THE SPATIAL AND TEMPORAL DISTRIBUTION OF AVIAN
STICK NESTS ACROSS A MANAGED FOREST
BY
SYDNEY LEE GOWARD

A Thesis Submitted in Partial Fulfillment of the Requirements for the
Degree of Bachelor of Natural Resource Science (Honours)
In the Department of Natural Resource Sciences at
Thompson Rivers University

THESIS EXAMINING COMMITTEE
Karl Larsen (Ph.D.), Thesis Supervisor & Professor, Dept. Natural Resource Sciences
John Karakatsoulis (Ph.D.), Senior Lecturer & Program Chair, Dept. Natural Resource
Sciences
Laura Trout (M.Sc., P.Biol.), Wildlife and Habitat Biologist, Hinton Wood Products

Defense: April 13th, 2018 in Kamloops, British Columbia, Canada

ABSTRACT
Stick nests (as created by several forest dwelling birds) are valuable habitat features.
Consequently, forest management practices in Western Canada often call for stick nests and
the surrounding habitat to be conserved where possible. I examined historical distributions
of stick nests across a working-forest landscape in west-central Alberta, to determine if
locations as amassed by forest workers (1999 -2017) appeared randomly-distributed across
the landscape or were biased towards specific habitat metrics, and if so, did these metrics
change over time? I worked with three sets of data compiled from 1999, 2003, and 20152017, respectively. Biologically relevant and important management habitat metrics were
compiled using the most relevant GIS layers corresponding to the years of stick-nest
reporting. These metrics were calculated at five spatial scales: 25 m, 50 m, 100 m, 250 m,
and 500 m. Identical data were collected from generated random (reference) sites paired
with each stick nest site.
I used conditional logistic regression to isolate the best predictors of stick nest occurrence in
each time period, at each spatial scale. Models were successfully fitted for four of five spatial
scales only in the 1999 time period. Deciduous cover was found to be a strong explanatory
variable for stick nest locations at the 25 m and 50 m scale. Increased area of land-use
(primarily oil and gas developments) and a high component of deciduous cover were found
significant at the 100 m scale. The model generated for the 500 m scale indicated an increased
likelihood of stick nests in areas with increased area of land-use, probably a result of both
nesting behaviour and observer effects. The results of this study did not support the notion
that habitat metrics associated with stick nests have remained constant (or changed)
between 1999 and 2017 in the forest management area. A consistent and more thorough
stick-nest monitoring program is likely required to fully understand the factors (natural and
anthropogenic) linked to the conservation of stick nests across a working-forest landscape.
Moreover, investing in these monitoring programs may help improve sustainable
management practices over time by enhancing understanding of the complex influences of
landscape management on raptor nesting behaviour.

i

TABLE OF CONTENTS
Abstract .......................................................................................................................................i
List of Tables ............................................................................................................................. iii
List of Figures ............................................................................................................................ iv
Acknowledgements.................................................................................................................... v
Introduction .............................................................................................................................. 7
Study Area & Stick Nest locations ........................................................................................... 13
Methods .................................................................................................................................. 15
Data Compilation ................................................................................................................. 15
Isolation of Putative Explanatory Variables & Modelling .................................................... 17
Results ..................................................................................................................................... 20
Discussion................................................................................................................................ 25
Works Cited ............................................................................................................................. 30
Appendix 1: Map of Nest locations used in data analysis ...................................................... 33

ii

LIST OF TABLES
Table 1: List of the common bird species observed using stick nests in the Hinton Forest
Management Area (Barnes 2005). ................................................................................ 9
Table 2: Justification for the selection of the five spatial scales (25m, 50m, 100m, 250m,
500m) analysed. .......................................................................................................... 16
Table 3: An inclusive list of variables initially measured within each of the five spatial scales
(25 m, 50 m, 100 m, 250 m, 500 m plot radii) around each nest site and reference
site. .............................................................................................................................. 18
Table 4: A summary of putative explanatory habitat variables entered into binary
conditional logistic regression analysis of five spatial scales, within three different
time periods of stick nest datasets. Variables that were part of significant models
appear in bold along with the corresponding P value of the model. ......................... 19

iii

LIST OF FIGURES
Figure 1: A photograph of a stick nest inventoried in the Hinton Forest Management Area in
2002 (left) shown again in 2017 during a re-visit (right). ........................................... 10
Figure 2: Author (bottom left) inventorying a stick nest (top right as indicated by arrow) in a
Trembling Aspen (Populus tremuloides) crown, occupied by one adult and two
juvenile Red-tailed Hawks (Buteo jamaicensis) in the 2017 season. Inset in the
bottom right corner is the stick nest as seen through the spotting scope................. 11
Figure 3: Location of the Hinton Forest Management Area (shaded in grey) in Alberta,
Canada. Source: Wade Gullason. ................................................................................ 14
Figure 4: Comparative box plots of deciduous cover at the 25 m (top), 50 m (middle), and
100 m (bottom) spatial scales, over all three time periods (1999, 2003, and 20152017). 1= Nest Site and 2= Reference Site. X = mean. ............................................... 22
Figure 5: Box plot of the area of land-use (primarily oil and gas developments) at the 100 m
(top) and 500 m (bottom) spatial scales, over all three time periods (1999, 2003, and
2015-2017). 1= Nest Site and 2= Reference Site. X = mean. ...................................... 23
Figure 6: A scatter plot showing the relationship between the two explanatory variables
(Area of Other Land-use (%) and Area of Deciduous Cover (%)) in the 1999 model of
the 100 m scale. .......................................................................................................... 24

iv

ACKNOWLEDGEMENTS
When I first started this project, I didn’t realize how large of a group effort writing an honours
thesis would constitute. I’d like to acknowledge the following people for their contributions
and support toward this project and express as much gratitude as possible.
To begin, I’d like to sincerely thank my honours examining committee, Karl Larsen, Laura
Trout, and John Karakatsoulis, for their commitment to this project, especially through the
endless drafts, office visits, and phone meetings.
A more specific thank you is due to Dr. Karl Larsen for mentoring and guiding me over the
past year on this project, and for creating an incredible work environment in your lab for all
us students. Thank you also for always being a voice of reason when I came into your office
confused and stressed about my ideas that even I didn’t understand. You truly went above
and beyond my expectations as a supervisor and as a research mentor.
Laura Trout, of Hinton Wood Products (HWP), deserves a mountain of gratitude for actively
participating throughout the entire project on both an employer and personal level. Your
contributions of providing feedback, even when I requested it with unreasonable time
frames, responding to (almost all) my excessive emails, seeding the original research idea,
and especially for being my first professional biologist mentor, did not go without much
gratitude.
A huge thank you is also due to Wade Gullason, GIS Analyst at HWP, for your time and effort
in the GIS data collection. It’s safe to say this project wouldn’t have been possible without
your talents and hard work, which I can assure will be re-paid in fishing trips.
Richard Briand, Woods Manager at HWP, deserves a big thank you as well, for giving HWP’s
full support in partnering in research and welcoming staff participation throughout the entire
process. Hinton Wood Products contributed all the data for this project, from the inventory
of stick nest sites to the parameters of measure. This project wouldn’t have been possible
without the hard field work of many foresters and summer students who took the time to
diligently inventory stick nests in the Hinton Forest Management Area over the past 18 years,
creating the bulk of the data set used in this project. Specifically, Joanne Mann, who did much
of the inventory work and Lindsay Barnes, who wrote a summary paper on all the work that
had been done between 1999 and 2005, deserve additional credit. In the early stages of this
project, Joanne was a great resource for me to understand how and why nests were
inventoried.
Finally, to all my friends (there truly are too many to name I but can’t go without giving a
specific acknowledgement to Ashley Quaal) and family (Mom, Dad, Dani, and Danny) that
supported me through all the highs and lows of taking on an honours thesis - you are the real
champions here. I don’t know if you’ll be more excited about the completion of this project
because you’re proud of me, or because it means I’ll finally stop complaining about it.
I couldn’t have dreamed of a better support group. Thank you for all you did in making this
capstone project of my degree possible.
v

“A conservationist is one who is humbly aware that with each stroke [of the axe] he is
writing his signature on the face of the land.”
-Aldo Leopold, A Sand County Almanac, 1949

vi

INTRODUCTION
Long-term monitoring of cost-effective bio-indicators is an important component of
sustainable resource management (Rodríguez-González et al. 2017). Investing in these
monitoring programs helps improve management practices over time and promotes a better
understanding of the ecology of specific ecosystems. Forest birds (birds that require forested
habitats for part or all of their life history) are of high biological importance to the ecosystems
they live in, as predators, prey, and structural modifiers (i.e. nest building - Wells 2011).
Raptors, defined as birds of prey, long have been of interest to scientists and forest managers.
Raptors are apex avian predators with strong and direct trophic influence on small mammals
(Marti et al. 1993; van Eeden et al. 2017) and other birds. The intrinsic value of these birds
has been revealed throughout research history. For example, declining populations of
Northern Spotted Owl (Strix occidentalis caurina) provoked changes to old-growth forestry
practices in the USA, with those changes filtering their way through Canadian practices
(SOPET 2007). Despite the acknowledged role these species play in ecosystems, several
populations and species of raptors in western North America are of special concern, facing
decreasing abundance and density due to habitat alterations, environmental pollutants, and
climate change (Smith and Francis 2012).
Few studies on raptors have attempted to understand the community and habitat
associations for boreal species at multiple spatial scales (Mahon et al. 2016). Spatial scale is
an important factor to consider when studying wildlife habitat relationships (Graf et al. 2005;
Wheatley 2010). Scale refers to the area of interest for both wildlife and researchers and is
important to consider because different habitat metrics may be important to a species at
different scales. Understanding scale is critical to interpreting ecological data and wildlifehabitat relationships (Graf et al. 2005; Boyce 2006; Wheatley 2010). With raptors, for
example, important habitat features may be different closer to the nest than those at the
stand level. Most research to date on managed, forested landscapes has focused on nesting
requirements of specific species (Olsen et al. 2006; Harrower et al. 2010); however, a coarsefilter ecosystem-based approach to management requires common denominators that allow
7

consideration for multiple species (Hunter 1999). This approach to wildlife/resource
management, advocated strongly in the 1980-90’s, has been recently re-emphasized in
Canada by the World Wildlife Fund as a means to target conservation of multiple species
(World Wildlife Fund 2017). Proceeding with effective coarse-management

science is

critically important, as multisector resource developments across North America continue to
rise at a rapid rate that correlates with the declining abundance and diversity of boreal bird
species (Mahon et al. 2016). Furthermore, even fewer studies have attempted to spatially
and temporally analyse raptor stick nest distributions in managed forests, even though it is
known that “effective conservation and management relies heavily on understanding the
relationship between wildlife behaviour and landscape conditions over space and time”
(Sorensen et al. 2015).
Collective monitoring of the ‘stick nest community’ of birds is one obvious coarse-filter
approach to avian conservation. Stick nests are large, semi-permanent structures built and
utilized by many species of birds (Figure 1; for examples see Table 1). Stand structure is of
particular importance in managed forests (Bonar et al. 2003; Hinton Wood Products 2015)
and stick nests can be used as indicators of forest structure, and are ideal for long-term
monitoring because of their conspicuous nature. Within upland forested landscapes, they are
typically found in the branches of both dead and live deciduous, and sometimes coniferous,
trees (Cornell Lab of Ornithology; National Eagle Center; Barnes 2005). The size of stick nests
varies significantly by species and age, but they average approximately one meter in
diameter by ½ meter in height (Cornell Lab of Ornithology). Stick nests provide breeding
habitat for a variety of species’ use over long periods of time (Figure 1), making them valuable
landscape features (Barnes 2005; Harrower et al. 2010). The use of one nest is not restricted
to one breeding pair of birds, with documented cases of multi-species use of individual nests.
For example, Great Grey Owls and Northern Goshawks have been observed using the same
nest during different breeding seasons, and Ospreys use the same nest repeatedly over
multiple years (Barnes 2005). Because of their persistence and ecological value, coupled with
the ability to be easily surveyed, stick nests present a potentially useful bio-indicator (Burgas
et al. 2016). The effectiveness of using raptor nests as a bio-indicator (to evaluate areas of
8

value for other, less conspicuous species) has been found to outperform other strategies,
especially when evaluating multiple specie’s nests at one time (Burgas et al. 2016).
Table 1: List of the common bird species observed using stick nests in the Hinton Forest
Management Area (Barnes 2005).
Common Name
Northern Goshawk
Red-tailed Hawk
American Crow
Common Raven
Great Grey Owl
Osprey

Latin Name
Accipiter gentilis
Buteo jamaicensis
Corvus brachyrhynchos
Corvus corax
Strix nebulosa
Pandion haliaetus

It can be challenging to understand selection of a community of stick nesters instead of an
individual species, but the logistics and cost of assigning each stick nest to a specific species
makes collective assessments of all nests more common. Hinton Wood Products (HWP), a
Division of West Fraser Mills Ltd., has been consistently cataloguing all stick nests reported
by forestry workers in the Hinton Forest Management Area (FMA) since 1999 (Figure 2). The
distribution of stick nests in relation to the forested landscape is an important element of the
coarse-filter approach. In addition to this, a total of 471 occupancy surveys were conducted
on the inventory nests in the FMA between 2000 and 2005, resulting in the identification of
six common species of stick nesting birds: Common Ravens, Great Grey Owls, Great Horned
Owls, Northern Goshawks, Ospreys, and Red-tailed Hawks (Table 1). Forestry practice
guidelines in Western Canada stipulate that stick nests be preserved during forest operations.
In Alberta, the government-approved Operating Ground Rules (OGRs) require a 100m noharvest buffer placed around any identified stick nest (Hinton Wood Products and
Government of Alberta 2011).

9

Figure 1: A photograph of a stick nest inventoried in the Hinton Forest Management Area in 2002 (left) shown again in 2017 during a
re-visit (right).

10

Figure 2: Author (bottom left) inventorying a stick nest (top right as indicated by arrow) in a Trembling Aspen (Populus
tremuloides) crown, occupied by one adult and two juvenile Red-tailed Hawks (Buteo jamaicensis) in the 2017 season. Inset
in the bottom right corner is the stick nest as seen through the spotting scope.

11

The current stick nest inventory at HWP has been largely the result of incidental sightings
reported by forestry workers. Potential biases stemming from this form of data collection
include the non-random distribution of workers throughout the forest, (i.e. linked to where
the company is planning harvest or harvesting), the locations of travel routes, the frequency
of worker travel, and the visibility of nests in various forest types among seasons. Exploring
these potential biases is essential to assessing the overall value of stick nest inventories as
indicators of ecosystem and forest structure, much less the natural distribution of stick nests
on the landscape, factor(s) affecting their location and longevity, and their reliability as
predictive habitat features that forest managers can consider in long-term habitat planning.
If biases are present, forest companies may need to consider a shift from an incidental
monitoring system to a stratified and formal monitoring system. Additionally, the
identification of predictor variables of nest occurrence may help forest planners anticipate
where they might find a stick nest and where they may need to spend additional time looking
for nests.
The objective of my study was to investigate the distribution of stick nests within the Hinton
Forest Management Area (FMA) in west-central Alberta, through an analysis of habitat and
landscape attributes drawn from Geographical Information System (GIS) spatial data. I
specifically examined whether the location of stick nests, as reported by forest workers
through different periods of time, were randomly located or instead associated with roads,
habitat, or other features on the landscape. To this end, the central research questions of my
study were (1) Are the habitat metrics associated with stick nests markedly different from
that generated randomly (2) do stick nest metrics remain consistent over time as the
landscape shifts under a forest harvest regime?

12

STUDY AREA & STICK NEST LOCATIONS
The Hinton FMA consists of nearly 1,000,000 hectares of provincial Crown land, under an
area-based forest management and timber harvest agreement, with Hinton Wood Products,
a Division of West Fraser Mills (HWP) as the sole forestry operator. With the town of Hinton,
Alberta, Canada (53.4037° N, 117.5718° W) at its center, the FMA is located in the Foothills
of the Canadian Rocky Mountains (Figure 3). The landscape features rolling hills with an
elevation range of 830-2340 MSL. Within this area, the primary forest type is pure conifer
[Lodgepole pine (Pinus contorta var. latifolia) and/or Spruce (Picea sp.)] followed by mixed
woods that include deciduous trees (primarily Populus tremuloides and Populus balsamifera)
(Beckingham et al. 1996; Wheatley et al. 2002). Common land base disturbances include a
historical fire regime (fires are actively suppressed by the Government of Alberta), forest
harvesting, and sub-surface resource extraction such as coal, oil and gas developments (Bott
et al. 2003). In general, HWP focuses on harvesting conifer species in proportions similar to
what would of naturally been affected by wildfire on the land base, consistent with an
ecosystem-based management strategy that aims to emulate a natural disturbance regime
(San-Miguel et al. 2017). The ecosystem-based approach to management within the Hinton
FMA has been a recent focus of staff at HWP (Hinton Wood Products and Government of
Alberta 2011). In the mid 2000’s, the FMA became subject to the Mountain Pine Beetle attack,
resulting in the Healthy Pine Strategy commencing in 2006. This shifted the focus of harvest
towards pine- dominated stands, and away from the historical mixed-forest.
The initiative for stick nest inventory and monitoring by HWP was the result of a new forest
management plan in the region, based on sustainability and quantitative analysis of nontimber values (Bott et al. 2003). Stick nests discovered by forest workers in the FMA have
been reported/documented since 1999. Using GPS locations included in each report, a survey
of the habitat features within 400 m2 of the nest tree (11.28m radius circular plot) is
conducted by a company biologist or field technician.

13

Figure 3: Location of the Hinton Forest Management Area (shaded in grey) in Alberta, Canada.
Source: Wade Gullason.
14

METHODS
Data Compilation
Three time periods for stick-nest data were assessed in this analysis: 1999, 2003, and 20152017 (data from the years 2015, 2016, and 2017 were combined due to small sample sizes).
These time periods were selected to best represent the earliest, middle, and most recent of
the larger dataset. Data entry and storage was conducted in Excel 2016 MSO (Microsoft
Corporation 2016). Nest sites were manually sorted in GIS software ([ESRI] Environmental
Systems Research Institute 2015). Each time period was analyzed separately.
I used a multi-scale approach to investigate habitat metrics associated with the locations of
nests reported in each time period (Graf et al. 2005; Boyce 2006; Wheatley 2010). In this
study, I isolated five different spatial scales around the center of each nest site: 25 m, 50 m,
100 m, 250 m, and 500 m radius. These scales were selected based on previous literature
regarding stick nests and edge effects (Table 2). GIS data coverage of the region was restricted
and did not encompass any land outside the FMA border; therefore, no scales were allowed
to overlap non-FMA land.
Reference sites were generated for each stick nest site by using a paired design and a
randomly generated compass bearing. I avoided overlap between the spatial scales of any
nest site or reference site within the individual time periods (Johnson et al. 2006) by forcing
the centre of each reference sites to be 1001 m from the stick-nest site. This was done to
maintain independent samples. If overlap between neighbouring nest site scales occurred, I
randomly removed one (or more) nests from the analysis. Sample sizes of stick nests within
my working data set were 1999 n=21, 2003 n=9, and 2015-2017 n=9. A map of stick nest
locations used in the analysis is provided in Appendix 1.

15

Table 2: Justification for the selection of the five spatial scales (25m, 50m, 100m, 250m,
500m) analysed.
Scale Size (m)
25

Selection Justification
Supporting Literature
- Represents the habitat characteristics in (Paton 1994; Bevers and
the immediate vicinity of the nest tree
Hof 1999)
- Edge effects usually show strong
impacts on nests at this distance

50

-

Furthest distance at which edge effects
have shown strongest influence on
nests

(Paton 1994; Bevers and
Hof 1999)

100

-

Current legislated fixed-width buffer
requirement for stick nests in Alberta

(Ministry of Agriculture and
Forestry; Hinton Wood
Products and Government
of Alberta 2011)

250

-

Nest stand characteristics
Intermediate between the 100m and
500m scales

(Harrower 2007)

500

-

Represents stand-level habitat
(Bull et al. 1988; Sorte et al.
characteristics
2004; Harrower 2007)
Standard post-fledging area of Northern
Goshawks
Encompasses part of broader home
ranges/nesting territory for various
species

-

16

The habitat metrics I selected for analysis were derived from the biological knowledge of
those stick-nest species within the study area (Table 3) (Cornell Lab of Ornithology; Barnes
2005; Olsen et al. 2006; Harrower 2007; Burgas et al. 2016). Ten variables of biological and
management value were selected for analysis, here after referred to by their abbreviated
name (Table 3). The quantification of these metrics was conducted in GIS ArcMap (version
10.2.2) software ([ESRI] Environmental Systems Research Institute 2015) primarily by a GIS
Analyst at HWP, using all West Fraser (HWP) databases available (Table 3). The historic GIS
data layers used in these analyses were chosen based on the closest overlap in time with the
stick nest database. In all cases except one, metrics were extracted from GIS data sets
available for the specific year of nest reporting. The one exception for time-specific data was
distance to road measurements (DIST_RD) as there was no way to separate roads by year,
however, it is likely that all roads were present at all time periods.

Isolation of Putative Explanatory Variables & Modelling
I used SPSS Statistics (version 24) to conduct my statistical analysis ([IBM] International
Buisness Machines 2016). All ten habitat variables (Table 3) initially were tested for
autocorrelation using two-tailed Pearson correlation coefficients. Significant correlations
were identified using α=0.05 and in these cases one variable was excluded from the analysis.
Determining which correlate to exclude was based on two factors: 1) consistency across
spatial scales and time periods and 2) biological knowledge and applicability to the central
research questions. Following these criteria, deciduous cover, age, and distance to road were
included for each model (with additional variables present where possible). The sole
exception to this was the 500 m scale analysis for 1999, where deciduous cover was excluded
because of correlations with other consistent variables, however, this exclusion was deemed
to not affect results.

17

Table 3: An inclusive list of variables initially measured within each of the five spatial scales
(25 m, 50 m, 100 m, 250 m, 500 m plot radii) around each nest site and reference site.
Variable of interest
Deciduous cover (m2)

Code
DECID

Coniferous cover (m2)

CONIF

Area harvested (m2)

A_CUT

Area of water bodies (m2)

A_WAT

Area of other land-use (m2)

A_LAND

Distance to nearest
waterbody (m)

DIST_WAT

Distance to road (m)

DIST_RD

Distance to nearest cutblock (m)

DIST_CUT

Dominant age of forest
cover (years)

AGE

Dominant height of forest
cover (m)

HEIGHT

Description
Polygons extracted from 2001 Aerial
Vegetation Inventory (AVI) data (most
accurate AVI data available)
Polygons extracted from 2001 Aerial
Vegetation Inventory (AVI) data (most
accurate AVI data available)
Cut-blocks, primarily clear cuts, that were
harvested at least one year prior to nest
inventory
Distance to water bodies represented as
polygons on maps, i.e. large or moderate
width streams/creeks; generally, areas <400m2
and non-permanent waterbodies not included
Primarily includes oil and gas infrastructure,
data inventory from Government of Alberta
Distance to water bodies represented as
polygons on maps, i.e. large or moderate
width streams/creeks; generally, areas <400m2
and non-permanent waterbodies not included
Nearest road; includes all temporary and
permanent roads
Only includes those harvested at least one
year prior to nest discovery, up to 10 years of
age; purpose for this was potential foraging
value in the habitat before trees were too
tall/dense following planting
Extracted from 2001 Aerial Vegetation
Inventory (AVI) data; most accurate AVI data
available)
Extracted from 2001 Aerial Vegetation
Inventory (AVI) data; most accurate AVI data
available)

18

Table 4: A summary of putative explanatory habitat variables entered into binary conditional
logistic regression analysis of five spatial scales, within three different time periods of stick
nest datasets. Variables that were part of significant models appear in bold along with the
corresponding P value of the model. Variables were deemed insignificant at P> 0.05.
TIME PERIOD OF STICK NEST DATASET
2003 n=9
2015-2017 n=9
DIST_WAT
DIST_WAT
DIST_RD
DIST_RD
AGE
AGE
DECID
DECID
A_CUT
A_LAND
DIST_CUT

SCALE
25 m

1999 n=21
DIST_WAT
DIST_RD
AGE
DECID
(P=0.008)

50 m

DIST_WAT
DIST_RD
AGE
DECID
(P=0.015)

DIST_RD
AGE
DECID
A_LAND
HEIGHT
DIST_CUT

DIST_RD
AGE
DECID
A_LAND

100 m

DIST_WAT
DIST_RD
AGE
DECID
A_LAND
(P=0.013)

DIST_WAT
DIST_RD
AGE
DECID
HEIGHT
DIST_CUT

DIST_RD
AGE
DECID
A_LAND

250 m

DIST_WAT
DIST_RD
AGE
DECID
A_WAT

DIST_WAT
DIST_RD
AGE
DECID
HEIGHT
DIST_CUT

DIST_WAT
DIST_RD
AGE
DECID
A_LAND
HEIGHT

500 m

DIST_WAT
DIST_RD
AGE
DIST_CUT
A_WAT
A_LAND
(P=0.049)

DIST_RD
DECID
AGE
HEIGHT
DIST_CUT

DIST_WAT
DIST_RD
AGE
DECID
A_LAND
A_CUT
HEIGHT

19

After identifying the final set of putative explanatory variables, I used binary response
conditional logistic regression (Hosmer et al. 2000; Johnson et al. 2006; Harrower et al. 2010;
Thaker et al. 2011) to analyse for differences between the stick nest sites and reference sites,
within each time period. For this analysis I used the COXREG package in SPSS with a forward
conditional entry method.
Descriptive statistics, in the form of boxplots, were created in Microsoft Excel version
16.0.4639.1000 (Microsoft Corporation 2016) to draw visual comparisons of significant
predictive variables across time.

RESULTS
Models distinguishing between stick nest and reference sites (revealing significant
explanatory variables) were only successfully fitted for one of the three time periods (1999),
and then for only four of five spatial scales (25 m, 50 m, 100 m, and 500 m) (Table 4). Models
were not successfully fitted at a 250 m scale in 1999 or at any spatial scale in 2003 or 20152017 (Table 4).
Area of deciduous cover (DECID) was a significant predictor of stick-nest locations in three of
the scales (25 m, 50 m, and 100 m) tested for the 1999 data set (Table 4). In general, the
boxplot diagrams indicated relatively consistent levels of DECID across time periods. At 25 m,
DECID at nest sites was always higher than that of the reference sites across all three time
periods (Figure 4). In the 50 m scale, the DECID at nest sites was higher than that of the
reference sites in both 1999 and 2003, but the reverse was detected in 2015-2017 when
reference sites showed slightly higher cover than nest sites (Figure 4). At the 100 m scale, the
amount of DECID at nest sights was higher than that of the reference sites in the earlier two
years (1999 and 2003), but in 2015-2017, the reference sites contained a slightly higher
deciduous cover than nest sites (Figure 4). Additionally, at the 100 m scale, a combination of
DECID and A_LAND was found to significantly explain stick nest presence.
The area of other land-use (A_LAND) was a significant predictor of stick-nest locations in two
of the scales (100 m and 500 m) tested for the 1999 data set (Table 4). Nest sites showed
20

more A_LAND than reference sites during the 1999:100 m analysis, but there was some
variation across all time periods at this same scale (Figure 5). When the two dependent
variables were plotted against each other, there did not appear to be a strong interaction
between these two habitat metrics (Figure 6). The 999:500 m model also showed A_LAND to
be significant and a comparison of the boxplots across time (Figure 5) indicated more A_LAND
around nest sites than reference sites in 1999, and almost no difference in 2003. However, in
2015-2017, A_LAND at nest sites was lower than at reference sites.

21

25 m Scale

50 m Scale

100 m Scale

Figure 4: Comparative box plots of deciduous cover at the 25 m (top), 50 m (middle), and 100
m (bottom) spatial scales, over all three time periods (1999, 2003, and 2015-2017). 1= Nest
Site and 2= Reference Site. X = mean.
22

100 m Scale

500 m Scale

Figure 5: Box plot of the area of land-use (primarily oil and gas developments) at the 100 m
(top) and 500 m (bottom) spatial scales, over all three time periods (1999, 2003, and 20152017). 1= Nest Site and 2= Reference Site. X = mean.

23

Nest Site (1)

Reference Site (2)

100
90
80

Area of Deciduous Cover (%)

70
60
50
40
30
20
10

0
-10
-10

0

10
20
30
Area of Other Land-use (%)

40

50

Figure 6: A scatter plot showing the relationship between the two explanatory variables (Area
of Other Land-use (%) and Area of Deciduous Cover (%)) in the 1999 model of the 100 m scale.

24

DISCUSSION
One of my primary goals was to identify habitat metrics associated with stick nests, and how
consistent of a role they played over time; however, I was only able to find explanatory
variables in the earliest data set (1999) in four cases where any of the habitat metrics I tested
were markedly different from the reference sites. Of second importance to the timing and
scope of model success, was the appearance of deciduous cover and area of other land-use
as the variables found to be important for predicting stick nest locations.
Deciduous cover significantly explained stick nest presence at the 25 m, 50 m, and 100 m
scales, suggesting a link during the early time period of my analyses. Raptors in the Hinton
FMA have historically shown a strong consistency with nesting directly in deciduous trees
(Barnes 2005), so this pattern can be confidently interpreted as nesting behaviour rather than
observer bias. The preference for deciduous cover was especially prominent at the 25 m
scale, which represents the habitat in the immediate vicinity of the nest tree. This is
consistent with observations of stick nests in small isolated patches of deciduous cover within
an otherwise conifer forest (Barnes 2005), however the relationship seemed to disappear in
the latter two time periods.
A combination of increased area of other land use and a high component of deciduous cover
was significantly important at the 100 m scale in 1999. This was most likely a result of area of
other land use edge effects (both observed effects and increased habitat diversity) and
deciduous cover as preferred deciduous nesting substrate occurring at the stand level by
nature. However, the impact of these factors varies a lot from species to species and therefor
explanations are difficult to interpret. Area of other land use was the sole predictive variable
at the 500 m scale in 1999, indicating an increased likelihood of stick nests in areas with
increased area of other land use. This relationship may have been caused by multiple factors
acting together or in isolation: Firstly, studies on edge effects (Bevers and Hof 1999) have
suggested that increasing the amount of edge ecotone on a landscape increase habitat
diversity and forage opportunities. Adversely, nest predation and disease are reportedly
higher near edges (Paton 1994), but this would only be a consideration if we were addressing
25

the 25 or 50 m scales. Edge effects have been found to have the greatest impact within 2550 m of a nest (Paton 1994), so given the size of this 500 m scale, I am more inclined to suggest
that the edges of the land-use developments are creating a broader diversity on the
landscape. Second, is the concept of observer effects, or in other words, an increased ability
for forest workers to see and therefor report stick nests. Again, the impact of these factors
varies considerably from species to species. Furthermore, in the Hinton FMA, oil and gas
developments (represented in this study by A_LAND) picked up substantially in 1999
(anecdotal from HWP staff), so the nests could have just been documented more as
developments noticed them more.
Aside from the aforementioned cases in the earlier data set, I could not identify consistent
predictive variables to explain stick nest occurrences. In regard to the 250 m scale in 1999,
there is the possibility that 250 m was an inappropriate scale size to assess for nest site
selection, i.e. no clear selection of features is occurring at 250 m scale in this group of birds.
The other possibility is that there is a transition occurring in terms of what the birds are
looking for at this scale (Boyce 2006; Wheatley 2010).
One of the more interesting results from my study was the lack of model success in the later
two time periods of the study. On one hand, this lack of model success in 2003 and 20152017 time periods may have been due to potentially insufficient sample sizes. Total nests
observed were much higher than the sample used in analysis, but some sites were eliminated
when habitat data for the site was incomplete. Given the reported ecological relevance of
deciduous cover at smaller scales (Schaffer 1998; Barnes 2005), it is interesting that it did not
remain a consistent predictive variable in the later time periods. I suspect this is due primarily
to limited sample sizes in those analyses, as the reference sites showed a consistent amount
of deciduous cover throughout time as indicated by the boxplot diagrams. On the other hand,
the absence of predictive variables in the latter time periods could be due to the complexity
of the landscape creating a need for more specific details and more thorough investigations
to reveal the larger number of factors at work.

26

Interestingly, distance to roads was included in every model attempt, but never surfaced as
a significant explanatory variable. At face value, this suggests that nest reports were not
biased by forestry workers operating primarily near roads, but it is important to note this data
may not have been entirely accurate given the previously mentioned lack of ability to stratify
road data to time period. Alternatively, the preponderance of roads throughout the FMA may
make it difficult for stick nests to be built or persist at a location truly distant from roads, even
across all the scales I used for comparison. This is possible, as the average distance from a
road for a site was 219 m and, similarly, the reference site was 220 m away on average, yet
the furthest nest site from a road in this study was 1432 m and the furthest reference site
was 1947 m. This shows that there are still some areas in the FMA with relatively less road
disturbance, but the majority has a higher density. Due to the paired study design, the degree
of variation between my plots is likely narrow, which could be another explanation for why
roads weren’t a significant predictor, and therefor observer effects are still entirely possible
in the inventory, even though my results did not confirm such effects. To that end, there is
conflicting literature on the positive and negative impacts of roads (Lambertucci et al. 2009;
Downing et al. 2015; Wiącek et al. 2015). Finally, another explanation is lack of time sensitive
road data limiting the accuracy of the analysis.
My secondary objective in this study was to look at if predictive metrics remain consistent
over time as the landscape shifts under a forest harvest regime. The results of this study do
not indicate that the habitat metrics associated with stick nests have remained constant (or
changed) between 1999 and 2017 in the Hinton FMA. Due to a lack of successful modelling
across time periods, I cannot comment on whether a shift in the nesting behaviour of birds
has occurred. The association with deciduous cover in the earlier time periods is consistent
with our understanding of preferred stick nest microhabitat, but otherwise limited
information was obtained. There are a few possible reasons for this lack of model success.
Sample size, which was smaller in the latter two time periods, was potentially limiting, but
there are other possible explanations. As time progresses, there is increased management
and land-use occurring on the landscape, making analysis like this difficult, especially with
species that may be exhibiting shifting behaviours. The lack of occupancy data also prevents
27

comment on which species and which nests are most subject to habitat alterations. Data
required to understand these issues will require longer term commitment to a monitoring
program involving occupancy surveys.
In an effort to maintain preferred nesting habitat on the landscape, I would recommend
managing for deciduous cover, possibly through deciduous patch retention and regrowth of
harvested trees. It is important to acknowledge that this study failed to detect deciduous
cover as an important variable in 2003 and 2015-2017, which could have resulted for several
reasons: 1) the lack of occupancy study could have skewed results (most likely), 2) deciduous
cover has been altered so much it’s no longer relevant (this is unlikely due to increased focus
on pine harvest with Mountain Pine Beetle outbreak and the life history of deciduous trees
which are pervasive and fast growing), 3) flawed study design (i.e., sample sizes were
inefficient-likely), or 4) deciduous cover is not actually important to stick nesters and previous
literature is incorrect (very unlikely as most nests are observed directly in deciduous trees).
Regardless, I would recommend using the precautionary principle and manage for deciduous
cover until further research can be conducted. Additionally, long-term monitoring of known
stick nest locations and a detailed inventory system (including occupancy data and spatially
balanced on the land base) is important to an effective study design to understand nesting
response to forest/land management. By inventorying stick nests diligently (and increasing
the sample sizes of nests) more significant results revealed in future analysis.
This study did not consider occupancy because 1) current occupancy data in the Hinton FMA
are sparse and inconsistent and 2) occupancy data are expensive and difficult to obtain. Such
data may have better represented the selection use over the changing landscape, therefor
the following recommendations are being made based on the results of this study. Firstly,
more effort should be put into occupancy studies during the breeding season, to determine
activity of nests and even specific species use, in order to better understand the response of
different birds to landscape changes. Such work would become an integral part of sustainable
forest management practices, by ensuring large-scale and long-term planning for habitat
availability for the community of raptors.

Finally, although I found no evidence that the

locations of stick nests were subject to biased reporting, it nonetheless would be valuable to
28

broaden the current inventory system (incidental discovery) to include intentional nest sweep
surveys in areas where active forestry operations are less prevalent.
The forest industry has been concerned with stick nests and their interactions with the
landscape for many years, due to the both biological value of nests and the economical
impacts they can have on local harvests (Hinton Wood Products and Government of Alberta
2011). As management practices evolve, the effectiveness of such practices will rely heavily
on understanding the relationship between wildlife behaviour and landscape conditions over
space and time (Sorensen et al. 2015), and bio-indicators such as raptors and/or stick nests
will likely be valuable tools (Burgas et al. 2016). This work, as an attempt to assess a
community of stick nests as opposed to single-species dynamics, is unique among current
literature in this field, which typically focuses on single-species research. Given the more
broad, course-filter approach of general forest management practices today, there is a need
for more community-level research moving forward. The results of my study emphasize the
importance of thorough long-term monitoring and detailed inventory work for stick nests in
forestry settings. Investing in these monitoring programs may help improve sustainable
management practices over time by enhancing understanding of the complex influences of
landscape management on raptor nesting behaviour.

29

WORKS CITED
[ESRI] Environmental Systems Research Institute. 2015. ArcMap.
[IBM] International Buisness Machines. 2016. SPSS Statistics.
[SOPET] Spotted Owl Population Enhancement Team. 2007. Northern Spotted Owl
population enhancement and recovery in British Columbia: proposed five-year
action plan.
Barnes L. 2005. Stick nest inventory and occupant surveys in the Hinton Wood Products
Forest Management Area. Hinton Alberta.
Beckingham JD, Corns IGW, Archibald JH. 1996. Field Guide to Ecosites of Northern Alberta.
Edmonton, Alberta: Natural Resources Canada, Canadian Forest Service.
Bevers M, Hof J. 1999. Spatially optimizing wildlife habitat edge effects in forest
management linear and mixed-integer programs. For. Sci. 24:249–258.
Bonar RL, Lougheed H, Andison DW. 2003. Natural disturbance and old-forest management
in the Alberta Foothills. For. Chron. 79.
Bott R, Murphy P, Udell R. 2003. Learning from the forest: a fifty-year journey towards
sustainable forest management. Seidlitz L, editor. Calgary Alberta: Fifth House Ltd.
Boyce MS. 2006. Scale for resource selection functions. Divers. Distrib. 12:269–276.
Bull EL, Henjum MG, Rohweder RS. 1988. Home range and dispersal of Great Gray Owls in
Northeastern Oregon. J. Raptor Res. 22:101–10.
Burgas D, Juutinen A, Byholm P. 2016. The cost-effectiveness of using raptor nest sites to
identify areas with high species richness of other taxa. Ecol. Indic. 70:518–530.
Cornell Lab of Ornithology. All about birds, Northern Goshawk life history. [accessed 2018
Mar 11]. https://www.allaboutbirds.org/guide/Northern_Goshawk/lifehistory.
Downing RJ, Rytwinski T, Fahrig L. 2015. Positive effects of roads on small mammals: a test
of the predation release hypothesis. Ecol. Res. 30:651–662.
van Eeden R, Whitfield DP, Botha A, Amar A. 2017. Ranging behaviour and habitat
preferences of the Martial Eagle: implications for the conservation of a declining
apex predator. PLoS One 12:1–22.
Graf RF, Bollmann K, Suter W, Bugmann H. 2005. The importance of spatial scale in habitat
models: Capercaillie in the Swiss Alps. Landsc. Ecol. 20:703–717.
Harrower WL. 2007. Nesting requirements of the Northern Goshawk (Accipiter gentilis
atricapillus) in southeastern British Columbia. MSc Thesis: University of Victoria.
Harrower WL, Larsen KW, Stuart-Smith KA. 2010. Movements and resource selection of
fledgling goshawks in montane forests of southeastern British Columbia. J. Wildl.
30

Manage. 74:1768–1775.
Hinton Wood Products. 2015. Hinton Wood Products natural disturbance strategy. Hinton,
Alberta.
Hinton Wood Products, Government of Alberta. 2011. Timber harvest planning and
operating ground rules.
Hosmer DW, Lemeshow S, Sturdivant RX. 2000. Applied logistic regression. 3rd ed. John
Wiley & Sons, Incorporated.
Hunter ML. 1999. Wildlife, forests, and forestry : principles of managing forests for
biological diversity. Englewood Cliffs, New Jersey: Prentice Hall.
Johnson C, Nielsen S, Merrill E, McDonald T, Boyce M. 2006. Resource selection functions
based on use–availability data: theoretical motivation and evaluation methods. J.
Wildl. Manage. 70:347–357.
Lambertucci SA, Speziale KL, Rogers TE, Morales JM. 2009. How do roads affect the habitat
use of an assemblage of scavenging raptors? Biodivers. Conserv. 18:2063–2074.
Mahon CL, Holloway G, Sólymos P, Cumming SG, Bayne EM, Schmiegelow FKA, Song SJ.
2016. Community structure and niche characteristics of upland and lowland western
boreal birds at multiple spatial scales. For. Ecol. Manage. 361:99–116.
doi:10.1016/J.FORECO.2015.11.007.
Marti CD, Steenhof K, Kochert MN, Marks JS. 1993. Community trophic structure: the roles
of diet, body size, and activity time in vertebrate predators. Oikos 67:6.
Microsoft Corporation. 2016. Microsoft Excel.
Ministry of Agriculture and Forestry. Forest Management Planning.
National Eagle Center. Eagle nesting & young. [accessed 2018 Mar 11].
https://www.nationaleaglecenter.org/eagle-nesting-young/.
Olsen BT, Hannon SJ, Court GS. 2006. Short-term response of breeding Barred Owls to
forestry in a boreal mixedwood forest landscape. Avian Conserv. Ecol. 1:1.
Paton PWC. 1994. The effect of edge on avian nest success: how strong is the evidence?
Conserv. Biol. 8:17–26.
Rodríguez-González PM, Albuquerque A, Martínez-Almarza M, Díaz-Delgado R. 2017. Longterm monitoring for conservation management: lessons from a case study
integrating remote sensing and field approaches in floodplain forests. J. Environ.
Manage. 202:392–402.
San-Miguel I, Andison DW, Coops NC. 2017. Characterizing historical fire patterns as a guide
for harvesting planning using landscape metrics derived from long term satellite
imagery. For. Ecol. Manage. 399:155–165.
31

Schaffer WW. 1998. Northern Goshawk (Accipiter gentilis) habitat characterization in
central Alberta.
Smith AC, Francis CM. 2012. The state of Canada’s birds.
Sorensen AA, Stenhouse GB, Bourbonnais ML, Nelson TA. 2015. Effects of habitat quality
and anthropogenic disturbance on Grizzly Bear (Ursus arctos horribilis) home-range
fidelity. Can. J. Zool. 93:857–865.
Sorte FA La, Mannan RW, Reynolds RT, Grubb TG. 2004. Habitat associations of sympatric
Red-Tailed Hawks and Northern Goshawks on the Kaibab Plateau. J. Wildl. Manage.
68:307–317.
Thaker M, Vanak AT, Owen CR, Ogden MB, Niemann SM, Slotow R. 2011. Minimizing
predation risk in a landscape of multiple predators: effects on the spatial distribution
of African ungulates. Ecology 92:398–407.
Wells J V. 2011. Boreal birds of North America: a hemispheric view of their conservation
links and significance. Berkeley, California: University of California Press.
Wheatley M. 2010. Domains of scale in forest-landscape metrics: Implications for specieshabitat modeling. Acta Oecologica 36:259–267.
Wheatley M, Larsen KW, Boutin S. 2002. does density reflect habitat quality for North
American Red Squirrels during a spruce-cone failure? J. Mammal. 83:716–727.
Wiącek J, Polak M, Kucharczyk M, Bohatkiewicz J. 2015. The influence of road traffic on
birds during autumn period: Implications for planning and management of road
network. Landsc. Urban Plan. 134:76–82.
World Wildlife Fund. 2017. Living Planet Report Canada: A national look at wildlife loss.

32

APPENDIX 1: MAP OF NEST LOCATIONS USED IN DATA ANALYSIS

33