You are on page 1of 37

Monitoring and Mitigation of Moose Rail Mortality in

Northern BC: 2017 Report


Telkwa and Fraser Subdivisions

November 2017

Prepared for:
Canadian National Railway Company
13477 116th Ave,
Surrey, BC V3R 6W4

Prepared by: Reviewed by:


Garrett Kerr, BSc, RPBio Brad T. Pollard, MSFM, RPBio, RPF
McElhanney Consulting Services Ltd. McElhanney Consulting Services Ltd.
Suite 1 5008 Pohle Ave. Suite 1 5008 Pohle Ave.
Terrace BC V8G 4S8 Terrace BC V8G 4S8

McElhanney File No. 2321-30297-00


Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

Executive Summary
Since the spring of 2007, the Canadian National Railway Company (CN) has been working with
a Moose Mortality Working Group to address rail-related moose mortality in northern BC. The
group, consisting of local stakeholders, provincial biologists and academic specialists, began their
efforts along CNs Telkwa Subdivision east of Smithers, and has since expanded to investigate
the Fraser Subdivision east of Prince George. Investigations over the past ten years have
included literature review, habitat mapping, hi-rail and aerial mortality surveys, and development
of survey correction factors. In addition, exclusion fencing has been installed at three sites along
the Telkwa Subdivision and monitored for effectiveness at reducing moose-train collision levels.
The mortality monitoring program was continued along the Telkwa Subdivision in 2016-2017, with
an aerial mortality survey conducted in early April. Application of the correction factor for Telkwa
Subdivision aerial surveys resulted in an estimate of 305 moose-train collisions from the winter,
57% lower than the average level observed over the previous six winters. The maximum snow
depth recorded at the Burns Lake airport this year was 31 cm; this level was 34% lower than the
lowest previous maximum seen during the study, and 52% percent lower than the average
maximum snow depth for the past six years. This represents the second consecutive winter with
mild snow conditions and moose-train collision levels substantially lower than in previous years.
Analysis of relationships between weather variables and annual variation in collision levels found
strong positive correlations to both snowfall and maximum snow depth, a correlation that has
been noted in three of the past four years analysis.
No new collisions were recorded within the three exclusion fence segments, and analysis of
spatial variation of collisions before and after fence construction continued to indicate that the
fences have lowered collision levels. However, opportunities for fence installation have been
severely restricted due to limitations on fence lengths and requirements to terminate fences at
existing rail bridge locations. In addition to mortality monitoring in 2016-2017, a prototype
apparatus designed to improve the effectiveness of exclusion fencing was tested at one end of
the Perow exclusion fence. This apparatus consisted of high-density, non-stick plastic sheeting
laid over the rail grade between and adjacent to the rails at the fences end. The site was
monitored using wildlife trail cameras, and while no moose were recorded interacting with the
apparatus, it held up to regular train traffic and snow plowing over the course of the winter.
Installation of the sheeting at the ends of all three exclusion fence sites could help to determine
its effectiveness at restricting moose movement. If successful, the apparatus could allow for
installation of exclusion fencing in high-collision areas previously considered unsuitable for
mitigation.
The Working Group also continued to investigate moose-train interactions along the Fraser
Subdivision this year. Efforts during the 2016-2017 field season were focused on examining areas
with high historical mortality levels, as well as continuing the development of the aerial survey
correction factor. Field investigations were conducted in August 2016 to assess environmental
factors influencing moose-train collisions and to evaluate sites that may be suitable for mitigation.
No sites were identified that would be considered suitable for exclusion fencing as installed along
the Telkwa Subdivision, in part due to high densities of public road-rail crossings in areas with
elevated collision risk. However, application of fencing in combination with non-stick sheeting may
allow for treatment of short stretches where a lack of rail bridges and high densities of road-rail
crossings would otherwise restrict potential for the installation of exclusion fencing.

ii November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

One hi-rail survey and one aerial survey were completed along the Fraser Subdivision the week
of April 10th. Application of the Fraser Subdivision correction factor to the aerial survey results
produced an estimate of 238 moose-train collisions. Although Environment Canada weather
station data were not sufficiently complete to conduct conclusive statistical analysis, snow
conditions do not appear to be strongly linked to annual collision variation along the Fraser
Subdivision, in contrast to the Telkwa Subdivision findings. This is likely a result of a lower
prevalence of suitable moose winter range along the rail corridor than was found along the Telkwa
Subdivision. Collision levels suggested moderate correlation to train traffic levels at McBride,
though the statistical power of the four-year sample size is limited. No specific collision hotspots
were identified this winter, though the spatial distribution of collisions this winter was consistent
with previous results.
Snow conditions for this years aerial and hi-rail mortality surveys created the best survey
conditions along the Fraser Subdivision of any survey years to date. However, the search
efficiency of the Fraser Subdivision aerial survey was lower this year than in years past, increasing
the relative uncertainty of the updated correction factor. Over the three years of data collection
along the Fraser Subdivision, the relative uncertainty of the aerial survey correction factor has
remained nearly unchanged, ranging between 35 and 38%. In contrast, the correction factor for
aerial surveys along the Telkwa Subdivision has a relative uncertainty of 15%. Search efficiency
during Fraser Subdivision aerial surveys is likely affected by higher variation in survey personnel
than has been seen along the Telkwa Subdivision. However, the main factor in the limited search
efficiency is likely the challenging flight conditions that are presented by the narrow right-of-way
and sloped topography along most of the Fraser Subdivision. Due to these characteristics, safely
conducting the aerial survey requires flying at higher altitudes and/or faster airspeeds than is
optimal for better search efficiency.

iii November 2017


Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

Acknowledgements
Reducing wildlife (and specifically moose) mortality along the railways of northern British
Columbia is an objective shared by CN, the hunting community, the provincial government, and
local citizens. This common objective ensured valuable input, feedback, and participation from a
number of individuals. Special thanks to CN Assistant Track Supervisors Ole Quam and Mike
Kowalski, as well as helicopter pilots Rob Altoft (Yellowhead Helicopters) and Rob Henderson
(Canadian Helicopters), who escorted us on mortality surveys to ensure safe track access and air
travel. Our thanks also to Dean Peard, Len Vanderstar and Adrian Batho for their assistance in
completing the mortality surveys, while Kenneth Child, Wayne Salewski, Mike Klaczek, Stephen
MacIver, Heidi Schindler, and Luanne Patterson from the Working Group provided valuable
insight, expert knowledge and constructive suggestions.

iv November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

TABLE OF CONTENTS
Introduction ................................................................................................................................. 1
Methods ....................................................................................................................................... 2
Telkwa Subdivision.................................................................................................................... 2
Aerial Survey ......................................................................................................................... 2
Analysis of Collision Incident Variation .................................................................................. 2
Mitigation ............................................................................................................................... 4
Fraser Subdivision..................................................................................................................... 5
Aerial and Hi-rail Mortality Surveys ........................................................................................ 5
Analysis of Collision Incident Variation .................................................................................. 7
Correction Factor Development ............................................................................................. 8
Mitigation Investigation .......................................................................................................... 8
Results ......................................................................................................................................... 9
Telkwa Subdivision.................................................................................................................... 9
Moose Train Collision Monitoring ....................................................................................... 9
Mitigation ............................................................................................................................. 11
Fraser Subdivision................................................................................................................... 12
Moose Train Collision Monitoring ..................................................................................... 12
Correction Factor Development ........................................................................................... 13
Mitigation ............................................................................................................................. 14
Discussion ................................................................................................................................. 15
Telkwa Subdivision.................................................................................................................. 15
Collision Incident Variation .................................................................................................. 15
Mitigation ............................................................................................................................. 16
Fraser Subdivision................................................................................................................... 17
Collision Incident Variation .................................................................................................. 17
Correction Factor ................................................................................................................. 18
Mitigation ............................................................................................................................. 18
Conclusions and Recommendations ...................................................................................... 19
Literature Cited ......................................................................................................................... 21
Appendix I. Fraser Subdivison Aerial Survey Correction Factor ......................................... 22
Methods................................................................................................................................... 22
Correction Factor Variables ................................................................................................. 22
Uncertainty Analysis ............................................................................................................ 23
Results .................................................................................................................................... 23
Fraser Subdivision Variables ............................................................................................... 23
Uncertainty Analysis ............................................................................................................ 24
Appendix II. Year 9 Moose-Train Collision Locations ........................................................... 25
Appendix III. Telkwa Subdivision Scatterplots ...................................................................... 28
Appendix IV. Fraser Subdivision Mortality Distribution and Two-Mile Segment Ratings.. 30

v November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

LIST OF FIGURES

Figure 1: Overview map showing the CN Telkwa Subdivision study area (red). .......................... 3
Figure 2. HDPE sheeting apparatus installed at the west end of the Mile 73.0 bridge on the Telkwa
Subdivision, immediately west of the Perow exclusion fence site. ............................................... 5
Figure 3. Map demonstrating the study area location along CNs Fraser Subdivision (red) between
McBride and Prince George, BC. .................................................................................................. 6
Figure 4. Moose collisions per mile in each of the 13 Telkwa Subdivision rail sections in Year 9
(solid) compared to the average for Years 3-8 (striped). ............................................................ 10
Figure 5. Average monthly snowfall, average and maximum snow depths, and number of days
with 40 cm snow depth compared to moose-train collision levels along the Telkwa Subdivision.
.................................................................................................................................................... 11
Figure 6. Average monthly snowfall, average snow depth and maximum snow depth for Years 6-
9 at the Prince George STP weather station, compared to historical norms and estimated moose-
train collision levels from hi-rail and aerial surveys along the Fraser Subdivision. ..................... 13
Figure 7. Areas from Mile 38-50 of the Fraser Subdivision reviewed during the August 2016 field
assessment for mitigation opportunities. Rail crossings (yellow pins), bridges (blue pins), a rail
siding (green line), and areas potentially suitable for exclusion fencing (red lines) are indicated.
.................................................................................................................................................... 15
Figure 8. Map of Year 9 winter moose mortalities (red pins) recorded along CNs Telkwa
Subdivision, between Endako and Smithers, during the April 6, 2017 aerial survey, as well as
exclusion fence segments (yellow lines). .................................................................................... 26
Figure 9. Map of the Fraser Subdivision, between McBride and Prince George, including Year 9
moose (red pins), deer (yellow), and elk (green) mortalities recorded during hi-rail and aerial
mortality surveys in April 2017. ................................................................................................... 27
Figures 10-14 Scatterplots of Telkwa Subdivision Variables vs. Collision Levels..28

LIST OF TABLES

Table 1. Proportion of Telkwa Subdivision moose-train collisions occurring within rail miles where
fencing was installed. Proportions were calculated prior to fence installation (Pre-Fence), in the
seven years since fence installation, and for Year 9 alone. ........................................................ 10
Table 2. Pearson's correlation coefficient (r), strength, and p-value for correlations of annual
variation in snow conditions and train traffic to collision levels along the Telkwa Subdivision. ... 11
Table 3. Application of Fraser Subdivision correction factor to results of Year 9 surveys. ......... 23
Table 4. Improvements in the relative uncertainty of correction factors for aerial mortality surveys
along the Fraser Subdivision as mortality data have accumulated from Year 6 forward. ........... 24
Table 5. Two-mile rail segments on the Fraser Subdivision rated very low (1-3 total mortalities),
low (4-6), moderate (7-9) and high (10+) moose collision risk, based on CN-reported data (1996-
2013) and hi-rail and aerial mortality survey records from Years 6-9 (2013-2017). Miles with no
recorded mortalities were not included. ...................................................................................... 30

vi November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

Introduction
In early 2007, the Canadian National Railway Company (CN) and several interested parties
identified a need to address train-related moose mortality in northern British Columbia. A Moose
Rail Mortality Working Group (Working Group) was formed, consisting of representatives of CN,
the Ministry of Forests, Lands and Natural Resource Operations (FLNRO), various stakeholders,
and academic specialists. The initial objectives were to collect reliable collision data, identify high-
mortality areas, and prioritize areas for mitigation along the Telkwa Subdivision, east of Smithers.
In the fall of 2008, McElhanney Consulting Services Ltd. (McElhanney) was contracted to develop
an approach that would meet the goals of the project. The Working Group spent the following
seven years completing a comprehensive study of moose-train interaction issues along the
Telkwa Subdivision. The study included literature review, habitat mapping, mortality surveys, and
installation and monitoring of 6.8 km of exclusion fencing. In addition, a mortality survey correction
factor was developed to allow moose-train collision levels to be estimated from the results of a
single, post-winter helicopter survey.
Analysis found that collision distribution was influenced by the presence of suitable moose winter
range characteristics such as browse availability and snow interception, while annual collision
levels varied with snow conditions that influenced the timing and magnitude of moose migration
to the valley bottoms. Exclusion fences were found to be effective at restricting moose access to
the rail grade, though potential fence locations were severely limited by requirements to end fence
segments at rail bridges, and by fence length restrictions intended to ensure migration routes
were not interrupted. Periodic fence damage and moose movement across rail bridges prevented
the fences from being 100% effective.
As a better understanding of moose-train collision issues along the Telkwa Subdivision was
gained, the Working Group turned their attention toward investigating other subdivisions where
moose mortality has been an issue. Expansion of the study to the Fraser Subdivision, east of
Prince George, provided potential for analysis of variables influencing moose-train interactions in
a region with topography and vegetation different from the Telkwa Subdivision. The first aerial
mortality survey on the Fraser Subdivision was completed in spring 2014. The following winter,
the Fraser Subdivision mortality survey program included a pair of hi-rail surveys and a spring
aerial survey to facilitate adaptation of the aerial survey correction factor to the Fraser Subdivision.
In addition to the mortality surveys, habitat mapping and landscape analysis were completed to
help evaluate spatial influences on collision distribution.
The first three years of mortality surveys along the Fraser Subdivision suggested that collision
levels were substantially lower than what has been recorded along the Telkwa Subdivision. In
addition, habitat mapping along the Fraser Subdivision indicated a lower prevalence of moose
winter range adjacent to the rail corridor. The highest collision levels in both historical data and
recent mortality survey data along the Fraser Subdivision were found between Miles 38 and 50,
though no specific hotspots showed collision levels comparable to those found along the Telkwa
Subdivision.
Data from the aerial and hi-rail surveys along the Fraser Subdivision were also used to develop
an aerial survey correction factor, adapted from the correction factor developed for the Telkwa
Subdivision. However, the low collision levels along the Fraser Subdivision limited the rate at
which the correction factor could be improved, with a relative uncertainty of 35% the best that had

1 November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

been achieved to date. Furthermore, variable snowpack persistence along the Fraser Subdivision
made timing the spring surveys to maximize carcass visibility a substantial challenge.
In Year 9 (2016-2017), mortality monitoring on the Telkwa Subdivision consisted of a single spring
aerial survey, followed by application of the correction factor, to estimate total winter collision
levels. Additional efforts along the Telkwa Subdivision involved the installation and monitoring of
an apparatus intended to restrict moose movement along the rails near the ends of existing
exclusion fencing.
On the Fraser Subdivision, field assessments were completed to identify variables that may
influence moose-train interactions and evaluate potential mitigation options. The spring aerial
survey along the Fraser Subdivision was preceded by a hi-rail survey to allow continued analysis
of environmental characteristics and further improvement of the correction factor.

Methods
Telkwa Subdivision
Aerial Survey
The Telkwa Subdivision aerial survey was conducted along the entire 125-mile (201 km)
subdivision, from Endako to Smithers (Figure 1). A pilot, a FLNRO navigator/recorder, and two
observers (one FLNRO biologist and one McElhanney biologist) flew along the rail corridor in a
Bell 206 Long Ranger helicopter equipped with rear bubble windows, at a height ranging from
approximately 75-100 m above ground level (AGL), with an airspeed ranging from 50-70 km/h.
All pieces of wildlife or livestock carcasses were called out by the observers, marked by the
navigator on a handheld GPS unit, and recorded on the aerial survey form. Information recorded
included the waypoint number, carcass species, age class (adult or calf), and carcass class (new
or old). Carcasses with flesh or hide remaining on them were identified as new (from the most
recent winter), while carcasses that showed evidence of sun bleaching or that were obscured by
dead foliage or leaf litter were classified as old (prior to this winter). Where the age of the carcass
was difficult to discern, or where crew members differed on their classification, a best estimate
was made and the uncertainty was noted.
In addition to carcasses recorded, live wildlife observed in and adjacent to the right-of-way (RoW)
were also noted during the survey. Species, age class, sex and GPS location were included, as
well as notes on animal condition (e.g. hair loss due to tick infestation). The navigator also
recorded personal, anecdotal and environmental conditions, as well as other observations that
may not have been identified by observers.

Analysis of Collision Incident Variation


Collison Distribution
The Telkwa Subdivision was divided into 12 10-mile segments (Miles 0-120) and one 5.2-mile
segment (Miles 120-125.2). The number of new carcasses identified during the 2017 aerial survey
was tallied within each segment, and compared to the distribution recorded over the past six years
of hi-rail and aerial survey data. Distributions were also compared to those observed in historical
data from CNs wildlife collision reporting program, between 1990 and 2008.

2 November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

Figure 1: Overview map showing the CN Telkwa Subdivision study area (red).
The proportion of new carcasses identified within miles treated with exclusion fencing was also
calculated and compared to previous results and the historical average. Collision proportions
within fenced miles were calculated for the 19 years prior to fence installation (1990-2008), for
the six years following fence installation (20102017), and for the Year 9 aerial survey alone.
Historical CN data were used to determine spatial collision proportions prior to fence installation,
while collision proportions occurring after fence installation were determined using hi-rail and
aerial mortality survey data. Hi-rail and aerial survey data were used to assess collision
proportions after fence installation for several reasons: 1) the GPS waypoints were considered to
more accurately reflect collision locations; 2) the collision levels recorded by the hi-rail and aerial
survey were believed to more accurately reflect actual collision numbers, as CN data are known
to under-represent collision numbers (McElhanney 2009); and 3) the higher collision numbers
recorded during hi-rail and aerial surveys provide a larger sample size for statistical analysis.
A one-tailed, two-proportion Z-test was completed to determine whether moose mortality levels
had seen a statistically significant decrease within the fenced miles, following fence construction.

3 November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

Collision data from 2009 were not incorporated in the analysis, as it was suspected that fence
construction activities would have influenced moose-train interactions that year.
Annual Collision Variation
Snow condition variables, including average and maximum snow depth, days with greater than
40 cm snow (McElhanney 2014), and total snowfall were calculated using data from two
Environment Canada weather stations near the Telkwa Subdivision rail corridor. The Burns Lake
Decker Lake station was located at the Burns Lake Airport, immediately adjacent to the rail
corridor near Mile 48. The Smithers A weather station was found at the Smithers Airport,
approximately 5 km north of the Telkwa Subdivisions west terminus and 30 m higher in elevation
than the rail grade in Smithers.
Daily Snow on the Ground measurements from both stations were averaged between December
1 and March 31 for each winter from Year 3 (2010-2011) through 9 (2016-2017); the two stations
data were then averaged to get a more complete picture of snow conditions along the entire
subdivision. The maximum snow depth recorded at either weather station was also recorded for
each winter. Additionally, the number of days that either weather station recorded a snow depth
of at least 40 cm was tallied for each winter; literature suggests that moose migration to winter
range is triggered near this snow depth (Modafferi 1991, Van Ballenberghe 1977, Coady 1974).
The total snowfall recorded at the Smithers A weather station between October 1 and April 30 of
each winter was also calculated; snowfall data were not available for the Burns Lake Decker
Lake weather station. Finally, train traffic levels at the Endako switch (at the Telkwa Subdivisions
east end) were tallied from January through March for each winter.
Annual variations in snow conditions and train traffic levels were compared to the collision levels
from each years surveys. Scatterplots comparing snow variables and train traffic to annual
collision totals for Years 3 through 9 were visually inspected. Pearsons correlation coefficients, r,
were calculated for each variables relationship with collision levels. Correlation coefficients
between 0 and 0.2 were rated very weak, 0.20.4 weak, 0.40.6 moderate, 0.60.8 strong, and
0.81 very strong (Evans 1996). For each correlation coefficient, p-values were calculated and
statistical significance was determined using =0.05.
For years in which both aerial and hi-rail surveys were completed along the Telkwa Subdivision,
collision levels were estimated based on the best estimate from the results of the two survey
methodologies. For years in which only aerial surveys were completed, the estimate was based
on the application of the correction factor to the aerial survey results.

Mitigation
In an effort to restrict moose movement across the rail bridge at Mile 73.0 and into the exclusion
fence to the east, an non-stick sheeting apparatus was constructed immediately west of the bridge
in October 2016. The apparatus consisted of a total of eight 4x8x3mm high-density polyethylene
(HDPE) sheets laid between and adjacent to the rails to cover the rail grade and create a non-
stick surface over which hooved animals would be unable to walk. Two sheets were laid between
the rails and fastened to the rail ties, covering a 16 stretch of the rail grade, while three sheets
were laid immediately outside each rail, covering an 8 width and 12 length outside either rail
(Figure 2).

4 November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

Figure 2. HDPE sheeting apparatus installed at the west end of the Mile 73.0 bridge on the Telkwa
Subdivision, immediately west of the Perow exclusion fence site.
Wildlife trail cameras were installed in late winter to monitor any potential occurrences of moose
interacting with the HDPE sheeting apparatus. Three cameras were installed at the edge of the
rail RoW on February 24th, 2017, and directed perpendicularly across the rail grade to catch
moose crossing the apparatus and diagonally to capture any moose approaching from the west.
Two cameras were set to record video footage, while the other one was set to collect still
photographs. The cameras were active until March 29th.

Fraser Subdivision
Aerial and Hi-rail Mortality Surveys
The Fraser Subdivision aerial survey was conducted along the entire 146-mile (235 km)
subdivision from Prince George to McBride (Figure 3). A pilot, navigator and two observers flew
along the rail corridor in a Bell 206 Jet Ranger helicopter equipped with rear bubble windows, at
an approximate height ranging from 100-200 ft AGL, with an airspeed ranging from 60-80 km/hr.
The navigator and one observer were McElhanney biologists, while the other observer was a
FLNRO biologist. Carcass recording procedures along the Fraser Subdivision were the same as
for the Telkwa Subdivision flight.

5 November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

Figure 3. Map demonstrating the study area location along CNs Fraser Subdivision (red) between
McBride and Prince George, BC.
The Fraser Subdivision hi-rail survey again covered the entire 146.1-mile track between McBride
and Prince George over two days. Two McElhanney biologists rode along in a hi-rail unita
pickup truck fitted with flanged steel wheels for rail travelnavigated by CNs track supervisors,
at a speed of 30 to 40 km/h. McElhanneys observers were situated on either side of the vehicle
to gain a vantage point over the rail grade, ditches and adjacent RoW on either side of the track.
Where any evidence of wildlife or livestock collisions was observed, including complete or partial
skeletons, soft tissue, hair or scavenger activity, the survey crew marked the site by rail mileage
and GPS waypoint and investigated the evidence on the ground. Data collected at each
suspected collision site included species, age and sex (if evident), carcass condition, evidence of
scavenging, and estimated time of death. Photographs were also taken at each carcass site.
Where several carcass pieces were scattered at one site, survey crews attempted to determine
the minimum number of individuals, as well as a best estimate for the number of carcasses.

6 November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

Determinations were based on the number and types of body parts present, evidence of carcass
scattering by trains or scavengers, and relative degree of decomposition. In addition, survey
crews estimated the aerial visibility of each carcass, to assist in analysis of waypoint overlap
between aerial and hi-rail surveys; aerial visibility was rated as poor, moderate or good.

Analysis of Collision Incident Variation


Mortality Survey Results Consolidation
Accurately assessing winter moose mortality along the Fraser Subdivision involved assessing
overlap between hi-rail and aerial survey results to determine which mortalities recorded by one
survey method were not detected by the other, and which mortalities were identified by both
surveys. In addition, results from the Year 9 surveys were compared to previous years data to
determine whether any carcasses may have been mistakenly identified as current year
mortalities.
Waypoints and photographs from the April 2017 aerial and hi-rail surveys were closely compared
to determine which carcasses were matches. Historical CN collision data were reported by track
mileage only, and were estimated to be accurate to 800 m. Hi-rail and aerial survey data were
recorded by GPS waypoint; hi-rail data were estimated to be accurate within 15 m, while aerial
data were estimated to have a precision of 100 m.
Moose carcass waypoints occurring within 50 m of one another were considered matches,
provided that photographs from the hi-rail surveys suggested the carcass could be visible aerially.
Waypoints occurring within 300 m of one another were considered potential matches, as flight
speed and communication timing limited waypoint precision during aerial surveys. Waypoint
descriptors, carcass condition, hi-rail survey photographs, and proximity of other carcasses were
used to determine the likelihood that waypoints had identified the same carcass. Waypoints
greater than 300 m from one another were not considered potential matches.
Any hi-rail-detected mortalities that were not identified by the aerial survey were included in the
total mortality count. In addition, any full or partial carcasses identified by the aerial surveys that
were missed by hi-rail surveys and could not be explained from previous years results were
added to the total mortality count. Carcasses that were identified by the aerial survey as current
year, but that fell within 50 m of carcasses identified by hi-rail surveys as old were not included in
the current year mortality count. In contrast, carcasses identified by the aerial survey as old that
matched closely with current-season carcasses identified during the hi-rail surveys were included
in the Year 9 mortality count, as the ability to stop at collision sites during the hi-rail surveys
allowed for closer examination of carcasses.
Where carcasses were identified during the Year 9 hi-rail survey as current year, but matched
Year 8 hi-rail data within 20 m or aerial data within 50 m, descriptions and photographs from both
years were examined to determine the likelihood that the carcass year assessment was incorrect.
Such comparisons were also completed where Year 9 aerial waypoints matched Year 8 data
within 50 m. Carcasses that were identified as moose during aerial surveys, but that matched the
location of deer or elk carcasses recorded by the spring hi-rail survey within 50 m were also
excluded.

7 November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

Collision Distribution
Wildlife collision data reported by CN between 1996 and 2013 were used to identify rail segments
along the Fraser Subdivision with high historical moose-train collision levels. The Fraser
Subdivision was dissected into 73 two-mile segments, for which all historically reported collisions
were tallied. Moose-train collision data from hi-rail and aerial mortality surveys completed in Years
69 were also tabulated for each two-mile segment and combined with the historical data. Ratings
were given to each two-mile section based on the combined number of collisions recorded;
segments with zero to three collisions rated very low, four to six rated low, seven to nine rated
moderate, and ten or more rated high (McElhanney 2016a). Mortality levels along the Fraser
Subdivision were also compared to historical levels along the Telkwa Subdivision, where ratings
were low (18 historical collisions), moderate (916), high (1724) and very high (2532).
Annual Collision Variation
Train traffic data from the McBride switch at the Fraser Subdivisions east end were compiled
from January through March for each of the past four winters. Collision levels over each of the
past four winters were then compared to train traffic levels; a Pearsons correlation coefficient was
calculated for the relationship, and the strength of correlation was rated on the same basis as the
Telkwa Subdivision correlations. The p-value of the correlation coefficient was again calculated
and statistical significance was determined using =0.05.
Weather data available from Environment Canada weather stations along the Fraser Subdivision
were not sufficiently complete to allow conclusive statistical analysis. However, some snow
characteristic data from the Prince George STP weather station could be tabulated using available
data for the past four winters and compared anecdotally to historical norms for that station
between 1981-2010 (Env Can 2017). Snowfall totals from October through April and average
snow depth from December through March were tabulated for each of the past four winters and
compared to the best estimate of moose-train collision levels using data from hi-rail and aerial
surveys along the Fraser Subdivision.

Correction Factor Development


Mortality survey analyses in Year 9 included the continued development of an aerial survey
correction factor for estimation of winter moose mortality levels along the Fraser Subdivision.
Methods and Results of the correction factor development are included in Appendix I.

Mitigation Investigation
Given previous findings that the stretch of the Fraser Subdivision between Miles 38 and 50
appeared to have the highest collision levels historically and over the past three years of surveys,
this stretch was proposed for field investigations. Rail segments within this stretch that could be
safely accessed on foot from nearby rail crossings or adjacent roads were targeted for field
assessments. Field surveys assessed environmental variables that may influence moose-train
interactions, and identified characteristics of the rail corridor that could affect suitability for
installation of mitigation measures. Environmental factors recorded included topography within
and adjacent to the rail RoW, ecosystem types in adjacent vegetation, and proximity of major
waterbodies. Rail corridor characteristics that were surveyed included occurrence of rail sidings,
proximity of public roads and presence of rail crossings, and dimensions and characteristics of

8 November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

bridges and overpasses, including length, clearance, and number and type of spans. Accessibility
for heavy machinery was also noted during field assessments.

Results
Telkwa Subdivision
Moose Train Collision Monitoring
The Year 9 moose mortality survey program for CNs Telkwa Subdivision involved one aerial
survey, conducted April 6, 2017. Visibility was good for the survey, with no cloud cover. Snow
cover was minimal, estimated at less than 5% coverage within the RoW, with maximum depths
of approximately 10 cm. Ditch flooding due to snowmelt was not identified as an issue during the
survey.
Sixteen carcasses were identified as current-winter mortalities during the Telkwa Subdivision
aerial survey. Of these, one mortality site was situated within 100 m of a Year 8 carcass, though
waypoint descriptions did not indicate whether the two sites may have been the same carcass,
and no field photographs were available; this carcass was included in this years mortality count.
Applying the correction factor developed for Telkwa Subdivision aerial mortality surveys resulted
in an estimate of 305 moose-train collisions in Year 9, the lowest collision estimate in any of the
past seven years.
Collison Incident Variation
Of the 16 current-year carcasses identified during the Telkwa Subdivision aerial survey, four were
identified as full carcasses or skeletons, three were recorded as partial carcasses, and nine were
not fully described. Only one of the carcasses was found on the east half of the subdivision,
between Endako and Topley, while the remaining 15 occurred on the west half, between Topley
and Smithers. Six carcasses were recorded between Miles 78 and 85.5, while another four were
identified in a 2-mile stretch near Mile 120.
With only one carcass recorded between Miles 0 and 70 in Year 9, corrected collision levels along
this stretch were 94% lower than the six-year average (Figure 4). West of Mile 70, corrected
collision levels were 20% lower than the average from Years 3 through 8, with above-average
collision levels from Mile 80-90 and Mile 120-125.2. Overall, corrected collision levels along the
Telkwa Subdivision in Year 9 were 55% lower than the six-year average. Locations of collisions
recorded along the Telkwa Subdivision can be found in Appendix II, Figure 8.

9 November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

1.2

1.0
Collisions/Mile

0.8

0.6

0.4

0.2

0.0

Rail Section (miles)

Figure 4. Moose collisions per mile in each of the 13 Telkwa Subdivision rail sections in Year 9
(solid) compared to the average for Years 3-8 (striped).
No new collisions were recorded within the exclusion fence segments this year. Overall, the
proportion of Telkwa Subdivision collisions occurring within the 3.9 miles of track now isolated
with fencing has decreased from 10.1% (46 of 455) prior to fence construction, to 3.3% (14 of
430) over the six years since installation (Table 1). A one-tailed, two-proportion Z-test on these
values returned a test statistic of 4.05, for a p-value of 2.5x10-5 (=0.05), indicating that there has
been a statistically-significant decrease in the proportion of Telkwa Subdivision collisions
occurring within the fenced miles.
Table 1. Proportion of Telkwa Subdivision moose-train collisions occurring within rail miles where
fencing was installed. Proportions were calculated prior to fence installation (Pre-Fence), in the
seven years since fence installation, and for Year 9 alone.

Fence Site Pre-Fence Years 3-8 Year 9 Post-Fence


(Mileage) (1990-2008) (2010-2016) (2016-2017) (2010-2017)

Endako (11.0-11.9) 3.5% 0.7% 0% 0.7%

Perow (70.1-73.0) 5.7% 2.7% 0% 2.6%

Barrett (94.7-95.0) 0.9% 0% 0% 0%

Fence Total 10.1% 3.4% 0% 3.3%

Variation in Weather and Train Traffic


In Year 9, all snow metrics along the Telkwa Subdivision were lower than in any previous year of
the hi-rail survey program. The average snow depth of 13 cm (December through March) was
60% below the six-year mean, the maximum snow depth at the Burns Lake-Decker Lake weather
station (31 cm) was down 52%, and total snowfall at the Smithers A weather station was 36%
below average; this was the first winter in the past seven in which the snow depth did not exceed
40 cm (Figure 5).

10 November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

80 180

70 160

140
60
120
Snow (cm), Days

50

Collisions
100
40
80
30
60
20
40

10 20

0 0
2010-2011 2011-2012 2012-2013 2013-2014 2014-2015 2015-2016 2016-2017

Smithers Snowfall Avg Snow Depth Max Snow Depth Days > 40 cm SOG Collisions

Figure 5. Average monthly snowfall, average and maximum snow depths, and number of days
with 40 cm snow depth compared to moose-train collision levels along the Telkwa Subdivision.
Visual inspection of the scatterplots comparing traffic and snow condition variables to annual
collision levels did not indicate any specific linear relationships between traffic or weather
variables and collision levels (Appendix III). Average snow depth and the number of days with at
least 40 cm of snow on the ground showed weak correlation with collision numbers (Table 2). The
maximum snow depth showed a strong positive correlation with collision numbers, though this
correlation was not statistically significant. A strong correlation was also found between the
snowfall at the Smithers A weather station and Telkwa Subdivision collision levels; this
relationship was statistically significant, at a p-value of 0.035. Train traffic levels showed a very
weak negative correlation to collision levels.
Table 2. Pearson's correlation coefficient (r), strength, and p-value for correlations of annual
variation in snow conditions and train traffic to collision levels along the Telkwa Subdivision.

Parameter Mean Snow Days >40 cm Maximum Snowfall Train


Depth Snow Snow Depth Traffic
Correlation Coefficient, r 0.25 0.27 0.69 0.79 -0.11
Strength weak weak strong strong very weak
p-value 0.59 0.56 0.085 0.035 0.81
Statistically significant? No No No Yes No

Mitigation
Wildlife trail cameras installed to monitor the HDPE trial sheeting at Mile 73.0 on the Telkwa
Subdivision did not record any photographs or video footage of moose interacting with the
apparatus. Furthermore, no evidence of moose presence was noted during site visits on March
3rd and 29th, and as such, no evidence was obtained to confirm that the structures further restricted

11 November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

moose movement into the exclusion fencing. However, the apparatus did not appear to affect
train traffic and held up through regular track maintenance and snow plowing.
Fraser Subdivision
Moose Train Collision Monitoring
The mortality monitoring program for the Fraser Subdivision consisted of one hi-rail and one aerial
mortality survey completed in April 2017. The hi-rail survey was completed over two days, April
10 and 11, while the aerial survey was conducted April 13. Snow conditions were good during
both surveys, with less than 25% coverage within the RoW. Patches of snow remained in the
ditches between Miles 55 and 65, and west of Mile 90. Where snowpack persistence within the
RoW was greater than 50%, snow depths were estimated at less than 25 cm. Ditch flooding was
restricted to areas between Miles 118 and 125, near Giscome, and Mile 140 near Willow River.
Visibility was good during the aerial survey, with high cloud cover that did not affect the survey.
Collision Incident Variation
The Fraser Subdivision hi-rail survey identified 14 moose carcasses from the past winter, as well
as two elk carcasses and four deer carcasses. Of the moose carcasses, five were identified as
adults and three as juveniles, while six could not be assigned an age class. One male could be
identified in the field, while the remaining 13 carcasses could not be visually identified to sex. Of
the two elk carcasses, one was determined to be an adult female, while the other could not be
classified to age or sex. None of the four deer carcasses could be classified to age or sex.
The aerial survey along the Fraser Subdivision recorded 11 moose carcasses from the past
winter. Comparison of the results from the two surveys found that five moose carcasses from the
past winter were accurately recorded by both the hi-rail and aerial surveys. Seven moose
carcasses and two elk carcasses that were identified during the hi-rail survey as having moderate
or good aerial visibility were not recorded during the aerial survey. In addition, two deer carcasses
recorded during the hi-rail survey were mistakenly identified as moose during the aerial survey,
and one old moose carcass (Year 8) was recorded as new (Year 9). Four moose carcasses that
were identified during the aerial survey had not been identified during the hi-rail survey.
Consolidation of the two surveys results suggested that a total of 17 new moose carcasses, two
elk and four deer were identified.
While no specific hot-spots of moose-train collisions were identified in Year 9, most of the
carcasses occurred in a 50-mile stretch near the centre of the subdivision (Appendix II, Figure 9).
Of the 17 new moose carcasses identified this year, 13 were found between Mile 33 (near
Crescent Spur) and Mile 82 (near Longworth). Review of historical CN-reported wildlife collision
data found nine two-mile segments with moderate moose-train collision levels and two with high.
Adding hi-rail and aerial survey data compiled in Years 6-9 found 12 segments with moderate
collision levels and four with high ratings (Appendix IV, Table 5). Miles 38-40, 44-46, 46-48 and
96-98 were each rated as high-collision segments, with 10 or more total collisions compiled in
historical records and recent surveys. Three segments (Miles 20-22, 38-40 and 44-46) have had
four or more mortalities recorded during aerial and hi-rail surveys over the last three winters, with
eight mortalities recorded between Miles 38 and 40.

12 November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

Variation in Weather and Train Traffic


While weather data in the region were not sufficiently complete to allow statistical analysis,
available snowfall and snow depth data from the Prince George STP weather station suggest that
snow conditions during the Year 9 winter, like the previous winter, were relatively mild. The
historical climate norms (1981-2010) for average monthly snowfall and average snow depth at
the Prince George STP are 20.2 cm and 16.8 cm, respectively. In comparison, Year 9 had an
average monthly snowfall of 21.0 cm (4% above the norm) and an average snow depth of 6.7 cm
(60% below the norm). Furthermore, the maximum snow depth in Year 9 was 38 cm, the lowest
maximum snow depth since the 2011-2012 winter. Average monthly snowfall (October through
April), average snow depth (December through March) and maximum snow depth for each of the
past four winters at the Prince George STP, as well as the historical norms, can be seen in
Figure 6 and compared to moose-train collision levels estimated from hi-rail and aerial surveys
along the Fraser Subdivision.
70 45

40
60
35
50
30
Snow Depth (cm)

40

Collisions
25

30 20

15
20
10
10
5

0 0
2013-2014 2014-2015 2015-2016 2016-2017

Monthly Snowfall Average Snow Depth Maximum Snow Depth


Snowfall Historical Norm Snow Depth Historical Norm Collisions

Figure 6. Average monthly snowfall, average snow depth and maximum snow depth for Years 6-9
at the Prince George STP weather station, compared to historical norms and estimated moose-
train collision levels from hi-rail and aerial surveys along the Fraser Subdivision.
Comparison of annual variation in collision levels over the past four winters to winter train traffic
levels at the McBride switch on the Subdivisions east end found a Pearsons correlation
coefficient, r, of 0.528. This suggested a moderate positive correlation, which is not statistically
significant at a p-value of 0.472.

Correction Factor Development


Comparison of results between the hi-rail and aerial mortality surveys conducted along the Fraser
Subdivision in April 2017 led to an updated search efficiency factor, s, of 0.550.13. Combining
this value with the carcass retention factor (r=0.970.02) and the carcass persistence factor
(p=0.910.10) resulted in a correction factor, FF, of 2.050.75 for Fraser Subdivision aerial

13 November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

surveys. Detailed results of the continued development of an aerial survey correction factor can
be found in Appendix I.

Mitigation
Investigations over the previous three years along the Fraser Subdivision identified the greatest
moose-train collision risks between Miles 38 and 50. The following areas along this stretch were
safely accessible on foot, and were assessed on August 26, 2017 for environmental variables
and RoW characteristics (Figure 7):
The private rail crossing at Mile 32.8, near the Crescent Spur heliski operation on Prospect
Road;
From the Morkill Forest Service Road (FSR) Overpass at Mile 35.0, 800 m south toward
an oxbow and adjacent floodplain;
From a private crossing on Murray Road at Mile 37.1, 700 m west along the rail grade
where a large wetland is situated to the south
From a public rail crossing at Mile 39.5 east to the rail bridge over Snowshoe Creek at
Mile 38.7
From the Ptarmigan Creek rail bridge at Mile 46.8 west to the Walker Creek FSR rail
crossing at Mile 47.7
The private rail crossing at Mile 32.8, near the Crescent Spur Heliski operation, appears to provide
suitable rail access for equipment if mitigation work is planned to the west. No specific
environmental or topographic characteristics that would increase moose-train collision levels were
observed at this site.
Suitable moose winter range was found in floodplain forests between an isolated oxbow and the
Fraser River south of the Morkill FSR overpass. Installation of exclusion fencing along this stretch
would require an alternative fence closure, as no rail bridges are situated nearby. Furthermore,
fencing at this site would have to be designed to avoid potentially directing moose toward the
FSR.
Advanced deciduous stands that provide suitable moose winter range were also found between
the rail grade and a large wetland between Miles 37 and 38. However, a rail siding along much of
this stretch likely reduces moose-train collision risk.
While the rail bridge over Snowshoe Creek at Mile 38.7 would be suitable for installation of
exclusion fencing on either side, and has sufficient clearance for use as a moose underpass,
suitable moose winter range here is restricted to a 200 m stretch where the rail grade crosses the
floodplain of the creek itself.
Between the rail bridge over Ptarmigan Creek at Mile 46.8 west to the public rail crossing on the
Walker Creek FSR, the rail corridor follows the floodplain of Ptarmigan Creek, with high value
moose winter range to the north of the RoW. The Ptarmigan Creek rail bridge is suitable for use
in the construction of exclusion fencing, though its clearance is relatively low for use as a moose
underpass. Exclusion fencing may be suitable along this stretch, though planning will have to
consider potential for fences to redirect moose toward the FSR.

14 November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

Figure 7. Areas from Mile 38-50 of the Fraser Subdivision reviewed during the August 2016 field
assessment for mitigation opportunities. Rail crossings (yellow pins), bridges (blue pins), a rail
siding (green line), and areas potentially suitable for exclusion fencing (red lines) are indicated.

Discussion
Telkwa Subdivision
Collision Incident Variation
For the second consecutive winter, all snow metrics along the Telkwa Subdivision (snowfall,
maximum snow depth, average snow depth and snowpack persistence) were milder than any of
the previous six years. Similarly, moose-train collision levels along the Telkwa Subdivision in
Year 9 were the lowest seen over the course of the study, 57% below the average for the previous
six years.
The timing and magnitude of major snowfall events can influence how many moose occupy the
winter range each year, while timing of snowpack accumulation and the persistence of a deep
snowpack can influence the length of time that moose densities are elevated on the winter range.
Because much of the Telkwa Subdivision runs through suitable moose winter range, snow
conditions that cause more moose to migrate to the winter range and/or that result in prolonged

15 November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

occupation of the winter range are likely to lead to higher moose-train collision levels. Over the
past seven years, annual moose-train collision levels along the Telkwa Subdivision have been
strongly correlated with both snowfall at the Smithers airport and with the maximum snow depth
at the Burns Lake airport, with the relationship between collision levels and Smithers snowfall
statistically significant. It is suspected that the mild winter conditions again this winter caused
fewer moose to migrate to winter range along the Telkwa Subdivision rail corridor, resulting in
fewer moose-train collisions than any of the previous six years.
The distribution of collisions in Year 9 was heavily skewed toward the west end of the subdivision
this winter, with only one mortality recorded between Endako and Topley. The east end of the
subdivision appears to have very condensed winter range that is very heavily utilized during
severe winters, but not during mild winters. In contrast, the west end of the subdivision has
widespread winter range between Topley and Smithers, and appears to have relatively steady
collision levels. Despite historically low snow load this winter, collision levels on the west half of
the subdivision were only 20% below the average from the past six years.

Mitigation
With no new collisions recorded within exclusion fence sites in Year 9, and no winter damage
recorded during the spring aerial survey, the proportion of moose-train collisions occurring within
fence sites continued to decrease. However, while the installation of exclusion fencing has
significantly reduced moose-train collision levels within fenced miles, fencing effectiveness has
not been as high as initially hoped. Bridges have not been as effective at restricting moose
movement into exclusion fences as was anticipated. This has been exacerbated by a recent move
to replace remaining wooden trestle bridges with concrete tub style bridges. While the wooden
trestle bridges have spaces between the rail ties in the deck and have been considered mostly
impassable to moose, the decks of the concrete tub bridges are filled with ballast and closely
resemble the surrounding grade.
In September 2016, an apparatus constructed of HDPE sheeting was installed immediately west
of the rail bridge at Mile 73.0 of the Telkwa Subdivision. This site constitutes the west terminus of
the Perow exclusion fence, and the rail bridge here is a concrete tub with relatively low clearance.
The HDPE sheeting was intended to prevent moose from moving along rails and crossing the
bridge into the exclusion fencing segment running between Miles 70.1 and 73.
While monitoring of the prototype HDPE site using wildlife trail cameras did not record any moose
interacting with the apparatus, regular winter train traffic and track maintenance activities did not
damage the apparatus and it remained functional throughout the winter. Installation of additional
HDPE apparatuses at other exclusion fencing termini may help to confirm their effectiveness if
collision levels decrease further within the exclusion fences.
Furthermore, the HDPE apparatus may allow fencing to be installed in areas previously
considered unsuitable. During planning for the installation of exclusion fencing along the Telkwa
Subdivision, the length of fence segments was restricted to minimize impacts on moose migration
and gene flow. As a result, areas considered suitable for fencing were limited to those sites where
existing rail bridges were situated in close proximity and could act as either fence ends or wildlife
underpasses. If HDPE sheeting is proven effective at restricting moose movement along the rail
grade into exclusion fencing segments, areas with high moose-train collision levels that do not
have rail bridges may be reconsidered as an option for mitigation.

16 November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

Areas of the Telkwa Subdivision that could benefit from application of exclusion fencing and/or
HDPE sheeting include:
the rail stretch between Miles 8 and 11, east of existing exclusion fencing from Mile 11-
11.9;
the stretch from existing exclusion fencing at Perow west to Houston; and
sites along the floodplain of the Bulkley River floodplain between Quick and Smithers.
These sites have had high mortality levels historically, but have been considered unsuitable for
exclusion fencing due to a paucity of rail bridges.

Fraser Subdivision
Collision Incident Variation
Aerial and hi-rail mortality surveys completed along the Fraser Subdivision in April 2017 identified
a total of 17 moose mortalities from the past winter, as well as two elk and four deer. Survey
conditions were good for both the aerial and hi-rail surveys along the Fraser Subdivision, with
survey timing the best that it has been for any of the past four years of spring surveys. Persistence
of the snowpack in some shaded areas of the RoW was balanced with minor flooding in ditches
elsewhere. Timing the aerial surveys along the Fraser Subdivision to maximize carcass visibility
remains a challenge, though local track personnel are becoming more familiar with suitable snow
conditions for the hi-rail and aerial surveys. Application of the Fraser Subdivision correction factor
to the results of the aerial survey produced a moose mortality estimate of 238. This represents
an increase from the estimate of 145 moose-train collisions in Year 8, but remains lower than
collision levels observed along the Telkwa Subdivision in each of the past seven years.
While weather data in the region were not sufficiently complete to allow conclusive statistical
analysis, snowfall and snow depth data from the Prince George STP weather station suggest that
the last three winters have been relatively mild, with this winter showing the lowest maximum
snow depth in the past five years. Anecdotal comparison of snow metrics and collision levels
found little evidence of a relationship between the two. Given that much of the Fraser Subdivision
runs through areas with low winter range suitability, it is possible that the severity of winter snow
conditions has limited impact on moose density along the rail corridor.
Other factors that could influence year-to-year variation in moose-train collision levels along the
Fraser Subdivision are also poorly understood. Collision levels have shown moderate positive
correlation with train traffic levels at the McBride switch, though this relationship has not been
statistically significant. The power of this analysis was limited by a small sample size including
only four winters of reliable data, though it is possible that train traffic levels may have a stronger
influence on collision levels on the Fraser Subdivision than along the Telkwa Subdivision due to
the isolation of most of the subdivision from the highway corridor.
No specific collision hot spots were identified from the Year 9 mortality survey results, though the
spatial distribution of mortalities in Year 9 was generally consistent with averages over the
previous two years of hi-rail and aerial surveys and the historical data. The rail stretch between
Miles 38 and 50 remains a priority area if site-specific mitigation measures are to be considered
along the Fraser Subdivision.

17 November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

Correction Factor
Despite improved snow conditions during this years survey, survey efficiency decreased from
s=0.640.14 in previous years (Years 6-8) to 0.550.13 using data from Years 6-9, resulting in
an increase in the relative uncertainty of the aerial survey correction factor developed for the
Fraser Subdivision. During the development of the aerial survey correction factor for the Telkwa
Subdivision, the relative uncertainty improved substantially over the course of the mortality survey
program, showing continuous improvement as data accumulated from year to year (McElhanney
2016a). Examination of a total of 363 mortalities during hi-rail and aerial surveys along the Telkwa
Subdivision allowed the relative uncertainty of the aerial survey correction factor to improve from
38% to 15% over six years of data accumulation. In comparison, the Fraser Subdivision mortality
survey program has examined a total of 62 mortalities over the past three winters, but the relative
uncertainty of the correction factor has seen almost no improvement over that time, remaining
between 35 and 38%.
It is suspected that two factors have had a substantial impact on the potential improvement of
survey efficiency: inconsistency in survey personnel, and flight conditions resulting from the
characteristics of the Fraser Subdivision RoW. The uncertainty in timing the Fraser Subdivision
aerial surveys to optimize snow conditions has required the flexibility to complete surveys on
relatively short notice. Thus, in each of the four aerial mortality surveys completed along the
Fraser Subdivision so far, two of the three survey crew members were new to the survey
methodology. Consistency in survey personnel would likely improve search efficiency
substantially. However, regardless of survey personnel experience, flight conditions along the
Fraser Subdivision continue to present a challenge for aerial surveys. Maintaining safe control of
the helicopter over the narrow RoW and sloped topography of the subdivision requires faster
airspeeds and higher ground clearance than would be ideal for mortality surveys.
Given the lack of improvement in the search efficiency of aerial surveys along the Fraser
Subdivision, further improvements to the precision of the correction factor are unlikely to occur in
the near term. However, given the relatively low mortality levels recorded along the Fraser
Subdivision in the past two years, the existing uncertainty level for the aerial survey correction
factor may be sufficient to estimate mortality levels for management purposes.

Mitigation
No specific collision hotspots were observed along the Fraser Subdivision in Year 9, though more
than 75% of carcasses observed were situated in a 50-mile stretch between Crescent Spur and
Longworth (Miles 35-85). This distribution was generally consistent with the results of the past
three years of aerial and hi-rail surveys along the Fraser Subdivision, as well as with historical
data. The combined historical and survey collision data continue to indicate that the highest
collision levels consistently occur between Miles 38 to 50, west of Crescent Spur, and field review
of this site suggested that several locations with suitable moose winter range on either side of the
rail likely contribute to high winter moose densities along this stretch of the rail corridor.
Potential for installation of exclusion fencing is complicated here by high densities of rural roads,
including roads running alongside the rail RoW, and a number of public rail crossings situated
between existing rail bridges. Installation of exclusion fencing along this stretch would thus
present a risk of funnelling moose onto public roads. If, however, short segments of exclusion
fencing can be installed in combination with HDPE sheeting, two stretches could be considered

18 November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

for treatment: Miles 46.8 to 47.7 between the Ptarmigan Creek rail bridge and Walker Creek FSR,
and Miles 34.2 to 34.6 south of the Morkill FSR overpass. These areas have higher collision rates
than other areas along the Fraser Subdivision, but overall total mortality levels in these areas
remain substantially lower than have been recorded on the Telkwa Subdivision. Thus, the costs
and benefits of mitigation measure installation on the Fraser Subdivision should be considered.

Conclusions and Recommendations


Corrected collision levels along the Telkwa Subdivision in 2016-2017 were the lowest they have
been over the seven years of detailed survey data. Similarly, Environment Canada weather
stations in Smithers and Burns Lake recorded the mildest snow conditions of any year of the
study. Collision levels were strongly correlated with both snowfall totals at the Smithers airport
and maximum snow depth at the Burns Lake airport. This suggested that mild snow conditions
this winter led fewer moose to migrate to high value winter range, resulting in lower moose
densities along the rail corridor.
No new mortalities were recorded within fence sites this winter, and there has been a statistically
significant decrease in the proportion of Telkwa Subdivision collisions occurring within fenced
miles. While monitoring of a HDPE sheeting apparatus installed at the end of one exclusion
fencing did not produce any evidence of influences on moose movement, this prototype may
present opportunities to improve the effectiveness and expand the application of exclusion
fencing.
Along the Fraser Subdivision, corrected mortality levels indicated an increase over last years
estimate, but numbers remained lower than any of the past six years along the Telkwa
Subdivision. Variables influencing year-to-year variation in collision levels along the Fraser
Subdivision remain difficult to analyze. Collision levels showed moderate correlation with train
traffic levels, while anecdotal comparison of collision levels and snow conditions suggests little or
no effect. An apparent lack of correlation between snow metrics and collision levels may be a
result of the limited prevalence of suitable winter range along most of the Fraser Subdivision rail
corridor.
No fine-scale collision hotspots were observed along the Fraser Subdivision in Year 9, though
more than 75% of carcasses observed were situated in a 50-mile stretch between Crescent Spur
and Longworth. This distribution was generally consistent with the results of the past three years
of aerial and hi-rail surveys along the Fraser Subdivision, as well as with historical data. The
combined historical and survey collision data indicate that the highest collision levels consistently
occur between Miles 38 to 50, immediately west of Crescent Spur. Field review of this stretch
suggested that several locations with suitable moose winter range on either side of the rail likely
contribute to comparatively higher winter moose densities along this stretch of the rail corridor.
Potential for installation of exclusion fencing here is complicated by high densities of rural and
resource roads, including roads running alongside the rail RoW and a number of public rail
crossings. If HDPE sheeting can be utilized to prevent moose movement into open fence ends,
the installation of shorter segments of exclusion fencing may be possible in high-collision areas
where the rail corridor and public roads are not closely proximal. Two sites were identified where
exclusion fencing with HDPE sheeting may be suitable.

19 November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

Addition of the hi-rail and aerial survey data collected in Year 9 to previous years data resulted
in an aerial survey correction factor for the Fraser Subdivision of 2.050.75, compared to a factor
of 1.770.62 developed in Year 8. Despite good snow conditions for this years survey, a decrease
in survey efficiency resulted in an increase in the factors uncertainty level. Persistently poor
survey efficiency is partly the result of subdivision characteristics that make aerial surveys difficult,
but would likely improve with more consistent survey personnel. Continued study of factors
influencing the accuracy of aerial surveys along the Fraser Subdivision is likely to have little effect
on correction factor uncertainty in the near term. However, given the low collision levels that have
been observed both historically and in the past four years, the existing uncertainty level may be
sufficient for management purposes.
The following recommendations are proposed for the continuation of winter moose mortality
monitoring and mitigation along the Telkwa and Fraser Subdivisions:
Continue to monitor the HDPE site for evidence of influences on moose movement along
the rail.
Consider installation of short sections of exclusion fencing in combination with HDPE
sheeting between Miles 8 and 10 on the Telkwa Subdivision.
If deemed necessary, consider installation of exclusion fencing with HDPE sheeting
between Miles 46.8 and 47.7 near Ptarmigan Creek and between Miles 34.2 and 34.6
south of the Morkill FSR overpass. Use of exclusion fencing must consider potential for
moose to be redirected onto public roads.
Consider alternative fence designs that may limit potential for entrapment of moose,
including fencing on only one side of the rail, with directional gates.

20 November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

Literature Cited
Agresti, A and B Coull. 1998. Approximate is better than exact for interval estimation of binomial
proportions. The American Statistician 52: 119126.
Coady, JW. 1974. Influence of snow on behavior of moose. Nat. Can. 101: 417-436.
Environment Canada. 2017. National Climate Data and Information Archive.
[http://www.climate.weatheroffice.gc.ca/climateData/canada_e.html] Accessed May
2017.
Evans, JD. 1996. Straightforward statistics for the behavioral sciences. Brooks/Cole Publishing.
Pacific Grove, CA.
Huso, MMP. 2011. An estimator of wildlife fatality from observed carcasses. Environmetrics 22:
318329.
McElhanney Consulting Services Ltd. (McElhanney). 2009. Moose Collision Assessment: Telkwa
Subdivision. Unpubl. Terrace, BC.
McElhanney Consulting Services Ltd (McElhanney). 2014. Winter Rail Mortality of Moose in
Northern BC: Year VI, Telkwa and Fraser Subdivisions. Unpubl. Terrace, BC.
McElhanney Consulting Services Ltd. (McElhanney). 2016a. Monitoring and Mitigation of Moose
Rail Mortality in Northern BC: 2016 Report. Unpubl. Terrace, BC.
McElhanney Consulting Services Ltd (McElhanney). 2016b. Monitoring and Mitigation of Moose
Rail Mortality in Northern BC: Year 7, Telkwa and Fraser Subdivisions. Unpubl. Terrace,
BC.
Modafferi, RD. 1991. Train-moose kill in Alaska: characteristics and relationship with snowpack
depth and moose distribution in Lower Susitna Valley. Alces 27: 193-207
Olson, DD. 2013. Assessing Vehicle-Related Mortality of Mule Deer in Utah. All Graduate Theses
and Dissertations. Utah State University. Paper 1994.
van Ballenberghe, V. 1977. Migration behavior of moose in southcentral Alaska. 13th Congress of
Game Biologists, pp. 103-109. Atlanta, GA.

21 November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

Appendix I. Fraser Subdivison Aerial Survey Correction Factor


Methods
Results from hi-rail and aerial surveys were compared and consolidated using the methods
developed in Year 6 (20132014), to develop subdivision-specific aerial survey correction factors
(McElhanney 2014).
Year 9 survey data were compared to previous years data, then incorporated into the total
database for Years 6 through 9 to refine the best estimate for the aerial survey correction factor.
No additional analysis was completed for the Telkwa Subdivision aerial survey correction factor
in Year 9.

Correction Factor Variables


Analysis of survey accuracy in Year 9 involved continued improvement of the correction factor for
aerial surveys along the Fraser Subdivision. During correction factor development, several
variables were identified that could influence survey accuracy; carcass retention, carcass
persistence and search efficiency were variables included in the correction factor, while carcass
identification accuracy and survey timing factors could not be included. The following correction
factor, F, was developed in Years 6 and 7 for aerial spring carcass surveys, based on previously
developed correction factors for bat and deer mortality surveys (McElhanney 2014, Huso 2011,
Olson 2013):
1
=

Where p represents the carcass persistence, r the carcass retention rate, and s the search
efficiency. Carcass persistence was defined as the proportion of carcasses recorded during hi-
rail surveys that were still visible during either the final hi-rail survey of the season or the aerial
survey. Search efficiency was defined as the proportion of carcasses within the RoW at the end
of the year that were detected by the spring aerial mortality surveys, accounting for both carcass
visibility and observer efficiency. The carcass retention rate was defined as the proportion of
collisions where wounded moose remained within the rail RoW such that they could be detected
by a mortality survey.
The Year 9 survey program consisted of one aerial survey and one hi-rail survey completed back
to back. This survey program facilitated further development of the search efficiency variable, but
did not allow for continued development of the carcass retention rate or carcass persistence
factor. Carcass persistence was calculated using data from the Year 7 and 8 surveys along the
Fraser Subdivision. The carcass retention rate developed using data from Years 3 through 6 on
the Telkwa Subdivision was used in development of the Fraser Subdivision correction factor this
year.
Margins of error for each variable were analyzed using the adjusted Wald method for binomial
proportions, with =0.05 (Agresti and Coull 1998). The relative uncertainties of each variable in
the correction factor were then added to determine the overall margin of error. While carcass
species identification error could not be included in the correction factor itself, the relative
uncertainty it presents (number of errors per carcass recorded) was added to the uncertainty level
of the correction factor. The potential impact of survey timing on spring survey accuracy, including

22 November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

variable snow conditions, was not incorporated into the correction factor, as its influence could
not be separated from that of other variables.

Uncertainty Analysis
The sensitivity of correction factor uncertainty was analyzed by comparing variation in the
uncertainty output from year to year and with the accumulation of mortality count data along the
Telkwa and Fraser Subdivisions (McElhanney 2016a). Correction factor outputs were developed
for the cumulative data compiled along the Telkwa Subdivision each year (Years 34, 35, 36,
etc.), and for the Fraser Subdivision data from Years 7 through 9. The relative uncertainties of
each correction factor output were then compared to the number of hi-rail recorded mortalities on
which each correction factor was based. This allowed uncertainty variation to be analyzed as
mortality data have accumulated over the course of the study.

Results
Fraser Subdivision Variables
Nine of the 14 carcasses recorded by hi-rail were not detected during the aerial survey; seven of
these were rated to have moderate or good visibility during the hi-rail survey, suggesting that they
should have been observed during the aerial survey. Combined with the previous two years of
analysis, this resulted in an aerial survey search efficiency, s, of 0.550.13. The carcass retention
factor, r, of 0.970.02, was carried over from previous analysis of Telkwa Subdivision data
(McElhanney 2014), while the carcass persistence, p, of 0.910.10, was carried over from
analysis along the Fraser Subdivision in Year 7 (McElhanney 2016b).
As in Years 6-8, carcass identification accuracy was analyzed but was not included in the
correction factor. One carcass was identified aerially as old but hi-rail results indicated that this
carcass was from Year 9. In two instances, hair piles that had been identified as deer carcasses
during the aerial survey were identified as moose carcasses during the aerial survey.
The preceding variables combined for a correction factor, FF, of 2.050.75 for Fraser Subdivision
aerial surveys. Application of the correction factor to the results of the aerial survey produced an
estimate of 238 moose-train collisions along the Fraser Subdivision in Year 9 (Table 3). This
estimate represents an increase from the Year 8 estimate of 145, but is within the margin of error
of the average for corrected collision levels along the Fraser Subdivision over the past three
winters (24.7).
Table 3. Application of Fraser Subdivision correction factor to results of Year 9 surveys.

Subdivision Count Method Result


Hi-rail 14
Aerial 11
Fraser
Hi-rail and Aerial 17
Aerial Corrected 238

23 November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

Uncertainty Analysis
During three years of correction factor development along the Fraser Subdivision, a total of 62
carcasses have been examined by hi-rail surveys. The error level of the correction factor
developed in Year 9 was 0.75. This represents a 37% relative uncertainty level, an increase from
the 35% relative uncertainty of the correction factor developed in Year 8, and a very slight
decrease from the 38% relative uncertainty in Year 7 (Table 4).
Table 4. Improvements in the relative uncertainty of correction factors for aerial mortality surveys
along the Fraser Subdivision as mortality data have accumulated from Year 6 forward.

Carcasses Correction Relative


Data Years
Examined Factor Uncertainty

Year 7 (2014-15) 37 1.810.68 38%


Years 7-8 (2014-16) 48 1.770.62 35%
Years 7-9 (2014-17) 62 2.050.75 37%

24 November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

Appendix II. Year 9 Moose-Train Collision Locations

25 November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

Figure 8. Map of Year 9 winter moose mortalities (red pins) recorded along CNs Telkwa Subdivision, between Endako
and Smithers, during the April 6, 2017 aerial survey, as well as exclusion fence segments (yellow lines).

26 November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

Figure 9. Map of the Fraser Subdivision, between McBride and Prince George, including Year 9 moose (red pins),
deer (yellow), and elk (green) mortalities recorded during hi-rail and aerial mortality surveys in April 2017.

27 November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

Appendix III. Telkwa Subdivision Scatterplots

Average Snow Depth Maximum Snow Depth


180 180
160 R = 0.0632 160
140 140
120 120
100 100
80 80 R = 0.4787
60 60
40 40
20 20
0 0
0 10 20 30 40 50 0 10 20 30 40 50 60 70 80

Days >40 cm Snow Total Snowfall


180 180
R = 0.0717
160 160
140 140
120 120
R = 0.6231
100 100
80 80
60 60
40 40
20 20
0 0
0 10 20 30 40 50 60 70 0 10 20 30 40 50

Figures 10-13. Scatterplots comparing weather variables from Environment Canada Smithers and Burns Lake weather stations (x-axes)
to annual moose-train collision levels (y-axes) along CNs Telkwa Subdivision in Years 3 (squares), 4 (exes), 5 (circles), 6 (diamonds),
7(triangles), 8 (long dashes) and 9 (short dashes).

28 November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

Train Traffic
180
160
140
120
100
80 R = 0.0122
60
40
20
0
0 200 400 600 800 1000 1200 1400 1600

Figure 14. Scatterplot comparing train traffic levels (x-axis) from Endako
switch at east end of Telkwa Subdivision to annual moose-train collision
levels in Years 3 (squares), 4 (exes), 5 (circles), 6 (diamonds), 7 (triangles),
8 (long dashes), and 9 (short dashes).

29 November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

Appendix IV. Fraser Subdivision Mortality Distribution and


Two-Mile Segment Ratings
Table 5. Two-mile rail segments on the Fraser Subdivision rated very low (1-3 total mortalities), low
(4-6), moderate (7-9) and high (10+) moose collision risk, based on CN-reported data (1996-2013)
and hi-rail and aerial mortality survey records from Years 6-9 (2013-2017). Miles with no recorded
mortalities were not included.
Mile Mile Historical Year 9 Years 6-9 Total Overall Telkwa
Start End Mortalities Mortalities Mortalities Mortalities Rating Rating
2 4 1 0 2 3 Very Low Low
8 10 2 0 2 4 Low Low
10 12 0 1 2 2 Very Low Low
14 16 2 0 2 4 Low Low
16 18 1 0 1 2 Very Low Low
18 20 4 0 1 5 Low Low
20 22 3 0 4 7 Moderate Low
22 24 2 0 0 2 Very Low Low
24 26 7 0 1 8 Moderate Low
26 28 3 0 0 3 Very Low Low
28 30 3 0 1 4 Low Low
30 32 8 0 1 9 Moderate Moderate
32 34 3 1 1 4 Low Low
34 36 7 1 1 8 Moderate Low
36 38 0 0 1 1 Very Low Nil
38 40 8 1 8 16 High Moderate
40 42 5 1 2 7 Low Low
42 44 6 0 3 9 Moderate Moderate
44 46 10 0 4 14 High Moderate
46 48 13 0 0 13 High Moderate
48 50 7 1 1 8 Moderate Low
50 52 2 0 2 2 Very Low Low
52 54 2 0 1 3 Very Low Low
54 56 0 1 2 2 Very Low Low
58 60 3 1 3 6 Low Low
60 62 3 1 1 4 Very Low Low
62 64 4 1 1 5 Low Low
64 66 1 0 0 1 Very Low Low
66 68 2 0 0 2 Very Low Low
68 70 3 0 0 3 Very Low Low
70 72 1 1 1 2 Very Low Low
72 74 1 0 0 1 Very Low Low

30 November 2017
Monitoring and Mitigation of Moose Rail Mortality in Northern BC: Year 9 (2017) Telkwa and Fraser Subdivisions

Mile Mile Historical Year 9 Years 6-9 Total Final Telkwa


Start End Mortalities Mortalities Mortalities Mortalities Rating Rating
74 76 8 0 0 8 Moderate Low
76 78 6 1 1 7 Low Low
78 80 7 1 1 8 Moderate Low
80 82 3 1 1 4 Very Low Low
82 84 4 0 0 4 Low Low
84 86 5 0 2 7 Moderate Low
86 88 4 0 1 5 Low Low
88 90 3 0 3 6 Low Low
90 92 0 0 1 1 Very Low Low
92 94 3 0 1 4 Low Low
94 96 5 0 0 5 Low Low
96 98 7 0 3 10 High Moderate
98 100 4 0 2 6 Low Low
100 102 2 0 2 4 Low Low
102 104 3 1 1 4 Very Low Low
104 106 4 0 0 4 Low Low
106 108 1 0 0 1 Very Low Low
108 110 1 0 0 1 Very Low Low
110 112 1 0 0 1 Very Low Low
112 114 1 0 0 1 Very Low Low
114 116 1 1 1 2 Very Low Low
116 118 3 0 0 3 Very Low Low
118 120 2 1 1 3 Very Low Low
124 126 1 0 0 1 Very Low Low
128 130 3 0 0 3 Very Low Low
130 132 4 0 0 4 Low Low
132 134 7 0 1 8 Moderate Low
134 136 1 0 0 1 Very Low Low
136 138 2 0 0 2 Very Low Low
138 140 2 0 0 2 Very Low Low
140 142 1 0 1 2 Very Low Low

31 November 2017

You might also like