure natural levels of CWD are maintained on the landscape . i Acknowledgements This project would not have been possible without the support and encouragement of the Aleza Lake Research Forest Society. Much thanks to Michael Jull who has been a mentor to me as a forester and as a forest scientist . Thanks to Samantha Gonzalez , Kailee MacKinnon , John Mainville , Alicja Muir , Rachelle Winsor , Keaton Freel , Saskia Hart , Marie Gamier , Autin Bartel , and Yuxiao Zhao for their assistance with data collection . Dr . Nicholas Coops and members of the UBC Integrated Remote Sensing Studio provided me time and space as I learned how to use lidar remote sensing data . Dr . Brian Menounos provided in- kind support for the acquisition and initial data processing of the lidar dataset . I am very thankful for the efforts of Dr . Che Elkin , who supervised , taught , challenged , and encouraged me throughout the entire project . I am grateful to the additional members of my examining committee: Dr . Roger Wheate, Dr . Hugues Massicotte , and John Pousette for their questions and review of this document . Financial support for the project was provided by the Sustainable Forestry Initiative and a UNBC research seed grant . Finally, I am grateful to my family, my partner in life Tina , our son Samuel , our parents Earle and Dorothy Chisholm , and Gordon and Mary Townsend for their support throughout this time of study and learning. ii Dedication I dedicate this work to my son Sam to encourage him to “ do hard things ” and to “ never stop learning ”. iii TABLE OF CONTENTS Abstract i Acknowledgements ii Dedication iii List of Figures vi List of Tables vi Chapter 1 Chapter 2 Chapter 3 Research background My motivation and central question Consideration of the issues 1 1 Project objectives 9 Long term impacts of forest harvesting on stand structure and vegetation in sub- boreal forests of British Columbia Introduction Methods Study location Sampling design and data collection Statistical processing Stand structure Species richness , diversity, and abundance Results Stand structure 2 Vegetation Discussion Conclusion Acknowledgements 11 11 13 13 15 16 16 16 18 18 22 25 29 30 Quantifying the recovery of coarse woody debris in British Columbia’s managed sub- boreal spruce forests Introduction Methods Study site Sampling design CWD field data collection 31 31 35 35 35 37 iv Chapter 4 Lidar analysis Results Empirical assessment of CWD Lidar detection of CWD Predictive mapping Predictive map efficacy Discussion Empirical differences Detection of CWD using lidar , and landscape management Management Opportunities Conclusion Acknowledgements 39 41 41 45 47 47 50 50 53 54 55 56 Synopsis 57 57 58 58 59 60 Introduction Forest disturbance and patterns of recovery Methods Indicators of forest recovery Recommendations Are we growing forests or growing trees? Bibliography Appendix A 61 62 Mean Species Abundance v 71 List of Figures 2.1 2.2 The Aleza Lake Research Forest highlighting the landscape units and stand development stages . Key metrics of stand structure with corresponding biodiversity metrics of understory vegetation comparing stand development stage and landscape 14 19 units . 2 2.3 Basal area ( m per hectare ) of species by stands development stage and landscape unit . 2.4 Basal area by diameter class . 2.5 Hellinger-transformed abundance of functional groups . 2.6 Linear discriminant analyses to determine if understory vegetation can be used to differentiate logging history. Map of the 9 , 000 hectare Aleza Lake Research Forest . One hectare plot layout including four 400m 2 sub- plots and associated 30 metre CWD transects . 3.3 Surface model highlighting CWD generated from lidar data using a 1.3m height cut-off . 3.4 Empirical metrics of CWD including volume , piece count , and decay class diversity. 3.5 Decay class abundance by stand development stage . 3.6 The comparison of detection rates of CWD from lidar demonstrates that larger pieces of CWD are more likely to be detected . 3.7 Observed vs. predicted coarse woody debris volume from leave-one-out validation from the random forest regression model . 3.8 Covariate importance in the Random Forest Model . 3.9 Predictive map of CWD across the Aleza Lake Research Forest . 3.10 Average predicted CWD volume within forest inventory polygons. 3.1 3.2 vi 20 21 23 24 36 38 40 42 44 45 47 48 49 50 List of Tables 2.1 Stand development stages examined in this study. 2.2 Summary of all linear discriminant analyses. 15 23 3.1 Summary of empirical CWD volume , piece count , and Shannon’s diversity index ( H ) . 3.2 Detection rates of CWD by diameter class across all sites treatments . 3.3 Detection rates of CWD by decay class. 3.4 Detection rates of CWD by treatment . 43 46 46 46 A. l Mean Hellinger-transformed species abundance from linear discriminant analysis. 72 vii Chapter 1 Research background My motivation and central question As an introduction for the motivation to undertake this research , I provide this personal reflection from my experience of working in the forests. I am a practicing forester and am accredited as a Registered Professional Forester with the Association of British Columbia Forest Professionals. Much of my career has focused on the regeneration of forests following clear-cut harvest , including management of operations to prepare sites for planting , overseeing the planting of millions of trees , and stand tending including competing vegetation management after planting . These activities took me across much of the Central Interior of British Columbia , from 100 Mile House north to Fort Ware , and Fraser Lake and east into Alberta . In Alberta , these activities extended from Ivananaskis north to Peace River and west to the provincial border . The story of forestry practices across all these regions is generally the same : harvest of old natural forest and regenerate planted trees. Through my career one question has continually returned to my mind : are we growing forests or are we simply growing trees? The following research has provided an opportunity to explore this question and present quantitative answers to the impact and patterns of recovery of forests on the landscape through the examination of key metrics of biodiversity, following harvest and planting . 1 Consideration of the issues The conservation of biodiversity in forested lands is critically important for the long-term sustainable management of forest ecosystems ( Gao et ah , 2014 ; Harrison et ah , 2014 ) . Old- growth forests , those that are old in age and free from human disturbance, are frequently identified as areas with high biodiversity value ( Kane et ah , 2010a ; Mladenoff et ah , 1993) . This richness in biodiversity is assumed to derive from the fact that these areas are highly complex systems ( Filotas et ah , 2014; Messier et ah , 2013) though attempts to quantify forest complexity to support the ecological theory that forests are complex adaptive systems is highly challenging ( Parrott , 2010 ) . The oldest forest stands , generally those with the largest trees , are assumed to have emergent properties that support a greater variety of species in addition to species that require an old forest state ( Burton , 2013; Levin , 1998 ) . Key physical attributes that make up the old-growth forests , characterizing the physical structures that make them unique, include higher levels of large dead wood in the form of coarse woody debris ( CWD ) ( Harmon et al . , 1986b ; Stokland et ah , 2012 ) , standing snags ( Spies et ah , 1988 ) , and live stand characteristics including canopy gaps ( DeLong et ah , 2003) , high variability in canopy shape , and old large trees ( Kane et ah , 2010a ) . With the increasing availability of 3-dimensional remote sensing data from lidar , opportunities to assess a plethora of forest metrics across whole landscapes are now accessible ( Ahmed et ah , 2015; Coops et ah , 2016; Filotas et ah , 2014; Gao et ah , 2014 ) ; it is expected that this novel remote sensing data , paired with empirical data , will provide new insights into characterizing and quantifying forest structures that are indicative of biodiversity and forest ecosystem complexity. For decades there has been the recognition that forests provide a richness in biodiversity and that they should be managed for numerous values or ecosystem services ( Franklin , 1989 ; Harrison et ah , 2014 ) . Cardinale et ah ( 2012 ) provides a succinct definition : 2 “ Biodiversity is the variety of life , including variation among genes , species and functional traits .” Examining biodiversity is critically important when considering ecosystem services to humanity as well as ecosystem function . Organisms are known to influence their own local environment and generally, with increases in biodiversity, greater biomass production and nutrient cycling are afforded . Additionally, biologically diverse systems are considered to be more stable while remaining dynamic providing for higher ecosystem resilience ( Cardinale et ah , 2012 ; Parrott and Lange , 2013) . While recognizing the importance of biodiversity, governments, non-governmental organizations , as well as industrial resource- based companies make commitments to the conservation of biodiversity. For example , Canadian federal law provides for the conservation of Biodiversity through The Species at Risk Act , the Canadian Environmental Protection Act , the Migratory Birds Convention Act , and the Canada Wildlife Act . In addition , numerous voluntary forest certification systems or organizations market the conservation of biodiversity as a priority and companies that manufacture products from forest harvest include these certification standards into their business model . Gulbrandsen ( 2004 ) reviews the potential of forest certification regimes to ‘fill the gaps ’ in legislation , though the effectiveness of these systems is debated ( Castka , 2016 ; Gullison , 2003; Kalonga et ah , 2016 ) . In British Columbia , Canada forest lands are dominantly publicly owned ( 95% ) and managed by the government ( Canadian Council of Forest Ministers , 2017 ) . The conservation of biodiversity is given high priority and is one of the resource values protected under the Forest and Range Practices Act . Of particular importance for landscape level planning is the establishment of Biodiversity Orders which are set at provincial and regional scales , that is Forest Districts or Timber Supply Areas ; these ‘ orders ’ issued under the Government Action Regulation set minimum old forest target levels ( MSRM , 2004 ) . A specific example of the Aleza Lake Research Forest has government targets set to conserve 30 percent of the forest lands in an old forest state ( ILMB , 2009 ) . 3 Forest ecosystems are theorized to be complex adaptive systems ( CAS , Levin , 1998; Parrott , 2010 ) . This concept comes from “ Complexity Systems Science ” , originally from the fields of non-linear physics and information theory, and is being used to understand and describe ecosystems - even suggesting that it could unify the study of global forests providing a common understanding for all forest types from tropical to boreal ( Filotas et al . , 2014) . Forest ecosystems are excellent examples of complex systems as they are heterogeneous , self-organizing , dynamically optimizing their internal components and processes and thereby attaining maximum complexity. This ‘ maximum-complexity ’ is considered to be an emergent and distinct attribute of an ecosystem ( Messier ct ah , 2013) . Forest disturbance occurs both naturally and from anthropogenic sources and varies in scale and severity. Large scale disturbances occur in the 1000 to 100 , 000s of hectare range and can include impacts by fire or insect outbreaks ( Burton , 2013) . These large scale disturbances provide a “ legacy of structure ” in the forms of standing dead trees burnt or otherwise , an influx of CWD , and small patches of undisturbed or moderately disturbed refugia ; long return disturbance intervals occur in the 300-1000 year range ( Franklin et ah , 2002 ; Spies et ah , 1988) . In periods between major disturbances , smaller scale events occur creating canopy gap dynamics. For example , wind events may uproot or snap small clusters of trees often weakened by fungal pathogens or rots . These small scale natural disturbances also leave high levels of physical structures as a legacy. In contrast , conventional timber harvesting methods generally remove most of the biomass from a site , reducing structures that may be essential for ecosystem development . In landscapes with a mosaic of old natural , “ naturally disturbed ” and “ management disturbed ” stands , it is important to consider the complex dynamics occurring . CAS theory suggests that naturally disturbed stands experience shift from a state of theoretical maximum complexity to one of increased disorder . While managed stands , in particular those that are artificially regenerated and managed for a particular tree species , move towards a highly ordered or homogenized system ( Messier et ah , 2013) . Following these 4 disturbances , forest systems will begin returning to a state of maximum complexity. However , it is unclear if there are different ecological processes at work between the natural and managed stands and if they are moving towards similar or unique maximum diversity states . While concepts of the CAS theory may be indicative of ecosystem health and resilience , studies examining this theory are limited ( Parrott , 2010 ) . These central and related conceptual pillars forest as complex adaptive systems and - biodiversity- are useful but represent thousands of species , their interactions and their habitats . It is neither feasible nor practical to quantify in its entirety the biodiversity at work and to monitor the development of forest complexity. However , it is possible to examine key metrics of biodiversity and to examine if these are indicative of CAS . At large scales , climate and natural disturbance patterns compose the major drivers of biodiversity and the processes driving forest complexity. Areas with high energy input are ( e . g . equatorial / tropical sites ) more diverse than those at higher latitudes ( Kimmins , 1996 ) . The availability or access to resources needed for growth , reproduction , and the ultimate survival of an organism within a given ecology depend on numerous factors. In a very broad sense , this can be placed into two categories : a ) sustenance - sources of energy and nutrients essential for survival and b ) a safe space to live and reproduce. These two categories are highly dependent on the structures that form habitats and the climate that surround the habitat . What then are the essential structures driving ecological processes? Soils form a substrate: a safe place for numerous organisms to further colonize and develop . The local topography of a site will influence the availability of nutrients at any given site . Water flows from ridge tops to depressions or valley bottoms and consequently, this local topological structure has a strong influence on the water availability for organisms. Trees that form the forest stands grow to maturity and provide incredible opportunities for vertical structure that may enhance habitat for some organisms while reducing available resource availability for others ( e .g . creating elevated structures for birds while reducing light availability in the lower canopy ) . It is this variety of physical structures within forests 5 that provide an array of diverse habitats, microhabitats and niche spaces for organisms to interact and live ( complexity and biodiversity ) . The examination of vegetation provides a direct measure of diversity for one kingdom of organisms. Plant species diversity has also been found to correlate well with overall diversity, as demonstrated by Gao et al . ( 2014 ) . Understanding that plant communities are both a measure of one group of organism diversity and an indicator of overall biodiversity, one must consider how plant communities function in forested ecosystems . These systems are complex not simply in the sense of many processes happening simultaneously but in the sense of complexity theory described above . Plant communities develop and in many ways the ‘ whole exceeds the sum of their parts ’. McElhinny et al . ( 2005 ) provide a review of physical structural complexity and suggest a definition and recommendations for measuring it . Their work focuses on the ability to use structure as an indicator of biodiversity at the forest stand scale with the ultimate goal of generating a metric that could be used to classify various forest types. Forest structural characteristics they suggest as important include: canopy cover , understorv vegetation , deadwood , and forest tree sizes , abundance and distributes . However , it should be noted that , for many of the attributes listed ( a list generated from their literature review ) , links to biodiversity were assumed with few studies giving definitive linkages. As such , there seems to be a general sense in the scientific community that forest structure is strongly linked to biodiversity, yet the authors themselves suggest that this assumption may need significant substantiation . Harmon et al. ( 1986b ) provided a thorough examination of CWD in the Pacific Northwest : it highlighted the use of CWD by numerous organisms as well as examined sources of CWD and rates of decomposition . More recently, Stokland et al . ( 2012 ) provided a comprehensive exploration of the ecological role played by dead wood structures in ecosystems with a focus on biodiversity. The authors demonstrate the importance of dead 6 wood to the conservation of biodiversity, estimating that between 400 , 000 and a million different species use these structures for habitat or food sources . The contribution of dead wood to biodiversity is striking. The sources of dead wood , that is the recruitment of these materials, are dependent on the ecological processes that cause individual trees to die . Sources of CWD come from two pools : the pre-disturbance stand ( i . e. pre-fire / stand replacement disturbance ) and from within an existing stand ( Spies et ah , 1988) . Generally, following a major disturbance (e.g . fire , clear-cut harvest ) , some remnant pieces may provide large individual pieces that begin a slow decay. This is more distinctly the case following a fire where , although much of the material may have burnt , remaining dead trees will provide CWD . These large pieces will decay relatively slowly, given their low surface to volume ratios . In contrast , in maturing forest , the recruitment of CWD significantly depends on forest stand dynamics: patterns of growth and mortality of trees . Following a stand-replacing disturbance, forest re-establishes ( naturally or through management intervention ) and as the young trees mature , they compete with one another ( e . g. at crown closure ) , entering a stem-exclusion phase where intra-tree competition will cause some individuals to die as they are outcompeted for resources. The dead stems are relatively small as these stands have not yet matured , they have a higher surface to volume ratio and will decay relatively quickly - thereby contributing CWD for a shorter period . For example , second growth plantations in the central interior of British Columbia will enter a stem exclusion phase at approximately 30 years of age and individuals may only be 15 - 20cm diameter at breast height ( author ’s professional experience ) . At later ages , through ongoing stem exclusion processes or additional impacts by wind , pathogens, and localized insect attacks , large trees will be recruited to CWD . These large structures provide high quality habitat especially for larger organisms that require larger sites for roosting and cavity nesting . Additionally, large dead wood takes longer to decay given low surface to volume ratios and thereby provide relatively long lasting structure . Parrott ( 2010 ) suggests the following approaches to measure or examine complexity 7 in forest ecosystems: ( a ) Temporal Measures - change of state over time ; ( b ) Spatial Measures - describe the configuration of the system at a point in time ( raster based data from remote sensing or modelling is often used here ) ; ( c ) Spatiotemporal measures - involving 3-dimensional modeling of information layers and time ; and ( d ) Structural Measures ( Ecological topology ) - describing the organization and relationships in the system. Aerial Laser Scanning ( ALS ) / Light detection and ranging ( lidar ) is a remote sensing technique that provides 3-dimensional point cloud data of the area surveyed . This system is an active sensor , usually mounted to either a helicopter or airplane , emitting lasers towards the ground and capturing a return signal that provides a datapoint with a realworld coordinate ( easting , northing , and elevation ) with sub- metre accuracy. Additional data includes signal intensity, and meta-data ( e . g . time , signal return sequence , flight path identifier ) . Numerous models are generated from this type of data - two of the most common are digital terrain models and crown height models ( White et ah , 2016 ) . The enhancement of traditional forest inventories through the use of ALS / lidar has been well documented ( Wuldcr et ah , 2008a , 2013) . Forest inventory modelling is traditionally done through airphoto interpretation where human operators evaluate stereo images to identify similar forest stands and provide general forest attributes based . In contrast , lidar provides opportunities to directly measure forest stand in an automated system . Metrics from lidar point cloud data are highly correlated with numerous forest timber metrics including: tree height , stand height , basal area , and volume ( White et ah , 2016 ) . More recently, studies have begun to examine the use of lidar for mapping areas for biodiversity monitoring. Guo et ah ( 2017) examined the capacity of lidar to regionally map vegetation structure for biodiversity monitoring . This research team was able to automatically classify nine distinct forest types based across the forested ecologies of Alberta . In a companion study, a forest structure habitat index and model of avian species 8 richness was developed ( Coops et al . , 2016 ) . Additional studies have examined structural complexity ( Kane et al. , 2010b , a ) , correlated lidar metrics for avian ( Melin et al. , 2016 b; Vauhkonen and Imponen , 2016 ) and moose ( Melin et al . , 2016a ) habitat . Another study has examined similarities and differences between old natural stands and mature second growth stands (Sverdrup-Thygeson et al. , 2016 ) . A discussion paper has been put forward to examine the use of lidar data for the classification of ecosystems ( Campbell et al. , 2017) . Regarding CWD there has been some analysis using lidar to indicate fire fuels loading ( Kramer et al . , 2014 ) and gap analysis (Tanhuanpaa et al . , 2015) . Following a review of literature to date , the following gaps or areas for future research are noted: • Attempts to quantify forest complexity supporting ecosystem as complex adaptive systems are limited . With the increasing availability of lidar datasets , opportunities to assess a plethora of forest metrics now exist which may provide insights into quantifying forest complexity and biodiversity. • Regional landscapes (sub- boreal plateau ) : sub- boreal plateau landscapes are subtle . Subtle in the sense that very minor changes in topography can create very significant differences in ecology both in how an ecology presents as well as the underlying processes in a forest stand . • In the central interior of British Columbia , there are no known examinations of forest ecology using lidar . • Remote sensing of coarse woody debris is very limited . Project objectives Using indicators of biodiversity and forest ecosystem complexity, including empirical and remotely sensed metrics (e.g . forest stand , live tree , snags , coarse woody debris 9 and understory vegetation ) , this thesis will examine and provide indices to quantify biodiversity and forest ecological complexity. The following chapters address gaps in the literature and are written as stand-alonepapers that are intended for publication . Chapter 2 examines the response of vegetation communities across a chronosequence of forest stands that encompass examples from the entire history of industrial forest harvesting in the Central Interior of British Columbia . Key metrics from this study include using measures of biodiversity and quantitative ecology ( note that lidar data is not used ) . Chapter 3 examines coarse woody debris ( CWD ) as a key indicator of biodiversity and ecosystem processes . Comparing empirical and remotely sensed lidar data , it is possible to quantify and characterize CWD . Questions around the ability to detect and characterize CWD , and any limitations for doing so are answered . Additionally, the examination of different stand histories provides distinct insights into the impacts of disturbance history on CWD quantity and quality indicative of biodiversity. 10 Chapter 2 Long term impacts of forest harvesting on stand structure and vegetation in sub- boreal forests of British Columbia Introduction Forest management in British Columbia is thought to be sustainable. However , recent publications and reports highlight concerns that forest management may be having long- term impacts on forest biodiversity and threaten the ecological sustainability of this industry ( Price ct ah , 2020; Gorley and Merkel , 2020 ) . Literature highlights that long term impacts of forest harvesting in the Canadian context are not fully known given the relatively short age of the forest industry ( Venier et ah , 2014; Hart and Chen , 2006 ) . Our study provides insight in the long term impacts of forest harvesting in the sub- boreal forests of British Columbia comparing key biodiversity metrics of vegetation across a sequence of managed stand development stages to natural old-growth . DeLong ( 2007) highlights how the Province of British Columbia has developed harvest patterns similar to wildfire disturbance in an effort to mimic patch size and serai stage distributions of historical wildfires across a range of ecologies ( e . g . short-fire return interval to long fire return interval forest types ) . “ In effect , wildfire is the disturbance process we are generally attempting to 11 replace with harvesting.” - DeLong ( 2007 ) However , recent studies emphasize that clear-cut harvesting does not fully mimic natural ecological processes . DeLong ( 2007) had recognized that harvesting would not mimic wildfire in many aspects but creating the patterns of disturbance would address other forest management issues ( e.g . excessive road building and forest fragmentation ) . Of particular concern is that disturbance by wildfire is primarily chemical disturbance where in contrast , forest harvesting is dominantly a physical disturbance; these key differences result in different site conditions and successional pathways of vegetation and concern that biodiversity is impacted . ( Hart and Chen , 2006 , Venier et al . ( 2014 ) ) . Similarly, in their study in Alberta , Macdonald and Fenniak ( 2007) also found that vegetation community composition were likewise altered across a variety of partial harvest systems suggesting that impacts may not be limited to clear-cut harvesting. Metrics of plant species diversity provides a direct measure of biodiversity for that kingdom and are considered indicative of total biodiversity ( Gao et al . , 2014 ) . Boutin et al . ( 2009 ) suggests specific means for measuring biodiversity in plants which include trends in the abundance and distribution of species . In this study we examine the response of past harvesting disturbance on forest plant communities to examine how they respond across a sequence of stand development stages from juvenile , immature , and mature sites , and how these plant communities compare with natural old-growth forests . Key questions examined include: 1. Are there measurable losses or changes in biodiversity. 2 . How does stand development stage impact the abundance , richness , and diversity of vegetation . 3. Do managed stands become natural with time. 12 Methods Study location Investigation took place at the Aleza Lake Research Forest ( ALRF , V.54°03.8' , IF.122°04.3/ ) . Forest stands within the ALRF are coniferous-leading dominated by hybrid spruce ( Picea glauca ( Moench ) Voss x engelmannii Parry ex Engelm . ) , and sub-alpine fir ( Abies lasiocarpa ( Hook . ) Nutt .) , and small dispersed numbers of douglas fir ( Pseudotsuga menziesii ( Mirb . ) Franco) and western hemlock ( Tsuga heterophylla ( Raf . ) Sarg. ) . Deciduous tree species include paper birch ( Betula papyrifera Marsh . ) , trembling aspen ( Populus tremuloides Michx.) , and cottonwood ( Populus trichocarpa Torr . & A . Gray ) . The birch is a common but minor portion of the conifer-leading stands while the poplar species may form pure stands or have dispersed individuals in the common conifer-leading stands . The area is within the sub-boreal spruce wet-cool ecological unit ( DeLong , 2003) . Soils are fine textured lacustrine ( DeLong et al. , 2003) . The ALRF was established as a research site in 1924 with a key objective of demonstrating sustained yield forestry. Since then it has been actively managed for timber harvesting amongst other research endeavours. Pedersen ( 2003) describes how various harvest systems were used historically throughout British Columbia highlighting that thin-from-above harvest systems ( i .e . intermediate utilization ) were used in the first half of the century and focused on targeting larger diameter trees , with a transition in the 1960s to clear-cut harvesting which remains the dominant harvest system to today. The ALRF generally followed this pattern with with the earlier harvest systems focused on the removal of large spruce ( Aleza Lake Research Forest Society , 2019 ) . 13 555000 560000 565000 Figure 2.1: The Aleza Lake Research Forest highlighting the landscape units and stand development stages . 14 Table 2.1: Stand development stages examined in this study. Class Name juv imm Juvenile Immature Mature Old-growth mat OG wOG Height Range ( m ) Managed Forest 5 - 19.9 20 - 29 > 29 > 29 20 - 29 Yes Yes Yes No No Forested Wetlands Sampling design and data collection The research forest was divided into two main landscape units ( see Figure 2.1) . The Knolls , in the northern portion of the forest , is gently rolling with slopes ranging between 8 and 24 % (1st and 3rd quartile) . The Plateau unit is distinctly flatter with slopes ranging from 0 to 9% . Both landscape units are drained by deeply incised gullies . Sets of sample plots of lha were established across each landscape unit using stratified random sampling. Stratification was based on wall-to-wall lidar metrics to place the forest into height classes of relative stand maturity ( table 2.1) . Height classes were based on minimum height thresholds. Heights were extracted from a lm 2 resolution lidar derived crown height model ; a minimum of 1% the pixels had to exceed the height threshold to be considered within a given class. Known harvest history polygons were used to further stratify the forests between managed and natural stands . Within each plot , 4 nested plot centres were established to collect the following : tree species identity and diameter at breast height ( dbh ) : for trees larger than lOcnr dbh , a circular 200m 2 was used , and a 50m 2 plot was used for smaller trees : 2.0 - 9.9cm dbh . Vegetation identity and abundance was measured by pooling the data from two lm2 quadrats located at four metres north and four metres south of nested plot centre . All plant species within the quadrat were identified including an ocular estimate of percent cover . 15 Statistical processing Data processing and statistical analyses were completed in R version 4.02 ( R Core Team , 2020 ) . Statistical processing for species richness , abundance , and diversity used the vegan package ( Oksanen et ah , 2019 ) . Linear discriminant analyses were completed following methods in Borcard et al. ( 2018 ) using the R libraries vegan ( Oksanen et ah , 2019 ) , MASS ( Venables and Ripley , 2002 ) , and additional scripting provided by Borcard et ah ( 2018) . Stand structure Forest stand structure was evaluated for each treatment comparing basal area by : total , leading species , and diameter class . Tree density was standardized to total stems per hectare . ANOVA was used to compare results across the stand development stages ( SDS ) and landscape units ( i . e . the treatments ) . Species richness , diversity, and abundance The biodiversity of understory vegetation was examined to compare the differences between the SDS and landform influences ( i . e . the treatments ) . Key metrics of richness , abundance ( i.e . percent cover ) , and diversity were generated for all species which were compared using ANOVA . Abundance of vegetation communities was compared by functional groups ( bryophytes , forbs , shrubs , grasses , and lichen ) and further by abundance of individual species. Abundance metrics used the Hellinger transformation ( Legendre and Gallagher , 2001) . To compare functional group abundance , a MANOVA framework was used as per Warton and Hudson ( 2004) and more recently Hart et ah ( 2019 ) . To highlight specific differences and the potential drivers of differences , linear discriminant analyses ( LDA ) were completed . LDA analysis examined if SDS could be determined using vegetation functional group abundance and further by individual species abundances. A series of LDAs using individual species were completed using Hellinger-transformed abundance 16 of observed species . Four LDA runs were made with a minimum of 5 , 10 , 20 , and 40 species observations ; this was done so that unique observations would not dominate the discrimination process . Further , results were leave-one-out validated as recommended by Borcard et al. ( 2018 ) . Summary tabular results are provided for each of the LDAs , while detailed results for the LDA using 40-observation are presented here . The Hellinger transformation is the square root of the proportional abundance within a given site where i is the index of the site and j is the index of the species ( i.e . where the data is organized by sites in rows and species in columns) . As such Uij is the abundance of the individual species within a site , and ]j ; is the total abundance within the one site or rowa . %= ' yji Vi “ As described by Legendre and Gallagher ( 2001) 17 Results Stand structure Initially, following harvest ( i . e . juvenile stage ) , total stems per hectare was elevated while total basal area was reduced . Stems per hectare decreases with SDS with both landscape units following similar patterns ( F (4 , 181 )= 19.378 , p = 0.000 ) . Total basal area recovers to natural ranges ( i .e. compared to old-growth ) by the immature or mature SDS ; in addition , the pattern of basal area recovery differed between the landscape units ( Figure 2.2 ) . In the Knolls landscape units the immature , mature , and old-growth stands all had similar basal areas. In the Plateau the pattern of basal area recovery differed with a pattern of basal area increasing from juvenile through to mature and reduced basal area in the old-growth sites. Figure 2.2 highlights the significant differences in basal area . ANOVA demonstrated that these differences were a result of SDS , landscape , and the SDS-landscape interactions ( F ( 3, 183) = 3.667, p = 0.013 ) . Coefficient of basal area variance indicated significant difference based on SDS with the immature having higher variability than all others ( F (4 , 39 ) = 4 - 445, p = 0.005 ) Tree species richness and diversity did not show significant differences across the treatments. However , differences in species composition by basal area indicated that the dominant leading species changed over this chronosequence . On average juvenile stands were spruce leading , immature sites were sub-alpine fir leading , while mature and old-growth sites were spruce leading ( Figure 2.3) . Figure 2.4 highlights the increase in frequency of larger diameter class trees . Old-growth sites have the largest individuals with trees in the 90 and > 100cm dbh classes . 18 Overstory structure and understory biodiversity metrics 100 Knolls - Plateau ^ 50 75 25 4» - 5000 4000 3000 2000 0 1000 t > t f - 0- 250 200 150 - 100 - 50 - 0 - 10 - #o 1 ^ juv 4 T+ ^^ s irnm mat O'G 6G juv W irnm I ^ mat I O'G wd)G Stand Structural Stage Figure 2.2 : Key metrics of stand structure with corresponding biodiversity metrics of understory vegetation comparing stand development stage and landscape units. Total basal area ( m2 per hectare ) recovers to natural levels . Likewise stems per hectare decrease over time towards natural levels. Vegetation response, using total abundance ( percent cover ) and total species richness ( number of unique species ) , shows little response. 19 Basal Area of Species imm juv mat OG wOG 60 50 40 Knolls 30 2 m per hectare 20 10 0 60 50 YLL^ A 40 Plateau 30 20 10 0 - ! 4 n 4-r T Sx Bl Ep Fd C A Sx Bl Ep Fd C A Sx Bl Ep Fd C A Sx Bl Ep Fd C A Sx Bl Ep Fd C A Figure 2.3: Basal area ( m2 per hectare ) of species by stands development stage and landscape unit . Species include spruce ( Sx ) , sub-alpine fir ( Bl ) , birch ( Ep ) , other conifers ( C ) , and poplars ( A ) . 20 Diameter class basal area by SDS and landscape unit Points are the basal area from individual sub−plots juv imm mat OG wOG 60 40 60 i 40 Jl i * Jig§1 ii ! i!8 i If 10 20 30 40 50 60 70 80 90 100+ 10 20 30 40 50 60 70 80 90 100+ ii t hi !• P 20 • • i* 10 20 30 40 50 60 70 80 90 100+ !•: & !! Is# :.... 10 20 30 40 50 60 70 80 90 100+ »;• ! i*l; 10 20 30 40 50 60 70 80 90 100+ 2 . I ** * K 20 m per hectare . •• Diamter class (cm at breast height (1.3m) Figure 2.4 : Basal area by diameter class. Each point represents the basal area contribution within each sub- plot . OG sites are the only sites with trees in the 90 and 100cm dbh classes . 21 Vegetation No significant differences were seen in total species diversity while a few differences are noted in the total abundance and richness - specific differences were determined through Tukey post hoc analysis ( Figure 2.2 ) . Juvenile stands in the Plateau had lower total abundance than the old-growth and mature sites of both landscape units , and the immature and forested wetland sites in the Plateau ( F (4 , 176 ) — 4 - 910 , p — 0.001 ) . Two significant differences in species richness were noted with juvenile sites in the Knolls unit had higher richness while the forested wetlands sites had lower richness ( F ( 3, 176 ) = 3.631 , p = 0.014 ) . Abundance by functional group MANOVA analysis of Hellinger-transfornred functional group abundance demonstrated that there are significant differences in the understory vegetation cover by SDS ( F (4 , 179 ) . = 5.454 , p = 0.000 ) , though no differences were attributed to the landscape units . Figure 2.5 highlights that in the juvenile stages , bryophyte cover was reduced . Linear discriminant analysis was used to determine if stand development stages could be separated ; results of the LDA are provided in Figure 2.6 with A1 providing the biplot and A 2 providing the confusion matrix . The biplot suggests high levels of confusion when using functional groups to determine stand development stage . The confusion matrix highlights that the juvenile sites are classified with an accuracy of 73% ; however , accuracy for other stand development stages is low . Species abundance A series of LDA were completed to determine if the abundance of individual species could be used to classify the sites into SDS . Table 2.2 provides a summary of the LDA results . The more species used in the LDA , the higher the accuracy of classifying the sites correctly. However , this was at a cost of higher mean-misclassification error ( nrmce ) in 22 Hellinger transformed abundance 0.75 0.50 CD £ - 0.25 - -a 5 0.00 - 0.75 - 0.50 - 0.25 - 0.00 - B F S G L B F S G L B F S G L B F S L G B F S G L Functional Group Figure 2.5 : Hellinger-transformed abundance of functional groups arranged by stand development stage and landscape unit . Functional groups include Bryophytes ( B ) , Forbes ( F ) , Shrubs ( S ) , Grasses ( G ) , and Lichens ( L ) . Table 2.2 : Summary of all linear discriminant analyses. When more species are used accuracy increases ; however there is little change to the mean misclassification error ( mince ) . Accuracy Min. Observations Species Observed FuG 5 10 20 40 all grouped 98 74 58 39 Percent of total Species juv imm mat OG wOG all 0.73 0.98 0.95 0.86 0.86 0.47 0.9 G 0.93 0.78 0.76 0.52 0.93 0.80 0.82 0.70 0.00 0.94 0.84 0.72 0.72 59% 45% 35% 23% 23 0.45 1.00 1.00 0.95 0.95 mince cv-mmce 0.57 0.04 0.09 0.17 0.20 0.60 0.43 0.44 0.45 0.47 A1 LDA ordination using functional groups B1 LDA ordination using species S* OG wOG LD1 ( 76.18%) A2 juv imm mat OG wOG juv 32 0 0 juv 38 5 1 0 0 juv 28 7 0 9 21 12 0 3 imm 2 34 2 3 4 imm 6 24 5 maf 8 11 23 0 2 mat 2 4 31 6 1 mat 4 7 19 12 OG 8 13 10 0 1 OG 0 2 7 23 0 OG 0 5 12 15 0 wOG 0 6 5 0 9 wOG 0 0 0 19 wOG 1 4 1 2 3 0 2 Figure 2.6 : Linear discriminant analyses to determine if understory vegetation can be used to differentiate logging history. Site classification using vegetation functional groups including biplot ( Af ) and confusion matrix ( A 2 ) . Classification of the sites using species with a minimum of 40 observations including biplot ( Bf ) , confusion matrix ( B2 ) , and leave-on-out validated confusion matrix ( B3) 24 the leave-one-out validated models. All LDA runs demonstrate that sites can be classified correctly the majority of the time . In Figure 2.6 , B1- B3 provide LDA results for the run using species that were observed within a minimum of 40 individual quadrats. The biplot shows graphically good separation of the juvenile and forested wetlands sites . This plot graphically displays the discrimination of the SDS with a total of 83.7% of the variance explained ( LD1: 55.3% and LD 2 : 28.4 % ) . Despite not being graphically separated , the confusion matrix ( B2 ) highlights that immature , mature, old-growth , and forested wetlands are also accurately classified . B3 provides the leave-one-out validated confusion matrix highlighting that strong separation remains for all SDS . Old-growth shows the lowest accuracy ( 65% ) and is most often confused with mature and then immature sites . Species mean Hellinger transformed abundance are listed in descending order of influence in Appendix A . l . Discussion No differences in biodiversity metrics were associated with the trees ( i . e. Shannon’s H\ and richness ) ; however , there are distinct changes in stand structure . Stand structure changes from initial harvest disturbance through to maturity follows well established stand dynamics. Initial stratification of the stands types was based on differences in height ; total basal area increased with stand development with full site occupancy developing in the immature stage ; and total tree density fell from the youngest to the oldest sites. These changes in the physical characteristics of the stand can affect light and moisture , and influence understory plant communities ( Venier et al . , 2014 ) . Importantly is the fact that mature and old-growth are indistinguishable from each other using the stand structure metrics selected here . One component of stand structure that was not examined here were the dynamics of dead wood , namely : standing snags and coarse woody debris ( CWD ) may separate managed from unmanaged stands (see Chapter 3 for an examination 25 of CWD ) . In their review of stand structure dynamics Brassard and Chen ( 2006 ) found that CWD was reduced in managed stands and that this could potentially be detrimental to biodiversity. Vegetation response, based on total abundance , richness , and diversity showed few differences. Total species richness was highest in the juvenile sites in the Knolls, while seemingly contradictory, vegetation cover was lowest in the juvenile Plateau . Reduced cover may have been a result of closed canopy often associated with earlier stand development stages , and this would be consistent with other studies ( Kumar et ah , 2018b ) . However , this appears to be in conflict with higher cover and significantly higher species richness found in the Knolls juvenile sites , a point consistent with Venier et al . ( 2014) who found vascular plant diversity increased following harvesting. Significantly lower species richness was also found in the wOG sites . This is believed to be a response to the different site conditions and productivity of these forested wetlands. Further examination of vegetation response , using abundance by functional group , does clarify that juvenile sites are distinct from the other stand types. The juvenile is classified correctly 70 % of the time ( based on Table A 2 in Figure 2.6 ) . In the juvenile sites bryophyte cover was lowest and shrub cover was highest . This is consistent with Kumar et al . ( 2018b ) and Venier et al . ( 2014 ) who found that increased shrub cover resulted in the suppression of non-vascular plants ( e . g . bryophytes) . Kumar et al. ( 2018b ) further suggests that this may be a function deciduous leaf little causing physical inhibition or an allelopathic effect on the non-vascular species. A challenge to using the functional group analysis is that there is no separation of the other stand types . Examination of abundance using species , as opposed to functional groups , highlights that there are differences between all of the stand development stages ( Figure 2.6 B1- B3 ) . The juvenile and forested wetland sites are the most distinct with accuracy in excess of 85% and 95% respectively. The immature , mature , and old-growth sites are also accurately 26 classified at a minimum of 70% and as high as 93% (see table 2.2 ) . These results highlight that differences in disturbance history and differences in local , site levels , environmental conditions are measurable and likely long lasting . When classification errors does occur it is most often associated with sites in neighbouring stand development stages . This suggests that there is a progression of vegetation community composition from early to late development stage . Using species to differentiate the sites is more effective than using functional groups as species have differing ecological niches. Specifically, species abundance in response to disturbance history is influenced by light and moisture availability ( Barbier et ah , 2008; Kumar et ah , 2018b ) . For example , two shrubs Lonicera involucrata ( black twinberry ) and Oplopanax horridus ( Devil’s club ) , are respectively shade intolerant and shade tolerant with L . involucrata most abundant in the juvenile sites and O . horridus abundant in the mature and old-growth sites. Similarly for the forbs group , the ferns Dryopteris expansa and Gymnocarpium disjunctum are shade tolerant and most abundant in mature and old-growth sites ; in comparison , the Galium spp ( Bedstraw ) are more abundant in the juvenile sites . Differences in the bryophytes highlight ecological differences Sphagnum moss , a hydrophylic species , is most abundant in the forested wetlands with little to no representation in other sites. Beaudry et al . (1999 ) or Klinkeuberg ( 2021) provide descriptions of specific species . Cross validation highlights that the LDA results are robust demonstrating similar trends to the raw results ; sites are classified into their stand development stages , though with reduced accuracy. Where classification into SDS errors do occur , these are most often placed in a neighbouring SDS . This suggests a continuum or progression between the stand development stages and that the managed forests are developing towards old-growth condition . This work focused on the impacts from management disturbance . In comparison , natural 27 disturbances , particularly those from fire are known to have more distinct differences from managed stands . A primary cause of differences is that fire- based disturbances are primarily chemical in nature versus physical . The chemical nature of disturbance by fire causes increased soil disturbance , releases nutrients that were immobile in organic compounds and raised pH ( Hart and Chen , 2006 ) . Fire origin stands and those harvested have different plant communities , with burned areas supporting pyrophilic species and young harvested areas showed greater similarity to old forest plant communities. On a longer time scale natural spruce-sub-alpine fir stands similar to those in this study have a long fire return interval . DeLong ( 2011) states an average return interval of 220 years . With active harvest management there is concern that shortened disturbance return intervals could negatively impact biodiversity ( Vcnier et ah , 2014 ) Works discussing the theory of forest recovery from disturbance suggest that this development towards the old-growth condition is not direct or determinant and that the complexity of ecological factors could cause forests to not return to original forest condition ( Messier et ah , 2015; Parrott , 2010 ) . The fact that mature and old-growth can be separated , despite having similar stand structure , suggests that the return to natural condition is either incomplete or that a new condition is being established . Similarly, Venier et al. ( 2014 ) suggests that harvest with truncated harvest rotation cause concern for long term biodiversity. Management of forest should take into consideration that harvest disturbance changes understory plant vegetation communities and that full recovery of these communities takes significant time and may not return to previous conditions . This study suggests that mature sites remain somewhat distinct from old-growth forest condition . In order to mitigate long-term impacts to the biodiversity of plant communities and ensure that natural distributions of plant communities remain representative , unharvested forest reserves , that represent the ecology of the harvested areas , should be maintained adjacent 28 to the site. In addition , consideration should be given to harvest rotation length. The oldest managed stands in this study were most similar to old-growth . Future work should: a ) Continue to monitor the development of second growth plantations. In this study the immature and mature managed stands originated from thin-from-above partial harvest and relied on natural regeneration . In contrast , the juvenile sites were planted , with regular spacing , as such the stand structure development may differ and likewise may cause alternate vegetation community responses than those observed here , b ) Consider additional stand structure metrics including the abundance of dead wood and canopy closure . Direct crown closure measurements may provide greater insights into the distributions of plant communities. Conclusion This study examined a series of stand development stages with and without historic forest harvest and demonstrated that there are long term impacts to understory vegetation communities . Vegetation communities remain distinct , despite forest stand structure returning to nearly natural old- growth conditions. Juvenile stands and forested wetlands are most distinct , highlighting that stand structure and localized environmental conditions are important factors influencing community composition . Vegetation communities in the later stand development stages converge towards natural old-growth but still provide unique signatures. Forest managers should be mindful that changes in vegetation communities due to harvesting are long lasting and options to mitigate potential loss in biodiversity include extending harvest rotation length and provide unharvested reserves that represent the local ecology of the areas . 29 Acknowledgements This project would not have been possible without the support of the Aleza Lake Research Forest . Much thanks to Kailee MacKinnon , Alicja Muir , John Mainville, Keaton Freel , Samantha Gonzalez , and Austin Bartel for their help during data collection. We acknowledge funding support was provided by the Sustainable Forestry Initiative . 30 Chapter 3 Quantifying the recovery of coarse woody debris in British Columbia’s managed subboreal spruce forests Introduction Coarse woody debris ( CWD ) is an integral component of forest ecosystems , providing key habitat features for a wide range of species ( Stokland et al. , 2012 ; Harmon et al. , 1986 b ) . It is recognized that forest management impacts on CWD need to be considered when developing stand and landscape level plans given that managed forests have a deficit of coarse woody debris compared to natural stands ( Brassard and Chen , 2008; Spies et ah , 1988 ) . To address this issue various CWD targets in the form of minimum acceptable volumes of CWD immediately following harvest , while at the landscape scale targets are set for minimum percent areas in old forest condition ( SFI , 2015; Forest Practices Board , 2012 ; Muller and Butler , 2010 ; Berry et al . , 2018) . Here I examine the differences in the quantity and character of CWD in a sequence of stands comparing managed and old-growth natural forests , evaluating how CWD recovers over time , both within stands and how it is distributed across the landscape . CWD has been shown to be critical for maintaining biodiversity, as well as supporting a broad range of ecological functions ( Arsenault , 2003; Stokland et al . , 2012 ) . Saproxylic 31 species depend on dead wood as a structure for reproduction or as a direct source for nutrient uptake ( Langor et al. , 2008; Maiiak and Jonsell , 2017) . Fauteux et al. ( 2012 ) highlighted the importance of CWD for use by small mammals in eastern Canada . In western Canada Keisker ( 2000 ) provides a review of functional use of CWD by numerous animals. The character of the CWD changes through time due to the biological degradation / decay processes , with concomitant impacts on the ecological functions that the CWD play ( Langor et al. , 2008 ) . Harmon et al . (1986 b ) and more recently Stokland et al . ( 2012 ) provided thorough overviews of the functions provided by CWD . In addition to the biological use of CWD , these woody materials provide a valuable carbon sink ( Fredeen et al. , 2005; Bois et al . , 2009 ) . Ehnstrom ( 2001) stresses the usefulness of CWD by claiming that : “ Dead wood is perhaps the most important substrate for the maintenance of the diversity of insects , cryptogams and fungi in the boreal forests .” The importance of high volume and large piece size CWD , as described by diameter and piece length , is often stressed when evaluating the importance of CWD inputs. This is partially due to the fact that large pieces remain in the environment for the longest periods of time, contribute large carbon and nutrient pools , and provide large structural habitat features (Stone et ah , 1998; Harmon et al. , 1986b ) In addition to size , the state of decay is important , and ranges from material that is very firm with little decay to pieces that have high levels of decay ( i .e from biological degradation ) where wood cellular structures have been broken down creating soft material . These differences in the type of material are critical for the ecological function that the material provides ( Kumar et al . , 2018a ; Siitonen , 2001 ) . In natural forest stands CWD is generally input into the ecosystem through disturbance events such as wind , fire , and the influences of pests and pathogens on the live stand . CWD sources include materials: from before disturbance , generated as a result of , and 32 trees that remain following a disturbance (Spies et al . , 1988) . In contrast , managed stands , and in particular those that are clear-cut harvested , lose the vast majority of their biomass. Siitonen ( 2001) highlights that though there may be an initial flush of CWD in a newly harvested juvenile stand this is relatively short and then the harvest areas lacks opportunities for CWD input . While potentially not as intensely disturbing as clear-cut harvesting, partial cutting may also have a large impact on CWD volume and attributes . Lee et al . ( 2017) demonstrates that partial retention harvest systems can support natural saproxylic. beetle assemblages. However , they raise concern that even with partial retention there may not be enough CWD recruitment to support native diversity. Historically, in British Columbia , diameter limit cutting, a thin-from-above silvicultural system , was common practice from the early 1900s through to the 1960s (Stevenson et al . , 2011; Pedersen , 2003) . Diameter limit harvesting removed the largest and most desirable trees ; while smaller diameter trees commonly in the sub-canopy and understory were left to regenerate the site . With the removal of the main canopy the potential input of large diameter CWD is removed ; large diameter CWD recruitment is thereby delayed until the stand matures further . In our region , many of the stands that were diameter limit logged are now mature and viable for a second harvest . Understanding the current CWD loading and distribution of these previously harvest sites ( i . e . understanding the impact of harvesting legacy ) is therefore important as it will assist in determining future impact on CWD inputs. The importance of CWD volume and character for biodiversity and ecosystem functioning emphasize the need to manage CWD at the landscape scale , and be able to assess and project CWD inputs following management interventions ( Brassard and Chen , 2008) . Remote sensing via Aerial laser scanning (lidar ) techniques have been used to describe stand structure for timber ( White et al . , 2017; Naesset , 2002 ) and characterize stand 33 structure as indicators of biodiversity ( Guo et al. , 2017; Zellweger et al. , 2014 ) . In their review Davies and Asner ( 2014 ) encourages the research community to use ALS to characterize critical habitats. Detection of CWD has been attempted using direct and indirect lidar metrics ( Joyce et al . , 2019; Tanhuanpaa et al . , 2015) . In addition to providing metrics to characterize stands, modelling empirical data with lidar allows for the generation of predictive maps across landscapes ( Hengl et al . , 2018 ; White et al . , 2013) . In this study, the influence of recent and historic harvest patterns on CWD quantity and characteristics are examined . This work compares natural forest stands with forest stands of varied harvest legacies , from 20 - 90 years since disturbance . My aim is to quantify the impacts of forest harvesting on the CWD structures within each stand type and across the forested landscape . Known harvest histories provide a chronosequence of stands that were clear-cut , diameter limit harvested , or are natural old-growth . CWD is evaluated using a combination of field plots and lidar providing opportunities to characterize CWD within stands and across the landscape . Key structural components of the CWD are examined including size , volume , and decay class. Lidar provides detail on the detection of CWD and allows for predictive mapping of CWD across the landscape . My work aims to address these main questions : 1. What are the differences in CWD volume and characteristics across different forest management histories? 2 . Does CWD recover over time and specifically within timber rotation period ? 3. Can lidar be used as an effective tool for detecting CWD and provide predictive mapping useful for meeting landscape level objectives? 34 Methods Study site This study was completed at the University of Northern British Columbia’s Aleza Lake Research Forest ( Ar.54°03.8/ , W.122°04.3') . This 9 , 000 hectare forest is dominated by hybrid white spruce ( Picea glauca ( Moench ) Voss x engelmannii Parry ex Engelm . ) and sub-alpine fir ( Abies lasiocarpa ( Hook . ) Nutt . ) with minor components of Douglas-fir Pseudotsuga menziesii ( Mirb . ) Franco ) , trembling aspen ( Populus tremuloides Mich ) , and paper birch ( Betula papyrifera Marsh. ) . Topographically the area is part of a larger plateau landscape with elevations ranging from 660 - 720 metres resulting from flooding following the end of previous glaciation ( i . e . Glacial Lake Prince George ) as described by Tipper (1971) . Topography is generally flat with gently rolling terrain in the north with lacustrine soils that are incised by various streams. The area was established as a experimental station in 1924 with the key objective “ to demonstrate sustained yield, forestry at a practical level ” ( Schmidt , 1992 ) . Since then , the forest has been actively managed for timber and other values ( Aleza Lake Research Forest Society , 2019 ) . Sampling design Eight forest stand types were sampled in & 2 x 2 x 3 design : defined by two landscape units, two harvest histories ( i .e . harvested or old-growth ) , and three stand height classes. The landscape units included first the Knolls in the northern portion of the forest distinguished by gently rolling terrain with moderately well drained soils (silt-clay loams to clay-loams) , and slopes ranging from 2- 25% . Second , the Plateau , a generally flat landscape with slopes outside of stream gully areas ranging from 0-10 % with moderately to poorly drained soils ( clay loams to clays ) . See Chapter 2 , Figure 2.1 for a map of the landscape units. In each landscape unit plots were randomly located in each harvest history and height strata . 35 ^Victoria Vancouver Figure 3.1: The 9 , 000 hectare Aleza Lake Research Forest is the focus of this study. This forest was established in 1924 to demonstrate sustained yield harvesting. 36 This was done based on known harvest history records with stand heights determined from a lnr resolution crown height model derived from lidar . The natural forests were divided into old-growth ( OG ; heights > 29 m ) , and natural immature sites ( iV7; heights 20 — 29 m ) . Areas with harvest history ( HH ) were divided into three height classes : juvenile ( juv ; 5 — 19.9m ) , immature ( mm ; 20 — 28.9m ) , and mature stands ( mat ; > 29m ) . Younger stands with heights less than 10m were not examined in this study. Harvest history records are available for the research forest dating back to the 1920s . Prior to 1966 the most common harvest system were methods of intermediate utilization or diameter limit cut ( DLC , Pedersen , 2003) . These DLC sites are distinct from the current dominant regional practice of clear-cut management followed by tree planting , used from the late 1960 ’s to present , though planting programs were not common until the 1980s . In contrast , the DLC sites were thin-from-above silviculture retaining small diameter stems ( less than either 12 or 9 inches at dbh ) , and sites were left to regenerate naturally. These differences in silvicuture systems are considered when examining the results of this study. Immature and mature managed stands were harvested in the era of intermediate utilization . CWD field data collection A series of four 30nr long linear transects were established to measure CWD from within four sub-plots ( Figure. 3.2 ) . Transect locations were established in the forest using the iSXBlue II (sub-metre accuracy GPS ) . This was done to ensure that empirical data could be matched with the lidar data . All CWD larger than 7.5cm in diameter at the point they intersect with the transect were measured . For each piece of CWD the length , diameter , decay class , and location along the transect were recorded . The decay class of the CWD was placed into a discrete scale from 1 to 5 as described in British Columbia ( 2010 ) . A decay class of 1 has no significant decay to 5 where the wood 37 Figure 3.2 : One hectare plot layout including four 400m 2 sub-plots and associated 30 metre CWD transects. is highly decomposed . This transect survey did not consider CWD that had decomposed to the point that it was considered part of the soil profile as defined by > 50% of the CWD piece being embedded in the soil . The volume of CWD was calculated using the Van Wagner formula Equation : ( 3.1) as described in Van Wagner (1968) and Fraver et al . ( 2018 ) where d is the piece diameter in metres and L is the length of the transect in metres. V 4 E £) i2 x 10 , 000 ( 3.1) Statistical analysis was done using R 3.6 . 2 ( R Core Team , 2019 ) . CWD attributes were evaluated using ANOVA testing and post- hoc comparisons were made using Tukey Honest Significant Differences. A significance level of 0.05 was used throughout . Eta-squared ( 7 f ) was used to quantify effect size or strength of relationships . Shannon’s diversity index was used to compare the diversity of decay classes as suggested by Brassard and Chen ( 2008 ) . 38 Lidar analysis On May 11, 2015 the study area was Aerial Laser Scanned (lidar ) using UNBC ’s Riegl VQ-580 lidar scanner mounted on a fixed wing plane . The timing of this data collection provided for snow free and leaf off conditions. The plane flew at an elevation of approximately 500 m above the ground . Scan frequency was 380 kHz with a scan angle of 30° from vertical ( 60° field of view ) . Laser divergence was 0.2 mrad equaling a 10cm on-ground footprint ( vertical pulses ) . The resulting data provided 10 first pulse returns per square metre. Lidar data was processed using LASTools software ( fsenburg , 2016 ) and the the R package lidR ( Roussel and Auty , 2020 ) . In order to determine to what extent individual CWD pieces were identifiable from ALS , a 0.25m 2 resolution near-ground-surface-model ( NGSM ) was generated using a height cutoff of 1.3m ( i .e. ALS data higher than 1.3m above the ground was removed ) . This was visualized in QGIS ( QGIS Development Team , 2019 ) with a digital representation of the transect location . Empirical transect data including piece size and location along the transect was cross referenced with a digital version of the transect overlaying the NGSM and where possible CWD pieces were identified ( Figure 3.3) . Wall-to-wall lidar coverage of the study area provided the opportunity to model CWD across the landscape . To do this standard ALS area based approach metrics ( ABA ) were generated corresponding to the transect data collection (White et ah , 2013; White et ah , 2017; Naesset , 2014 ; and Rousse, 2018 ) . In addition , a metric of near-ground-point-density ( NGPD ) was used to characterize the amount of material in the CWD layer , that is non-ground classified lidar points with elevations of 0.1m to 1.3m above ground were used (equation ( 3.2 )) . Figure 3.3 illustrates the area around the empirical transect for which the ABA metrics were generated ( 30nr long transect x 13.33m wide , a 400m 2 area ) . A random forest regression model was fit to the empirical data and then used to predict the volume of CWD across the landscape. Key processing packages include: mlr ( Bischl 39 Figure 3.3: Surface model highlighting CWD generated from lidar data using a 1.3m height cut- off . Identifiable individual pieces of CWD that crossed the transect were digitized and labelled corresponding to the piece in the empirical data . The digital transect includes tick- marks every 2.5m to compare with piece locations in the empirical data . The boxed area represent the area from which the ABA lidar metrics were generated . 40 et al. , 2016 ) , ranger ( Wright and Ziegler , 2017) , stars ( Pebesma , 2020 ) , and sf ( Pebesma , 2018) . Prediction accuracy was assessed using the r 2 coefficient . The predictive model was validated using the leave - one - out method described by Tompalski et al. ( 2019 ) NGPD = CWDiso CWD13o + G ( 3.2 ) Equation ( 3.2 ) : To generate the near-ground-point-density ( NGPD ) the lidar data is height limited between 0.1 and 1.3m above ground ( CWDi 30 ) and normalized against the sum of this CWDi 30 and the ground classified points ( GO - To evaluate the utility of the predictive map zonal statistics were conducted using existing forest inventory polygons greater than 5 hectares in size. A spatial union of the ALRF ’s harvest history database and provincial forest inventory polygons1 was completed . A negative 20nr buffer was applied to all polygons to reduce potential edge effects . Mean raster values from the predicted CWD layer were generated for each polygon using zonal statistics process in QGIS . Managed stands younger than 30 years were considered juvenile , age 30 - 59 as immature, and over 60 years as mature . Old-growth stands were all those that did not have a harvest history. In addition , old-growth polygons were reviewed against aerial photography to remove sites that were distinctly forested wetlands . Results Empirical assessment of CWD I tested if there were differences in CWD volume between the Knolls and Plateau areas . No significant differences attributed to landscape unit were found ( F ( l , 182 ) = 1.749, p = 0.1877 ) . As such the results presented are in relation to stand history and development Data ’s Vegetation Resource Inventory 41 stage ( F (4 , 182 ) = 20.0121 , p = 0.000 ) . During field data collection it was noted that the natural immature sites in the Knolls landscape unit were previously unmapped harvest areas as evidenced by old cut stumps. These plots were subsequently classified with the managed immature. In the Plateau landscape unit the natural immature sites were unique in that they were not upland productive forest but were forested wetlands as evidenced by their herb and shrub plant communities ( unpublished data ) . These were reclassified as old- growth forested wetlands { wOG). CWD Metrics by Stand History CWD Volume Piece Count 40 400 30 300 - - 100 ,Vi juv imm mat ? 10 - 0 - OG wOG 1.5 - 1.0 - 0.5 - 0.0 - JU ’ T - - 20 - JL 200 0 Decay Class Diversity - juv imm mat OG wOG juv imm mat OG wOG Figure 3.4: Empirical metrics of CWD highlight that younger stands have lower total volume ( m? / ha. ) , lower piece counts in transects ( pieces / 100 m ) , and diversity across decay classes (Shannon’s H ' ) . Letters above boxplots indicate treatments that are statistically similar . There was a distinct pattern in the volume of CWD across harvest histories with the lowest volume in the younger stands increasing in the mature forest sites. The old-growth 42 Table 3.1: Summary of empirical CWD volume, piece count , and Shannon ’s diversity index ( H ) . Treatment Plot N C'WD Volume v se Piece Count pc se juv imm 11 mat OG wOG 11 10 5 101.23“ 102.87a 155.95a 310.01b 186.99;l 20.12 18.07 23.36 11.74c 14.72"1 18.79 d 30.58® 27.67® 1.74 1.41 1.75 1.86 1.22 12 27.52 5.69 Decay Class Shannon ’s H 3.34 2.77 2.52 2.77 2.89 0.976f 1.192s 1.36s 1.405s 1.412fg H se 0.128 0.119 0.038 0.037 0.067 sites had distinctly more volume than the previously harvested stands ( F(4, 44 ) = 15.483, p = 0.000 , rf = 0.585 ) . Tukey Honest Statistical Difference post-hoc analysis indicated there were significant differences between the old-growth and all of the other stand types . This is highlighted in Figure 3.4 where stands that are statistically similar have the same letter ( p > 0.05 ) . Mean CWD volume in the old-growth sites was 310m 3 , twice that of the mature HH sites , and three times higher than the juvenile and immature HH sites . Although HH sites were statistically indistinguishable a trend of increasing volume corresponding to stand development appears to exist . Table 3.1 provides a summary of the CWD volume, piece counts , average decay class , and average piece size. Piece count shows a similar pattern with significant differences between the old-growth and the managed forest sites ( Figure 3.4; F ( 4, 44) = 21.654 , p = 0.000 , rf = 0.663) . Based on post- hoc analysis the mature stands were shown to have significantly more pieces of CWD than the juvenile though they were not significantly different than the immature stands. Analysis of piece size across treatment showed some significant differences. However , there was no distinct pattern and the strength of these differences was weak ( rf = 0.023) . Shannon’s diversity index ( H ) was used to determine what differences exist in the character / types of CWD . A similar pattern to volume and piece count is seen as diversity increases with stand development . Significant differences between the diversity indices were observed ( Figure 3.4 ; F ( 4 , 44 ) = 3.721, p = 0.011, if = 0.253) . Post-hoc analysis showed that juvenile and immature stands were statistically similar , and lower than the mature and 43 Piece count by decay class Figure 3.5: Decay class abundance arranged in panels of stands development stages. Old-growth sites have CWD in all decay class with the greatest amount in decay class 3. In comparison , juvenile stands have small counts in decay class 1 and 2 ; while immature and mature sites have elevated levels of decay class 1. 44 old-growth sites. Patterns in the different CWD decay classes are visible in Figure 3.5. The old-growth stands have the greatest about of CWD in decay class three with a smaller number of pieces in the other decay classes. In comparison , the harvest history sites have different patterns especially in the classes 1 and 2 ( newly recruited CWD ) . The juvenile stands are effectively void of decay class 1 materials and have lower proportions of class 2 than that of the old-growth sites ; in comparison , data from the immature and mature stands suggest that these classes are accumulating. Lidar detection of CWD On average the detection rate of individual CWD pieces from lidar was 23.7% . However , detection rates are dependent on the size of individual pieces ( Figure 3.3) , decay class ( Figure 3.6 ) , and stand canopy characteristics ( Table 3.2 ) . To examine the influence of decay on detection rates , only logs of 30cm in diameter or larger were considered ; Table 3.3 highlights that CWD with higher levels of decay were less likely to be detected . To examine the influence of stand development stage , large logs with little decay were used ( i . e. logs 30cm - 45cm in diameter in decay classes 1- 3) ; Table 3.4 shows that stand structural differences associated with stage development stage affect the ability to detect individual pieces. Mature and old-growth sites had individual piece detection rates that were more than double that of juvenile stands . Lidar detection rates of CWD by diameter class jj | (60,lnf] O (30,45] 1 , (15 30] g (0,15] 0.0 ! O2 04 0.6 ^ 08 Figure 3.6 : The comparison of detection rates of CWD from lidar demonstrates that larger pieces of CWD are more likely to be detected . 45 Table 3.2 : Detection rates of CWD by diameter class across all sites treatments. Diameter 0-15 15-30 30-45 45-60 60 + Field Measured Lidar detected Detection Rate 370 469 246 49 8 31 120 91 25 5 0.08 0.26 0.37 0.51 0.62 Table 3.3: Detection rates of CWD by decay class across all treatments. For this table only CWD between > 30 cm in diameter was used . Decay Class 1 2 3 4 5 Field Measured Lidar detected Detection Rate 20.0 50.0 95.0 96.0 42.0 17.0 27.0 50.0 20.0 7.0 0.85 0.54 0.53 0.21 0.17 Table 3.4: Detection rates of CWD by treatment . For this examination only CWD of 30-45cm diameter and in decay classes 1-3 were used . Stand Type Field Measured Lidar detected Detection Rate 21 16 38 76 14 6 6 25 53 4 0.29 0.38 0.66 0.70 0.29 juv imm mat OG wOG 46 Predictive mapping In order to provide a prediction of CWD across the landscape , area- based-approach lidar metrics were used . Results from the leave-one-out validated random, forest regression model provides an r 2 = 0.2221 ( Figure 3.7) . Variables that were determined to be the most important for providing the prediction included key stand structural characteristics of height ( i .e . zq90 , zq95 , zq85 , zq25 , zmax) and stand height variability ( zsd , zkurt , Rumple indices) . In addition , the NGPD metric was listed as the fourth most important variable ( Figure 3.8) . Using the model a wall-to- wall predicted CWD map was generated ( Figure 3.9 ) . Observed versus predicted CWD Volume m3/ha. - -s - •• 0 * predicted Figure 3.7: Observed vs . predicted coarse woody debris volume from leave-one-out validation from the random forest regression model . Predictive map efficacy Mean predicted CWD volume was determined for the forest inventory polygons and showed a clear separation between all stand types ( F(4 , 184 ) , p = 0.000 , if = 0.607 ) . 47 Covariate importance in the Random Forest Model Rumple zsd zmax cwd2ground zq90 zq95 zq85 RumpleXYZ zq25 zkurt ' VCI 2.5 pzabovezmean zq80 - zq75 ' zentropy • zpcum6 ' zmean • zq30 ‘ GapFraction137 ' zpcum2 • 0.00 0.02 0.04 0.06 Figure 3.8: Random forest models allow for the determination of which of the covariates were most important in generating the model . Key metrics for predicting CWD include stand structural metrics of canopy variability ( e.g . Rumple , RumpleXYZ , zsd ) , height ( e.g. zmax , zq90 , zq95 ) , and the custom near-ground- point-density ( i . e . cwd 2ground ) . 48 Predicted Volume of Coarse Woody Debris ^ Figure 3.9 : A predictive map of CWD across the Aleza Lake Research Forest . Polygons with known disturbance histories highlight the CWD volumes differences between the treatments consistent with empirical analysis. Polygon labels include the stand type with the mean predicted CWD volume per hectare for the polygon . Tukey Honest Statistical Difference post-hoc analysis demonstrated that all treatments were significantly different from one another ( Figure 3.10 ) . At the stand level , predicted mean CWD volumes ranged from 103m3 in the juvenile stands to 209m 3 in the old- growth stands . 49 Predicted CWD within Forest Inventory Polygons 250 t jS 200 E O 8 I 150 - 100 juv mat imm OG Figure 3.10 : Average predicted CWD volume within forest inventory polygons . Discussion Empirical differences Volume of CWD My empirical results clearly demonstrate that CWD volumes in managed stands , regardless of time since disturbance and method of logging , have distinctly lower levels of CWD than natural stands , by a factor of 2-3. This difference is not only in the volume of CWD but is also reflected in a lower number of pieces , and shifts in the character and diversity of the CWD material ( i . e . decay class ; Figure 3.4 ) . The trend that managed stands have less CWD than natural stands findings is consistent with numerous studies ( Spies et ah , 1988 ; Siitonen , 2001; Brassard and Chen , 2008 ) . CWD material can be classified as inputs that were present pre-disturbance , and CWD that originated from the remaining trees following a disturbance ( Spies et al. , 1988; 50 Sturtevant et al. , 1997 ) . Stands in earlier stages of stand development lack sources of CWD . As the stands mature small stems may die as a result of self-thinning . However , these smaller stems are relatively low in volume and due to their small size are expected to decay faster ( Freschet et al. , 2012 ; Harmon et ah , 1986a ) . Mature stands , those that were partially harvested , indicate a trend of increasing CWD as shown by stands having 50 percent more CWD than immature sites . However , this increase still remains half of what is in natural stands. As these stands are mature, and thereby available for harvest , it raises the question if a return to natural levels is actually possible. If harvest scheduling is primarily based on the maturity of the forest then the natural disturbance pattern will be shortened and as a result CWD loading will not return to natural levels . The additional levels of CWD in these sites may be a function of natural self -thinning processes . Our data is inconclusive as far as partial cutting , causing an increase in CWD volume . Character of CWD The character of CWD in the managed stands differs from that of the old-growth. Results from Shannon’s diversity index shows a trend of low decay class diversity increasing with each stand developmental stage ; Figure 3.5 provides insight on the abundance of each decay class . Old-growth sites have CWD represented in all decay classes with decay class 3 having the most . In contrast , juvenile sites are generally devoid of decay classes I and 2 ( materials with low levels of decay ) . Juvenile sites , having been clear-cut harvested , have effectively no new potential inputs of CWD except perhaps along the mature edges surrounding these sites . Immature , and mature sites have elevated levels of decay class 1 which may be indicative of stem exclusion stand dynamics as suggested by Brassard and Chen ( 2006 ) . Though the Shannon H' diversity index values between mature and old-growth are effectively the same the actual abundance within each class differs ( as noted above ) . 51 All sites have decay class 3 as the most abundant class . This is likely due to the fact , that as stems decay they remain in decay class 3 the longest period of time , consistent with finding of Newberry et al . ( 2004 ) who examined time since death based on CWD decay metrics. All sites show lower abundance of decay class 4 and 5 ( highly decomposed materials ) . This is likely due to a lower amount of time that material would remain in decay class 4 relative to decay class 3. Our sampling method likely discounted the amount materials in decay class 5 , as logs that were 50 % of greater embedded in the forest soil profile were not measured. The managed mature and immature forest stands do have similar decay class distributions , however , the abundance of material remains far less and there are numerous decades in which the managed stands have lower volume and an altered distribution of CWD decay classes. The old-growth forested wetlands share some similar CWD characteristics with the upland forest old-growth stands in terms of piece counts and diversity of decay classes ; they differ however in that the total CWD volume is lower . The lower volume compared with the old-growth sites is attributed to lower site productivity ( the site was not capable of producing larger trees and thereby high total volume on the site ) . Evidence suggests that the ecological functions provided by CWD debris may be com- promised in managed forests given that total amount of material is significantly reduced and the character of that material takes significant time to recover . Brassard and Chen ( 2006 ) note that a varied decay class may be just as important as the volume of CWD . Our study does suggest that decay classes differ and that there may be several decades in which not all decay classes are represented ( i . e . juvenile sites ) . Kumar et al . ( 2018a ) found that decay class had a strong influence on epixylic plant communities . Langor et al . ( 2008 ) found that saproxylic insects communities were likewise influenced by decay class . Given that there are strong differences between old-growth and stands with harvest histories , to ensure that ecosystem integrity is met , it is important to identify where areas 52 of high CWD are on the on the landscape. Detection of CWD using lidar , and landscape management Review of CWD transects in the near-ground surface models highlight that CWD can be detected using lidar . However , successful detection of individual pieces is reliant on a number of factors . Individual CWD pieces influences detection rate where large pieces with lower levels decay ( i . e . decay classes 1-3) have higher detection success rates ( tables 3.2 and 3.3 ) . The structure of the live forest canopy also influences detection rate . Detection rates are highest in old-growth stands and lowest in the juvenile sites , as the latter had closed canopies and lacked canopy lift such that tree branches remained visible in surface model and obscured visualization of the near-ground surface model . Likewise , immature stands had slightly higher detection rates though it is suspected that the degree of canopy closure reduced detection rates . Highest detection rates are in mature and old-growth stands where crown lift is highest , and the canopy is more transparent due to stand dynamics ( e . g . gap generation ) . Direct CWD detection in this study was done through manual comparison of CWD surface model with empirical data . Image segmentation algorithms should be investigated to automate this process and provide opportunities to map individual CWD across a landscape . Area based metrics were used to generate models of CWD and demonstrate that it is possible to generate predictive maps of CWD across a landscape. Figure 3.8 highlights that metrics of stand structural diversity are predictive of CWD . These include standard deviation ( zsd ) , and rumple indices2 , which are measures of crown surface roughness , and metrics of stand height ( zmax and zq90 ) . Kane et al . ( 2010 b ) highlights that these metrics are consistent with stand structural development where increasing values of these metrics corresponds with increasing stand structural diversity. Finally, the custom NSPD 2 Rumplepts used Delaunay triangulation of lidar points , and Rumplechm. used a raster based method following Jenness ( 2004 ) 53 metric was highlighted as a variable useful for determining CWD . The Random Forest model provides an r 2 — value = 0.2221 which is strong enough to demonstrate distinct trends but not adequate enough to provide pixel-for- pixel certainty. The forest stand polygon analysis demonstrates that , despite a lower r 2 value , there is a clear separation of the stand development stages . This confirms that there is strong utility to highlight where areas of CWD are on the landscape. Some caution is needed in that modelled CWD volumes tallied at the polygon stand level differ from values expected from the empirical data . Juvenile and immature stands had predicted volumes that fell within the empirical ranges expected ( e.g. juvenile : 113; immature: 145 ; mature: 166 ; and old- growth : 209 ; see Figure 3.10 ) . Our predictive map ( Figure 3.9 ) shows a clear distinction between areas with harvest history and old-growth . More careful examination of Figure 3.9 shows that the recovery of CWD volume is visible on the landscape ; highlighted areas show increasing amounts of CWD corresponding with their stand development stages . Management Opportunities Ecological Differences Results from this study are consistent with others suggesting that high volumes of CWD can be considered a good indicator of old forest condition ( Kunttu et ah , 2015 ; Ulyshen et ah , 2018 ) . However , ecological drivers including natural disturbance patterns need to be considered . For example , areas with higher frequency forest fire may still see low levels of CWD in old forest stands . Ulyshen et al . ( 2018 ) examined longleaf pine forests with frequent fire return intervals and determined that following fire , there was a large input of CWD in the stand development stage immediately following fire disturbance and over time CWD levels fell and were lowest in the oldest stands . However , one could argue that such landscapes never have the opportunity to become old ( Arsenault , 2003) . 54 Bauhus et al. ( 2009 ) suggests that silvicultural manipulation of maturing and mature second growth stands could allow for the recruitment of old forest structures. This is supported by Sandstrom et al. ( 2019 ) who reviewed numerous studies covering temperate and boreal forests of North American and Europe , and concluded that artificially increasing volumes of CWD significantly increased abundance and richness of saproxylic insects and wood-inhabiting fungi , although they caution that manipulating CWD while providing volume does not address potential issues relating to mimicking natural distribution of decay classes . Designated areas Providing areas reserved for natural processes , including natural disturbance , may be critically important to ensure that islands of natural CWD volume and distribution of varied CWD decay classes remain on the landscape as these may be essential for the survival of some saproxylic species . Predictive mapping of CWD , as used in this study, provides an opportunity to highlight areas to either implement silvicultural treatments to enhance CWD and to consider for reserve designation . In considering future research , predictive mapping methods could be to highlight other forest attributes or create indices for ecological function . Conclusion My research compared natural old-growth fforests to managed stands , and found that managed stands have much lower CWD volume and differ in the character of CWD ( i . e . decay class) . Old-growth stands have twice as much volume than the mature managed stands and while mature managed stands did show an increase in CWD compared with the immature stands , indicating that recruitment of CWD has started , more time will be needed to return to the natural levels . The character of CWD , in its various decay 55 classes , also differs in managed stands; of particular concern are the juvenile sites which were nearly devoid of decay class 1 and 2 . Recovery of CWD volume and character to natural levels does not occur within timber rotation period , raising concerns that the lack of CWD could impact the biodiversity of numerous taxa requiring CWD in its various forms. Without management intervention to create CWD it will therefore not be possible to see natural levels of CWD within the managed forests. Managing CWD at the landscape , including identifying where stands rich in CWD are located , provide options for management to maintain stands rich in CWD on the landscape . The use of lidar in this study demonstrated that CWD can be detected . Modelling of empirical data with the lidar allows for the production of predictive maps of CWD , which could be used by decision makers to manage CWD at the landscape scale. Acknowledgements This work would not have been possible without the support of the Aleza Lake Research Forest Society. Much thanks to John Mainville , Kailee Mackinnon , Rachelle Winsor , Samantha Gonzalez for their assistance with data collection . Much thanks to Dr . Brian Menounos for his support in coordinating the acquisition and initial processing of the lidar data including use of his lidar equipment . I acknowledge the financial support of the Sustainable Forestry Initiative and UNBC Research Office seed grant . 56 Chapter 4 Synopsis Introduction The forest landscapes of Canada are being progressively changed by forest management ( Venier et ah , 2014) . Although regarded as sustainable , concerns have been raised over the last number of decades about the impacts of forestry practices on biodiversity and ecological integrity ( Harmon ct ah , 1986a ; Hart and Chen , 2006 ; Brassard and Chen , 2006 ) . In their consideration of forest disturbance and recovery, Messier et al . ( 2015) promote the concept of forests as complex adaptive systems and that forest stands develop from a stand initiating event ( e . g . harvest or fire ) and converge towards old-growth condition though this may not be linear or determinative. For the purpose of measuring forests as complex adaptive systems , Parrott ( 2010 ) encourages the research community to develop tools to measure how ecosystems are developing , to use remote sensing to supplement research , and to be spatially and temporally explicit in measurements . Remote sensing from aerial laser scanning ( lidar ) provides excellent detail of the 3-dimensional structure of forests ( Wulder et ah , 2008b ) . Davies and Asner ( 2014 ) encourages the use of lidar to describe habitat condition and ecosystem function . My previous chapters have demonstrated the use of research tools to quantitatively measure the impacts of forest management disturbance and patterns of recovery. Here , based on findings of Chapters 2 and 3 I consider : 57 1. Are the forest stands converging towards old forest condition ? 2 . Provide recommendations for forest land management . 3. Highlight areas for further research. 4. Are we growing forests or are we growing trees? Forest disturbance and patterns of recovery Methods In the previous chapters impacts on forest biodiversity were quantitatively investigated . Chapter 2 provided direct measurements to biodiversity by examining plant response to changes in forest structure . Metrics of biodiversity included impacts to stand structure and diversity : tree species composition , sizes , and density. Understory vegetation community composition was measured using total species diversity, richness , and abundance. Analysis of vegetation abundance was considered as total abundance, abundance by functional group , and individual species abundance. Chapter 3 provided indirect measurements of biodiversity by examining coarse woody debris ( CWD ) as key structual features in forests essential for the support of numerous taxa especially those that are considered epixylic or saprophylic ( Stokland et ah , 2012 ) . This chapter measured both the character and quality of CWD . In addition , the use of lidar provided opportunity to evaluate differences in CWD volume across the landscape. In both chapters, the same series of forest stands was used covering two landscape units and a range of stand development stages , stratified by height class , and presence or absence of harvest history. Managed stands included plantation forest stands : juvenile, and stands that were thinned from above: immature, and mature. Natural stands included old-growth and forested wetlands . With the exception of the forested wetlands this series of stands formed a chronosequence allowing for insight into the impacts and recovery of 58 key biodiversity metrics both spatially and temporally. Indicators of forest recovery The results from Chapter 2 suggest that there is very little impact from forest management on metrics of biodiversity. Managed forest stand structure was distinct from natural through the juvenile and immature stages while the mature stands , based on the metrics used , were indistinguishable from the old-growth . Results showed that total species diversity and richness for the main canopy and the understory vegetation showed only slight differences linked to management and these returned to natural levels by the immature stage . Vegetation community abundance by functional groups indicated impacts from forest harvesting on the juvenile stage of stand development with recovery by the immature stage. The examination of abundance by individual species is where a signature of long term impacts forest harvesting were discernible . All harvested stands could uniquely be identified . Patterns from the linear discriminant analysis do suggest that vegetation communities are converging towards old forest condition as the mature harvested sites and old-growth sites showed overlap and some classification confusion ( Figure 2.6 ) . Chapter 3 found distinct differences in the total volume of CWD . Volume of CWD in the juvenile and immature sites were one third that of old-growth . Mature sites did see some recruitment of CWD but remained half that of the old-growth . The number of CWD pieces showed a pattern of recovery but all managed stand types were lower than old-growth stands. The types of CWD , indicated by decay class , was also altered , being less diverse in the managed stands ( Figure 3.4) . This chapter also demonstrated that it was possible to use lidar to detect and model the spatial distribution of CWD on the landscape providing a visual tool to identify where CWD pools are most impacted and where opportunities for conserving areas high in CWD are located . These chapters demonstrate that there are patterns of recovery. Vegetation community 59 composition converges to old-growth with stand maturity. The greatest area of concern in the management of CWD , as patterns indicate a trajectory towards old-growth ; however , managed stands remain distinct and well below natural conditions even at stand maturity. Recommendations The forest manager needs to be mindful that the impacts forest management has on vegetation and CWD are long lasting. Although there are indications of some return to natural conditions managed forests are distinct from natural old-growth . Given that mature stands , as their name implies , are ready for a subsequent harvest , this raises further concern that full recovery of these stands will never be achieved as they will be harvested prior to full convergence with natural conditions ( Vcnier et al . , 2014 ; Hart and Chen , 2006 ) . Options for the full recovery to natural condition requires more time or , as Bauhus et al. ( 2009 ) suggests, treatment of maturing stands could be conducted to encourage the development of old-growth attributes. Alternatively, action can be taken in the spatial management of old-growth or development of mature stands towards old-growth . Areas of reserve , with the objective of conserving or restoring old-growth condition should be established. This can be done at both the stand level and landscape level . Additional research to refine my findings is also needed . The series of stands used in this study included plantation forests and those that were thinned from above . Future work needs to continue to monitor the development of plantation forests which may develop different vegetation communities and patterns of CWD recruitment . Further , lidar metrics could be used with the community vegetation data to examine in greater detailed the impacts of stand structural conditions. For example, lidar can provide a direct measure of canopy closure impacting light conditions in the understory. 60 Are we growing forests or growing trees? We are growing forests but perhaps not in their entirety. Through my research it is evident that managed forests and in particular mature managed forests continue to differ from natural old-growth . The vegetation communities though somewhat distinct are converging towards natural stand condition . The greater concern appears to be the long-term management of CWD on the landscape . In the management of forest lands , indications are that it will be possible to see the return of natural levels of CWD . However , as a land manager I need to be mindful that this will not occur without action . Rotation forestry, harvesting stands as soon as they are mature , will never allow the return of old-growth attributes and the services they provide . To manage for CWD , it must be done spatially to either conserve existing areas or define areas for recruitment of natural CWD condition . 61 Bibliography Ahmed , 0 . S. , Franklin , S. E. , Wulder , M . A . , and White, J . C . ( 2015 ) . Characterizing stand-level forest canopy cover and height using Landsat time series , samples of airborne LiDAR , and the Random Forest algorithm . ISPRS Journal of Photogrammetry and Remote Sensing , 101:89-101. Aleza Lake Research Forest Society ( 2019 ) . Aleza Lake Research Forest Management Plan # 3 ( 2019 - 2029 ) . Arsenault , A . ( 2003) . A note on the ecology and management of old-growth forests in the Montane Cordillera . The Forestry Chronicle , 79 ( 3) : 441 454 . Publisher: Canadian Institute of Forestry. - Barbier , S . , Gosselin , F . , and Balandier , P. ( 2008) . Influence of tree species on understory vegetation diversity and mechanisms involved — A critical review for temperate and boreal forests . Forest Ecology and Management , 254 (1) :1—15. Bauhus, J . , Puettmann , K . , and Messier , C . ( 2009 ) . Silviculture for old-growth attributes . Forest Ecology and Management , 258 ( 4) : 525-537. Beaudry, L . , Coupe , R . , DeLong , C . , and Pojar , J . (1999 ) . Plant indicator guide for northern British Columbia: boreal , sub- boreal , and subalpine biogeoclimatic zones ( BWBS , SBS , SBPS , and northern ESSE ). Number 46 in Land management handbook . British Columbia Ministry of Forests Research Program , Victoria , BC . OCLC : 255262857. Berry, A . , Lavers , A . , and Mitchell , L . ( 2018) . Old forest policy and regulatory frameworks in Nova Scotia and New Brunswick with a comparison to British Columbia . The Forestry Chronicle , 94 ( 01) :13-19 . Publisher : Canadian Institute of Forestry. Bischl , B . , Lang , M . , Kotthoff , L . , Schiffner , J . , Richter , J . , Studerus, E . , Casalicchio, G . , and Jones , Z . M . ( 2016 ) . mlr : Machine learning in r . Journal of Machine Learning Research, 17 (170) : l 5. - Bois , C . H . , Janzen , D . T. , Sanborn , P. T. , and Fredeen , A . L . ( 2009 ) . Contrasting total carbon stocks between ecological site series in a subboreal spruce research forest in central British Columbia . Canadian Journal of Forest Research, 39 ( 5 ) :897 907. - Borcard , D . , Gillet , F. , and Legendre , P. ( 2018 ) . Numerical Ecology with R . Use R! Springer International Publishing : Imprint : Springer , Cham , 2 nd ed . 2018 edition . Boutin , S . , Hauglrland , D . L . , Schieck , J . , Herbers , J . , and Bayne , E . ( 2009 ) . A new approach to forest biodiversity monitoring in Canada . Forest Ecology and Management , 258:S168 S175. - 62 Brassard , B . and Chen , H . ( 2008) . Effects of Forest Type and Disturbance on Diversity of Coarse Woody Debris in Boreal Forest . Ecosystems, 11 ( 7) :1078 1090 . Brassard , B . W . and Chen , H . Y. H . ( 2006 ) . Stand Structural Dynamics of North American Boreal Forests . Critical Reviews in Plant Sciences , 25 ( 2 ) :115 137. British Columbia , editor ( 2010 ) . Field manual for describing terrestrial ecosystems. Number 25 in Land management handbook . B. C . Ministry of Forests and Range , Victoria , 2 nd ed edition . OCLC : ocn679520302 . Burton , P. ( 2013) . Exploring complexity in boreal forests. In Messier , C . , Puettmann , K . , and Coates , K . , editors , Managing forests as complex adaptive systems . Building resilience to the challenge of global change . Routledge , London, pages 79-109. Routledge. Campbell , L . , Coops , N . C. , and Saunders , S . C . ( 2017) . LiDAR as an Advanced Remote Sensing Technology to Augment Ecosystem Classification and Mapping. Journal of Ecosystems and Management, 17 (1) . Canadian Council of Forest Ministers ( 2017) . Forest governance in the Province of British Columbia . Cardinale, B . J . , Duffy, J . E . , Gonzalez , A . , Hooper , D . U . , Perrings , C . , Venail , P. , Narwani , A . , Mace , G . M . , Tilman , D . , Wardle , D . A . , Kinzig , A . P. , Daily, G . C . , Loreau , M . , Grace , J . B. , Larigauderie , A . , Srivastava , D . S. , and Naeem , S . ( 2012 ) . Biodiversity loss and its impact on humanity. Nature , 486 ( 7401) : 59 67. - Castka , P. ( 2016 ) . Certification and biodiversity: how voluntary certification standards im.pact biodiversity and human livelihoods. IUCN , Gland . OCLC : 994866403. Coops , N . C . , Tompaski , P. , Nijland , W . , Rickbeil , G . J . , Nielsen , S. E . , Bater , C . W. , and Stadt , J . J . ( 2016 ) . A forest structure habitat index based on airborne laser scanning data . Ecological Indicators, 67: 346 357. - Davies , A . B. and Asner , G . P. ( 2014 ) . Advances in animal ecology from 3D-LiDAR ecosystem mapping . Trends in Ecology & Evolution , 29 (12 ) : 681 691. DeLong , C . ( 2003) . A field guide to site identification and interpretation for the southeast portion of the Prince George Forest Region. Number 51 in Land management handbook . British Columbia , Forest Science Program , Victoria . OCLC : ocm52108974 . DeLong , S . C . ( 2007 ) . Implementation of natural disturbance-based management in northern British Columbia . The Forestry Chronicle , 83( 3) : 338 346 . - DeLong , S . C . ( 2011) . Land units and benchmarks for developing natural- disturbance based forest management guidance for Northeastern British Columbia. British Columbia , Ministry of Forests and Range , Forest Science Program , Victoria . OCLC : 551884808 . DeLong , S . C . , Arocena , J . M . , and Massicotte , H . B . ( 2003) . Structural characteristics of wet montane forests in east-central British Columbia . The Forestry Chronicle , 79 ( 2 ) :342 351. - 63 Ehnstrom , B . ( 2001 ) . Leaving Dead Wood for Insects in Boreal Forests - Suggestions for the Future . Scandinavian Journal of Forest Research, 16 : 91 98. Publisher: Taylor & - Francis Ltd . Fauteux , D . , Imbeau , L. , Drapeau , P. , and Mazerolle , M . J . ( 2012 ) . Small mammal responses to coarse woody debris distribution at different spatial scales in managed and unmanaged boreal forests. Forest Ecology and Management , 266 :194 205 . - Filotas , E . , Parrott , L . , Burton , P. J . , Chazdon , R . L . , Coates , K . D . , Coll , L . , Haeussler , S . , Martin , K . , Nocentini , S . , Puettmann , K . J . , Putz , F . E . , Simard , S. W . , and Messier , C . ( 2014) . Viewing forests through the lens of complex systems science . Ecosphere , 5 (1) . Forest Practices Board ( 2012 ) . Conserving Old-growth Forests in BC: Implementation of old-growth retention objectives under FRPA , Special Investigation . Franklin , J . ( 1989 ) . Towards a new forestry. American Forests, 95 (11-12 ) . Franklin , J . F. , Spies , T . A. , Van Pelt , R . , Carey, A . B . , Thornburgh , D . A . , Berg , D . R . , Lindenmayer , D . B . , Harmon , M . E . , Keeton , W . S . , Shaw , D . C . , and others ( 2002 ) . Disturbances and structural development of natural forest ecosystems with silvicultural implications , using Douglas-hr forests as an example. Forest Ecology and Management , 155 ( l ) :399 423. - Fraver , S. , Ducey, M . J . , Woodall , C. W. , D’Amato, A . W . , Milo , A . M . , and Palik , B . J . ( 2018) . Influence of transect length and downed woody debris abundance on precision of the line-intersect sampling method . Forest Ecosystems , 5 (1) . Fredeen , A . L . , Bois , C . H . , Janzen , D . T . , and Sanborn , P. T . ( 2005 ) . Comparison of coniferous forest carbon stocks between old-growth and young second-growth forests on two soil types in central British Columbia , Canada . Canadian Journal of Forest Research, 35 ( 6 ) :1411 1421. Publisher : NRC Research Press . - Freschet , G . T . , Weedon , J . T . , Aerts , R. , van Hal , J . R. , and Cornelissen , J . H . C . ( 2012 ) . Interspecific differences in wood decay rates : insights from a new short-term method to study long-term wood decomposition : New method to assess wood decay dynamics and rates . Journal of Ecology , 100 (1) :161 170 . - Gao , T. , Hedblom , M . , Emilsson , T . , and Nielsen , A . B. ( 2014 ) . The role of forest stand structure as biodiversity indicator . Forest Ecology and Management , 330:82 93. - Gorley, A . and Merkel , G . ( 2020 ) . A new future for old forests : A strategic review of how British Columbia manages for old forests within its ancient ecosystems . Technical report , Province of British Columbia , Victoria , BC . Gulbrandsen , L . H . ( 2004 ) . Overlapping public and private governance : Can forest certification fill the gaps in the global forest regime? Global Environmental Politics , 4 ( 2 ) : 75 99. - Gullison , R . E . ( 2003) . Does forest certification conserve biodiversity ? Oryx, 37 ( 02 ) . 64 Guo , X . , Coops , N . C . , Tompalsld , P. , Nielsen , S . E . , Bater , C . W. , and John Stadt , J . ( 2017) . Regional mapping of vegetation structure for biodiversity monitoring using airborne lidar data . Ecological Informatics, 38:50-61. Harmon , M . , Franklin , J . , Swanson , F . , Sollins , P. , Gregory, S . , Lattin , J . , Anderson , N . , Cline , S. , Aumen , N . , Sedell , J . , Lienkaemper , G . , Cromack , K . , and Cummins , K . (1986a ) . Ecology of Coarse Woody Debris in Temperate Ecosystems . In Advances in Ecological Research , volume 34 , pages 59-234 . Elsevier . Harmon , M . E .. Franklin , J . F . , Swanson , F. J . , Sollins , P. , Gregory, S . V . , Lattin , J . D . , Anderson , N . H . , Cline , S . P. , Aumen , N . G . , Sedell , J . R. , Lienkaemper , G . W. , Cromack Jr . , K . , and Cummins , Iv . W. (1986b ) . Ecology of Coarse Woody Debris in Temperate Ecosystems. In Ford , A . M . a . E . D . , editor , Advances in Ecological Research , volume 15 , pages 133 302 . Academic Press . - Harrison , P. , Berry, P. , Simpson , G . , Haslett , J . , Blicharska , M . , Bucur , M . , Dunford , R . , Egoh , B. , Garcia-Llorente , M . , Geamana , N . , Geertsema , W . , Lommelen , E. , Meiresonne , L . , and Turkelboom , F . ( 2014 ) . Linkages between biodiversity attributes and ecosystem services : A systematic review . Ecosystem Services , 9:191-203. Hart , S . A . and Chen , H . Y . H . ( 2006 ) . Understory Vegetation Dynamics of North American Boreal Forests. Critical Reviews in Plant Sciences , 25 ( 4 ) : 381-397. Hart , S . C . , Massicotte , H . B . , Rutherford , P. M . , Elkin , C . M . , and Rogers , B. J . ( 2019 ) . Early response of understory vegetation to wood ash fertilization in the sub boreal climatic zone of British Columbia . The Forestry Chronicle , 95 ( 02 ) :135 142 . - Hengl , T . , Nussbaum , M . , Wright , M . N . , Heuvelink , G . B. M . , and Graler , B. ( 2018 ) . Random forest as a generic framework for predictive modeling of spatial and spatio. , 6:e5518. Publisher: PeerJ Inc. temporal variables. Peer! ILMB ( 2009 ) . Landscape Biodiversity and PG TSA Research Forests Guidance . Isenburg, M . ( 2016 ) . LAStools - efficient LiDAR processing software. Jenness , J . S . ( 2004 ) . Calculating landscape surface area from digital elevation models. Wildlife Society Bulletin, 32 ( 3) :829 839 . - Joyce , M . J . , Erb , J . D . , Sampson , B . A . , and Moen , R . A . ( 2019 ) . Detection of coarse woody debris using airborne light detection and ranging ( LiDAR ) . Forest Ecology and Management , 433: 678 689 . - Kalonga , S . K . , Midtgaard , F . , and Klanderud , K . ( 2016 ) . Forest certification as a policy option in conserving biodiversity: An empirical study of forest management in Tanzania. Forest Ecology and Management , 361:1 12 . - Kane , V . R . , Bakker , J . D . , McGaughey, R . J . , Lutz , J . A . , Gersonde , R . F . , and Franklin , J . F . ( 2010a ) . Examining conifer canopy structural complexity across forest ages and elevations with LiDAR data . Canadian Journal of Forest Research, 40 ( 4 ) : 774 787. - Kane , V . R . , McGaughey, R . J . , Bakker , J . D . , Gersonde , R . F . , Lutz , J . A . , and Franklin , 65 J . F . ( 2010 b ) . Comparisons between fielcl-and LiDAR-based measures of stand structural complexity. Canadian Journal of Forest Research , 40 ( 4 ) : 761 773. - Keisker , D . G . ( 2000 ) . Types of wildlife trees and coarse woody debris required by wildlife of north- central British Columbia. Number 50 in Working paper . British Columbia , Ministry of Forests Research Program , Victoria , BC . OCLC: ocm46855461. Kimmins, J . P. ( 1996 ) . Forest ecology. Number Book , Whole . Prentice Hall , Upper Saddle River , N . J , 2 nd edition . Klinkenberg , ed . , B. ( 2021) . E- Flora BC: Electronic Atlas of the Flora of British Columbia [eflora.be.ca]. Lab for Advanced Spatial Analysis , Department of Geography, University of British Columbia , Vancouver , BC . Kumar , P. , Chen , H . Y. H . , Thomas , S . C . , and Shahi, C . ( 2018a ) . Epixylic vegetation abundance , diversity, and composition vary with coarse woody debris decay class and substrate species in boreal forest . Canadian Journal of Forest Research. Publisher : NRC Research Press. Kumar , P. , Chen , H . Y. H . , Thomas , S . C . , and Shahi , C . ( 2018b ) . Linking resource availability and heterogeneity to understorey species diversity through succession in boreal forest of Canada . Journal of Ecology , 106 ( 3) :1266 1276 . eprint : https : // besjournals.onlinelibrary. wiley. com / doi / pdf / 10.1111/1365- 2745.12861. - Kunttu , P. , Junninen , K . , and Kouki , J . ( 2015 ) . Dead wood as an indicator of forest naturalness : A comparison of methods. Forest Ecology and Management , 353:30 40 . - Langor , D . W . , Hammond , H . J . , Spence , J . R . , Jacobs , J . , and Cobb , T . P. ( 2008) . Saproxylic insect assemblages in Canadian forests: diversity, ecology, and conservation. The Canadian Entomologist , 140 ( 4 ) : 453 474. - Lee , S.-I. , Spence , J . R . , and Langor , D . W . ( 2017) . Combinations of aggregated and dispersed retention improve conservation of saproxylic beetles in boreal white spruce stands. Forest Ecology and Management , 385:116 126. - Legendre, P. and Gallagher , E . D . ( 2001) . Ecologically meaningful transformations for ordination of species data . Oecologia, 129 ( 2 ) : 271 280. - Levin , S . A . (1998) . Ecosystems and the biosphere as complex adaptive systems . Ecosystems , l ( 5 ) :431 436. - Macdonald , S . E . and Fenniak , T . E . ( 2007 ) . Understory plant communities of boreal mixedwood forests in western Canada : Natural patterns and response to variableretention harvesting. Forest Ecology and Management , 242 (1) : 34 48 . - Manak , V. and Jonsell , M . ( 2017) . Beetle diversity in two types of fine woody debris : lessons for bioenergy harvest . Forestry: An International Journal of Forest Research,, 90 ( l ) :82 87. Publisher : Oxford Academic . - McElhinny, C . , Gibbons , P. , Brack , C . , and Bauhus , J . ( 2005 ) . Forest and woodland stand 66 structural complexity : Its definition and measurement . Forest Ecology and Management , 218 (1-3) :1—24 . Melin , M. , Matala , J . , Mehtatalo , L . , Pusenius , J . , and Packalen , P. ( 2016a ) . Ecological dimensions of airborne laser scanning — Analyzing the role of forest structure in moose habitat use within a year . Remote Sensing of Environment , 173: 238-247. Melin , M . , Mehtatalo , L . , Miettinen , J . , Tossavainen , S . , and Packalen , P. ( 2016b ) . Forest structure as a determinant of grouse brood occurrence An analysis linking LiDAR data with presence / absence field data . Forest Ecology and Management , 380: 202 211. - - Messier , C . , Puettmann , K . , Chazdon , R . , Andersson , K . , Angers , V . , Brotons , L . , Filotas , E . , Tittler , R . , Parrott , L . , and Levin , S. ( 2015) . From Management to Stewardship : Viewing Forests As Complex Adaptive Systems in an Uncertain World : From management to stewardship. Conservation Letters, 8 ( 5 ) :368 377. - Messier , C . C . , Puettman , K . J . , and Coates , K . D . , editors ( 2013) . Managing forests as complex adaptive systems: building resilience to the challenge of global change . The Earthscan forest library. Earthscan from Routledge, Abingdon , Oxon , first edition edition . OCLC : 768171495. Mladenoff , D . J . , White , M . A . , Pastor , J . , and Crow , T . R . (1993) . Comparing Spatial Pattern in Unaltered Old-Growth and Disturbed Forest Landscapes . Ecological applications , 3( 2 ) : 294-306 . Muller , J . and Butler , R . ( 2010 ) . A review of habitat thresholds for dead wood : a baseline for management recommendations in European forests . European Journal of Forest Research, 129 ( 6 ) : 981-992 . MSRM ( 2004 ) . Prince George Timber Supply Area Biodiversity Order Background Report . pdf . Newberry, J . E. , Lewis , K . J . , and Walters , M . B . ( 2004 ) . Estimating time since death of Picea glauca x P. Engelmannii and Abies la.siocarpa in wet cool sub- boreal spruce forest in east-central British Columbia . Canadian Journal of Forest Research, 34 ( 4 ) : 931-938. Naesset , E . ( 2002 ) . Predicting forest stand characteristics with airborne scanning laser using a practical two-stage procedure and field data . Remote Sensing of Environment , 80 ( l ) :88 99. - Naesset , E . ( 2014 ) . Area-Based Inventory in Norway From Innovation to an Operational Reality . In Maltamo , M . , Naesset , E . , and Vauhkonen , J . , editors , Forestry Applications of Airborne Laser Scanning , number 27 in Managing Forest Ecosystems , pages 215 240 . Springer Netherlands. - - Oksanen , J . , Blanchet . F. G . , Friendly, M . , Kindt , R . , Legendre , P. , McGlinn , D . , Minchin , P. R . , O ’ Hara , R . B . , Simpson , G . L . , Solymos , P. , Stevens , M . H . H . , Szoecs , E . , and Wagner , H . ( 2019 ) . vegan: Comm/ unity Ecology Package https : // cran .r-project . org , https : // gitlrub . com / vegandevs / vegan . , 67 Parrott , L . ( 2010 ) . Measuring ecological complexity. Ecological Indicators , 10 ( 6 ) :1069— 1076 . Parrott , L . and Lange , H . ( 2013) . An introduction to complexity science . In Messier , C . , Puettmann , K . , and Coates , K . , editors , Managing forests as complex adaptive systems. Building resilience to the challenge of global change . Routledge , London, pages 79-109. Routledge . Pebesma , E . ( 2018 ) . Simple Features for R: Standardized Support for Spatial Vector Data . The R Journal , 10 ( l ):439 446. - Pebesma , E . ( 2020 ) . stars: Spatiotemporal Arrays , Raster and Vector Data Cubes. https : // r-spatial . github . io / stars / , https : // github . com / r-spatial / stars / . Pedersen , L . ( 2003) . Allowable annual cuts in british Columbia: The agony and the ecstasy : Address. Jubilee lecture series. Faculty of Forestry, University of British Columbia . Price , K . , Holt , R . , and Daust , D . ( 2020 ) . BC ’s old growth forest : A last stand for biodiverity. Technical report , Sierra Club BC . QGIS Development Team ( 2019 ) . QGIS Geographic Information System. Open Source Geospatial Foundation . R Core Team ( 2019 ) . R: A Language and Environment for Statistical Computing . R Core Team ( 2020 ) . R : A Language and Environment for Statistical Computing . R Foundation for Statistical Computing, Vienna , Austria . Rousse , J .-R . ( 2018) . lidR : Airborne LiDAR Data Manipulation and Visualization for Forestry Applications . Roussel , J .-R . and Auty, D . ( 2020 ) . lidR : Airborne LiDAR Data Manipulation and Visualization for Forestry Applications. R package version 3.0 . 3. Sandstrom , J . , Bernes , C . , Junninen , K . , Lohrnus , A . , Macdonald , E . , Muller , J . , and Jonsson , B . G . ( 2019 ) . Impacts of dead wood manipulation on the biodiversity of temperate and boreal forests . A systematic review. Journal of Applied Ecology , 56 ( 7) :1770 1781. - Schmidt , R . (1992 ) . The Aleza Lake Experiment Station (1920’s and 1930’s) . Technical report , BC Ministry of Forests , Victoria , B . C . SFI ( 2015 ) . SFI 2015-2019 Standards and Rules: Standards , Rules for Label Use , Procedures and Guidance . Siitonen , J . ( 2001) . Forest Management , Coarse Woody Debris and Saproxylic Organisms : Fennoscandian Boreal Forests as an Example. Ecological Bulletins , ( 49 ) :11 41. - Spies , T . A . , Franklin , J . F . , and Thomas , T. B . (1988) . Coarse Woody Debris in Douglas- Fir Forests of Western Oregon and Washington . Ecology , 69 ( 6 ) :1689 1702 . - Stevenson , S. K . , Armleder , H . M . , Aresenault , A . , Coxson , D . , DeLong , S . C . , and Jull , M . ( 2011) . British Columbia ’s inland rainforest : ecology , conservation , and management. Number Book , Whole. UBC Press , Vancouver . 68 Stokland , J . N . , Siitonen , J . , and Jonsson , B . G . ( 2012 ) . Biodiversity in Dead Wood . Ecology, Biodiversity and Conservation . Cambridge University Press , Cambridge . Stone , J . N . , MacKinnon , A . , Parminter , J . V . , and Lertzman , K . P. (1998) . Coarse woody debris decomposition documented over 65 years on southern Vancouver Island . 28: 9 / Sturtevant , B . R . , Bissonette , J . A . , Long , J . N . , and Roberts , D . W . (1997) . Coarse Woody Debris as a Function of Age , Stand Structure , and Disturbance in Boreal Newfoundland . Ecological Applications, 7 ( 2 ) : 702 . Sverdrup-Thygeson , A . , 0rka , H . 0 . , Gobakken , T . , and Naesset , E . ( 2016 ) . Can airborne laser scanning assist in mapping and monitoring natural forests? Forest Ecology and Management . 369 :116 125. - Tanhuanpaa , T . , Kankare , V . , Vastaranta , M . , Saarinen , N . , and Holopainen , M . ( 2015) . Monitoring downed coarse woody debris through appearance of canopy gaps in urban boreal forests with bitemporal ALS data . Urban Forestry & Urban Greening , 14 ( 4 ) :835 843. - Tipper , H . (1971) . Glacial Geomorphology and Pleistocene History of Central British Columbia . Bulletin 196 . Tompalski , P. , White , J . C . , Coops , N . C . , and Wulder , M . A . ( 2019 ) . Quantifying the contribution of spectral metrics derived from digital aerial photogrammetry to area-based models of forest inventory attributes. Remote Sensing of Environment , 234:111434 . Ulyshen , M . D . , Horn , S. , Pokswinski , S . , McHugh , J . V . , and Hiers , J . Iv . ( 2018) . A comparison of coarse woody debris volume and variety between old-growth and secondary longleaf pine forests in the southeastern United States. Forest Ecology and Management. 429 :124 132 . - Van Wagner , C . E . ( 1968) . The Line Intersect Method in Forest Fuel Sampling. Forest Science , 14 ( l ) :20-26. Vauhkonen , J . and Imponen , J . ( 2016 ) . Unsupervised classification of airborne laser scanning data to locate potential wildlife habitats for forest management planning. Forestry , 89 ( 4 ) : 350 363. - Venables , W . N . and Ripley, B. D . ( 2002 ) . Modern Applied Statistics with S. Springer , New York , fourth edition . ISBN 0-387-95457-0 . Venier , L . , Thompson , I . , Fleming , R . , Malcolm , J . , Aubin , I. , TrofymowJ . A , LangorD , SturrockR , PatryC , OuterbridgeR . O , HolmesS . B , HaeusslerS , GrandpreL , D . , ChenH . Y . H , BayneE , ArsenaultA , and BrandtJ . P ( 2014 ) . Effects of natural resource development on the terrestrial biodiversity of Canadian boreal forestsl . Environmental Reviews. Publisher : NRC Research Press . Warton , D . I. and Hudson , H . M . ( 2004 ) . A MANOVA statistic is just as powerful as distance- based statistics, for multivariate abundances. Ecology, 85 ( 3) :858 874 . - 69 White , J . , Canada , Natural Resources Canada , and Canadian Wood Fibre Centre ( 2017) . A model development and application guide for generating an enhanced forest inventory using airborne laser scanning data and an area- based approach. Canadian Wood Fibre Centre , Victoria , B . C . OCLC : 1021855287. White , J . , Wulder , M . A . , Varhola , A . , Vastaranta , M . , Coops , N . C . , Cook , B . D . , Pits , D . , and Woods , M . ( 2013 ) . A best practices guide for generating forest inventory attributes from airborne laser scanning data using the area- based approach. Information Report FI-X-10 . Natural Resources Canada , Canadian Forest Service . White , J . C . , Coops , N . C . , Wulder , M . A . , Vastaranta , M . , Hilker , T . , and Tompalski , P. ( 2016 ) . Remote Sensing Technologies for Enhancing Forest Inventories: A Review. Canadian Journal of Remote Sensing , 42 ( 5) :619 641. - Wright , M . N . and Ziegler , A . ( 2017) . ranger : A fast implementation of random forests for high dimensional data in C ++ and R . Journal of Statistical Software , 77 (1) :1—17. Wulder , M . , Coops , N . , Morsdorf , F . , Nelson , R . , Newnham , G . , and Vastaranta , M . ( 2013 ) . Status and prospects for LiDAR remote sensing of forested ecosystems. Canadian Journal of Remote Sensing , 39 ( S1) :S1 S5. - Wulder , M . A . , Bater , C . W. , Coops , N . C . , Hilker , T . , and White , J . C . ( 2008a ) . The role of LiDAR in sustainable forest management . The Forestry Chronicle , 84 ( 6 ) :807 826 . - Wulder , M . A . , Bater , C . W . , Coops , N . C . , Hilker , T . , and White , J . C . ( 2008b ) . The role of LiDAR in sustainable forest management . The Forestry Chronicle , 84 ( 6 ) :807 826 . - Zellweger , F . , Morsdorf , F . , Purves , R . S . , Braunisch , V . , and Bollmann , K . ( 2014) . Improved methods for measuring forest landscape structure: LiDAR complements field- based habitat assessment . Biodiversity and Conservation, 23 ( 2 ) : 289 307. - 70 Appendix A Mean Species Abundance 71 Table A. l : Mean Hellinger-transformed species abundance from linear discriminant analysis using the species that were observed a minimum of forty times . Fn Genus Species SCode Bryophytes Bryophytes Bryophytes Bryophytes Bryophytes Bracliythecium Hylocomium Plagiomnium Pleurozium Ptilium spp splendens medium schreberi crista-castrensis imm mat OG wOG BRACHSPP HYLOCSPLEN PLAGIMEDIU PLEURSCHRE PTILICRIST 0.168 0.024 0.122 0.183 0.121 0.169 0.098 0.078 0.191 0.243 0.188 0.078 0.123 0.158 0.185 0.189 0.133 0.102 0.123 0.214 0.167 0.091 0.161 0.168 0.376 RHIZOGLABR RHYTITRIQU SPHAGSQUAR 0.040 0.322 0.082 0.024 0.133 0.135 0.077 0.085 0.302 0.331 0.185 0.014 0.009 0.289 0.014 0.002 0.016 0.120 0.149 0.006 Bryophytes Rhizomnium Bryophytes Rhytidiadelphus Bryophytes Sphagnum Bryophytes Timmia Forbes Aralia triquetrus squarrosum austriaca nudicaulis TIMMIAUSTR ARALINUDIC 0.047 0.139 0.013 0.023 0.128 Forbes Forbes Forbes Forbes Forbes filix-femina uniflora canadensis expansa arvense ATHYRFILIX CLINTUNIFL CORNUCANAD DRYOPEXPAN EQUISARVEN 0.106 0.074 0.246 0.024 0.018 0.053 0.178 0.083 0.105 0.086 0.115 0.147 0.157 0.197 0.038 0.104 0.085 0.012 0.011 0.003 0.121 0.000 0.143 0.111 0.047 EQUISSYLVA GALIUTRIFI GALIUTRIFL GYMNODISJU LINNABOREA 0.023 0.012 0.042 0.135 0.086 0.032 0.016 0.005 0.011 0.005 0.025 0.151 0.226 0.058 0.037 0.006 0.004 0.017 0.197 0.039 0.080 0.000 0.015 0.101 0.016 0.036 0.021 0.048 0.031 0.089 Athyrium Clintonia Cornus Drvopteris Equisetum Equisetum glabrescens juv Forbes Forbes Forbes Forbes Forbes Galium Galium Gymnocarpium Linnaea sylvaticum trifidum triflorum disjunctum borealis Forbes Forbes Forbes Forbes Forbes Lycopodium Mitella Petasit. es Prosart. es Rubus clavatum nuda palmatus hookeri pedatus LYCOPCLAVA MITELNUDA PETASPALMA PROSAHOOKE RUBUSPEDAT 0.014 0.025 0.063 0.056 0.022 0.047 0.048 0.018 0.027 0.120 0.032 0.036 0.009 0.085 0.105 0.038 0.017 0.000 0.000 0.146 Forbes Forbes Forbes Shrubs Shrubs Rubus Streptopus Tiarella Alnus Lonicera pubescens lanceolatus trifoliata viridis involucrata RUBUSPUBES STREPLANCE TIARETRIFO ALNUSVIRID LONICINVOL 0.038 0.039 0.010 0.087 0.135 0.152 0.125 0.151 0.228 0.080 0.068 0.014 0.192 0.071 0.033 0.026 0.151 0.183 0.062 0.058 0.056 0.028 0.178 0.083 0.084 Shrubs Shrubs Shrubs Shrubs Shrubs Oplopanax Ribes Rosa Rubus Spiraea horridus lacustre acicularis parviflorus betulifolia OPLOPHORRI RIBESLACUS ROSAACICU RUBUSPARVI SPIRABETUL 0.029 0.079 0.177 0.192 0.038 0.040 0.057 0.049 0.110 0.055 0.039 0.073 0.209 0.138 0.186 0.193 0.107 0.067 0.053 0.045 0.101 0.004 0.041 0.005 0.016 Shrubs Shrubs Shrubs Shrubs Spiraea Vaccinium Vaccinium douglasii menibranaceum ovalifolium Viburnum edule SPIRADOUGL VACCIMEMBR VACCIOVALI VIBUREDULE 0.127 0.104 0.017 0.018 0.025 0.074 0.061 0.109 0.030 0.091 0.055 0.067 0.093 0.074 0.066 0.061 0.179 0.038 0.139 0.013 72