UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The abundance, distribution and brood parasitism of upland-breeding warbling vireos in a fragmented forest… Fonnesbeck, Christopher James 1998

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

Item Metadata

Download

Media
831-ubc_1998-0438.pdf [ 5.61MB ]
Metadata
JSON: 831-1.0088477.json
JSON-LD: 831-1.0088477-ld.json
RDF/XML (Pretty): 831-1.0088477-rdf.xml
RDF/JSON: 831-1.0088477-rdf.json
Turtle: 831-1.0088477-turtle.txt
N-Triples: 831-1.0088477-rdf-ntriples.txt
Original Record: 831-1.0088477-source.json
Full Text
831-1.0088477-fulltext.txt
Citation
831-1.0088477.ris

Full Text

T H E ABUNDANCE, DISTRIBUTION AND BROOD PARASITISM OF UPLANDBREEDING WARBLING VIREOS IN A FRAGMENTED FOREST LANDSCAPE by CHRISTOPHER JAMES F O N N E S B E C K B . S c , The University of British Columbia, 1996  A THESIS SUBMITTED IN PARTIAL F U L F I L M E N T OF T H E REQUIREMENTS FOR THE D E G R E E OF M A S T E R OF SCIENCE  In THE F A C U L T Y OF G R A D U A T E STUDIES D E P A R T M E N T OF Z O O L O G Y  We accept this thesis as conforming to the required standard,  T H E UNIVERSITY OF BRITISH C O L U M B I A July 1998  © Christopher James Fonnesbeck 1998  In presenting this thesis in partial fulfilment  of the  requirements for an advanced  degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department  or by  his  or  her  representatives.  It  is  understood  that  copying or  publication of this thesis for financial gain shall not be allowed without my written permission.  Department of  "?ooli  The University of British Columbia Vancouver, Canada Date  DE-6 (2/88)  ABSTRACT  In a two-year study of the Warbling Vireo Vireo gilvus, / explore patterns of distribution, habitat, and brood parasitism along an elevation gradient in the South Okanagan region, British Columbia using landscape analyses. The Warbling Vireo is threatened at low elevations in the Okanagan by intense parasitism pressure by Brown-headed Cowbirds Molothrus ater. Vireo persistence may depend upon neighbouring source populations in the montane forests overlooking the valley. In the first part of the study, Ifound brood parasitism to be significantly lower in higher elevation forests compared to the valley. Cowbirds are less and parasitise fewer species upland. Of sampled nests, only the Warbling Vireo is parasitised frequently, with approximately 30 percent of all nests suffering parasitism in the forest. Such low levels of cowbird activity are likely due to food limitation, since host species dominate the songbird community. Levels of upland parasitism were not linearly correlated with proximity to the valley, either due to sampling error or to local-scale ecological factors. Forest cover appears to affect parasitism levels, with most parasitism occurring on sites of moderate tree density. The second part of my study employs variables derivedfrom digital forest inventory maps that are used to predict Warbling Vireo distribution and avian species richness across the Okanagan Forest at two spatial scales. The presence of vireos is positively correlated with early serai stage variables at both the local stand level and the landscape level. At the landscape scale, species richness is associated positively with habitat diversity, and negatively with mature U  forest area in a multiple regression model. Though sustained industrial forestry in the  Okanagan Forest may enhance avian diversity and Warbling Vireo habitat alike, the frequenc  of vireo detections is significantly lower at higher elevation. This reduces the likelihood that montane forest is a source of Warbling Vireos for the valley, in spite of decreased parasitism pressure in upland habitat.  iii  TABLE OF CONTENTS  ABSTRACT  ii  T A B L E OF CONTENTS  iv  LIST O F T A B L E S  . vii  LIST O F FIGURES  ix  ACKNOWLEDGEMENTS  xii  CHAPTER 1  PATTERNS OF B R O W N - H E A D E D COWBIRD B R O O D PARASITISM IN A W E S T E R N UPLAND CONIFEROUS FOREST  1  Introduction  2  Brood Parasitism of Warbling Vireos in the South Okanagan, B C Methods  ,  5 5  Location and Biogeography  6  Study Sites  8  iv  Nest Monitoring  9  Census Techniques  10  Cattle Grazing  10  Remote Sensing Data  11  Statistical Analysis  12  Results  12 Nesting and Parasitism Data  15  Census Data  16  Logistic Regression  16  Discriminant Analysis  17  Discussion20 Warbling Vireo Parasitism  22  Conclusions  37  CHAPTER 2  ASSOCIATIONS OF FOREST S T R U C T U R E W I T H W A R B L I N G V I R E O H A B I T A T AND SONGBIRD RICHNESS IN A N U P L A N D CONIFEROUS FOREST  Introduction  37  37  Effects of Industrial Logging on Songbird Populations  38  The Warbling Vireo: A Case Study  41  v  Methods  :  43  Study Location  43  Data Collection  44  Data Analysis  45  Results  47 Forest Stand Models  48  Landscape Models  49  Discussion  50  Stand-level Models  51  Landscape Models  53  Conclusions  57  LITERATURE CITED  67  APPENDIX  72  vi  LIST O F T A B L E S  Table 1 Elevation and distance from the Okanagan Valley of study sites used in 1997  24  Table 2 Estimated tree density and 95% confidence limits for all 1997 study plots. Adjusted values due to clumping bias in brackets (see text)  24  Table 3 Number of nests discovered in 1996 by site and host type. See text for description of host categories  25  Table 4 Number of nests discovered in 1997 by site and host type. See text for description of host categories  25  Table 5 Number of birds detected in plot censuses by site and host type  26  Table 6 Parasitism levels (percent) recorded in this study for all species in 1997 and 1996 (upland only), compared to data from Ward and Smith (1999, Unpublished data). The Ward and Smith data are from lower elevation upland sites (900-1100 m). Nest sample size for each species are provided in brackets, and non-zero parasitism values are in bold  27  Table 7 Total-sample correlation coefficients for discriminant landscape variables  28  Table 8 Pooled within canonical structure of discriminant function  28  Table 9 Description of input variables for local-scale analysis of Warbling Vireo detections and songbird diversity  59  vii  Table 10 Maximum likelihood analysis of variance table for the categorical data model (PROC C A T M O D ) testing the effect of two forest stand variables on Warbling Vireo occurrence. 59 Table 11 Parameter estimates for biogeoclimatic zone categories in Warbling Vireo categorical data model. More positive values indicate association with a higher probability of vireo presence  60  Table 12 Analysis of variance table resulting from general linear model for mean number of species detected per station. F-values are based on Type III sums of squares  VIII  60  LIST O F F I G U R E S  Figure 1 Relative productivity of parasitised versus non-parasitised Warbling Vireo nests in the Okanagan Forest. The average clutch size is pooled over all nests  29  Figure 2 Map showing biogeoclimatic zone distribution across the study region (MS = Montane spruce, IDF = Interior Douglas-fir, ESSF = Engelmann spruce-Subalpine fir, PP = Ponderosa pine, I C H = Interior Cedar/ Hemlock) Figure 3 Map showing stand age distribution across the Okanagan Forest  30 31  Figure 4 Map showing layout of study plots in 1996 (green circles) and 1997 (blue rectangles). The 1996 sites are rough approximations of utilised area, whereas the 1997 plots are more precisely plotted using GPS co-ordinates  32  Figure 5 Parasitism by elevation and host type in 1996. Other species include SOSP and O C W A (one parasitised nest each). Sample size for each series indicated above each column . . 33 Figure 6 Parasitism levels summarised by site and species for 1997 data. Other host species include D U F L , WIFL, OCWA, and D E J U (One parasitised nest each). Sample size for each series indicated above each column  33  Figure 7 Comparison of W A V I parasitism along an elevation gradient. Historical data are pre1978 from Cannings et al. (1987). Valley and mid-elevation data includes nests from this  ix  study and from Ward and Smith (1999, unpublished data). Sample size for each series is in parentheses  34  Figure 8 Estimated cowbird egg laying dates (1997) relative to observed cattle presence and absence for West (top), North (middle) and East (bottom) sites  35  Figure 9 The relative occurrence of the most abundant species on each site. Refer to Appendix for songbird species codes  36  Figure 10 The relative occurrence of male and female Brown-headed Cowbirds across the study sites. The ratio of female cowbirds to hosts is included on the secondary Y-axis  36  Figure 11 Relationship of stand age to stand height in the Okanagan Forest. This relationship pools stands with respect to species  61  Figure 12 Range of Warbling Vireo densities by habitat type for the Bulkley Valley, British Columbia. Estimates are derived from point count estimates; sample size is shown in parentheses. Taken from Campbell et al. (1997) Figure 13 Map showing the placement of transects across the study region  61 62  Figure 14 Power analysis for general linear model of species richness. This illustrates the power of detecting an effect size of 20 percent  63  Figure 15 Results of power analysis of logistic regression for Warbling Vireo occurrence. This models the probability of detecting a probability change of 0.60 from 0.40 when the response variable is increased by one standard deviation  63  Figure 16 Power analysis results for species richness multiple regression of landscape variables. This illustrates the probability of detecting a 20 percent effect size  x  64  Figure 17 Relative occurrence of the 15 most abundant species detected on all routes. Blue bars indicate the proportion of stations recording at least one individual, while the brown bars represent the mean number of detections per station (n = 193)  65  Figure 18 A comparison of the rank abundance of the 15 commonest species on transect censuses, relative to plot censuses in the same region. The straight line represents a perfect correlation  65  Figure 19 Residual plot of species richness regression model. Absence of a trend implies a good fit of data to the model  66  xi  ACKNOWLEDGEMENTS  I would like to give acknowledge the wisdom and experience of my thesis supervisor, Jamie Smith. His criticism, advice, and guidance bolstered my confidence throughout the project and contributed immeasurably to the quality of this work. I would also like to recognise the contributions of my thesis committee members Kathy Martin, and Tony Sinclair. This work would not have been possible without the hard work of my field crew over two seasons in the Okanagan. My gratitude extends to Dawn Cooper, Jeanne Roy, Jennifer Franklin, Shannon Beglaw, Tatiana Martinovic, Cameron Wright, and Rima Sultan. I would also like to thank Jerry Maedel for his GIS expertise, Glenna Boughton of the B.C. Ministry of Environment, Lands, and Parks, Mamie Taggart from Weyerhauser Canada, Inc., Susan Lidner of the B.C. Ministry of Forests, Darryl Atkinson of the Science Council of B.C., Judy Myers, Werner Kurz, Harald Steen, Leanna Warman, Gordon "Coop" Cooper, Anna Lindholm, Edward Gregr, Shelley Dyer, and Nick. I also wish to acknowledge the moral and financial support of my family throughout my postgraduate program. This study was funded by a Forest Renewal, B.C. grant to James N . M . Smith.  xii  CHAPTER 1 PATTERNS OF B R O W N - H E A D E D COWBIRD BROOD PARASITISM IN A WESTERN UPLAND CONIFEROUS FOREST  Introduction  The Brown-headed Cowbird Molothrus ater is an obligate avian brood parasite, that breeds coast to coast in North America, from the southern Yukon Territory to Texas and Florida. Prior to the 19th century, this species was more modestly distributed through the short-grass plains of the Dakotas and Montana (Mayfield 1965), as far South as Oklahoma, and East to the Mississippi River (Rothstein and Robinson 1994). European settlement of the West cleared large tracts of forested land, allowing the cowbird to expand into newly opened habitat. The Brown-headed Cowbird uses open urban or agricultural habitat as a preferred foraging location, where it is often commensal with domestic cattle. Cowbirds' use of densely forested land may be limited by the local availability of feeding habitat (Donovan et al. 1999). Thus, conversion of contiguous forest to urban or agricultural land on a continental scale facilitates cowbird range expansion. Due to their parasitic lifestyle, female Brown-headed Cowbirds are not compelled to remain near a nest to attend young. This allows an individual female to optimise breeding and 1  2 foraging independently by commuting between feeding areas and preferred laying sites, often at forest-field ecotones or woodlot edges (Brittingham and Temple 1983, Steidl et al. 1997). Commuting distances can be 6-7 kilometres (Robinson et al. 1995a, Rothstein et al. 1984), or more (Goguen and Mathews 1997), though the modal distance is likely 1-2 km (Thompson 1994). Fragmentation of forests may elevate cowbird density by converting contiguous forest to ecotones and edge habitat, and by bringing potential cowbird foraging and breeding sites closer together. The degree to which fragmentation exacerbates community levels of brood parasitism is influenced greatly by the landscape composition and land use (Thompson et al. 1999). Where forest fragments are situated in a matrix devoid of agriculture and other preferred cowbird foraging habitat, increased parasitism is not expected, since food becomes limiting. Also, forests of dense, narrow trees (e.g. Lodgepole pine Pinus contorta) with little understory will not support large populations of preferred cowbird host species (i.e. small, insectivorous birds that build open cup nests), and therefore parasitism levels should remain low. Other factors, however, may enhance cowbird activities in forests. Grazing in regenerating forest is thought to provide foraging opportunities for Brown-headed Cowbirds. For example, Kellner et al. (1998) showed that the presence of a single active stable within an otherwise intact Virginia forest elevated the local frequency of parasitism.  Brood Parasitism of Warbling Vireos in the South Okanagan, BC The Okanagan Valley of British Columbia is a unique ecosystem in Canada (Cannings and Cannings 1996). It represents the northernmost extent of the Great Basin, and thus the  3 northern limit of the range of many plant and animal species. Because of this, the region is a provincial and national focus for conservation; much of the Okanagan Valley flora and fauna is found nowhere else in Canada. This fertile valley, embraced by upland coniferous forest to the East and more open forest and grassland to the West, has been dramatically reshaped during the past three decades by expanding urban development and changing agricultural practices (Cannings et al. 1987). This has placed terrific pressure on grassland and riparian bird species persisting in dwindling or degraded habitat. The Okanagan Valley avifauna has been well-documented during the past 100 years. Of roughly 96 species of songbirds known to breed in the region, 39 are known to have been parasitised by Brown-headed Cowbirds (Cannings et al. 1987). Nineteen species of this subset experience parasitism at appreciable (> 20%) levels. It is likely that Brown-headed Cowbirds have occupied this region for several centuries or more, as the morphometry of this race is distinct from populations in neighbouring regions (Ward and Smith 1998). The general effect of recent (< 30 years) human development in the valley on cowbird activity has not been quantified. At least one passerine species is thought to be threatened with local extirpation due to cowbird parasitism. The Warbling Vireo Vireo gilvus suffers parasitism levels approaching 80% (Cannings et al. 1987). This species is particularly vulnerable to the deleterious effects of brood parasitism; cowbird young hatch earlier and grow faster than vireo nestlings, thus quickly overcrowding the small, pendulate nest at the expense of host young. Warbling Vireos rarely fledge young from parasitised nests (Figure 1). A simple population viability analysis suggests that this species is not self-sustaining in the Okanagan Valley with current parasitism rates (Ward  4 and Smith, unpub. manuscript). Ward and Smith suggest that local Warbling Vireo populations are sinks being sustained by immigration from neighbouring source demes. Warbling Vireos are known to nest at moderate to high elevations in the Okanagan Forest, and it is possible that these birds are a source population. The size and reproductive output of the forest population has not been estimated, and the effects of human-induced changes in upland breeding habitat due to forestry are unknown. Given the deleterious influence of forest fragmentation on songbird populations shown in other systems (reviewed by Faaborg et al. 1995), it is possible that the Okanagan Forest songbird community, including the Warbling Vireo, is threatened. The effects of fragmentation on nest predation and parasitism vary strongly among landscapes (Faaborg et al. 1995). Thus, fragmentation may not have the same impact on songbird populations in Northwestern coniferous forests as it does in Midwestern and Eastern deciduous forests. Nest predation rates in the Okanagan upland forests are negatively correlated with distance-to-edge (Campbell 1995), but the extent of parasitism is not known. In fact, only a small number of studies have examined the impact of cowbirds in managed western coniferous forests (Naef 1997, Hejl and Young 1999). This aspect of cowbird ecology has been identified as a critical research need (Robinson et al. 1995a). This study addresses the impact of cowbirds in a western coniferous forest that has been intensively managed for timber harvest for forty years. I use data collected in 1997 to quantify cowbird abundance and activity at varying distances from chronic anthropogenic disturbances along the valley, where Brown-headed Cowbirds are common. These data are compared to data  5 from moderate-elevations collected by my co-workers and myself in 1996, and to nesting data recorded by Ward and Smith in 1992-95 (Ward and Smith 1999). Additionally, I examine forest stand characteristics and landscape features which may be correlates of cowbird abundance and parasitism in this region. These are analysed at local and landscape scales. The local analyses test whether cowbird activity varies with tree stand density, and with a selection of stand metrics. The landscape-level analysis seeks associations between Warbling Vireo parasitism and landscape features at varying (1-7 km) distances from the nest plot.  Methods  Location and Biogeography This work took place in the South Okanagan region of British Columbia, Canada during the summers of 1996 and 1997. All sites in both years of the study were located in managed crown land within the Okanagan Forest, from 5 to 25 kilometres East of the town of Okanagan Falls (49°20' N/119°20' W). In 1996, the study was restricted to sites within Tree Farm Licence (TFL) 15, administered by Weyerhauser Canada. The work in 1997 was expanded to include land up to 14 kilometres North of TFL 15. This land is actively harvested by the same company. The area straddles three biogeoclimatic zones (Figure 2): Interior Douglas Fir (IDF) at low elevation (<1100m), Montane Spruce (MS) at medium elevation (1100 to 1400m), and Engelmann spruce-Subalpine fir (ESSF) at high elevation (1400 to 1700m). Most of the area  included in this study is in the Montane Spruce zone. Lower elevation sites are dominated by a Douglas fir Pseudotsuga menziesii overstory, interspersed with stands of Trembling aspen Populus tremuloides, Red alder Alnus rubra, and Western larch Larix occidentalis. Dominant ground cover is variable, from Grouseberry Vaccinium scoparium, Nootka rose Rosa nutkana, and Labrador tea Ledum groenlandicum to Willow Salix spp. and young aspen and alder. As the land rises, Douglas fir is replaced with denser Lodgepole pine Pinus contorta latifolia stands or older stands of Engelmann spruce Picea engelmannii. Higher still, spruce stands are complemented with Subalpine fir Abies lasciocarpa. Pine stands are characterised by tall, slender trees growing at high density, with little or no understory. Deadfall is common. Fir-spruce patches are generally more open, with an understory which includes Bunchberry Cornus canadensis, Labrador tea, and Arctic Lupine Lupinus arcticus. Like most managed Western forests, the landscape consists largely of disturbed forest stands at various stages of regrowth (Figure 3). Clearings of 40-100 ha. perforate the landscape, with occasional larger clearings of up to about 900 ha. Larger clear-cuts consist primarily of Lodgepole pine saplings (from silviculture), Red alder, Trembling aspen, and willow in wetter areas. Blue grasses Poa spp. and Kinnikinnick Arctostaphylos uva-ursi dominate the ground cover.  Study Sites In 1996, four sites were established ad hoc along the major active logging roads in TFL 15 (Figure 4). In the interest of maximising the sample size of nests, these areas were chosen in  7 areas favourable for finding songbird nests. All selected sites were moist, open mixed stands with abundant deciduous understory. I covered three upland sites in the study. For comparison, one site on the valley bottom was also monitored at Vaseaux Lake (Vaseaux Wildlife Centre). To facilitate analysis by a geographic information system (GIS), selection of sites in 1997 was more systematic (Figure 4). As Brown-headed Cowbirds commute distances of up to 6-7 km between foraging and laying habitat (Robinson et al. 1995a), I ensured at least 7 km separation between study areas. The first of three sites (hereafter, West site) is situated between K m 9 and K m 14 of Shuttleworth Creek Road, thus overlapping two of the 1996 locations. A second (hereafter, North site) is north of West site, 3 to 14 kilometres outside of TFL 15, while the final site (hereafter, East site) is located between K m 25 and K m 31 of Shuttleworth Creek Road. Physiographic characteristics for these areas are summarised in Table 1. Nest searching and census efforts among sites were standardised by estabkshing six plots within each study location. The six plots (900 x 450 m, or -40 ha) were chosen to capture a range of habitat types, from young, open clearcuts to dense, undisturbed forest patches. They were chosen non-randomly at locations easily accessible by the existing logging road network, and marked with fluorescent flagging tape at 150-m intervals, resulting in a 28-point grid. I estimated tree density for each plot to allow quantitative comparisons among plots. I used T-square sampling (Krebs 1989) to estimate the density of coniferous or deciduous trees greater than 2 metres in height (Table 2). Samples were taken at random points along transects within the plots. Due to the landscape patterns resulting from clear-cut logging, all plots had a clumped distribution of trees. This biases the density estimates. To correct this bias, I adjusted  8  density estimates by a factor of -21%, in accordance with bias estimates for T-square sampling derived from Monte Carlo simulation (Engeman et al. 1994).  Nest Monitoring In both years, all sites were surveyed for nests of all songbird species. Effort was equalised among plots by following a plot rotation schedule. All surveys took place between 6:00 A . M . and 2:00 P . M . As songbird activity declines during the late morning and afternoon, survey times for all sites changed throughout the season. In 1997,1 restricted nest-searching activities to within 100 metres of the marked plots, so that nesting data could be related to census data from the same plots. All nests were marked with flagging tape placed no closer than 10 metres from the nest. Additionally, workers recorded nest U T M co-ordinates using a Magellan GPS 2000XL® hand-held global positioning system unit (I later converted these co-ordinates to the Albers Equal-Area Conical projection, facilitating their use with British Columbia Ministry of Environment, Lands, and Parks (MoELP) digital forest cover maps). Contents and incubation status of each nest were recorded, as well as nest height, tree species, and parental activities (if applicable). Nests located up to 12 m in the canopy were checked with bicycle mirrors attached to extendible aluminium poles, often in combination with climbing or using binoculars. Nests out of the reach of our equipment were surveyed for nestling or feeding activity to verify active status. All nests were revisited every 2-6 days until fledging or nest failure occurred. Areas around empty nests occupied during the previous visit were surveyed for activity of fledglings.  9 I define a successful nest as one that fledges at least one chick, regardless of species. Empty nests were scored as successful if they were known to contain at least one healthy nestling less than three days prior to the expected fledging date. Fledging dates were anticipated using Ehrlich et al. (1988). This will result in an upwardly biased estimate of nest survival, since some nests found empty during this period may have failed. Nests of ambiguous status were scored neither as successful nor unsuccessful. A nest was scored as parasitised if it was known to contain at least one Brown-headed Cowbird egg or nestling at any time. By this criterion, parasitism rates for species known to eject cowbird eggs will be underestimated, since ejection may occur within hours of laying, before the nest is found or checked.  Census Techniques Songbird censuses were conducted during the 1997 season only. Each study plot was censused twice during the summer, using 50-metre fixed-radius point counts (Hutto et al. 1986). As described above, a plot was defined by a grid of 28 flagged points (4 x 7), whereby each point served as a counting station. Fourteen points were counted in the first census, the complement in the second census. Each census was completed by one of four trained observers. Each count was eight minutes in duration, preceded by a one-minute pause upon arrival to the station. During the count, all vocalising birds were identified and classified as occurring either inside or outside a 50-metre radius from the observer. Juvenile and female vocalisations were ignored, with the exception of those of the Brown-headed Cowbird. A l l cowbird vocalisations were identified as male or female. To reduce observer bias, each of the four  10 workers were used equally among the three study sites, and all plots were surveyed by more than one observer.  Cattle Grazing Grazing of commercial cattle on crown land is permitted throughout the province of British Columbia. In the Okanagan Forest, grazing is commonplace throughout the songbird breeding season. Cattle are sent to pasture at low elevation during the late spring. As the season progresses they move upland following grass and shrub growth, reaching distances of 20 km from the valley floor, and elevations of greater than 1600 m (personal observation). As grazing may affect local brood parasitism levels (Coker and Capen 1995), cattle presence and absence was noted on all sites every third day during the 1997 season. Cattle presence was recorded if at least one cow was seen or heard on or near any plot. In my analysis, I compared temporal patterns in cattle movements with estimated cowbird laying dates in parasitised nests to explore associations between Brown-headed Cowbird habitat use and local grazing.  Remote Sensing Data The Kamloops Regional Office of the British Columbia Ministry of Environment, Lands, and Parks (BC MoELP) and Weyerhauser Canada, Inc. provided all digital data. I used Terrain Resource Inventory Maps (TRIM) and Forest Cover (FC1/FIP) maps to derive information about elevation, stand age and height, crown cover, land use, aspect and slope. All maps are 1:20000  11 scale, projected using the M o E L P standard Albers Equal-Area Conic projection in ArcView shapefile (.shp) format. The coverage includes nine mapsheets in the 082E mapsheet group: 082E043-45, 082E033-35, and 082E023-25. All spatial data was visualised and manipulated using ArcView 3.0a (ESRI, Inc. 1997). The B C Ministry of Forests Resource Inventory Branch collects forest cover data on an ongoing basis. The database for TFL 15 is maintained by the commercial licensee, Weyerhauser Canada. All stands from this database were inventoried between 1963 and 1996 using a variety of techniques (e.g. boring, photo interpretation). All stands inventoried before 1996 were projected forward in age and height. As this region is heavily inventoried, estimates in the database should be reliable (R. Woods, B C Ministry of Forests Resource Inventory Branch, personal communication).  Statistical Analysis I calculated daily nest survival rates using a maximum likelihood Mayfield estimator (Krebs 1989). These estimates were stratified by site, plot, parasitism status and nest type. A l l calculations were carried out using the program M A Y F I E L D (Krebs 1989). I compared parasitism among plots of varying tree density using a mixed-model logistic regression. In this analysis, I tested for an effect of plot tree density on nest parasitism. To test for spatial correlations between plots, the site (North, East, or West) on which each nest was located was entered as a second factor. This analysis was performed using PROC PROBIT in SAS/STAT (SAS Institute, Inc. 1996).  12 The effect of local stand characteristics on parasitism was estimated using a multiple discriminant analysis (MDA) procedure. This analysis was restricted to Warbling Vireo nests only, as it is the primary host species of interest, and the only one for which we obtained an adequate sample size of nests. The discriminant variables were derived from the forest cover GIS database. Each nest was geocoded into the GIS using GPS co-ordinates recorded in the field. A simple query determined the polygon that intersected each nest for each attribute theme. Queries were performed on seven themes: elevation, slope, height, and crown closure class. Crown closure class was coded as four dummy variables representing five classes. Stand age was omitted due to a high correlation with height. This analysis was performed using PROC CANDISC and PROC DISCRIM in SAS/STAT. Statistical power analyses indicate extremely low power at a = 0.05 with 104 degrees of freedom. The power of detecting a standardised effect size of 0.05 increases from 0.43 to 0.58 when a is raised to 0.10. In the interest of balancing Type I and II error probabilities, a = 0.10 was used for all statistical tests in this study.  Results  Nesting and Parasitism Data I discovered and monitored 92 nests during the 1996 breeding season. These nests are summarised by host type and site in Table 3. Host type status is assigned according to the species'  13 regional frequency of parasitism (Cannings et al. 1987) and response to cowbird eggs. The categories are defined as follows: Host+:  Primary hosts; frequently parasitised (>30%), can rear cowbird young.  Host:  Secondary hosts; infrequently parasitised (10-30%), can rear cowbird young  Ejector:  Ejects cowbird eggs  Non-host:  Seldom parasitised (<10%), avoided due to large size, inappropriate food, or nest type  Suitable host species dominated the total number of nests found across all sites (73 percent on all sites, 97 percent upland sites only). Due to the attention paid to host species during the study, this pattern may be biased. For example, cavity-nesting species such as Mountain Chickadees Parus gambeli are both difficult to locate and of little interest to the study. The number of nests of ejector species is high in the valley site relative to the upland sites, due to the presence of Grey Catbirds Dumetella carolinensis on the valley site. The 1997 field season yielded 111 nests (Table 4). As in 1996, the data set is dominated by primary host nests. Census data (see below) mirrors this distribution, and should not be susceptible to a host bias (Table 5). The most notable difference is the increased number of secondary host nests (40 percent in 1997, compared to 14 percent in 1996). Despite the dominance of suitable host species in the Okanagan Forest avifauna, cowbird parasitism was infrequent on all upland forest sites (Table 6), relative to rates reported in the valley (Ward and Smith 1999). Not only are community parasitism rates low at all upland  14 locations in both years (only 9 percent of 66 nests in 1996, 11 percent of 111 nests in 1997), but the suite of species utilised by Brown-headed Cowbirds has been drastically reduced. In 1996, four of the six parasitised nests discovered were Warbling Vireo nests (Figure 5). Similarly, 75 % of all parasitised nests in 1997 were built by Warbling Vireos (Figure 6). No other species was parasitised on more than one occasion in either year. Multiple parasitism did not occur in 1996, and was infrequent in 1997. Two Warbling Vireo nests were doubly parasitised in the second year, one of which collapsed with two large nestlings inside; the other successfully fledged both cowbird chicks. Contrary to my predictions, the pattern of parasitism in upland forests appears not to be influenced by a site's proximity to urban development and agriculture. In both years, there was no trend of increasing parasitism toward the valley. In 1996, the highest rate of parasitism was recorded at the most distant site. The differences in parasitism among sites were not statistically significant for the songbird community (X  2 =  (X = 2.917, 2 d.l,p> 2  1.354, 2 d.f, p > 0.5), nor for the Warbling Vireo  0.2). In 1997, parasitism differed between sites for all species (X = 5.307, 2  2 d.f.,/? = 0.07), but not for the Warbling Vireo alone (X = 1.386, 2 d.l,p = 0.5). 2  Despite being the only species used commonly as a cowbird host, Warbling Vireos suffered less parasitism in upland habitat than in the Okanagan Valley. Figure 7 shows an approximate two-fold drop in parasitism at higher elevation when compared to both current and historical records in the valley (Cannings et al. 1987, Ward and Smith 1999). Across upland sites, Warbling Vireo parasitism does not appear to be influenced by proximity to the valley.  15 The relationship between the presence of grazing cattle and parasitism events is illustrated in Figure 8. Differences in cattle arrival dates among sites are related to the distance of each site to the valley (Table 1), where cattle begin their up-slope movement. With the exception of the West site (i.e. lowest elevation), there was no coincidence between the arrival of cattle and the onset of cowbird laying activity. Herds attended the West site throughout the season, as it was nearest to the commercial ranch property. The earliest record of cattle on the North site is an outlier, as there was no consistent cattle presence there until approximately two weeks later.  Census Data The mean number of birds detected on the study plots in 1997 differed according to site (F  2 1 5  = 9.64, p = 0.002). In particular, the West site supported more birds than the other sites  (Tukey's HSD). The mean number of species detected per plot did not differ between sites (F ^ 2  = 2.l7,p = 0.\48). Host species predominated throughout the region, and the Dark-eyed Junco was the modal species at all sites (Figure 9). The Warbling Vireo was among the ten most abundant species on each site. The mean number of cowbirds observed on each site differed significantly for females (F  2jl5  = 10.51,/) = 0.001) and both sexes pooled (F  2]5  = 6.57, p = 0.009).  More cowbirds were detected on the West site than on the East site, but neither differed from the North site (Tukey's HSD). Additionally, there were more females on the West site than either the East or North sites (Tukey's HSD). The large number of detections of male cowbirds on the West site (cowbirds are the fifth most commonly detected species there) may indicate the presence of more females than we detected. A useful parasitism metric is the ratio of female  16 cowbirds to hosts (Robinson et al. 1999). This ratio was very low for all sites, not exceeding 0.03 in any case (Figure 10). A ratio of 0.05 or greater is associated with a high level of parasitism (Robinson et al. 1999).  Logistic Regression Tree density was negatively correlated with the probability of nest parasitism (% = 3.01, 2  1 d.f, p = 0.083). However, the maximum likelihood estimate of the coefficient for tree density is very small (b = 0.003), which reduces the importance of this variable. The study site in which the nest was located was also influential (X = 4.89, 2 d.f.,p = 0.087); a nest was less likely to be 2  parasitised if it was located on the West (b = 0) site than the North site (b = 0.848), and less likely still if on the East site (b = -1.487).  Discriminant Analysis The strongest univariate correlations among discriminant variables were between the high crown-closure classes and stand height (Table 7). These correlations are moderate and positive (up to r = 0.59, n = 27). In contrast, correlations between low crown closure classes and height are generally negative. A test for homogeneity of within covariance matrices suggested that the matrices for the two groups were unequal (X = 182.32, 36 d.f.,/? <0.001), therefore the 2  matrices were not pooled for the discriminant function. Also, Shapiro-Wilk tests for normally distributed discriminant variables revealed a non-normal distribution of stand height observations  17 (d.f.= 16, p = 0.01). This renders any test of the significance of the discriminant function unreliable. As this is a simple two-group discriminant function analysis, only the first discriminant function is analysed. The function has a low eigenvalue (0.32). The canonical structure suggests that no single variable was strongly correlated to this function (Table 8). Of all variables, the low crown-closure class was most strongly correlated. The derived Fisher's function categorised 84.2% of all nests correctly as parasitised or non-parasitised. Thirteen of nineteen unparasitised nests were correctly classified, along with all eight parasitised nests. This is a high success rate, and significantly different from random (% ~ 10.56, 1 d.f.,p <0.001). 2  Discussion  The strongest pattern emerging from this study is a substantial reduction in cowbird activity in the Okanagan Forest, relative to the neighbouring valley. This reduction manifests itself in two ways: Lower levels of parasitism in the avian community, and a reduction in the number of species parasitised. This pattern cannot be explained by a lack of nesting opportunity in upland habitats. Though the songbird community composition changes with elevation, host species are dominant at all locations. Moreover, several commonly-parasitised host species that are present at low elevations appear in equal or greater numbers in the upland forest (e.g. Warbling Vireo, Chipping Sparrow Spizellapasserina).  18 Cowbirds are thought to be limited by food, rather than hosts, in contiguously forested habitat (Robinson et al. 1995a). It is possible that opening up forest landscapes with clearcut forestry enhances feeding opportunities for cowbirds, facilitating their penetration into new habitat. This is supported by some evidence of higher forest cowbird abundances in clearcuts (Thompson et al. 1992). However, clearcuts may not provide prime foraging opportunities for cowbirds, relative to agricultural or urban settings (Robinson et al. 1995a). Still, the relative richness of food in forest clearings compared to known cowbird foraging habitat has not been estimated. A possible mechanism to boost cowbird foraging opportunities is cattle grazing. Cattle are frequently utilised by cowbirds to stir up large insects and spiders while they graze. Thus, feed lots and pastures are often selected as foraging habitat by Brown-headed Cowbirds. Though grazing occurs on all parts of the upland Okanagan Forest at different times during the spring and summer, my results show that cowbird laying activities are not well correlated with cattle movement. It is possible that because these herds are highly mobile, cowbirds perceive them as ephemeral food sources, and do not use them regularly. In contrast, permanent livestock structures such as stables can significantly elevate local levels of parasitism (Kellner et al. 1998). This explanation may depend on the territoriality of female cowbirds, since non-territorial birds would be free to follow cattle across the landscape. Brown-headed Cowbird territoriality appears to vary geographically (Yokel 1989, Teather and Robertson 1984, Darley 1983). Despite generally lower levels of nest parasitism in the upland forest, there is no consistent linear trend of decreased parasitism with either elevation or valley proximity. This may  19 be the result of a poorly randomised sampling design and small sample size, resulting in noisy relationships, bias and sampling error. Since the study plot selection was non-random, biases are inestimable. Alternately, though proximity to the valley may limit the number of breeding cowbirds, it is possible that local factors contribute to variance in parasitism levels. In 1997, most of the recorded parasitism occurred on the North site, where fewer cowbirds and fewer hosts were detected, and there was a lower cowbird:host ratio than at the low-elevation west site. Further, at the East site, which is at a similar elevation and distance from valley compared to the North site, I only recorded one parasitism event. This result suggests that site effects are important. The statistical landscape models hint that landscape variables influenced local parasitism. Most sample plots suffering parasitism had moderate tree density and cover. Very dense habitat in M S and ESSF zones are dominated by lodgepole pine, characterised by tall, thin, branchless stems. There is little understory in this forest type, making it poor songbird nesting habitat. We discovered few host nests in mature Lodgepole pine stands, and no parasitism was recorded there. In comparison, very open clearcuts supported many host species. However, parasitism was also rare in cleared forest, perhaps due to a lack of adequate cowbird perches from which to survey for hosts. This has been cited as a possible explanation for low grassland parasitism rates, even in habitat with high densities of both cowbirds and hosts (VanderHaegen and Walker 1997 contra N . Mahony, unpublished data).  20 Warbling Vireo Parasitism On my study sites, cowbirds do not have much impact on the avian community, with the exception of the Warbling Vireo (two of three Orange-crowned Warbler Vermivora celata nests were parasitised over two years, but this sample is tiny). The majority of parasitised nests discovered in both years of the study were of Warbling Vireos. There are two plausible explanations for the apparent preference of cowbirds for this host species. First, cowbirds may select Warbling Vireo nests over others. Vireos are excellent hosts which often fledge cowbirds successfully, and whose own young compete poorly with parasitic nest-mates. As a taxon, vireos may be the most intensely parasitised of all cowbird hosts (Grzybowski et al. 1986, Ward and Smith, unpub. manuscript). This apparent preference is likely the result of natural selection for hosts that raise cowbird fledglings successfully, though the degree to which cowbirds actively select hosts is unclear. As cowbird densities are low in upland forests, they may be more selective for hosts than in lowland habitat, where competition for nesting opportunities is more intense. Alternatively, Warbling Vireos may be victimised more frequently because they are more easily detected by cowbirds. Males are noted for singing loudly while incubating, making their nests easy to find for humans (and probably cowbirds). If cowbirds are limited by perches, or otherwise prevented from searching efficiently in fragmented forests, nests of Warbling Vireos may be the most readily discovered. The discriminant function successfully predicted vireo parasitism status, particularly those nests which were parasitised. The success of this model is mitigated by three important caveats. First, the non-normality of the data compromises the reliability of the classification error  21 rate. Second, since the data were used to derive the model and to test it, the error rate could likewise be underestimated. A larger sample size would be required to use two independent data sets for fitting and testing. Finally, the use of dummy variables (cover class categories) in an analysis with small sample sizes can cause the model to over-fit the data (V. Lemay, personal communication). Also, the notable lack of association between the function and the original variables makes it unlikely that they should be such efficient discriminators as a group. I do not believe that this discriminant function would be as successful in classifying an independent data set. In spite of this, aspects of this result make biological sense. If parasitism is limited by cowbirds' ability to detect nests, as I have suggested, factors such as slope and crown cover should discriminate parasitised from non-parasitised nests. The pattern of Warbling Vireo parasitism from the Okanagan Valley to the adjacent montane forest is consistent with a source-sink population model. The valley population is not likely to be self-sustaining, due in large part to strong parasitism pressure (Ward and Smith, unpub. manuscript). If this is true, it is important that source populations be identified and monitored. Warbling Vireos are abundant in the Okanagan Forest, and do not appear to be threatened by brood parasitism. Perhaps of greater concern is the status of vireo habitat in the Okanagan Forest. The early-sucessional deciduous regrowth and reduction of stand age associated with clearcut tree harvesting should increase Warbling Vireo habitat availability in actively harvested forest. This should apply particularly in the Lodgepole pine-dominated old growth of the Okanagan forest. However, silviculture and management practises in this region involve the thinning and removal of young aspen and alder in cleared patches scheduled for  .  2  2  replanting, and riparian areas are not conserved (personal observation). Such activities have the long-term effect of reducing the habitat available to breeding Warbling Vireos. In my study, three nests were located in small Lodgepole pine. Warbling Vireos are not documented to nest in conifers (Ehrlich et al. 1988, Campbell et al. 1997), therefore, this choice of nest site selection could be driven by a lack of preferred nest trees. All three nests in pine failed to fledge young. More information about the movement, reproductive output, and dispersal of Warbling Vireos in upland forests is required before this region should be considered a true source population. The balance of this thesis is concerned with the characterisation of Warbling Vireo habitat in the Okanagan forest using census data and GIS-derived forest structure variables. This needs to be augmented with juvenile survival and behavioural data in future studies. If dispersal distances of yearling vireos can estimated, a metapopulation model could be constructed to predict population dynamics over time, and forecast the welfare of the Warbling Vireo in the South Okanagan.  Conclusions  I found lower parasitism levels overall in upland forest habitat in the South Okanagan, relative to the adjacent Okanagan Valley, both community-wide and among primary host species. The Warbling Vireo was virtually the only host affected by parasitism. This result is contrary to predictions, as decades of forestry have produced a patchy, open landscape, thought to be favoured by breeding female Brown-headed Cowbirds (Robinson et al. 1995b), and this is augmented by an avifauna dominated by host species, and large numbers of grazing cattle moving  23 upland throughout the spring and summer. Cowbirds may be limited by foraging opportunities in the forest, or alternately, by inefficient nest searching due to a lack of survey perches. Warbling Vireos, though they are the modal host for cowbirds in this region, experienced reduced parasitism pressure in upland habitat. This leaves open the possibility that the montane forest population acts as a source of vireos for the Okanagan Valley. Multivariate models used to describe patterns of parasitism in the region were somewhat successful. The probability of parasitism was weakly and negatively associated with forest density as modelled by logistic regression, and site effects were important. A discriminant function was able to correctly classify 82 percent of all Warbling Vireo nests parasitised or not, based on landscape variables. However, the model was not highly correlated with any variables, and may have been subject to over-fitting.  Table 1 Elevation and distance from the Okanagan Valley of study sites used in 1997. Site  Mean Elevation (m)  Distance from Valley (km)*  West  1252  7  North  1538  15  East  1590  17  * Measured to valley bottom (i.e. Okanagan River) Table 2 Estimated tree density and 95% confidence limits for all 1997 study plots. Adjusted values due to clumping bias in brackets (see text). Site  Tree Densitv  Lower  Urmer  1  188 (155)  174(143)  204 (168)  2  480 (396)  3  520 (429) 404 (333)  374 (309)  567 (468) 439(362)  4  120 (99)  111 (91)  131(108)  5  811(670)  749 (619)  884 (730)  6  252 (208)  233 (192)  275 (227)  1  11(9)  10(8)  12(9)  2  533(440)  492 (406)  581(480)  3  21 (17)  19(15)  23 (19)  4  278(229)  258 (213)  302 (249)  5  441 (364)  407 (336)  481 (397)  6  993(820)  914 (755)  1087 (898)  1  137(113)  129 (106)  146(120)  2  47 (38)  43(35)  51(42)  3  226 (186)  214(176)  240 (198)  4  259 (214)  241(199)  280 (231)  5  779 (643)  722 (596)  846 (699)  6  1947 (1609^1  1804(1490^)  2115 fl747)  Plot  West  North  East  25  Table 3 Number of nests discovered in 1996 by site and host type. See text for description of host categories. Site  host+  host  ejector  nonhost  Total  Valley  12  1  13  0  26  Upland Low  9  5  1  0  15  Upland M i d  15  2  3  2  22  Upland High  18  5  5  1  29  Total  54  13  22  3  92  Table 4 Number of nests discovered in 1997 by site and host type. See text for description of host categories. Site  host+  host  ejector  nonhost  Total  West  22  15  4  1  42  North  17  13  2  1  33  East  17  12  7 '  0  36  Total  56  40  13  2  111  26  Table 5 Number of birds detected in plot censuses by site and host type. Site  host+  host  ejector  nonhost  unknown  Total  North  139  236  5  95  10  485  East  168  266  15  137  9  595  West  273  328  27  215  8  851  Total  580  830  47  447  27  1931  27 Table 6 Parasitism levels (percent) recorded in this study for all species in 1997 and 1996 (upland only), compared to data from Ward and Smith (1999, Unpublished data). The Ward and Smith data are from lower elevation upland sites (900-1100 m). Nest sample size for each species are provided in brackets, and non-zero parasitism values are in bold. Species  1997 (This study)  1996 (This study)  1992-95 (Ward and Smith)  Warbling Vireo Dark-eyed Junco  30 (27) 6(16)  40 (10) 0(2)  53 (55) 0(5)  American Robin Chipping Sparrow  0(9)  0(3)  0(9) 11(9) 25 (4)  0(6) 0(2)  0(47) 42 (19)  Hermit Thrush Dusky Flycatcher Song Sparrow Swainson's Thrush Willow Flycatcher  0(4) 0(4)  0(2)  19 (31) 61(59) 50 (2)  0(7)  25 (16)  Cedar Waxwing Nashville Warbler  0(2) 0(2) 50 (2)  0(7) 0(2)  0(16)  0(1) 0(1) 0(1)  0(1)  0(1)  —  0(1)  0(2)  Orange-crowned Warbler Common Yellowthroat Lincoln's Sparrow Northern Waterthrush Savannah Sparrow Townsend's Solitaire Varied Thrush  0(4)  White-crowned Sparrow  0(1) 0(1)  Wilson's Warbler Western Wood Peewee  0(1) 0(1)  0(10) 33 (3)  100 (1)  —  —  —  —  41 (36)  Table 7 Total-sample correlation coefficients for discriminant landscape variables. Variable  SLOPE  SLOPE  1.00000  AGE  0.03235  AGE  ELEV'N  HEIGHT  CC1  CC2  CC3  CC4  1.00000  ELEV'N  -0.52192 -0.26351  HEIGHT  -0.01765  1.00000  0.96340 -0.27752  1.00000  CC1  0.02123 -0.29976 -0.26161 -0.32707  CC2  0.18907  0.04056 -0.35559  0.05795 -0.23452  CC3  -0.21502  0.37827 -0.10017  0.59380 -0.23452 -0.08000  CC4  0.01346  0.63914 -0.08765  0.51005 -0.29315 -0.10000 -0.10000  Table 8 Pooled within canonical structure of discriminant function. Variable  CAN1  SLOPE  0.397554  AGE  0.333837  ELEVATION  0.1418  HEIGHT  0.235484  CC1  -0.50891  CC2  0.316709  CC3  -0.215764  CC4  0.399826  1.00000 1.00000 1.00000 1.00000  Figure 1 Relative productivity of parasitised versus non-parasitised Warbling Vireo nests in the Okanagan Forest. The average clutch size is pooled over all nests. Sample sizes are indicated with each column.  Figure 2 Map showing biogeoclimatic zone distribution across the study region (MS = Montane spruce, IDF = Interior Douglas-fir, ESSF = Engelmann spruce-Subalpine fir, PP = Ponderosa pine, ICH = Interior Cedar/ Hemlock).  Figure 3 Map showing stand age distribution across the Okanagan Forest.  *  Penticton  '  North  0 B ^ West  r~~| P~l  East o  1997 Plots 1996 Sites Roads  ® S N  10  0  10  20 Kilometers  Figure 4 Map showing layout of study plots in 1996 (green circles) and 1997 (blue rectangles). The 1996 sites are rough approximations of utilised area, whereas the 1997 plots are more precisely plotted using GPS co-ordinates.  Low  Medium  High  Site WAVI  Other host  All species  Figure 5 Parasitism by elevation and host type in 1996. Other species include SOSP and O C W A (one parasitised nest each). Sample size for each series indicated above each column  West  North  East  Site WAVI  Other Host  All Species  Figure 6 Parasitism levels summarised by site and species for 1997 data. Other host species include DUFL, WIFL, OCWA, and DEJU (One parasitised nest each). Sample size for each series indicated above each column.  100  | I < 600 m  [J  600 - 1400 m  > 1400 m  Figure 7 Comparison of WAVI parasitism along an elevation gradient. Historical data are pre-1978 from Cannings et al. (1987). Valley and mid-elevation data includes nests from this study and from Ward and Smith (1999, unpublished data). Sample size for each series is in parentheses.  10-May  30-May  20-M ay  10-May  19-Jun  09-Jul  29-Jul  19-Jun  09-Jul  29-Jul  09-Jun  30-May  •  C attle  B H C O Egg  Figure 8 Estimated cowbird egg laying dates (1997) relative to observed cattle presence and absence for West (top), North (middle) and East (bottom) sites.  36 120 100  A  80 -\ cn o O  60 40 -| 20  -I  0 -  DEJU  WAVI  YRWA MOCH  RCKI  SWTH OCWA  CHSP  DUFL  BHCO  PISI  HETH  SOSP  WIWA  Species West  North  East  Figure 9 The relative occurrence of the most abundant species on each site. Refer to Appendix for songbird species codes.  West  Male  North Site Female  East  / \  Female BHCO: Host  Figure 10 The relative occurrence of male and female Brown-headed Cowbirds across the study sites. The ratio of female cowbirds to hosts is included on the secondary Y-axis.  CHAPTER 2 ASSOCIATIONS OF FOREST STRUCTURE WITH WARBLING VIREO HABITAT AND SONGBIRD RICHNESS IN AN UPLAND CONIFEROUS FOREST  Introduction  The South Okanagan, British Columbia is a unique ecosystem in Canada. Its flora and fauna are diverse because of the wide range of habitat types in the region, from semi-arid "pocket deserts" in the valley to subalpine coniferous forest upland. The Okanagan is the Northern extent of the Great Basin, so many species occurring here are found nowhere else in Canada. The valley is also a popular resort and retirement destination, and has experienced intense urban development in recent decades. Wetlands, grasslands, lowland forest and riparian habitat have been converted to vineyards, orchards and residential or commercial development. The extensive loss of habitat in the valley has placed enormous pressure on the rarer and habitat-sensitive species, threatening some with local extinction (e.g. Yellow-Breasted Chat Ictera virens, Chocolate Lily Fritillaria lanceolata, Great Basin Spade-foot Toad Scaphiopus intermontanus). The Warbling Vireo, though relatively common continent-wide, is one species in danger of extirpation in the South Okanagan Valley. This is probably the result of a high rate of brood parasitism by Brown-headed Cowbirds. In Chapter One, I showed that a pattern of reduced vireo parasitism in upland habitats is consistent with the hypothesis that a small but persistent valley 37  38 population is supported by iirimigration from a more productive upland forest population. However, the population status of the Warbling Vireo in the Okanagan Forest is unclear. For the forest to act as a source for the flagging valley population, it must provide enough habitat to support a sizeable vireo deme. An accurate prediction of population size is difficult because of the landscape dynamics associated with the heavily managed Okanagan Forest. Harvesting and silviculture have shaped and are changing the composition and structure of the forest. To assess the health of the population, habitat associations must be established, and habitat availability must be estimated. Extensive information about vegetation cover and landscape characteristics exists for most regions in the province of British Columbia, in the form of digital forest cover maps and terrain resource inventory maps (TRIM) at 1:20 000 and 1:50 000 scales (Kurz et al. 1998) This database is updated and maintained by provincial government agencies and private tree farm licence (TFL) holders (forest product companies). It represents a potentially valuable source of information for landscape analyses of wildlife populations. To date, such data have not been extensively utilised by population biologists. In this study, I use digital forest cover data to look for associations between habitat characteristics and Warbling Vireo occurrence in the Okanagan Forest.  Effects of Industrial Logging on Songbird Populations The effect of logging on songbird populations has been the focus of much research (reviewed in Thompson et al. 1995). A synthesis of the studies at hand reveals few generalities.  39 Among them is a mixed response of species to forest fragmentation (Freemark et al. 1995, Hejl et al. 1995). There may also be a differential effect according to migratory status, with resident species affected most profoundly, followed by long distance migrants, then short distance migrants. However, work to date has been dominated by studies of local phenomena, mainly the influence of tree removal. In addition, such work has been concentrated in Eastern and Midwestern USA. In Western forests, the effects of fragmentation by logging upon breeding songbirds are unclear. There have been many studies on the impacts of forestry, but these tend to be short-term population comparisons of the distribution and abundance of species in cut and uncut forests (Hejl et al. 1995). A common characteristic of coniferous forests managed for timber harvest is the relatively rapid regeneration of cleared land to forest. This regrowth may be due to natural reseeding or facilitated by silvicultural prescriptions. In the Okanagan Forest, stands are approximately five metres in height just 25 years after harvest, and up to 17 metres after 50 years (Figure 11). Conclusions based on short studies of species abundance in clearcuts may not apply beyond the first few seasons after harvest. Some key characteristics typify a coniferous forest managed for tree harvest. First, clearcutting (the predominant silviculture system in North America) tends to yield a landscape mosaic of even-aged stands over time (Thompson et al. 1995). This altered mix of serai stages benefits some songbird species and harms others (Thompson et al. 1992). Second, forests harvested for maximum sustained yield have a truncated serai stage distribution, with the gradual elimination of old-growth (> 100 years) stands. Second-growth monocultures with fewer foliage  40 strata and less diverse structural characteristics may reduce avifaunal diversity in regrowth under shorter harvest rotations (Bunnell and Kremsater 1990). Finally, managed forests are invariably subjected to fire suppression. Fire is an important natural disturbance in western forest systems (Hejl et al. 1995). It differs importantly from harvesting disturbance in that significant structure is retained after burning; forests are usually opened up rather than cleared. This serves to prevent the natural establishment of monocultures. Though cavity-nesting species in particular benefit from burning, the effect of fire on passerine populations is variable (Rotenberry et al. 1995). Avian population dynamics change with stand succession. Regenerating stands first experience an increase in avian species diversity (Thompson et al. 1995), followed by a decrease until mid-succession, as understory is eliminated by competition for light. Diversity will increase again if trees are thinned or gaps form. During succession, there is a marked turnover of species. Ground foraging and nesting species typical of clearcuts are replaced by shrub and sapling gleaners when saplings appear, which are in turn replaced by canopy nesters, cavity nesters, and understory species in mature stands. Thus, a mosaic forest of patches at different serai stages is diverse when measured at the appropriate landscape scale. Although clearcut silviculture regimes tend not to generate true habitat isolates characteristic of classical metapopulations, the patchwork of habitat types produced by clearcutting can affect the distribution of species within a forest (Freemark et al. 1995, Freemark et al. 1995, Dunning et al. 1992). In particular, the juxtaposition of patches of varying quality can set up source-sink dynamics (Dias 1996). Attributes such as patch size, forest type, crown cover, and edge to interior ratio can determine whether a given stand is of good or poor quality.  41 Despite a well-developed theory of landscape ecology, empirical studies (particularly manipulative experiments) of the influence of forest landscape structure on animal populations are scarce. Myriad measures of landscape structure and composition have been identified (Freemark et al. 1995), partly due to the advent of remote sensing and GIS technologies. Yet, very few of these have been tested for their utility in quantifying relationships between landscape heterogeneity and populations. The large spatial scale required for such analyses make experiments difficult, and as a result, computer modelling often replaces experimentation. There are many indices of landscape structure and composition, and individual species may respond to them differently. Adding to this complexity, these specific effects are often nested in biogeographic context.  The Warbling Vireo: A Case Study The Warbling Vireo is a good candidate for a landscape analysis, not only for its conservation significance in the South Okanagan, but because there are ecological data on the effects of forestry and forest structure on this species. This information is used to make predictions for this study. Warbling Vireos are primarily forest songbirds, occupying and breeding in a range of habitats, from deciduous and coniferous stands to clearings and suburban land (Ehrlich et al. 1988, Campbell et al. 1997). In British Columbia, they inhabit aspen stands, mixed forest and clearcuts at varying densities (Figure 12). Preferred forest types include aspen, spruce-fir, Ponderosa pine, and mixed conifer (Hejl et al. 1995). The Warbling Vireo is thought  42 to favour open habitat and forest-clearing edges over dense, contiguous stands (Campbell et al. 1997). It is a neotropical migrant, wintering in the Southeastern U S A and Northern Mexico. Warbling Vireos respond positively to increased edge habitat and are more abundant in coniferous stands that are adjacent to hardwood stands (Rosenberg and Raphael 1986). Thus, herbicide applications and removal of non-commercial deciduous trees may affect vireos negatively. Hejl et al. (1995) summarised the results of several studies comparing avian species' responses to cut versus uncut forest. In this meta-analysis, Warbling Vireos initially responded negatively to clearcut habitat in areas dominated by low shrub. As tall shrub became dominant, the abundance increased strongly. Partially cut (thinned) forest elicits a slight positive response. Using independent variables derived from forest cover maps, I test some of the aforementioned habitat associations in the Okanagan Forest. Information available from digital forest inventory maps includes stand age, stand height, biogeoclimatic zone, crown closure, and habitat diversity. These measures are correlated both with one another, and with other ecological factors. For example, age is related to tree size, foliage volume and stratification, and volume of coarse woody debris. The biogeoclimatic zone classification yields information about forest composition, structure, and rate of succession. The GIS variables are analysed using multivariate statistical models as predictors of Warbling Vireo habitat use. Given what is known about Warbling Vireo patterns of diversity, and the characteristics of the Okanagan Forest, I predicted that vireo abundance should be higher on early serai stage stands, resulting in a higher probability of vireo detection on point count censuses in this habitat. Second, vireo abundance should be highest in habitat with low crown closure, reflecting a  43 preference for open forest. Finally, vireos should prefer high habitat diversity because the juxtaposition of differently-aged stands will result in increased edge. Overall, one would expect this species to respond well to disturbance by logging: Undisturbed mature growth in the Okanagan Forest is dominated by high density Lodgepole pine Pinus contorta which offers no understory or structural diversity. Creating gaps in such a forest should create Warbling Vireo habitat. As a parallel analysis to the single species analysis of the Warbling Vireo, I analyse patterns of species richness using the same landscape variables. In doing so, one can follow how changes in vireo abundance compares to changes in avian community richness throughout the forest. As well, patterns of changes in bird species number can be identified in relation to changes imposed by forestry.  Methods  Study Location This work was carried out in the region described in Chapter One. Data collection took place in all four biogeoclimatic zones of the Okanagan Forest: Ponderosa Pine (PP), Interior Douglas Fir (IDF), Montane Spruce (MS), and Engelmann Spruce/ Subalpine Fir (ESSF). However, the sampling space extended further East, North, South and West than the study plots employed in Chapter One.  44  Data Collection Songbird censuses for this study were carried out in 1997, at stations distributed along non-permanent line transects (Figure 13). Transects originated at random points along established logging roads, and extended in randomly selected 8-point compass directions. Trained field workers conducted 50 metre fixed-radius point counts at 300 metre intervals along a given transect, following the methodology of Hutto (Hutto et al. 1986). Censuses began between 6:30 and 7:30 A M , and continued no later than 12:00 P M (11:00 A M on hot days). Data collection was halted when rain or strong wind interfered with census efforts. All point counts were eight minutes in duration, and were preceded by a one minute silent interval upon arrival at the station. Starting points were geo-referenced with hand-held global positioning system (GPS) units, and locations of subsequent points were estimated by adding 300 metre intervals in the appropriate direction. In addition to the line transects, counts were conducted along four roadside census routes. Beginning at randomly selected starting points, point counts were conducted every 300 metres along active logging roads. Since travel between points was by vehicle, rather than on foot, more roadside counts were completed per census than for the corresponding line transects. Inter-point distances were estimated using the automobile odometer, and all points were georeferenced with a GPS unit. Thirteen points comprised an independent data set collected to quantify observer bias. These censuses were taken in the IDF biogeoclimatic zone, and were done simultaneously by all  45 four observers employed in the study. The average numbers of birds per point detected by each observer were compared using a contingency table (PROC FREQ in SAS/STAT).  Data Analysis Two analyses of the effect of forest structure on Warbling Vireo abundance and songbird diversity was carried out using statistical models. Input variables were selected from a range of landscape metrics based on their presumed relevance to forest structure and their potential influence on songbird abundance (for example, see Freemark et al. 1995). These were extracted from M o E L P forest cover digital maps (see Methods, Chapter One) using a Geographic Information System, ArcView (ESRI, Inc. 1997). Response variables were derived from the census data using the GIS methods described in the Chapter One. Eighty-eight point counts were omitted from the data set before analysis. Seven fell outside of the coverage of the digital forest cover maps, while 81 points were collected by a disqualified observer, whose counts were shown to statistically differ from those of other observers (from analysis of the simultaneous censuses described above). In total, 182 counts were included in the data set. In all analyses, census points are considered as independent observations. The first analysis used a categorical data model to test for an association between local forest stand characteristics and the presence or absence of Warbling Vireos. Five input variables were selected to derive a linear model using PROC C A T M O D in SAS/STAT (SAS Institute, Inc. 1996). In a geographic information system (GIS), the geo-referenced counting stations were mapped onto the forest cover digital map. The attributes (crown closure class, stand height class, biogeoclirnatic zone, recent clearcut (< 15 years), and estimated stand age) corresponding to the  46 forest cover polygon that contained each counting station were extracted and associated with each point. These variables are summarised and described in Table 9. Stand height and stand age were reduced to ordinal class variables for this analysis. Likewise, Warbling Vireo abundance was reduced to a binary categorical variable representing presence or absence at each census station. The same variables were tested for their influence on the variation in the number of species detected (species richness). For this analysis, a general linear model was constructed (PROC G L M in SAS/STAT) rather than a categorical data model because the response variable is quantitative. A power analysis of the general linear model indicated that a standardised effect size of 20% would be detected with a probability of close to 1.00 using a = 0.05, therefore this value was used for these tests (Figure 14). The second analysis of stand characteristics used similar variables, but at a larger spatial scale. In this instance, quantitative differences in the composition of the surrounding landscape were tested as correlates of Warbling Vireo abundance and songbird diversity. From the full data set, 28 census points were selected by restricted randomization, to ensure equal representation of points with and without vireo detections. In the GIS, buffers of 500-metre radius were placed around each selected point. These buffers were used to query the forest cover database for three landscape characteristics: The percentage of buffer area characterized by low crown closure (< 25 percent), the area containing trees of age 50 or greater (i.e. area of mature growth), and the area recently disturbed (within 15 years) by harvesting. All values were standardized by dividing by the total buffer area. Additionally, a Simpson's index of habitat diversity was entered as a fourth landscape variable, calculated from the summed proportion of all habitat types within each buffer  47 (Krebs 1989). A polygon was identified as a distinct habitat type if it possessed a unique stand identification number. Stands are identified as a contiguous cluster of trees of approximately the same species and structural composition. The effect of these landscape variables on the presence of Warbling Vireos was tested using a multiple logistic regression (PROC LOGISTIC in SAS/STAT), while their influence on songbird species diversity was evaluated in a multiple linear regression (PROC REG). Forward stepwise model selection was used in both analyses, with variable entry and exit criteria of 0.20. Power analyses for these models show low power at a = 0.05 (about 25 and 50 percent, respectively). In the interest of balancing the probabilities of Type I and II error, all statistical tests used a = 0.10. This boosts power estimates by approximately 15 percent (Figures 15 and 16).  Results  Twenty-eight species were detected in the Okanagan Forest along line transects and roadside counts. The 15 most abundant species (Figure 17) are consistent with the census data presented in Chapter One, except that the modal species on the transect counts is the Yellowrumped Warbler Dendroica coronata, rather than the Dark-eyed Junco. These censuses also show the Warbling Vireo to be moderately abundant. In fact, 13 of the 15 most abundant species in both studies are identical, highlighting a strong overall relationship between counts along random transects and non-randomly selected study plots (Figure 18).  48  Forest Stand Models The derived categorical data model of Warbling Vireo habitat use (Table 10) fits the data well, since the likelihood ratio goodness offittest is non-significant (X = 4.08, 3 d.f., p = 0.25). The final model included only two of the five candidate variables, biogeoclimatic zone and recent forestry disturbance. The inclusion of any other variables resulted in a poor model fit. While the effect of logging is not a significant variable in the model (% = 0.86, 1 d.f.,p = 0.35), 2  biogeoclimatic zone significantly influences the probability of Warbling Vireo presence or absence, albeit narrowly (% = 7.86, 3 d.f., p = 0.048). Parameter estimates for each zone indicate that the highest probability of Warbling Vireo detection is associated with the PP forest type, and the lowest with M S (Table 11). Between the remaining two zones, IDF forest is more positively correlated with vireo detection than ESSF. In addition, the model suggests a higher probability of detection on sites recently disturbed by clearcutting. Two forest stand variables, biogeoclimatic zone and height class, were included in the best-fit model describing the variation of species number among points (Table 12). Disturbance, age class and crown closure were removed from the original set to yield this model, which was a significant but very weak predictor of species number (R = 0.0645, F =2.43, p = 0.037). Of the 2  modelled variables, only stand height class statistically influences variation in species number (F = 3.41, 2 d.f.,p = 0.035); higher species counts are associated with lower stand height. In addition, there was a trend for higher species detections on IDF stations, and fewest in ESSF habitat.  49  Landscape Models The stepwise procedure included only one landscape variable, low crown coverage (open) habitat, in the final logistic regression model of Warbling Vireo occurrence. This model is a significant predictor of vireo presence or absence (X  =  4.320, 1 d.f, p = 0.038). Using the  parameter estimates, the logit of the probability of detection is estimated as:  logit(p) = 1.4925 + 3.727 l(low crown area)  The parameter estimate for the crown closure variable (p = 3.7271) indicates that a greater proportion of very open habitat (anthropogenic or natural) is associated with a higher probability of Warbling Vireo detection. The residual chi-square is not significant (X  =  3.5422, 3  d.f.,/? = 0.32). The multiple regression model of species richness was derived with two variables, percentage of mature forest and habitat diversity. At a = 0.10, this regression is significant (R = 2  0.2552, F = 4.28, p = 0.03). The model predicts species number per census point by the following equation:  species = 6.455 - 5.706(mature) + 9.666(simpson)  50 The parameter estimate for the proportion mature forest (P, = -5.706) associates more species with less area of old growth or mature second growth; the effect of this variable is highly significant (p = 0.007). Species richness is positively correlated with habitat diversity (P = 2  9.666). This variable is narrowly significant (p = 0.06). The fit of the model is supported by a residual plot showing no obvious trends (Figure 19).  Discussion  The relative numbers of species detected on line and road transects varied little from detections on the study plots used in Chapter One. This suggests that the study plots, though selected non-randomly, are not a significantly biased sample of the forest as a whole. The only notable difference is that thrush species (Swainson's Thrush Catharus ustulatus, Hermit Thrush Catharus guttatus, American Robin Turdus migratorius and Townsend's Solitaire Myadestes townsendi) were more common on transects than on plots. There is no obvious habitat-related explanation for this pattern, since each species uses very different breeding habitats. For example, American Robins are habitat generalists, while Townsend's Solitaires prefer steep rocky slopes at moderately high altitudes. Interestingly, the Okanagan Forest avifauna shows a striking similarity to that of the boreal forest more than 1500 km northeast of the study site. Compared to censuses of the forest near Kluane Lake (Folkard and Smith 1995), six of the top twelve species in this study are identical, while a seventh, the dominant pariform Mountain Chickadee Parus sclateri, is replaced  51 in the Yukon by the Boreal Chickadee Parus hudsonicus. This pattern attests to the dramatic ecological shift that takes places with an increase in altitude in the Okanagan: While the valley, dominated by Ponderosa pine and sagebrush grassland, carries a similar suite of species to that of the Great Basin, moving just 1400-1600 m upland reveals a boreal-like avifauna. Apparently, the effect of altitude in the Okanagan Forest rnirnics that of latitude in the Northern boreal forest. Both regions share a generally dry climate, and similar forest structure (primarily coniferous vegetative cover, interspersed with patches of Trembling aspen Populus tremuloides).  Stand-level Models The results from the categorical data model of Warbling Vireo occurrence weaken the hypothesis that the Okanagan Forest is a demographic source for vireo populations in the valley. This is because the highest probabilities for Warbling Vireo detection occurred in the Ponderosa pine and Interior Douglas-fir biogeoclimatic zones (i.e. the two low altitude zones of the region). If this difference reflects the relative size of the breeding populations in these areas, there may not be enough vireos in high elevation habitats to act as a demographic source for the threatened valley population, despite lower parasitism upland (see Chapter One). Alternately, the source may be located in an elevational band of Interior Douglas-fir forest (approximately 900-1300 m), where the most Warbling Vireos were detected. The trend for more Warbling Vireo detections in cut stands agrees with predictions from a meta-analysis (Hejl et al. 1995). If, as in past studies, vireos in the Okanagan similarly avoid very new clearcuts, a non-linear relationship between time-since-logging and vireo detection  52 would exist. This logging disturbance variable may be confounded by silvicultural treatments applied after harvest. Specifically, some plots are mechanically or chemically stripped of deciduous understory in preparation for replanting. Doubtless, this would interfere with shrub and tree nesting songbirds on such patches. Species richness was most strongly correlated with low height class, suggesting that songbird species number peaks at early successional stages. This early peak is characteristic of several forest types, including aspen and oak (Thompson et al. 1995). However, Raphael (1991) showed that species richness did not differ among serai stages of Douglas-fir forests in Northern California. It is difficult to associate avian diversity and habitat use to stand age because birds are likely responding to forest structure, rather than age per se. Thus, in the general linear model presented here, height class is significant, while age is not. Modelling the effect of understory, or other structural characteristics, was not possible, because such data are not collected by the B.C. Ministry of Forests or lease-holding private companies. The digital database is used primarily for economic purposes, therefore many potentially useful ecological variables are not included. The general linear model also suggests a relationship of species number with biogeoclimatic zone, though it is weak. According to the least squares means estimates for this categorical variable, species richness peaks in the Interior Douglas-fir zone. This is a logical result, as the IDF is an interface between many of the valley species (e.g. Western Tanager Piranga ludoviciana, Swainson's Thrush, American Goldfinch Carduelis tristus) and several montane species (e.g. Townsend's Solitaire, Hermit Thrush, Mountain Bluebird Sialia currucoides).  53 The most notable outcome of these models is the ineffectiveness of most GIS-derived stand variables in predicting either Warbling Vireo occurrence or species number. The proportion of variance in the data set explained by the omitted variables was extremely low, with F-values sometimes approaching zero. Additionally, though the G L M is a statistically significant model, the tiny R-square value implies that the variables are unimportant contributors to the variance in species richness. At least two confounding factors may be responsible for this ineffectiveness. First, the accuracy of the forest cover data may be inadequate. Though extensively mapped, many stands in the region were surveyed long ago, and their characteristics projected forward using growth models (Richard Woods, B.C. Ministry of Forests, personal communication). These simple models likely do not account for complex forest dynamics. The second factor concerns the grain at which individual birds perceive the landscape. Grain is defined as the smallest scale at which individuals resolve a heterogeneous environment (Kotliar and Weins 1990). If songbirds respond to a scale of heterogeneity smaller than that resolved by forest cover polygons, the models will be ineffective. For example, although a large stand may be defined as mature Lodgepole pine second growth, it may contain small patches of aspen, or marshes. These may not be large enough to warrant a distinct polygon on a map, but they may be large enough to support songbirds not otherwise attracted to mature Lodgepole pine.  Landscape Models The landscape-level models reinforce the predictions of the forest stand models. The logistic regression describing the probability of Warbling Vireo detection as a function of variables  54 measured on a landscape scale predicts more detections in habitat close to large patches of very open forest or clearcuts. Thus, the trend of vireo use of disturbed, regenerating habitat discovered in the categorical data model is strengthened by analysis at a larger spatial scale. A significant result at this scale may be attributable to the use of a 500 m buffer to measure habitat, which makes the analysis robust to geo-referencing error. A small GPS error in any direction might place a census point in an incorrect forest cover polygon, yielding spurious habitat information in the stand-level model. This type of error would be absorbed in the landscape analysis. The model also confirms that Warbling Vireos use a variety of habitats. The correlation with early serai stage habitats was stronger on the landscape scale than for the forest stand. This result may indicate that, though open habitat is favoured, vireos can be found in other nearby forest stand types. A n individual-based behavioural study of Warbling Vireo habitat use is required to verify this hypothesis. This result concurs with the observation by Rosenberg and Raphael (1986) in California that Warbling Vireos were most abundant in Douglas-fir stands adjacent to hardwoods. The strongest correlation between landscape composition and species richness is a strong negative association between species number and the area of mature forest within 500 m. This is a departure from the prediction from surveys of avian responses to stand succession dynamics (Thompson et al. 1995). Diversity and richness are usually high in mature forest stands. Even those forest types which show a species richness peak early in succession experience a second peak at the mature forest stage. The particular characteristics of the montane Okanagan Forest  55 may be responsible for this anomaly. Mature second growth and old growth in the region is dominated by dense Lodgepole pine. Shade tolerant Engelmann spruce and Subalpine fir eventually top the Lodgepole pine canopy, but this happens over a very long cycle. In the Okanagan Forest, Lodgepole pine is the favoured commercial species, and it is planted and maintained as a monoculture. Replanted stands are thinned to facilitate stand growth, and are harvested at a younger age. Mature "dog's hair" forest is too closed, with too little structural diversity or understory to support many forest songbird species. This is in contrast with the spruce-fir stands which are more open, and more structurally complex. In fact, the pine stands more closely resemble the structure of immature pole-sized stands of other forest types. This immature stage is characteristically a low point in vertebrate species diversity (Thompson et al. 1995). A pattern was detected of higher species richness at census points with more stand types represented within the buffer area . A more diverse landscape composition should contain more species overall, since birds with differing habitat requirements can be accommodated; this includes not only single-habitat specialists, but also those species that require multiple habitat types. Conversely, a landscape composed of diverse habitats will likely exclude species which require large home ranges with a minimum core area of a single forest type. The Northern Spotted Owl is one such example (Thomas et al. 1990). However, the Okanagan Forest has been subject to a regular disturbance regime, both natural (e.g. fire) and anthropogenic, for a long time, so species with large, contiguous habitat requirements are not regionally common. In any case, the dependence of avian species on a diversity of habitat types is still unclear (Freemark et al. 1995).  56 Habitat diversity may interact with other measures of landscape structure not incorporated into this analysis. Specifically, the configuration of habitat types within a given area may contribute to patterns of habitat use by birds (Freemark et al. 1995). For example, the degree of connectivity between patches of similar composition may affect the occupancy and stability of those patches, in a metapopulation-like fashion (Freemark and Collins 1992). In addition, the juxtaposition of particular habitat types with one another will produce edges with varying characteristics (Hansen and diCastri 1992). As with true forest fragments, there can be edge effects caused by the adjacency of differently-aged patches. These effects include increased nest predation and parasitism by Brown-headed Cowbirds.  Analyses like those conducted here are useful for relating forest structure to the abundance and distribution of avian species and the structure of the avian community. The use of an ubiquitous type of digital forest cover map and a commercial GIS with one season's census data was able to resolve some interesting and plausible patterns. However, there are important limitations in both the independent and dependent variables used here. The digital forest inventory databases were compiled primarily for economic purposes. This may have resulted in very different maps than if ecological analysis was the goal. One must remember that any map is simply an interpretation of a real geographic place, and interpretations vary according to what the map is intended to show (Monmonier 1991). This study is also limited by the field data collected. First, indices of abundance and distribution, while convenient for building a large data set, are not reliable for inferring habitat  57 suitability or quality (Garshelis 1997, Freemark et al. 1995). Better measures of habitat quality are estimates of reproductive success or survival of organisms using the habitat. There may not be a close relationship between patch occupancy and patch quality (Garshelis 1997); population sinks may contain an appreciable number of birds. These would be misinterpreted by simple measures of abundance or presence-absence. For example, edge habitat is a preferred nesting site for many neotropical migrants, yet individuals using it may suffer high predation and parasitism rates (Gates and Gysel 1978). Second, the data presented here were collected over only one field season. It is difficult to show the effects of stand dynamics on avian species and communities with only one year of data. Any short-term study is susceptible to year effects, particularly in Western forests, where bird abundance can fluctuate significantly from year to year (Hejl et al. 1995). However, as long-term studies of breeding success and survival conducted at the scale required for a landscape analysis are not trivial undertakings, the use of preliminary surveys and analyses as presented here can be a good and inexpensive starting point. These may facilitate a more efficient large-scale study by uncovering potentially important patterns and trends.  Conclusions  Census data revealed a strikingly similar avifauna on random transect roots to that of the non-random study plots in Chapter One. Contrary to predictions, there appears to be fewer vireos in high-elevation upland habitat than lower down, as most detections occurred in Interior Douglas-fir or Ponderosa pine stands. This reduces the likelihood of the existence of an  58 important source population at high elevation, where cowbird parasitism is low. Statistical models of forest stand and landscape structure variables confirm a predicted Warbling Vireo preference for open, early serai stage forest stands. This is mirrored by a pattern of low species richness in areas dominated by mature growth. In addition, higher species richness is strongly associated with a high habitat diversity index. In general, landscape-level variables appear to be more important than stand-level variables in describing ecological patterns of forest birds. Forest practices in the South Okanagan appear not to impact Warbling Vireos nor the songbird community negatively. The variegated landscape pattern produced by short-rotation clearcutting increases the diversity of habitat types within a given area, and contains less mature Lodgepole pine. These characteristics are associated with higher songbird species richness and a higher probability of vireo detection.  59 Table 9 Description of input variables for local-scale analysis of Warbling Vireo detections and songbird diversity. Variable  Type  Description  Crown closure  Quantitative, discrete  Discrete classes of forest closure, from open (0) to heavy (7)  Stand height  Quantitative, discrete  Height classes, based on estimated continuous values; 0 = 0 to 9, 1 = 10 to 20, 2 = 21 and higher  Biogeoclimatic zone  Categorical  ESSF, M S , IDF, PP  Disturbance  Boolean  Recent anthropogenic disturbance (< 10 years)  Stand age  Quantitative, discrete  Continuous estimate of stand age, however, values tend to be artificially clumped around factors of 10  Table 10 Maximum likelihood analysis of variance table for the categorical data model (PROC C A T M O D ) testing the effect of two forest stand variables on Warbling Vireo occurrence. DF  Chi-squared  Probability  Intercept  1  13.45  0.0002  B G C zone  3  7.86  0.0489  Logged  1  0.85  0.3524  Likelihood Ratio  3  4.08  0.2529  Source  Table 11 Parameter estimates for biogeoclimatic zone categories in Warbling Vireo categorical data model. More positive values indicate association with a higher probability of vireo presence. B G C Zone  Parameter Estimate  ESSF  -0.5055  IDF  -0.024  MS  -0.6683  PP  1.1978  Table 12 Analysis of variance table resulting from general linear model for mean number of species detected per station. F-values are based on Type III sums of squares. Source  DF  F Value  Pr > F  Model  5  2.43  0.0371  BGC  3  2.05  0.1074  Height class  2  3.41  0.0354  61  100  150  200  Stand A g e (years)  Figure 11 Relationship of stand age to stand height in the Okanagan Forest. This relationship pools stands with respect to species.  _30 o 25  (10)  -£ 20  (4)  CD  E  J15  r  tr>  § 10 Q  I  5  (6)  (6)  (6) •  Clearcuts  Mature aspen Sapling aspen  Mixed conifer-aspen O l d aspen  Habitat Type Figure 12 Range of Warbling Vireo densities by habitat type for the Bulkley Valley, British Columbia. Estimates are derived from point count estimates; sample size is shown in parentheses. Taken from Campbell et al. (1997).  300  62  Transect Routes Roadside Transect A y Roads Biogeoclimatic Zone | | MS | | IDF | |ESSF | | PP | | Bare Ground | 1 ICH N  10  10  20  Kilometers S  Figure 13 Map showing the placement of transects across the study region.  0  — i  20  1  40  1  1  1  60 80 100 S a m p l e Size (N)  1—'  200  Figure 14 Power analysis for general linear model of species richness. This illustrates the power of detecting an effect size of 20 percent. 1  1  I  (I — 1  1  20  40  60 80 100 S a m p l e Size (N)  200  Figure 15 Results of power analysis of logistic regression for Warbling Vireo occurrence. This models the probability of detecting a probability change of 0.60 from 0.40 when the response variable is increased by one standard deviation.  0  1  20  40  60 80 100 S a m p l e Size (N)  200  Figure 16 Power analysis results for species richness multiple regression of landscape variables. This illustrates the probability of detecting a 20 percent effect size.  0.6  Figure 17 Relative occurrence of the 15 most abundant species detected on all routes. Dark bars indicate the proportion of stations recording at least one individual, while the light bars represent the mean number of detections per station (n=193V 25  -i  0 -I 0  ,  1  1  2  1  1  4  1  1  '  !  1  !  1  1  6 8 10 12 Transect Rank Abundance  i  1  14  i  1  16  Figure 18 A comparison of the rank abundance of the 15 commonest species on transect censuses, relative to plot censuses in the same region. The straight line represents a perfect correlation.  66  RESD IUAL  fffjff-ffff-ffff-ffff  ffff-ffff-ffff-ffff'ffff-ffff-ffff-ffff-ffff-ffff-ffff-ffffff-t  -  ffff-ffff~ffff-ffff-ffff-ffff-ffff-ffffffff~ffff-ffff-ffffff<E  &ffffff-ffffffff~ffff~ 3.00  3.25  3.50  3.75  4.00  4.25  4.50  4.75  5.00  5.25  Predicted Vau le of SPECE IS  5.50  5.75  6.00  6.25  6.50  Figure 19 Residual plot of species richness regression model. Absence of a trend implies a good fit of data to the model.  67  L I T E R A T U R E CITED  ArcView GIS, Version 3.0a. 1997. (Redlands, C A : Environmental Systems Research Institute, Inc.). The SAS System for Windows, Version 6.12. 1996. (Cary, NC: SAS Institute, Inc.). Brittingham, M . C . and S.A. Temple. 1983. Have cowbirds caused forest songbirds to decline? Bioscience 33:31-35. Bunnell, F.L. and L . L . Kremsater. 1990. Sustaining wildlife in managed forests. Northwest Environment Journal 6:243-269. Campbell, R.W., N . K . Dawe, I. McTaggart-Cowan, J. Cooper, G.W. Kaiser, M . C . E . McNall and G.E.J. Smith. 1997. Birds of British Columbia vol 3 Passerines: Flycatchers through Vireos. Vancouver, B C : U B C Press. Campbell, V.E.A. 1995. Patterns of nest predation and nest predator abundance in a fragmented Engelmann Spruce/Subalpine Fir Forest. M.Sc. Thesis (University of British Columbia). Cannings, R.A., R.J. Cannings and S.G. Cannings. 1987. Birds of the Okanagan Valley, British Columbia. Victoria, B.C.: The Royal British Columbia Museum. Cannings, R.J. and S.G. Cannings. 1996. British Columbia: A Natural History. Vancouver, B C : Greystone Books. Coker, D.R. and D.E. Capen. 1995. Landscape-level habitat use by Brown-headed Cowbirds in Vermont. Journal of Wildlife Management 59:631. Darley, J.A. 1983. Territorial behaviour of the female Brown-headed Cowbird (Molothrus Ater). Canadian Journal of Zoology 61:65-69. Dias, P. 1996. Sources and sinks in population biology. Trends in Ecology and Evolution ll(8):326-330. Donovan, T.M., F.R. Thompson III and J. Faaborg. 1999. Ecological trade-offs and the influence of scale on Brown-headed Cowbird distribution. In Ecology and Management of Cowbirds. (J.N.M. Smith, T. Cook, S.K. Robinson, S.I. Rothstein, and S.G. Sealy, Editors). Austin, T X : University of Texas Press. In press.  68 Dunning, J.B., B J . Danielson and H.R. Pulliam. 1992. Ecological processes that affect populations in complex landscapes. Oikos 65:169-175. Ehrlich, P.R., D.S. Dobkin and D. Wheye. 1988. The Birder's Handbook: A Field Guide to the Natural History of North American Birds. New York, N Y : Simon and Schuster Inc. Engeman, R . M . , R.T. Sugihara, L.F. Pank and W.E. Dusenberry. 1994. A comparison of plotless density estimators using Monte Carlo simulation. Ecology 75(6):1769-1779. Faaborg, J., M . C . Brittingham, T . M . Donovan and J.G. Blake. 1995. Habitat fragmentation in the temperate zone. In Ecology and Management of Neotropical Migratory Birds. (T.E. Martin and D . M . Finch, Editors). New York, N Y : Oxford University Press, pp 357-380. Folkard, N.F.G. and J.N.M. Smith. 1995. Evidence for bottom-up effects in the boreal forest: Do passerine birds respond to large-scale experimental fertilization? Canadian Journal of Zoology 73:2231-2237. Freemark, K . E . and B. Collins. 1992. Landscape ecology of birds breeding in temperate forest fragments. In Ecology and Conservation of Neotropical Migrant Landbirds. (J.M. Hagen III and D.W. Johnston, Editors). Washington, DC: Smithsonian Institution Press, pp 443454. Freemark, K . E . , J.B. Dunning, S.J. Hejl and J.R. Probst. 1995. A landscape ecology perspective for research, conservation, and management. In Ecology and Management of Neotropical Migratory Birds. (T.E. Martin and D . M . Finch, Editors). New York, N Y : Oxford University Press, pp 381-427. Garshelis, D.L. 1997. Delusions in habitat evaluation: Measuring use, selection, and importance. International School of Ethology, 12th Workshop. Research Techniques in Ethology and Animal Ecology: Uses and Misuses (Erice, Italy, 1996) D . Mainardi, Editor. In press. Gates, J.E. and L.W. Gysel. 1978. Avian nest dispersion and fledging success in field-forest ecotones. isco/ogy 59:871-883. Goguen, C.B. and N.E. Mathews. Cowbird parasitism and behavior in a grazed and ungrazed landscape in New Mexico. Research and Management of the Brown-Headed Cowbird in Western and Eastern Landscapes (Sacramento, CA, 1997). Grzybowski, J.A., R.B. Clapp and J.T. Marshall. 1986. History and current population status of the Black-capped Vireo in Oklahoma. American Birds 40:1151-1161. Hagen III, J.M. and D.W. Johnston. (Editors). 1992. Ecology and Conservation of Neotropical Migrant Landbirds Washington, DC: Smithsonian Institution Press.  69  Hansen, A.J. and F. diCastri. (Editors). 1992. Landscape Boundaries: Consequences for Biotic Diversity and Ecological Flows (Ecological Studies vol 92 New York, N Y : SpringerVerlag). Hejl, S.J., R.L. Hutto, C R . Preston and D . M . Finch. 1995. Effects of silviculture treatments in the Rocky Mountians. In Ecology and Management of Neotropical Migratory Birds. (T.E. Martin and D . M . Finch, Editors). New York, N Y : Oxford University Press, pp 220-244. Hejl, S.J. and J.S. Young. 1999. Brown-headed Cowbird distribution in conifer forest landscapes in the Northern Rocky Mountains. Research and Management of the Brown-Headed Cowbird in Western and Eastern Landscapes (M.L. Morrison and L. Hall, Editors). In press. Hutto, R.L., S.M. Pletschet and P. Hendricks. 1986. A fixed-radius point count method for nonbreeding and breeding season use. The Auk 103:593. Kellner, C.J., J.R. Karr and L . Rowe. Avian reproductive success and nest parasitism by Brownheaded cowbirds: Comparison of two landscapes. North American Ornithological Conference (St.Louis, MO, 1998). Kotliar, N.B. and J.A. Weins. 1990. Multiple scales of patchiness and patch structure: A hierarchical framework for the study of heterogeneity. Oikos 59:253-260. Krebs, C.J. 1989. Ecological Methodology New York, N Y : HarperCollins Publishers. Kurz, W.A., C.J. Fonnesbeck, M . Promislow and D.C.E. Robinson. 1998. The Status of Inventory Information in the Omineca-Peace Region: Towards a Strategic Plan for Improvements. Omineca-Peace Inventory Consortium (Vancouver, B C : ESSA Technologies, Ltd.). Mayfield, H.F. 1965. The Brown-headed Cowbird with new and old hosts. Living Bird 4: 13-28. Merriam, G. 1988. Landscape dynamics in farmland. Trends in Ecology and Evolution 3:16-20. Monmonier, M.S. 1991. How to Lie With Maps Chicago, IL: University of Chicago Press. Naef, V . L . Brown-headed Cowbird distribution patterns in agricultural and contiguouslyforested landscapes of the Skagit Valley, Washington State. Research and Management of the Brown-Headed Cowbird in Western and Eastern Landscapes (Sacramento, CA, 1997).  70 Opdam, P. 1991. Metapopulation theory and habitat fragmentation: a review of holarctic breeding bird studies. Landscape Ecology 5:93-106. Raphael, M . G . 1991 Vertebrate Species Richness Within and Among Serai Stages of DouglasFir Forests in Northwestern California Wildlife and Vegetation of Unmanaged DouglasFir Forests General Technical Report Technical coordinators Ruggiero L F, Aubry K B , Carey A B and Huff M H (Portland, OR: Pacific Northwest Research Station). Robinson, S.K., J.P. Hoover, and J.L. Herkert. 1999. Effects of tract size, habitat, cowbird density, and nesting stratum on cowbird parasitism levels. In Ecology and Management of Cowbirds. (J.N.M. Smith, T. Cook, S.K. Robinson, S.I. Rothstein, and S.G. Sealy, Editors). Austin, T X : University of Texas Press. In press. Robinson, S.K., S.I. Rothstein, M.C. Brittingham, L.J. Petit and J.A. Grzybowski. 1995a. Ecology and behavior of cowbirds and their impact on host populations. In Ecology and Management of Neotropical Migratory Birds. (T.E. Martin and D . M . Finch, Editors). New York, N Y : Oxford University Press, pp 428-460. Robinson, S.K., F.R. Thompson III, T . M . Donovan, D.R. Whitehead and J. Faaborg. 1995b. Regional Forest Fragmentation and the Nesting Success of Migratory Birds. Science (267): 1987. Rosenberg, K . V . and M . G . Raphael. 1986. Effects of forest fragmentation on vertebrates in Douglas-fir forests. In Wildlife 2000: Modeling Habitat Relationships of Terrestrial Vertebrates. (J. Verner, M . L . Morrison and C.J. Ralph, Editors). Madison, WI: University of Wisconsin Press, pp 263-272. Rotenberry, J.T., R.J. Cooper, J.M. Wunderle and K . G . Smith. 1995. When and how are populations limited? The roles of insect outbreaks, fire, and other natural perturbations. In Ecology and Management of Neotropical Migratory Birds. (T.E. Martin and D . M . Finch, Editors). New York, N Y : Oxford University Press, pp 55-84. Rothstein, S.I. and S.K. Robinson. 1994. Conservation and coevolutionary implications of brood parasitism by cowbirds. (The Ecology and Management of Cowbirds Austin, Texas, U S A November 1993) Trends in Ecology & Evolution 9(5): 162. Rothstein, S.I., J. Verner and E. Stevens. 1984. Radio-tracking confirms a unique diurnal pattern of spatial occurance in the parasitic Brown-headed Cowbird. Ecology 65:77-88. Steidl, R.J., J.P. Hayes and E. Schauber. 1997. Statistical power analysis in wildlife research. Journal of Wildlife Management 61(2):270. Teather, K . L . and R.J. Robertson. 1984. Female spacing patterns in Brown-headed Cowbirds. Canadian Journal of Zoology 63:218-222.  71 Thomas, J.W., E.D. Forsman, J.B. Lint, E.C. Meslow, B.R. Noon and J. Verner. 1990 A Conservation Strategy for the Northern Spotted Owl Report of the Interagency Scientific committee to Address the Conservation of the Northern Spotted Owl (Washington, D C : US Government Printing Office). Thompson III, F.R., S.K. Robinson, T . M . Donovan, J. Faaborg and D.R. Whitehead. 1999. Biogeographic, landscape, and local factors affecting cowbird abundance and host parasitism levels. In Ecology and Management of Cowbirds. (J.N.M. Smith, T. Cook, S.K. Robinson, S.I. Rothstein, and S.G. Sealy, Editors). Austin, T X : University of Texas Press. In press. Thompson III, F.R. 1994. Temporal and spatial patterns of breeding brown-headed cowbirds in the midwestern United States. Auk 111(4):979-990. Thompson III, F.R., W.D. Dijak, T.G. Kuloweic and D.A. Hamilton. 1992. Breeding bird populations in Missouri Ozark forests with and without clearcuts. Journal of Wildlife Management 56:23-30. Thompson III, F.R., J.R. Probst and M . G . Raphael. 1995. Impacts of silviculture: Overview and management reccomendations. In Ecology and Management of Neotropical Migratory Birds. (T.E. Martin and D . M . Finch, Editors). New York, N Y : Oxford University Press, pp 201-219. •i  •  Turner, M . G . and R.H. Gardner. (Editors). 1991. Quantitative Methods in Landscape Ecology . New York, N Y : Springer-Verlag. VanderHaegen, W . M . and B. Walker. 1997. Parasitism by Brown-headed Cowbirds in shrubsteppe communities of Eastern Washington. Research and Management of the Brown-Headed Cowbird in Western and Eastern Landscapes (Sacramento, CA, 1997). Ward, D. and J.N.M. Smith. 1998. Morphological differentiation of Brown-headed Cowbirds in the Okanagan Valley, British Columbia. The Condor 100(1): 1-7. Ward, D. and J.N.M. Smith. 1999. Inter-habitat differences in parasitism frequencies by Brownheaded Cowbirds in the Okanagan Valley, British Columbia. In Ecology and Management of Cowbirds. (J.N.M. Smith, T. Cook, S.K. Robinson, S.I. Rothstein, and S.G. Sealy, Editors). Austin, T X : University of Texas Press. In press. Yokel, D.A. 1989. Intrasexual aggression and the mating behaviour of Brown-headed Cowbirds: Their relation to population densities and sex ratios. The Condor 91:43-51.  72  APPENDIX  Four-letter codes, common and scientific names for all species in this study Code AMCR AMGO AMRO BHCO CAVI CEWX CHSP CLNU CONI CORA COSN COYE DEJU DUFL FOSP GCKI GRJA HETH LISP MOBL MOCH NAWA NOFL NOWA OCWA OSFL PIGR PISI RBNU RCKI RWBL SVSP SOSP  Common Name American Crow American Goldfinch American Robin Brown-headed Cowbird Cassin's Vireo Cedar Waxwing Chipping Sparrow Clark's Nutcracker Common Nighthawk Common Raven Common Snipe Common Yellowthroat Dark-eyed Junco Dusky Flycatcher Fox Sparrow Golden-crowned Kinglet Gray Jay Hermit Thrush Lincoln's Sparrow Mountain Bluebird Mountain Chickadee Nashville Warbler Northern Flicker Northern Waterthrush Orange-crowned Warbler Olive-sided Flycatcher Pine Grosbeak Pine Siskin Red-breasted Nuthatch Ruby-crowned Kinglet Red-winged Blackbird Savannah Sparrow Song Sparrow  Scientific Name Corvus brachyrynchos Carduelis tristus Tardus migratorius Molothrus ater Vireo solitarius Bombycilld cedrorum Spizella passerina Nucifraga columbiana Chordeiles minor Corvus corax Gallinago gallinago Geothlypis trichas Junco hyemalis Empidonax oberholseri Passerella iliaca Regulus satrapa Perisoreus canadensis Catharus guttatus Melospiza lincolnii Sialia currucoides Parus gambeli Vermivora ruficapilla Colaptes aurates Seiurus noveboracensis Vermivora celata Contopus borealis Pinecola enucleator Carduelis pinus Sitta canadensis Regulus calendula Ageleius phonecius Passerculus sandwichensis Melospiza melodia  Code STJA SWTH TOSO TOWA TRSW VATH VESP WAVI WBNU WCSP WEBL WIFL WIWA WWPE YRWA  Common Name Steller's Jay Swainson's Thrush Townsend's Solitaire Townsend's Warbler Tree Swallow Varied Thrush Vesper Sparrow Wabling Vireo White-breasted Nuthatch White-crowned Sparrow Western Bluebird Willow Flycatcher Wilson's Warbler Western Wood Peewee Yellow-rumped Warbler  Scientific Name Cyanocitta stelleri Catharus ustulatus Myodestes townsendii Dendroica townsendi Tachycineta bicolor Ixoreus naevius Pooecetes gramineus Vireo gilvus Sitta carolinensis Zonotrichia leucophrys Sialia mexicana Empidonax traillii Wilsonia pusilla Contopus sordidulus Dendroica coronata  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0088477/manifest

Comment

Related Items