UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Bird-habitat associations and simulated effects of logging on bird habitat in the aspen boreal mixedwood Sharp, Nyree Elizabeth 1998

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

Item Metadata


831-ubc_1998-0617.pdf [ 3.69MB ]
JSON: 831-1.0075428.json
JSON-LD: 831-1.0075428-ld.json
RDF/XML (Pretty): 831-1.0075428-rdf.xml
RDF/JSON: 831-1.0075428-rdf.json
Turtle: 831-1.0075428-turtle.txt
N-Triples: 831-1.0075428-rdf-ntriples.txt
Original Record: 831-1.0075428-source.json
Full Text

Full Text

BIRD-HABITAT ASSOCIATIONS A N D S I M U L A T E D EFFECTS OF L O G G I N G O N BIRD H A B I T A T IN T H E A S P E N B O R E A L M I X E D W O O D  by  N Y R E E E L I Z A B E T H SHARP  B.Sc.(Hon), University of Alberta, 1994  A T H E S I S S U B M I T T E D IN P A R T I A L F U L F I L M E N T O F THE REQUIREMENTS FOR T H EDEGREE OF M A S T E R OF SCIENCE in T H E F A C U L T Y OF G R A D U A T E STUDIES Centre for Applied Conservation Biology Department of Forest Sciences Faculty of Forestry  We accept this thesis as conforming to the required standard  T H E U N I V E R S I T Y O F BRITISH C O L U M B I A September,  1998  © Nyree Elizabeth Sharp, 1998  In  presenting  degree freely  at  this  the  available  copying  of  department publication  of  in  partial  fulfilment  University  of  British  Columbia,  for  this or  thesis  reference  thesis by  this  for  his thesis  and  study.  scholarly  or for  her  of  Po Ccj^r  The University of British Vancouver, Canada  financial  DE-6  (2/88)  6cCe^y Ce~S> Columbia  I further  purposes  gain  the  requirements  I agree  shall  that  agree  may  representatives.  permission.  Department  of  It not  be is  that  the  Library  permission  granted  by  understood be  for  allowed  an  advanced  shall for  the that  without  make  it  extensive  head  of  my  copying  or  my  written  Abstract It has recently become profitable in the Alberta mixedwood boreal forest to harvest aspen trees for pulp and paper production. Logging practices have the potential to alter the physical structure and resident vertebrate communities of the forest. Simulation models are an important tool in the study of such broad-scale habitat alterations. S I M F O R is a spatially-explicit model that integrates the predicted effects of forest harvest on specific habitat attributes and the response to these effects by vertebrate species. I explored these effects by focusing on the resident bird community. I first grouped the bird species into habitat-dependency groups, or guilds, based on their life history requirements. Then, using data provided by an extensive study in northern Alberta, I developed statistical relationships between these guilds and certain habitat attributes which life history characteristics suggest would be important for those species. I also developed projections of the availability of these attributes over time, using existing data and modelling. I entered these trends and relationships into S I M F O R , along with maps and harvest schedules for a representative township in Alberta. This allowed me to project the amount of suitable habitat that will be available over time for each guild, and also the spatial distribution of that habitat. The bird guilds exhibited a variety of responses to simulated logging. Those with more specific habitat requirements, such as primary cavity nesters, were predicted to decline under current forest management plans. Those with less specific habitat requirements and a preference for early serai stages were predicted to respond positively to logging in the short term. These results suggest that a variety of stand ages should be maintained over the management area, with a particular emphasis on providing a continuous supply of those attributes of old stands with  ii  which some guilds are closely associated. Adaptive management would seem the most efficient way of evaluating the effects of management decisions on bird habitat and allowing alternatives to be explored.  iii  Table of Contents ABSTRACT  ii  LIST O F T A B L E S  v  LIST O F FIGURES  vi  ACKNOWLEDGEMENTS  vii  CHAPTER 1 INTRODUCTION  1  1.1 T H E B O R E A L M I X E D W O O D  2  1.2 T H E S T U D Y A R E A  3  1.3 H A B I T A T M O D E L L I N G  4  1.4 S I M F O R  5  C H A P T E R 2 GUILDING AND H A B I T A T ASSOCIATIONS  8  2.1 I N T R O D U C T I O N  8  2.2 M E T H O D S  9  2.2.1 The Guilds : 2.2.2 Habitat Associations 2.2.3 Forest Structure Dynamics  9  2.3 R E S U L T S  9 11  11  2.3.1 The Guilds 2.3.2 Habitat Associations 2.3.3 Forest Structure Dynamics  11 13 15  2.4 D I S C U S S I O N  19  2.4.1 Habitat Associations 2.4.2 Forest Structure Dynamics  19 22  C H A P T E R 3 SIMULATION PROJECTIONS  24  3.1 I N T R O D U C T I O N  24  3.2 M E T H O D S  25  3.2.1 Simulation modelling 3.2.2 Randomization and Monte Carlo  25 27  3.3 R E S U L T S  28  3.3.1 Simulation modelling 3.3.2 Randomization and Monte Carlo  28 34  3.4 D I S C U S S I O N  35  3.4:1 Simulation modelling 3.4.2 Randomization and Monte Carlo  35 39  C H A P T E R 4 CONCLUSIONS AND M A N A G E M E N T RECOMMENDATIONS  40  LITERATURE CITED  44  APPENDIX 1  49  APPENDIX n  52 iv  List of Tables Table 1. Multiple regression equations relating bird guild abundance to habitat attributes  13  Table 2. Distributions of habitat attributes and predicted large primary cavity nester abundance for each of three age classes  35  List of Figures Figure 1. CART diagram for secondary cavity nesters  15  Figure 2. CART diagram for tree nesters  15  Figure 3. Density of conifer stems (dbh >10 cm) in the canopy over time  16  Figure 4. Density of live trees (dbh >10 cm) in the canopy over time  17  Figure 5. Ratio of coniferous to deciduous tree density over time  17  Figure 6. Percent cover of the low shrub layer over time  18  Figure 7. Median snag dbh (dbh >10 cm) over time  19  Figure 8. Area of high quality habitat for large primary cavity nesters over time  29  Figure 9. Area of high quality habitat for weak primary cavity nesters over time  30  Figure 10. Area of high quality habitat for secondary cavity nesters or ground nesters over time  31  Figure 11. Quality of habitat available for secondary cavity nesters over time  32  Figure 12. Quality of habitat available for ground nesters over time  32  Figure 13. Area of high quality habitat for tree nesters over time  33  Figure 14. Area of high quality habitat for shrub nesters over time  33  Figure 15. Area of high quality home range for large primary cavity nesters over time  34  vi  Acknowledgements There are many people I need to thank, without whose help this thesis would not be sitting in front of you. I would first like to thank my committee: Fred Bunnell, David Tait, Phil Lee and Ralph Wells for their assistance and the different perspective each brought to my research. I am greatly indebted also to Phil Lee, Jim Schieck, Larry Roy and other researchers at the Alberta Research Council (formerly the Alberta Environmental Centre) in Vegreville, Alberta for their generous sharing of data and time spent helping me track it all down. Alberta Pacific Forest Industries (AlPac) provided maps and harvest schedules for my model and provided useful input on an earlier stage of my work. For funding I am grateful to N S E R C for a postgraduate scholarship and the Network of Centres for Excellence in Sustainable Forest Management. I am very grateful to Glenn Sutherland and Ralph Wells for various support along the way and programming changes to S I M F O R which allowed me, among other things, to use multivariate models. I would specifically like to thank Glenn for providing me with numerous references, reviewing some of my thoughts, and helping me through the C A R T process. Pierre Vernier also took much time to help me with sundry bits of computing and to pull me up by my "bootstraps". I would also like to thank T i m Shannon for his time spent converting GIS files into a format S I M F O R could understand. I also owe a great deal to Steve Cumming for his time spent modelling, analyzing several townships, and providing insight into boreal mixedwood stand dynamics, as well as providing me with background and various references. I would also like to thank Fiona Schmiegelow for her assistance in interpreting the results from my bird data analyses, and providing me with background for my research. Special thanks go to Bob Mooney, James McCormick, Scott Harrison, and Samantha Berry for their support, editing and technical suggestions, without which I would not have made it through this. Also to Cathy Sharp for many evenings spent inspiring each other in the trailer, and to the rest of my family and friends for their invaluable support. I also greatly appreciate the encouragement, camaraderie, and advice of fellow students and researchers at the Centre for Applied Conservation Biology, and the assistance provided by Jackie Johnson, who keeps it all together for us. I would like to dedicate this thesis to Agnes Wood (Nan) Sharp, a constant source of support and inspiration.  vii  C H A P T E R 1 Introduction In northern Canada, logging activities have become an increasingly common factor in large-scale modification of natural habitats. Strategies for logging in a given area are often chosen without full consideration of the various options available and their potential impacts on wildlife species, i f these are even known. The mixedwood region of the boreal forest of northern Alberta was largely untouched by harvest practices until the last decade, when improved technologies in the processing of hardwood fibre made it economically profitable to produce pulp and paper from aspen trees (Schmiegelow and Harmon 1993; Stelfox 1995). This lack o f logging history provides uncommon flexibility to explore different configurations of harvest type, intensity of harvest, and the spatial arrangement of cutblocks, all of which can have substantial effects on wildlife habitats (e.g., Bunnell et al. 1997, 1998). Logging practices alter both the age distribution and physical structure o f forested areas. Areas of old forest are often reduced and specific habitat attributes important to the survival o f some vertebrate species may be lost. This alteration in habitat structure and availability can alter the competitive balance between species of a given taxonomic group (Welsh 1981), often leading to the displacement o f species which depend on old growth or certain habitat features by more generalist species. Fragmentation of suitable habitat that is left behind also creates opportunities for generalist or open-area species to displace those species preferring interior forest habitat (Franklin and Forman 1987; Schmiegelow et al. 1997). The effects of forest fragmentation are particularly apparent for bird species. Edge habitat can lead to increased nest parasitism and predation, and certain groups such as cavity nesters have been identified as being particularly sensitive to fragmentation and loss of older  1  forest habitat (Schmiegelow 1997), although the immediate effects appear to vary with species (Schmiegelow et al. 1997). Neotropical migrants, which make up the greatest number of bird species in the boreal mixedwood, are also thought to be sensitive (Schmiegelow and Hannon 1993) and to have less time than resident species to adapt to changes in this region, as they are only present for a few months of the year (Schmiegelow 1997). Bird species are the richest vertebrate taxonomic group in the boreal mixedwood forest (Schmiegelow and Hannon 1993), and have been used as indicators of change in forest condition in the eastern region of this biome (Welsh 1981). For these reasons, I chose to focus on birds to examine the effects of forest harvest on wildlife habitat in the boreal mixedwood forest. M y broad objective is to develop a methodology for projecting vertebrate responses to forest practices in the boreal mixedwood. Within this context, I have two specific objectives: 1) to develop relationships between bird abundance and forest habitat attributes; 2) to assess the probable effects o f logging on bird abundance. The first objective was predicated on the observation that bird species appear to be more closely associated with specific vegetation structures, which provide cover, foraging and nesting sites, than with the age of a given stand (Welsh 1981; Bunnell et al. 1998). T o address the second objective, I used simulation modelling to project these attributes through time so that the relationships established under the first objective could be used to predict expected bird abundance over the landscape.  1.1.The Boreal Mixedwood The boreal mixedwood forest is an extremely complex ecosystem, stretching from northeastern British Columbia across the central and northern parts of Alberta and Saskatchewan into south-western Manitoba. The landscape is a mosaic o f stands that vary in composition, age and  2  size (Peterson and Peterson 1992). Fire is the primary natural disturbance that has shaped this ecosystem, with insect outbreaks and windthrow also playing roles. Fire cycles that vary in both temporal and spatial scale have created natural heterogeneity across this landscape, resulting in a high diversity of plant, mammal, bird and insect species in the region (e.g., Cumming et al. 1996). Current forestry guidelines in Alberta for the mixedwood region suggest clear-cut logging with a harvest rotation age of 60 to 70 years (Alberta Pacific Forest Industries Inc. 1992). The possible effects of intensive, short-rotation logging on the integrity of the boreal mixedwood ecosystem are a cause for concern. Older forest stands have been found to be the most structurally diverse, with the highest variety of plant species (Lee et al. 1995). This diversity in vegetation and structure is reflected in high richness of animal species in old stands, because the diversity of microhabitats provides suitable nesting or denning and feeding sites for a variety of animals. Peterson and Peterson (1992) have suggested that rotation ages shorter than 80 years could lead to a decline in abundance of bird species dependent on old forests.  1.2 T h e Study A r e a  T o address my objectives, I required data for both vegetation and wildlife parameters over a sizeable area. The Alberta Environmental Centre (now the Alberta Research Council), in Vegreville, Alberta, carried out an extensive biodiversity study in the aspen-dominated boreal mixedwood of Alberta between 1991 and 1994. Their research covered a wide variety of plant, bird and mammal families, and provided extensive information on the tree species of the system (Stelfox 1995). Data on forest structure and wildlife abundance from this Biodiversity Study  3  were made available to me by researchers at the Centre, allowing me to develop relationships between bird species and specific attributes of the forest. The data were collected from twelve stands with aspen canopy cover of at least 80%. A l l stands were of fire origin and greater than 75 ha in size. The stands were located near Athabasca and L a c L a Biche, Alberta, in the southern boreal forest, on mesic sites with sandy clay loam soils (Stelfox 1995). Three ages of stands were chosen: young (20-30 years), mature (50-65 years) and old (120 years or older). Four stands of each age were selected and six random sites of 100-m radius were chosen in each stand, giving a total of 72 sites (see Introduction in Stelfox 1995 for exact locations o f sites). Young stands were characterized as having substantial predisturbance relics, such as large dead trees and downed woody material. Mature stands had the lowest structural diversity, with closed canopies, low plant species diversity and little understory. They were also of about the standard age of harvest rotation in the boreal mixedwood (Stelfox 1995). O l d stands were the most diverse, with a wide range of sizes of live and dead trees, canopy gaps and downed woody material. The detailed vegetation information collected in the three age classes gave me the base from which to develop projections o f specific habitat attributes over time.  1.3 Habitat Modelling  Forest management and efforts to sustain biological diversity necessarily encompass large areas (Bunnell 1992, 1997). Taking a landscape or regional perspective greatly adds to the complexity o f dealing with managed systems. Models provide an important means of both simplifying and visualizing broad scale problems. Experimental studies at the landscape scale are often logistically impossible to perform or replicate, and may be prohibitively expensive even 4  i f feasible (Turner et al. 1995). Models allow researchers to perform 'simulated' experiments, where the possible outcomes of many scenarios can be generated over a very short time frame. Experimental manipulations in the field can then focus on exploring those scenarios which appear to best suit management and conservation goals. Models which relate wildlife populations to attributes of forested habitats have been in use for at least three decades (Bunnell 1989). Habitat suitability models were developed to explore the response of wildlife species to changes in their habitat. However, the relationships among environmental factors and wildlife species were often poorly developed and the spatial configuration of habitat patches ignored (Van Home and Wiens 1991). Recognition that the landscape mosaic of cover type and habitat patches is essential to the ecological interpretation of model results has led to the development of spatially explicit models to enhance our understanding of complex ecosystem interactions. The context in which habitat patches are located, their size, and degree of isolation from other suitable habitat all affect the population dynamics of species using that habitat. Smaller patches isolated by longer distances and less favourable surroundings are less likely to support viable populations of wildlife species, and reduce opportunities for successful dispersal of individuals from one patch to another (Dunning et al. 1995). The implication is that the area of suitable habitat is effectively smaller, due to these effects, than that predicted by models which are not spatially explicit.  1.4 S I M F O R  S I M F O R is a spatially explicit model that directly addresses the two main criticisms V a n Home and Wiens (1991) had of habitat suitability models. It can incorporate explicit empirical descriptions of the relationships between wildlife species and their habitat, in the form of 5  regression equations, and it provides information on the landscape context and spatial configuration of stand types and habitat patches. It was also the first model to integrate both forestry practices (i.e., large-scale modifications to the landscape) and measures of habitat suitability for wildlife species (e.g., Daust and Bunnell 1992; Daust and Sutherland 1997). There are two modules in S I M F O R (Nelson and Hafer 1996). The landscape module deals with changes in forest structure from natural or anthropogenic disturbance. Maps in raster format (which divides the landscape into cells, generally of 1 ha for S I M F O R ) are provided by the user to describe the current composition (stand type, age) and spatial arrangement of patches. Harvest schedules are generally provided from separate harvest scheduling software, such as A T L A S (Nelson et al. 1996), and are also spatially explicit. These inputs are used to calculate the composition and size distribution of forest patches and the edge effects between them. Projections of the provision o f specific forest habitat attributes over time (supply curves), as well as the use of these attributes by species or groups o f species (response curves), provide input for the habitat module. S I M F O R uses the supply curves to project forest structure over time in logged and unlogged stands, and the response curves to determine the suitability o f individual raster cells and of larger habitat areas for the wildlife species of interest. Data describing the response of species or groups of species to changes in forest structure were generally not available before this study, so earlier studies with S I M F O R used categories of dependence based on expert opinion and general information (e.g., Daust and Sutherland 1997). The extensive dataset made available to me allowed me to develop more specific empirical representations of these relationships for use in S I M F O R . The second chapter of this thesis describes how I quantified relationships between bird species and specific habitat attributes, as well as the projections of these attributes through time.  6  In the third chapter, I describe how I used these predictions in S I M F O R to simulate the effects that logging would have on the supply of habitat attributes and, thus, on bird abundance. Conclusions and management recommendations are summarized in the final chapter.  7  C H A P T E R 2 Guilding and Habitat Associations  2.1 Introduction  Habitat associations have long been considered a useful tool for predicting bird abundance (Hansen et al. 1995). Detailed demographic information (fine-filter approach) is often not available for individual species unless they have been of particular research or conservation interest, and a focus on conserving ecosystems (coarse-filter approach) may overlook important factors. If the goal is conservation of all species representative o f an area, the most efficient method is thus to protect those specific habitats which appear most important to the range of species (Hansen et al. 1993). This habitat can not necessarily be described by reference to a single age category, such as old growth. Specific attributes of the forest habitat are known to be important to certain species. For example, large downed wood provides shelter and nesting sites for small mammals, and large trees or snags are required by cavity nesting birds. These attributes may occur to varying degrees in different serai stages, or under different silvicultural treatments (Bunnell and Kremsater 1990, 1994). Such relationships of wildlife species to particular vegetation or habitat attributes provide another means o f simplifying analysis and management. Animal species can be grouped into habitat-dependency groups, or guilds, based on their relationships to these attributes. Previous studies have used statistical methods to create these groups (e.g., Jaksic 1981; Verner 1984; Jaksic and Medel 1990; Simberloff and Dayan 1991; Daust and Sutherland 1997). I chose to group bird species based on their dependency on certain attributes assumed to be necessary for 8  their nesting or feeding success, such as large snags or downed wood. These assumptions represent current understanding of the species' natural history. I chose to employ natural history instead o f statistical associations to avoid the circularity of creating and testing the relationships using the same data set. The resulting guilds are more easily managed in the field than individual species, and simplify simulation modelling.  2.2 Methods  2.2.1 The Guilds The Biodiversity Study provided abundance data on 57 bird species over 1992 and 1993. T o facilitate modelling, I grouped the bird species, as it would be unwieldy to model all species individually. I classified the bird species into two sets of guilds based on their natural history (see Appendix I). T o do this, I obtained information on the nesting and feeding preferences o f each species from Erlich et al. (1988), Semenchuk (1992) and the Biodiversity Study report (Schieck and Nietfeld 1995). For winter wren and red-eyed vireo there was some uncertainty over which nesting guild would be most appropriate, so I sought expert opinion to determine the best classification for these species in this system (F. Bunnell, F . Schmiegelow, pers. comm.).  2.2.2 Habitat Associations I chose habitat attributes from the complete set of vegetation variables from the Biodiversity Study (see Lee et al. 1995), based on their assumed importance to the bird guilds. For instance, cavity nesters might be expected to be dependent on snag density or snag size. Initially, I performed simple regressions in Excel (Microsoft 1994) of the abundance of each guild against individual attributes that were considered potentially important to that group o f 9  species. Guild abundance was defined as the total number of observations for all member species of the guild over the two years. Community composition was similar between the two years (J. Schieck, pers. comm.). Once I determined that the nesting guilds were providing stronger models than the foraging guilds, I chose to focus on these guilds for the purposes of the simulation. I refined and improved the habitat associations for the nesting guilds by including multiple habitat attributes. I ran stepwise multiple regressions in Systat (Wilkinson et al. 1992) for each guild against all attributes that could be reasonably presumed to be important to that guild, based on current understanding of natural history. Those attributes and constants that were significant at the a=0.05 level were kept in the models. In the case of the secondary cavity nesters, only one attribute remained statistically significant, and its biological significance was unclear, so I used C A R T analysis to examine other potential relationships. C A R T , or Classification and Regression Tree analysis, is a method of exploring data where one response variable is predicted by several classification variables. It can be used to elucidate structure in data and to produce classification rules by sequentially bisecting the dataset into homogeneous subsets, based on the predictor values (Clark and Pregibon 1992). The results are most easily interpreted graphically, with each branch in the tree representing a split and each value at that split representing the value predicted for the response variable i f that path is chosen. I also used C A R T analysis to further explore what biological mechanisms might underlie the results obtained for the tree nesters.  10  2.2.3 Forest Structure Dynamics I used a combination of methods to develop the supply curves over time for each habitat attribute chosen. I used data on the abundance of each attribute from the Biodiversity Study for each of the three age classes (young, mature and old) to obtain average values for years 25, 57.5 and 120. I chose the values for year zero based on discussion with researchers familiar with the system, taking into account the fact that the primary forestry operator in the area has chosen to leave approximately 5 to 6% of merchantable timber in clumps of live and dead trees (Alberta Pacific Forest Industries Inc. 1993). The timelines were extended to 200 years (the projection period for the S I M F O R simulations) using stand dynamics models developed for the boreal mixedwood system (Cumming 1997).  2.3 R e s u l t s  2.3.1 The Guilds Classification based on nesting requirements resulted in six guilds. The cavity nesters showed different trends with respect to habitat attributes in exploratory regressions, and thus were split into three guilds: large primary cavity nesters, those birds that create their own cavities for nesting; weak primary cavity nesters, those birds that create their own cavities but require a more advanced stage of decay; and secondary cavity nesters, those birds that build their nests in pre-existing cavities. I placed winter wrens, which nest in the root wads of fallen trees or snags (Salt and Salt 1976; Campbell et al. 1997) in the large primary cavity nesting guild because they depend on the same attributes (large, old dead trees) as do other members of this group.  11  I defined three other nesting guilds: tree nesters, those species that nest in the crowns o f trees (as opposed to cavities); shrub nesters, which nest in the shrub layer; and ground nesters, which build their nests on the ground. Strong models could not be developed for coniferous and deciduous tree nesters separately, so these were left as a single guild. This finding may occur because few birds are restricted solely to coniferous or deciduous trees, and even small amounts of either conifers or hardwoods in mixtures increase vertebrate species richness (Willson and Comet 1996). I included red-eyed vireo in the shrub nester guild because they nest in shrubs and saplings more than in mature trees in this area (Salt and Salt 1976). A complete list of the species in each guild is presented in Appendix I. The second set o f guilds consisted of five groups defined by the foraging habits of the bird species. Bark gleaners were defined as those species which remove insects from the bark of tree trunks and branches. Foliage gleaners obtain invertebrates from leaves and occasionally branches. A s with the nesting guilds, foliage gleaners associated with deciduous or coniferous tree species did not differ from each other in their response to habitat variables and were left as a single guild. Aerial foragers included species which hawk, hover, or capture prey while in flight. The borers were all woodpecker species, obtaining sap or insects by excavating or drilling into bark. The final group was the ground foragers. In initial exploration, I discovered that the guilds based on nesting preference showed stronger relationships with habitat attributes than did those based on feeding preference. The best fit simple regressions using nesting guilds had higher coefficients of determination (r ) than 2  did those using foraging guilds. For example, the linear relationship of bark gleaners to mean tree dbh (one of the stronger foraging guild models) had a value of r =0.168, whereas almost all 2  nesting guild models had values exceeding r =0.20. Also, birds in the same nesting guild tended 2  12  to be more highly correlated with one another across surveys than did those in the same foraging guild. I therefore chose to use the nesting guilds in developing the habitat association models.  2.3.2 Habitat Associations I chose multiple regression models over single models as the most accurate and best representative of the species-habitat relationships. The proportion of variation accounted for by the multiple models was higher for all guilds than the best single model, and including more than one habitat attribute increased the biological realism as well, by representing more of the associations with habitat variables. The final multiple models included the following attributes: live tree density (LiveTree) and its reciprocal (1/LiveTree), canopy conifer density (Conifer), the ratio of coniferous to deciduous canopy tree density (ConifDec), percent low shrub cover (Shrub) and median snag diameter at breast height (MedSnagDBH). Density of birch trees, snag density, snag volume, volume of downed wood and canopy openness were not found to be significant in any of the models. Final regression equations for all models are shown in Table 1. Tolerance values were high for all models, indicating little correlation between independent (habitat attribute) variables.  T a b l e 1. Multiple regression equations relating bird guild abundance to habitat attributes. Number of sites is 72 and significance level is p<0.001 for all models Guild  Multiple Regression E q u a t i o n  R  Large Primary Cavity Nesters  6.259 ConifDec + 0.106 M e d S n a g D B H - 0.785  0.482  Weak Primary Cavity Nesters  0.032 M e d S n a g D B H + 0.010 Conifer  0.637  Secondary Cavity Nesters  0.0001466 LiveTree  0.503  Tree Nesters  13666.889 1/LiveTree + 0.195 M e d S n a g D B H  0.842  Shrub Nesters  3714.852 1/LiveTree - 10.845 ConifDec + 6.759  0.370  Ground Nesters  0.001 LiveTree + 0.270 Shrub + 18.536  0.265  13  2  The abundance of large primary cavity nesters was found to be positively correlated with both the ratio o f coniferous to deciduous tree density and median snag dbh (R =0.482). Weak 2  primary cavity nesters were positively correlated with median snag diameter as well, and with conifer density (R =0.637). Secondary cavity nesters were positively correlated with only live 2  tree density (R =0.503), and not with any measure o f snags. 2  Tree nesters showed a non-linear relationship to live tree density in the initial analyses, so I used the reciprocal of density to obtain a linear relationship. In the final regression model, tree nesters were positively correlated to this reciprocal measure and to median snag dbh (R =0.842). 2  Shrub nesters were also positively related to the reciprocal of tree density, and were negatively correlated with the ratio of coniferous to deciduous tree density (R =0.370). Ground nesters were 2  positively associated with live tree density and shrub cover (R =0.265). 2  C A R T analyses were performed for both secondary cavity nesters and tree nesters. The results for secondary cavity nesters (Figure 1) provide more detail on the positive association with tree density, as well as indicating a preference for sites with large snags. The diagram for tree nesters (Figure 2) confirms a preference for old stands, as the predicted abundance is higher where tree density is quite low.  14  Tree Density < 4040 stems/ha  Tree Density > 4040 stems/ha  Birch Density < 21 stems/ha Birch Density > 21 stems/ha  Tree Density < 5550 stems/ha  Tree Density > 5550 stems/ha  Median snag dbh < 20.175 cm Median snag dbh < 11.5 cm  Tree Density < 718.5 stems/ha  Median snag dbh > 20.175 cm Tree Density > 718.5 stems/ha  Median snag dbh > 11.5 cm  Figure 1. C A R T diagram for secondary cavity nesters. Values in the ellipses are the mean guild abundance per site at a given node.  Tree Density < 1066 stems/ha  Tree Density > 1066 stems/ha  Snag Density < 52.5 stems/ha /  X  Snag Density > 52.5 stems/ha  32.180  )  3  MedLargeDW < 47.782 m /ha  Q8.920J)  •  MedLargeDW > 47.782 m /ha  (^231^)  Figure 2. C A R T diagram for tree nesters. Values in the ellipses represent the mean guild abundance per site at a given node.  2.3.3 Forest Structure Dynamics The habitat attributes chosen for the models showed a variety of trends in availability over time. The density of conifer trees in the canopy increased over time in data of the Biodiversity Study (years 25-120). I extended this trend from year 120 to 200, the end of the  15  simulation period. Conifers were not found in the young stands of the study (year 25) and thus I set their initial density (year 0 of the simulation) at zero as well (Figure 3).  140  Years since disturbance Figure 3. Density of conifer stems (dbh >10 cm) in the canopy over time. L i v e tree density decreased with stand age over the course of the study. I set the density at year zero at the same level as year 25 because aspen trees regenerate extremely rapidly by suckering, to a height of a few metres in the first 15 years (Peterson and Peterson 1992). Simulations of tree density over time in aspen stands (see Section 2.2.3) suggested that the density would increase somewhat after year 120 as younger trees are recruited into gaps in the canopy. The formation of canopy gaps stabilizes after year 150 according to Cumming's model, so I assumed the density of live trees was constant from year 150 until the end of the simulation period (Figure 4).  16  0  50  100  150  200  Years since disturbance  Figure 4. Density of live trees (dbh >10 cm) in the canopy over time. I calculated the ratio of coniferous to deciduous tree density from the curves developed for conifers and for deciduous trees (the latter of which make up the bulk of, and thus follow the same pattern as that of the live trees, shown in Figure 4). Because conifers were not present for the first 25 years of the simulation, the ratio was zero for this time period, then increased until year 120 as conifers increased and deciduous trees decreased in density. The ratio dropped from years 120 to 150 as deciduous density increased, then rose again as deciduous density stabilized and conifer density continued to increase (Figure 5).  0.12 „  Years since disturbance  Figure 5. Ratio of coniferous to deciduous tree density over time.  17  Percent low shrub cover and median snag dbh both showed U-shaped relationships over the three age classes examined by the biodiversity study, with values in young stands being somewhat lower than those in old stands, and mature stands having the lowest values. For both of these attributes, values at year 0 were assumed to be essentially the same as those for year 25. I set the initial value for shrub cover as being equal to that at year 25 because shrubs recover rapidly after disturbance (Figure 6). The value for median snag size at year zero was set slightly higher than the value for year 25 as the large snags that are left behind in young stands after a fire begin to fall and small snags are produced by self-thinning in this time period (Figure 7). Shrub cover was believed to be stable after year 120, because the amount of available light would remain fairly constant. Median snag dbh was believed to decrease until year 150 before stabilizing, because the decrease in tree size after year 120 due to recruitment of young trees into canopy gaps would bring down the average size of snag produced (S. Cumming, pers. comm.).  o 20 > O  15  c<u  10  o  o  50  100  Years since disturbance F i g u r e 6. Percent cover of the low shrub layer over time.  18  150  200  2.4 Discussion  2.4.1 Habitat  Associations  The use of stepwise multiple regression models allowed for more refined and accurate models than did the initial simple regressions. The addition of other habitat variables increased both the statistical significance and biological realism of the relationships. There was often wide scatter in the single models, suggesting that the habitat variable was imposing an upper ceiling on the response variable, without restricting its distribution below the bound (Thomson et al. 1996). In other words, the primary habitat variable may be a limiting factor, with others determining the distribution of bird abundance below this upper limit. Multiple regression allowed me to explore these other factors. The attributes that remained significant in the multiple regression models reveal aspects of the ecology of the bird species in this system. For example, both of the attributes statistically associated with the abundance of large primary cavity nesters, conifer to deciduous ratio and  19  median snag dbh, are indicators of older stands. Conifers were found only in mature and old stands in the study and the largest snags were found in old stands, or as relics of old stands in young stands. Older stands would have not only the largest live trees, which some of the large cavity nesters use as nest sites (Erlich et al. 1988; Semenchuk 1992), but also the largest snags, and these too are the ones most likely to be in more advanced stages of decay, making them easier to excavate. These relationships would likewise explain why weak primary cavity nesters were also more abundant in stands with large snags and more conifers. Furthermore, older stands are more likely to have dead or dying trees and downed wood, which are all used as foraging sites by these guilds (Erlich et al. 1988). Working in mixedwood forests of British Columbia, Pojar (1995) also found cavity nesters to be associated with indicators of old stands, with abundance increasing with deciduous stem size and decreasing with increasing stem density. The model for secondary cavity nesters is not as easily interpreted, but C A R T analysis provided some insight into potential relationships. One would expect this guild to follow the same pattern as other cavity nesters, which create the cavities these species require. However, this guild had its highest abundance in stands with higher tree density and relatively closed canopies (see the right arm of the C A R T diagram in Figure 1), representative of younger stands. A possible explanation for this observation is that these younger stands contain large decayed relic snags (as indicated in the diagram) that already have cavities in them and thus are prime sites for these species, with less competition from the guilds that prefer older stands. It is also important to note that this guild is the smallest of the six, with only 27 observations between two member species (almost exclusively tree swallows), and stronger relationships often result as sample size increases (Hansen et al. 1995).  20  A s with the primary cavity nesting guilds, tree nesters demonstrated an association with old stands, being positively related to snag size. However, their U-shaped relationship to tree density in the initial analyses indicates some preference for young stands as well. They may avoid mature stands for a number of reasons, including canopy closure, lack of structural diversity, and lack o f larger trees. I ran a C A R T analysis for this group as well i n an attempt to confirm this (see Figure 2), but the results did not clarify the issue. The first split confirms a strong preference for older stands, but the attributes that seemed to be determining tree nester abundance at subsequent nodes were not ones that segregate clearly by age class, and thus it was not clear i f or why this guild was avoiding mature stands. The percent cover of low shrubs was not significant in the model for shrub nesters. This measure was highly variable, and included small saplings, which would be inappropriate as nest sites. Thus, the relationship would likely have been improved with data specific to the tall shrub layer, where these species often nest (Schmiegelow 1997). Such data were not available for all sites in the study. The strongest relationship for this guild (the reciprocal of tree density) was once again an indicator of stand age. In older stands, gaps created in the canopy by older trees which die and fall would allow for increased shrub growth over mature stands as well as providing a greater variety of foraging sites. Ground nesters were positively correlated with both tree and shrub density, which would be important in providing cover for this group, and are associated with young and old stands respectively. N o guild showed a positive association with mature (60 to 75 year old) stands. M y results appear to be consistent with analyses performed by Schieck and Nietfeld (1995) on the same dataset. Direct comparison to their results is not possible, as they examined correlations between individual species, as opposed to guilds, and functions of several habitat  21  attributes, as opposed to individual attributes. However, the attributes most significant in their functions were also significant in my analyses, and they found both species richness and the abundance of those species with at least ten observations to be highest in old stands, moderate in young stands, and lowest in mature stands.  2.4.2 Forest Structure Dynamics The general pattern o f succession in the aspen boreal mixedwood after a disturbance is initiated by a very rapid sprouting of many aspen suckers from the remnant root mass. The high density o f suckers leads to over-crowding and self-thinning, with density decreasing by 80% in the first 5 years (Peterson and Peterson 1992). White spruce, which is sensitive to direct sunlight, begins to recruit in mature stands with closed canopies. In some cases the stand may eventually be dominated by conifers. In others, gaps which form in older stands as senescent trees decay and fall allow for new aspen growth, thereby maintaining its presence in the canopy (Cumming 1997). This pattern o f succession is reflected in the time series developed for each attribute. A s expected, coniferous trees did not appear in the canopy until stands had reached maturity, and then continued to increase in numbers as the stand aged. The overall density o f live trees, o f which by far the greatest proportion was aspen, was initially very high, then dropped as selfthinning occurred. Once old growth was achieved (after year 120 in this case), the density increased slightly as some of the older trees began to die out, leaving gaps for a higher density o f smaller trees to exploit. This process soon reached an equilibrium, resulting in a relatively constant density of trees.  22  Median snag dbh was moderately high in naturally created young stands due to a legacy of large dead trees created when fire burned the previous, old stand. A s these snags collapsed and new, smaller snags were created by self-thinning, the median size of snags dropped and then began to increase again as the smaller snags fell and larger trees began to senesce. The snag size was greatest in old stands, where large trees would be dying and few small trees remain. Median snag diameter decreased again after this point, as tree size (and thus snag size) decreased due to gap dynamics. Shrub cover was also lowest in mature stands, where the closed canopy does not allow enough light for understory growth. It was somewhat higher in young stands, where the shorter trees do not restrict light but crowding is a problem. Shrub cover was at a maximum in old stands, where gap dynamics and greater light penetration would allow for greater understory growth (Peterson and Peterson 1992).  23  C H A P T E R 3 Simulation Projections  3.1 Introduction S I M F O R is a spatially-explicit simulation model of landscape dynamics through time (Daust and Sutherland 1997). It tracks specific habitat attributes through a given time period and translates their availability into habitat quality for animal species, based on algorithms supplied by the user. In this study, I used data from the Alberta Environmental Centre's Biodiversity Study in the aspen boreal mixedwood of north-eastern Alberta (Stelfox 1995). I used algorithms which I developed for guilds of bird species in this forest, as well as the supply of habitat attributes over time (see Chapter 2), to predict habitat quality for each guild. Quality of home range for the large primary cavity nester guild was also modelled. Maps for stand types and ages of a landscape of 9741 ha (township 71134 in Alberta) were supplied to S I M F O R in a raster format, where each raster represented one hectare. I chose this township because its stand composition and age structure were similar to those in the area from which the data for the habitat component were collected. Those data (from the Biodiversity Study) were collected from stands consisting of at least 80% aspen, so stands of this composition in the chosen township were the only ones manipulated in the simulation model. These stands occupied 3928 ha of the township. I applied a three-pass clear-cut logging system to the stands in the model. Under this regime the forest to be harvested is divided into cut blocks, each of which is assigned to one of the three passes such that all have been logged by the end of the third pass. The passes are generally 15 to 20 years apart and the rotation age in aspen forest is around 70 years (Alberta 24  Pacific Forest Industries Inc. 1992, 1993). The harvest schedule, specifying which areas were to be cut for each pass, was provided by Alberta Pacific Forest Industries (AlPac). Because the supply curves for the habitat attributes (trajectories of their availability through time) were generated using the average values for each age class from the Biodiversity Study, I also explored the effects that the variance associated with each of these values might have on the results generated by the model. I chose one representative guild for this process, and ran a Monte Carlo simulation to generate predicted distributions of guild abundance for each age class given the distributions associated with the habitat variables.  3.2 Methods  3.2.1 Simulation modelling The first steps I took in preparing the model were to enter supply curves for each of the forest attributes identified as significantly associated with one or more of the bird guilds (Figures 3-7 in Section 2.2.3 above) when entered as independent variables in the multiple regression models. These curves consist simply of the supply (e.g., density, size, percent cover) of each attribute as a function of time since disturbance. Response curves, describing the relationship between each guild and each attribute with which it was associated, were also entered into the model. Each curve consisted of a straight line with slope equal to the coefficient in the multiple regression model for that specific guild and habitat attribute. Response curves were also constructed for the constants in the models. These curves are used by S I M F O R to predict the abundance of each guild in each cell over the course of the simulations. The results are summarized as the number and location of  25  hectares of habitat of a given quality, or suitability, at each time step. Quality is defined arbitrarily, based on boundaries defined by the user. I determined quality of habitat for each guild individually. To determine the boundaries for low, medium and high quality habitat, I used the multiple regression model for a given guild to predict the expected abundance at each of the original 72 sites of the Biodiversity Study. I then ranked the sites based on the predicted abundance values, and calculated the thresholds for the quality categories by dividing the sites into quartiles, with the first quartile (18 sites) defining the upper boundary of the low category and the third quartile (54 sites) defining the lower boundary for high quality habitat. In the case of the large primary cavity nesters, for example, the multiple regression equation predicted an abundance of 0.5125 individuals for the 18th site and 0.5241 for the 19th, when the 72 sites were ranked in order of predicted guild abundance. I chose the average o f these two numbers (0.5188) as the threshold between low quality and medium quality habitat for this guild. Similarly, the average of the guild abundances predicted for the 54th and 55th sites was chosen as the threshold between medium and high quality habitat. I repeated this process for each guild individually, creating a series of thresholds unique to each guild. T o perform an individual simulation in S I M F O R , a particular guild and its associated habitat quality thresholds were selected. Each scenario was run over 200 years, with output consisting of the number of hectares of habitat of each quality being generated at every 10 year time step. The harvest schedule used was a three-pass clear-cut system with a 70 year rotation, such that the blocks assigned to the first pass were logged in year 5, the second pass in year 25 and the third in year 45. This was repeated for the second rotation at years 75, 95 and 115 and for the third at years 145, 165 and 185. A n y patches reaching an age greater than 200 years were assumed to have the same characteristics as a 200-year-old patch.  26  Home range quality for large primary cavity nesters was modelled in similar fashion, with the average habitat quality for the 40 ha surrounding a given cell determining the home range quality for that cell.  3.2.2 Randomization and Monte Carlo I explored the effects of variance in habitat attribute data on predicted guild abundance using large primary cavity nesters as an example. The attributes on which this guild depends showed different amounts of variance in the three age classes of the Biodiversity Study: there was no variance in the ratio of coniferous to deciduous trees in young stands, but high variance in median snag size; moderate variance in the ratio in mature stands and little in snag size; and large variance in both attributes in old stands. Large primary cavity nesters were also a logical choice because their conservation is an important management issue and their model was one of the strongest both statistically and biologically, as the attributes identified as important in the model correspond well with current understanding of the guild's natural history. The mean and standard deviation of the predicted guild abundance were generated within each of the three age classes. I used the N O R M I N V function in Excel (Microsoft 1994) in combination with the random number function to generate a value at random from the distribution of observed values for each attribute. The function takes the random number between 0 and 1 and returns the equivalent value from the cumulative normal distribution of the attribute in question. In those instances where the habitat attribute was not normally distributed within a given age class, I used an appropriate transformation to normalize the data. In one case, no transformation was appropriate, so bootstrapping was used to generate a new distribution.  27  The values generated by this process for each of the two attributes in a given age class were used in the multiple regression equation to generate a value for the predicted abundance of large primary cavity nesters. Where transformations had been used, I applied the opposite transformation to the values before using them. I repeated this process fifty times, generating a distribution for predicted abundance. I repeated the whole process for each age class, using the appropriate attribute distributions.  3.3 Results  3.3.1 Simulation modelling A t the beginning of the simulation, 713 of the 3928 ha of aspen forest (18.2%) were considered of high quality habitat for large primary cavity nesters (Figure 8). This level fluctuated slightly over the first 40 years of the simulation, reaching a peak o f 896 ha in year 40. B y year 50, 5 years after the last pass of the first harvest rotation, the area of high quality habitat for this guild had fallen to 359 ha, and continued a steep decline until year 90, when none of the habitat was of high quality. The level did not recover from zero for the rest of the simulated period (200 years). A l l other habitat over the course of the simulation was ranked as medium quality, so only the high quality habitat is illustrated.  28  C  N  ^  ^  O  O  O  O  —i  C  Year  N —<  l  ^ —  ^  O —  O  O  O  (N  Figure 8. Area of high quality habitat for large primary cavity nesters over time. Arrows indicate passes of the first rotation. The results were qualitatively different for weak primary cavity nesters. High quality habitat occupied 803 of the 3928 ha of aspen forest (20.4%) at the beginning of the simulation (Figure 9). This decreased slightly after the first pass of the first rotation, but generally increased until the third pass, when it fell sharply, from 1139 ha in year 40 to 519 ha in year 50. It recovered slightly over the second rotation but then stabilized at 598 ha for the rest of the simulation. A l l other aspen habitat was considered of medium quality for this guild over the course of the simulation.  29  CI CL 0  H—I  +-r-i  —  o  H r H o  1 o  1  1  1  o  1 o —  Year  1  1 o —  1  1 o —  1  1 o —  1  1 o —  h—I-  0  o tN  Figure 9. Area of high quality habitat for weak primary cavity nesters over time. Arrows indicate passes of the first rotation. Output maps produced by S I M F O R (Appendix II), which show the locations of the high quality rasters as well as which areas are harvested, indicate that the only high quality habitat available for this guild by the end of the second rotation is in the areas not slated for harvest (598 ha). These maps also indicate that this high quality habitat, although only 200 ha less than that available at the beginning of the simulation, is highly fragmented, with many patches only one hectare in size. The largest patch left unlogged was 18 ha. This would be the only patch large enough to encompass the territory size of members of this guild (Pravosudov and Grubb 1993; Smith 1993), leaving the other 580 ha in patches of insufficient size. The results for secondary cavity nesters and ground nesters showed a very different pattern of availability of high quality habitat from those of the primary cavity nesters (Figure 10). Although defined by different multiple regression models and habitat quality thresholds, the area of high quality habitat was identical for these two guilds over the course of the simulation, as the same harrow range o f stand age (25-30 years) was most suitable for both guilds. N o habitat was considered of high quality for these guilds at the beginning of the simulation. B y year 30, 25 years after the first pass of the first rotation, 965 ha were ranked as high quality, but only for 30  this interval. A t year 50, 25 years after the second pass, there are 1328 ha of high quality habitat, and at year 70, 25 years after the third pass, there are 1037 ha of high quality habitat. In the intervening years there was no high quality habitat. This pattern continued over subsequent rotations, with habitat only ranked as high quality for the time interval 25 years after each harvest pass.  1400  Year  Figure 10. Area of high quality habitat for secondary cavity nesters or ground nesters over time. Arrows indicate passes of the first rotation. These two guilds did differ somewhat in the classification of the remaining aspen habitat. Most of this lower quality habitat was ranked as medium quality over the course of the simulation for both guilds, but the secondary cavity nesters were predicted to encounter slightly more of low quality. L o w quality habitat was present for this group over the first 80 years of the simulation, with a peak of 456 ha in year 40 (Figure 11). Ground nesters experienced minimal low quality habitat for the first 30 years of the simulation, with approximately equal peaks of 173 and 156 ha at years 40 and 70 respectively (Figure 12). L o w quality habitat was negligible for this group from year 100 on.  31  Figure 11. Quality of habitat available for secondary cavity nesters over time. Arrows indicate passes of the first rotation.  Figure 12. Quality of habitat available for ground nesters over time. Arrows indicate passes of the first rotation. Tree nesters showed a substantial increase in high quality habitat over the first 10 years of the simulation, from 60 ha in year 0 to 317 ha in year 10, 5 years after the first pass of the first harvest rotation (Figure 13). The level increased slightly to year 20, then fell sharply to 58 ha in year 30, 5 years after the second pass. A peak of 456 ha was reached in year 40, but then the level declined sharply again in year 50, 5 years after the final pass of the first rotation, and declined to zero by year 80, with no recovery. Shrub nesters shared the same pattern in high  32  quality habitat availability over time with the exception of the first year, which for them had the maximum amount of high quality habitat (654 ha) over the simulation (Figure 14). This dropped to 317 ha in year 10 and then followed the same fluctuations and decline as in the case of the tree nesters. L o w quality habitat for the tree nesters and shrub nesters followed the same trend as that illustrated for the high quality habitat of the secondary cavity nesters and ground nesters.  .. 12  a  IS CS  J3  .8  en  10  _ 6  ~£  . 4  o a  x:  D. eft CS  .2  o a o > i~. u  OH  Year  Figure 13. Area of high quality habitat for tree nesters over time. Arrows indicate passes of the first rotation.  16  cs  14  j3  12  -C C CD B.  10 8 6 4 i2  CS  eft  cs  ~3 o o c  o  0 o  Year  Figure 14. Area of high quality habitat for shrub nesters over time. Arrows indicate passes of the first rotation.  33  The amount of high quality home range available to large primary cavity nesters shows a general increase over the first 60 years of the simulation, to a maximum of 94 ha or just under 2.5% (Figure 15). It then decreases rapidly, with no high quality home range available from year 80, five years after the first pass of the second rotation, until the end of the simulation.  Figure 15.  Area of high quality home range for large primary cavity nesters over time. Arrows indicate passes of the first rotation.  3.3.2 Randomization and Monte Carlo The predicted abundance of large primary cavity nesters was most variable in old stands, where median snag size, by far the stronger of the two contributing attributes, was also most variable (see Table 2). Predicted abundance was least variable in mature stands, where both predictor variables had low variance. Coefficients of variation ( C V ) for predicted abundance were lower than might be expected when compared to those for the habitat attributes. A s in the S I M F O R simulations, predictions of abundance over the threshold for high quality habitat for this guild (2.38) were only found in old stands in the Monte Carlo simulations.  34  Table 2. Distributions of habitat attributes and predicted large primary cavity nester abundance for each of three age classes  Conif/Decid Ratio  Median Snag DBH  Predicted Abundance  Young Mean  0  18.106  1.1048  Standard Deviation  0  6.485  0.59036  CV  undefined  0.35817  0.53434  Mean  0.012474  12.433  0.54660  Standard Deviation  0.031323  1.35  0.098065  CV  2.5111  0.10858  0.17941  Mean  0.10708  29.992  2.7407  Standard Deviation  0.15959  8.146  0.82807  CV  1.4904  0.27161  0.30214  Mature  Old  3.4 Discussion  3.4.1 Simulation  modelling  The supply of high quality habitat for large primary cavity nesters over the course of the simulation (Figure 8) suggests that this guild will likely suffer a substantial decline after the first rotation of logging. High quality habitat is available in relatively constant supply until the third pass of this rotation, after which it never recovers. None of the habitat is considered to be of low quality over the course of the simulation, with all of it falling into the medium quality category after year 80. However, given that this guild is perhaps the most dependent on appropriate nesting sites for reproductive success, having both specific site requirements and member species requiring larger trees (Harestad and Keisker 1989), the lack of high quality habitat may be especially detrimental to it. Those species which are already rare in the study area, such as pileated and hairy woodpeckers, may be extirpated entirely i f guild abundance drops below current levels.  35  This conclusion is further supported by the home range simulation for the large primary cavity nesters, which was run using 40 ha as a representative home range size for pileated woodpecker, the rarest member of the guild. The fact that less than 2.5% of the aspen habitat is ever considered to be of high quality given this home range size and that this is eliminated by the beginning of the second rotation suggests that the rarer members of this guild may be excluded from the area. It should also be taken into account that forty hectares is likely a conservative estimate of the home range size required by pileated woodpecker. In drier ecosystems, Bull and Jackson (1995) reported home ranges of 53 to 1056 ha. Robbins et al. (1989) found pileated woodpeckers to be dependent on large forested areas, with less than 50% probability of occurrence in patches smaller than 165 ha, although individuals were observed in patches of only 1 to 2 ha. The fact that the only high quality habitat available to weak primary cavity nesters by the end of the second rotation is in isolated unharvested stands is significant. The low, fragmented supply of high quality habitat may be somewhat mitigated by the fact that the harvested stands are considered medium quality habitat over the course of the simulation. However, many of these fragments were also separated from the next nearest high quality aspen patch by several hundred metres, and some from any other suitable habitat patch (of medium quality) by a few hundred metres or more. Other studies in the region have suggested that distances of 200 metres between suitable habitat fragments can be sufficient to isolate birds during the breeding season (Schmiegelow et al. 1997), so the spatial arrangement of these habitat patches may have a negative impact on breeding success (e.g., Appendix II). Tree nesters and shrub nesters revealed a slightly different pattern from that of the large primary cavity nesters over only the first 50 years of the simulation. For these guilds too there is  36  no high quality habitat available after the first rotation. A l l three of these guilds suffer because their highest abundance, as predicted by the multiple regression equations and supply of habitat attributes over time, occurs in stands of around 120 years old. Under the management scenario being implemented in the simulation model, none of the harvested stands will reach this age, and the stands that are not harvested will continue to age past this point. In the case of the weak primary cavity nesters, stands older than 120 years are also considered high quality, allowing members of this guild to find high quality habitat in the unharvested aspen stands. For all four of these guilds (both large and weak primary cavity nester groups, tree nesters, and shrub nesters), there seems to be a threshold reached in the harvested areas after the last pass of the first harvest rotation that has a significant impact on the amount of high quality habitat available. This effect has been observed in other studies (Turner 1989; Fahrig 1997), where certain bird species are not negatively affected by logging in an area until a critical area has been cleared. In the case of my simulation, the third pass reduces the remaining original forest in the cutblocks to clear-cut. Once this critical area has been cut, high quality habitat for these guilds will not regenerate in harvested stands within the time frame of the rotation age. The primary cavity nesting guilds, tree nesters and shrub nesters are all most abundant in older stands, where conifer density is highest, at an age that harvested stands are not able to reach under this scenario (120 years or older). Weak primary cavity nesters are the only guild able to find high quality habitat in unlogged areas, although suitable habitat for these four guilds may be available in surrounding conifer stands of sufficient age. Natural disturbances such as wildfire and insect outbreaks, however, which are not included in the model, could potentially further reduce the area of old aspen forest available to these guilds, as could the logging of conifer stands.  37  The secondary cavity nesters and ground nesters were the only two guilds to show a positive response to logging, as their highest abundance is reached in stands of 25 years or younger. The major predictor variable for both of these guilds is abundance of live trees, which is highest in young stands. Under the conditions of the simulation, stands of this age would only be created in harvested areas. The 965 ha harvested in the first pass are ranked as high quality habitat for 25 years after logging, and overlap with the 1328 ha logged in the second pass at year 30. Similarly, the second and third pass (1037 ha) overlap at year 50, producing the pattern of availability of high quality habitat seen in Figure 10. The apparent dependence of these two guilds on harvested areas for high quality habitat may in part-be an artefact of the modelling. Under natural conditions, young stands would also be created in the unlogged areas of the landscape through natural disturbance. The wide variety of guild responses to the creation of young stands through logging suggests that a broader mosaic of stand ages may need to be maintained over the landscape. Even under the assumption that the retention of some attributes from the original stands is incorporated into logging practices, the same areas that were ranked as high quality habitat for secondary cavity nesters and ground nesters were low quality habitat for tree nesters and shrub nesters. If some of the matrix of aspen forest is not left in a natural state with older stands, these latter two guilds and the large primary cavity nesters are predicted to be left without high quality habitat, and the weak primary cavity nesters are predicted to be restricted to isolated fragments of high quality habitat that may not be large enough to support viable populations. It is important to note that the model is only producing results for the portion o f the landscape with a composition of at least 80% aspen canopy cover. It may be that the surrounding conifer forest provides habitat for some of the species found in the aspen stands. It is likely,  38  however, that species assemblages differ between the stand types (Schmiegelow 1997), and some species tend to avoid coniferous habitat. Data would have to be collected on bird species abundances and specific habitat attributes in these other forest types, as well as harvest plans for the area, in order to incorporate these effects into the model.  3.4.2 Randomization and Monte Carlo The results from the Monte Carlo simulation suggest that, although variation in the input variables should be taken into account, it did not have a disproportionate effect on the output in this case. In fact, the variance associated with the predicted guild abundances was much less than that around the coniferous to deciduous ratio, and only slightly higher than that of the median snag dbh. The only exception was in young stands, where the coefficient o f variation for the coniferous to deciduous ratio was undefined. The effects of the variance in the predictor variables could be further explored by generating for each attribute a series of supply curves to be used in S I M F O R .  39  C H A P T E R 4 Conclusions and Management Recommendations The results of this study indicate that high quality habitat may well be scarce for most guilds in areas of aspen-dominated mixedwood that are harvested using the current management system. A three-pass clear-cut system with a 70 year rotation does not permit logged areas to reach a sufficient age to provide habitat attributes with which many bird species are strongly associated. If large areas of older forest are not present in the region to provide habitat for dependent species, these species are likely to decline in numbers. The proposed scale of industrial forestry in the region may result in only small, scattered patches of this habitat being available, which would be insufficient to support the current populations. The threat to large primary cavity nesters is of particular concern. Not only is this guild likely to decline in numbers, but individual species that are already rare will likely be lost as this occurs. It is also probable that the decline in overall numbers of large primary cavity nesters will have important ramifications for other cavity nesters which are dependent on this group to excavate holes (Martin and Eadie 1998). Weak primary cavity nesters are further at risk given their apparent dependence on stands older than the current rotation age for high quality habitat. Attempts by managers to mimic natural disturbance and to retain attributes of old stands in young stands may help mitigate these results, and various options could be further explored using this model. Large-scale natural disturbances such as fire and insect outbreaks do not completely consume the trees and other vegetation in a given area, so managers may choose to leave some material behind. Blocks may not be fully clear-cut, and the initial values of the habitat attribute supply curves could be modified to reflect varying levels of retention. For  40  instance, more large snags could be left standing, and this would be reflected in an increase in the initial median snag size. It has been suggested that other harvest options, such as a dispersed harvest or partial cutting, would have less of an impact on wildlife habitat than current practice (Peterson and Peterson 1992; Norton and Hannon 1997). In the case of a dispersed harvest, cutblocks are widely distributed across the landscape (Wallin et al. 1994), instead o f arranged in the checkerboard pattern o f a three-pass system. The predicted effects that this would have on habitat quality could be explored by using this plan as an alternative input to the model, and compared to the results of the original. Similarly, a plan involving reserves of aspen habitat or selection harvesting could be incorporated, and the results compared to see whether availability of high quality habitat for those guilds associated with older serai stages would improve. Modelling allows these options to be examined not just in regard to the amount of high quality habitat produced over time under each scenario. The spatial nature of models such as S I M F O R is a valuable tool for comparing not simply numerical output, but also the dispersion and arrangement o f individual patches, an essential part of the ecology of the system. In this way, the expected outcomes of a wide variety of management options can be explored in a relatively short time frame. These predictions would be further improved by incorporating natural disturbance, such as fire cycles, into the model. Expanding the model to include other forest cover types in the area would reduce some of the uncertainty around the role surrounding areas would play acting as a refuge for those species negatively impacted in harvested areas. M y model was also limited by the fact that data were only available to produce stand dynamics projections for naturally-regenerating stands. These projections were then applied to areas being subjected to logging, which may not develop in the same way. B y the same token,  41  assuming that initial conditions are the same between the first rotation and subsequent rotations is probably also in error. It is likely that even fewer snags, for instance, would be available after the first clear-cut rotation, which may result in even lower quality habitat for many guilds. Further, expressing habitat quality in terms of the amount of each of three categories available is somewhat arbitrary. In some cases the abundance predicted for a given guild at certain ages was close to the threshold between two categories, suggesting that slight variation in the supply of habitat attributes might have changed the projections of habitat quality. Although the exact results may not be entirely accurate given these limitations, the general trends in habitat quality are likely valid, and the overall methodology that I developed appears sound and is easily adapted to make use of new, improved information on the system. A s with any scientific hypothesis, the results achieved by modelling need to be verified by field data (Romesburg 1981; Oreskes et al. 1994) i f they are to be examined in any way other than by comparison to each other. True tests of the curves I developed to describe stand dynamics would require years to complete, but one alternative would be to use historical data on stand type and age distribution in the model to attempt to predict present conditions. This does not allow the testing of any forest stand compositions other than those that have occurred naturally, but it would allow for the testing of the basic algorithms of the model. The algorithms describing the predicted abundance of the guilds could be tested in the field, by making abundance estimates in the township used in the simulations. Verification and improvement of the model could be accomplished effectively in the framework of adaptive management (sensu Walters 1986), where researchers and managers work together to explore different scenarios on the ground as well as theoretically. Modelling allows for quick examination of the predicted outcome of numerous proposed management scenarios. If  42  carefully planned, these scenarios can then be experimented with, and the results used to refine and improve the models (Walters and Holling 1990). In this way, managers and researchers can direct each other's efforts to improve their understanding and use of the forested ecosystem.  43  Literature Cited Alberta Pacific Forest Industries Inc. 1992. Preliminary Forest Management Plan. Alberta Pacific Forest Industries, Edmonton, Alberta. Alberta Pacific Forest Industries Inc. 1993. Timber Harvest Planning and Operating Ground Rules. Pub. N o : Ref. 67, Alberta Pacific Forest Industries, Boyle, Alberta. B u l l , E . L . and J.E. Jackson.  1995. Pileated Woodpecker (Dryocopus pileatus).  North America, N o . 148 (A. Poole and F. G i l l , Editors).  In The Birds of  Philadelphia: The Academy of  Natural Sciences; Washington, D C : The American Ornithologists' Union. Bunnell, F . L . 1989. Alchemy and uncertainty: what good are models?  In Gen. Tech. Rep.  P N W - G T R - 2 3 2 . Portland, O R : U . S . Department of Agriculture, Forest Service, Pacific Northwest Research Station. 27 p. Bunnell, F . L . 1992. De mo'beta blues: coping with the landscape.  In Proceedings of the  Seminar on Integrated Resource Management. Forestry Canada, Maritimes Region. Information Report M - X - 1 8 3 E / F . Fredericton, N B , April 7-8. pp. 45-58. Bunnell, F . L . 1997. Operational criteria for sustainable forestry: focussing on the essence. For. Chron. 73(6): 679-684. Bunnell, F . L . and L . L . Kremsater.  1990. Sustaining Wildlife in Managed Forests. The  Northwest Environmental Journal 6: 243-269. Bunnell, F . L . and L . L . Kremsater.  1994. Tactics for maintaining biodiversity in forested  ecosystems. Proc. X X I I U G B Congr. 1: 62-72. Bunnell, F . L . , L . L . Kremsater and R . W . Wells. 1997. Likely consequences of forest management on terrestrial, forest-dwelling vertebrates in Oregon. Oregon Forest Resources Institute, Portland, Oregon. Bunnell, F . L . , L . L . Kremsater and M . Boyland. 1998. A n Ecological Rationale for Changing Forest Management on M a c M i l l a n Bloedel's Forest Tenure. O n the web, http://www.macblo.com/forestproject/docs/ubc_mbl.pdf. Campbell, R . W . , N . K . Dawe, I. McTaggart-Cowan, J . M . Cooper, G . W . Kaiser, M . C . E . M c N a l l and G . E . John Smith. 1997. The Birds of British Columbia. V o l . 3 . Royal British Columbia Museum, Victoria, B . C . Clark, L . A . and D . Pregibon. 1992. Tree-Based Models. In Chambers, J . M . and T . J . Hastie (Editors).  Statistical Models in S. Wadsworth & Brooks/Cole Advanced Books &  Software, Pacific Grove, California, pp. 377-419. 44  Cumming, S . G . 1997. Landscape dynamics of the mixedwood boreal forest. P h . D . thesis. University of British Columbia, Vancouver, B . C . Cumming, S . G . , P.J. Burton and B . Klinkenberg. 1996. Boreal mixedwood forests may have no "representative" areas: Some implications for reserve design. Ecography 19(2): 162-180. Daust, D . K . and F . L . Bunnell. 1992. Predicting biological diversity on forest land in British Columbia. The Northwest Environmental Journal 8(1): 199-200. Daust, D . K . and G . D . Sutherland. 1997. S I M F O R : Software for simulating forest management and assessing effects on biodiversity. In The Status of Forestry/Wildlife Decision Support Systems in Canada: Proceedings of a Symposium, Toronto, Ontario, 1994. I.D.Thompson (Editor). Published by Natural Resources Canada: Canadian Forest Service, Great Lakes Forestry Centre, Sault Ste. Marie, Ont. pp. 15-29. Dunning, J.B., Jr., D . J . Stewart, B.J. Danielson, B.R. Noon, T . L . Root, R . H . Lamberson and E . E . Stevens.  1995. Spatially explicit population models: Current forms and future uses.  Ecol. A p p l . 5(1): 3-11. Erlich, P.R., D . S . Dobkin and D . Wheye. 1988. The birder's handbook: A field guide to the natural history of North American birds. Simon & Schuster Inc., Toronto. Fahrig, L . 1997. Relative effects of habitat loss and fragmentation on population extinction. J. Wildl. Manage. 61: 603-610. Franklin, J.F. and R . T . T . Forman. 1987. Creating landscape patterns by forest cutting: Ecological consequences and principles. Landscape Ecology 1(1): 5-18. Hansen, A . J . , S.L. Garman and B . Marks. 1993. A n approach for managing vertebrate diversity across multiple-use landscapes. Ecol. A p p l . 3(3): 481-496. Hansen, A . J . , W . C . M c C o m b , R. Vega, M . G . Raphael and M . Hunter. 1995. Bird habitat relationships in natural and managed forests in the west Cascades of Oregon. E c o l . A p p l . 5(3): 555-569. Harestad, A . S . and D . G . Keisker. 1989. Nest tree use by primary cavity-nesting birds in south central British Columbia. Can. J. Zoo. 67: 1067-1073. Jaksic, F . M .  1981. Abuse and misuse of the term "guild" in ecological studies. Oikos 37: 397-  400. Jaksic, F . M . and R . G . Medel. 1990. Objective recognition of guilds: testing for statistically significant species clusters. Oecologia 82: 87-92.  45  Lee, P . C . , S. Crites and J.B. Stelfox.  1995. Changes in Forest Structure and Floral Composition  in a Chronosequence of Aspen Mixedwood Stands in Alberta. In Stelfox, J.B. (Editor). Relationships Between Stand Age, Stand Structure, and Biodiversity in Aspen Mixedwood Forests in Alberta. Jointly published by Alberta Environmental Centre ( A E C V 9 5 - R 1 ) , Vegreville, Alberta, and Canadian Forest Service (Project N o . 0001 A ) , Edmonton, Alberta, pp. 29-48. Martin, K . and J . M . Eadie. 1998. Nest webs: A community wide approach to the management and conservation of cavity-nesting forest birds. For. Ecol. Manage. In press. Microsoft Corporation. 1985-1994. Microsoft Excel Version 5.0c. Nelson, J.D. and Hafer, M . 1996. S I M F O R V2.91 Operations Manual. Unpublished User's Manual. Faculty of Forestry, University of British Columbia, Vancouver, B . C . Nelson, J.D., D . Gizowski and T . Shannon. 1996. A T L A S V2.87 Windows Edition. Unpublished User's Manual. Forest Operations Group, Faculty of Forestry, University of British Columbia, Vancouver, B . C . Norton, M . R . and S.J. Hannon. 1997. Songbird response to partial-cut logging in the boreal mixedwood forest of Alberta. Can. J. For. Res. 27: 44-53. Oreskes, N . , K . Shrader-Frechette and K . Belitz. 1994. Verification, Validation, and Confirmation of Numerical Models in the Earth Sciences. Science 263: 641-644. Pojar, R . A . 1995. Breeding bird communities in aspen forests of the sub-boreal spruce (dk subzone) in the Prince George forest region. Province of British Columbia, Victoria, B.C. Peterson, E . B . and N . M . Peterson. 1992. Ecology, management, and use o f aspen and balsam poplar in the prairie provinces, Canada.  In For. Can., Northwest Reg., North. For. Cent.,  Edmonton, Alberta. Spec. Rep. 1. Pravosudov, V . V . and T . C . Grubb Jr. 1993. White-breasted Nuthatch (Sitta carolinensis).  In  The Birds of North America, N o . 54 (A. Poole, P. Stettenheim, and F . G i l l , Editors). Philadelphia: The Academy of Natural Sciences; Washington, D C : The American Ornithologists' Union. Robbins, C . S . , D . K . Dawson and B . A . Dowell. 1989. Habitat area requirements of breeding forest birds of the middle Atlantic states. Wildl. Monogr. 103: 1-34. Romesburg, H . C . 1981. Wildlife Science: Gaining Reliable Knowledge. J. Wildl. Manage. 45(2): 293-313. Salt, W . R . and J.R. Salt. 1976. The Birds of Alberta. Hurtig Publishers, Edmonton, Alberta.  46  Schieck, J. and M . Nietfeld. 1995. Bird Species Richness and Abundance in Relation to Stand A g e and Structure in Aspen Mixedwood Forests in Alberta. In Stelfox, J.B. (Editor). Relationships Between Stand Age, Stand Structure, and Biodiversity in Aspen Mixedwood Forests in Alberta. Jointly published by Alberta Environmental Centre ( A E C V 9 5 - R 1 ) , Vegreville, Alberta, and Canadian Forest Service (Project N o . 0001A), Edmonton, Alberta, pp. 115-157. Schmiegelow, F . K . A .  1997. The effect of experimental fragmentation on bird community  dynamics in the boreal mixedwood forest. P h . D . thesis. University of British Columbia, Vancouver, B . C . Schmiegelow, F . K . A . and S.J. Harmon. 1993. Adaptive Management, Adaptive Science and the Effects of Forest Fragmentation on Boreal Birds in Northern Alberta. In Trans. 58th N . A . Wildl. & Natur. Resour. Conf. pp. 584-598. Schmiegelow, F . K . A . , C S . Machtans and S.J. Hannon. 1997. Are boreal birds resilient to forest fragmentation?  A n experimental study o f short-term community responses.  Ecology  78(6): 1914-1932. Semenchuk, G.P. (Editor).  1992. The Atlas of Breeding Birds of Alberta. Federation of Alberta  Naturalists, Edmonton, Alberta. Simberloff, D . and T . Dayan. 1991. The guild concept and the structure of ecological communities. Annu. Rev. Ecol. Syst. 22: 115-143. Smith, S . M . 1993. Black-capped chickadee. P. Stettenheim, and F. G i l l , Editors).  In The Birds of North America, N o . 39 (A. Poole, Philadelphia: The Academy of Natural Sciences;  Washington, D C : The American Ornithologists' Union. Stelfox, J.B. (Editor).  1995. Relationships between stand age, stand structure, and biodiversity  in aspen mixedwood forests in Alberta. Jointly published by Alberta Environmental Centre ( A E C V 9 5 - R 1 ) , Vegreville, A B , and Canadian Forest Service (Project N o . 0001A), Edmonton, A B . pp. 308. Thomson, J . D . , G . Weiblen, B . A . Thomson, S. Alfaro and P. Legendre. 1996. Untangling multiple factors in spatial distributions: Lilies, gophers, and rocks. Ecology 77(6): 16981715. Turner, M . G .  1989. Landscape Ecology: The Effect of Pattern on Process. Annu. Rev. Ecol.  Syst. 20: 171-197. Turner, M . G . , G.J. Arthaud, R . T . Engstrom, S.J. Hejl, J. L i u , S. Loeb and K . M c K e l v e y . 1995. Usefulness o f spatially explicit population models in land management. 12-16.  47  E c o l . A p p l . 5(1):  V a n Home, B . and J . A . Wiens. 1991. Forest Bird Habitat Suitability Models and the Development o f General Habitat Models.  In Fish and Wildlife Research 8. United  States Department of the Interior Fish and Wildlife Service. 31 p. Vemer, J. 1984. The guild concept applied to management of bird populations. Environ. Manage. 8(1): 1-14. Wallin, D . O . , F.J. Swanson and B . Marks. 1994. Landscape pattern response to changes i n pattern generation rules: land-use legacies in forestry. Ecol. A p p l . 4(3): 569-580. Walters, C.J. 1986. Adaptive Management of Renewable Resources. M a c M i l l a n Publishing Company, New York. Walters, C.J. and C S . Holling. 1990. Large-Scale management experiments and learning by doing. Ecology 7(6): 2060-2068. Welsh, D . A .  1981. Impact on bird populations of harvesting the boreal mixedwood forest.  In  R . D . Whitney and K . M . M c C l a i n , eds. Boreal mixedwood symposium. P r o c , September 16-18, 1980, Thunder Bay, Ontario. Environ. Can., Can. For. Serv., Great Lakes For. Res. Cent., Sault Ste. Marie, Ontario. C O J F R C Symp. Proc. O-P-9. pp. 155-167. Wilkinson, L . , M . H i l l , J.P. Welna and G . K . Birkenbeuel. 1992. S Y S T A T for Windows: Version 5 Edition. S Y S T A T Inc., Evanston, II. Willson, M . F . and T . A . Comet. 1996. Bird communities of northern forests: patterns o f diversity and abundance. Condor 98: 337-349.  48  Appendix I Table L A . Membership of the bird guilds (scientific names from Semenchuk 1992). Nesting guilds are listed first, followed by the foraging guilds. N is the total number of observations for each species.  Guild Name  Member Species  Scientific Name  Large Primary Cavity Nesters Hairy Woodpecker  Picoides villosus  6  Northern Flicker  Colaptes auratus  10  Pileated Woodpecker  Dryocopus pileatus  Winter Wren Yellow-bellied Sapsucker Weak Primary Cavity Nesters Black-capped Chickadee  Downy Woodpecker Red-breasted Nuthatch White-breasted Nuthatch Secondary Cavity Nesters  House Wren Tree Swallow  Tree Nesters  N  American Crow  Troglodytes troglodytes Sphyrapicus varius Parus atricapillus Picoides pubescens  Sitta canadensis Sitta carolinensis Troglodytes aedon Tachycineta bicolor Corvus brachyrhynchos  6 11 82 37  6 19 3 2 25 2  American Redstart Setophaga ruticilla 27 Black-billed Cuckoo Coccyzus erythropthalmus 1 Black-throated Green Warbler Dendroica virens 96 Blackpoll Warbler Blue Jay Broad-winged Hawk  Brown Creeper  Dendroica striata Cyanocitta cristata Buteo platypterus  1  Certhia americana  42 22 22 27  Brown-headed Cowbird  Molothrus ater  Cedar Waxwing Chipping Sparrow  Bombycilla cedrorum Spizella passerina  Common Raven  Gray Jay  Corvus corax Accipiter cooperii Regulus satrapa Perisoreus canadensis  Least Flycatcher Magnolia Warbler Olive-sided Flycatcher  Empidonax minimus Dendroica magnolia Contopus borealis  Cooper's Hawk Golden-Crowned Kinglet  Warbling Vireo  Vireo philadelphicus Carduelis pinus Carpodacus purpureus Pheucticus ludovicianus Vireo solitarus Vireo gilvus  Western Tanager  Piranga ludoviciana  White-winged Crossbill  Loxia leucoptera  Yellow-rumped Warbler  Dendroica coronata  Philadelphia Vireo Pine Siskin Purple Finch Rose-breasted Grosbeak Solitary Vireo  49  6 1  5 1 4 28  190 6 2 30 80  2 133 10 3  18 7  140  Table L A . (Continued)  Guild Name  Member Species  Scientific Name  Shrub Nesters  Alder Flycatcher  Empidonax alnorum Geothlypis trichas Vireo olivaceus Catharus ustulatus Dendroica petechia Mniotilta varia Wilsonia canadensis Chordeiles minor Oporornis agilis Junco hyemalis Catharus guttatus Melospiza lincolnii Oporornis Philadelphia Vermivora celata Seiurus aurocapillus Bonasa umbellus Vermivora peregrina Catharus fuscescens Zonotrichia albicollis  Common Yellowthroat Red-eyed Vireo Swainson's Thrush Yellow Warbler Ground Nesters  Black and White Warbler Canada Warbler Common Nighthawk Connecticut Warbler Dark-eyed Junco Hermit Thrush Lincoln's Sparrow Mourning Warbler Orange-crowned Warbler Ovenbird Ruffed Grouse Tennessee Warbler Veery White-throated Sparrow  Borers  Downy Woodpecker Hairy Woodpecker Northern Flicker Pileated Woodpecker Yellow-bellied Sapsucker  Bark Gleaners  Black and White Warbler Brown Creeper Magnolia Warbler Red-breasted Nuthatch White-breasted Nuthatch  Aerial Foragers  Alder Flycatcher Broad-winged Hawk Common Nighthawk Cooper's Hawk Least Flycatcher Olive-sided Flycatcher Tree Swallow  50  Picoides pubescens Picoides villosus Colaptes auratus Dryocopus pileatus Sphyrapicus varius Mniotilta varia Certhia americana Dendroica magnolia Sitta canadensis Sitta carolinensis Empidonax alnorum Buteo platypterus Chordeiles minor Accipiter cooper ii Empidonax minimus Contopus borealis Tachycineta bicolor  N 3 2 497 67 62 30 13 2 213 8 129 11 135 2 954 28 27 1 646 6 6 10 6 82 30 42 6 19 3 3 1 2 1 190 2 25  Table L A . (Continued)  Guild Name  Member Species  Scientific Name  N  Foliage Gleaners  American Redstart  Setophaga ruticilla  27  Black-billed Cuckoo  Coccyzus erythropthalmus  Black-capped Chickadee  Parus atricapillus  Black-throated Green Warbler Dendroica  Ground Foragers  virens  1 37 96  Blackpoll Warbler  Dendroica striata  Canada Warbler Cedar Waxwing Common Yellowthroat  Wilsonia canadensis Bombycilla cedrorum Geothlypis trichas  1  Golden-Crowned Kinglet House Wren Mourning Warbler  Regulus satrapa Troglodytes aedon Oporornis Philadelphia  4 2 135  Orange-crowned Warbler Philadelphia Vireo Pine Siskin Red-eyed Vireo Rose-breasted Grosbeak Ruffed Grouse Solitary Vireo Swainson's Thrush Tennessee Warbler  Vermivora celata Vireo philadelphicus Carduelis pinus Vireo olivaceus Pheucticus ludovicianus Bonasa umbellus Vireo solitarus Catharus ustulatus Vermivora peregrina  2 30 80 497 133 28 10 67 27  Warbling Vireo Western Tanager  Vireo gilvus Piranga ludoviciana  3 18  White-winged Crossbill Yellow Warbler  Loxia leucoptera Dendroica petechia  7 62 140  13 22 2  Yellow-rumped Warbler  Dendroica coronata  American Crow  Corvus brachyrhynchos  Blue Jay Brown-headed Cowbird  Cyanocitta cristata Molothrus ater  6 22  Chipping Sparrow  Spizella passerina  27  2  Common Raven  Corvus corax  Connecticut Warbler  Oporornis agilis  Dark-eyed Junco  Junco hyemalis  Gray Jay Hermit Thrush Lincoln's Sparrow  Perisoreus canadensis Catharus guttatus Melospiza lincolnii  28 129 11  Ovenbird  Seiurus aurocapillus  954  Purple Finch Veery  Carpodacus purpureus Catharus fuscescens  White-throated Sparrow  Zonotrichia albicollis  Winter Wren  Troglodytes troglodytes  51  5 213 8  2 1 646 11  Appendix II  


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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


Related Items