Introduction
Like roads, railroads affect wildlife through direct mortality, habitat loss and habitat
fragmentation (van der Grift 1999; Forman et al. 2003; Davenport and Davenport
2006). Direct mortalities result when trains strike wildlife. Strikes can be a significant
source of mortality for some wildlife populations and have been reported for decades
(Child 1983; Gundersen et al. 1998; Bertch and Gibeau 2010a). Studies have reported
strike rates for large mammals such as grizzly bears (Ursus arctos) (Bertch and
Gibeau 2010a, b) and moose (Alces alces) (Child 1983, 1991; Modafferi 1991).
In Canada’s Rocky Mountain National Parks, train strikes are a leading source of
mortality for grizzly bears (0.35 year
−1 ) and black bears (1.95 year
−1 ), and the
second largest source of mortality for deer (Odocoileus spp.), elk (Cervus elaphus)
and moose (Bertch and Gibeau 2010a, b) in the Banff and Yoho National Parks. It is
likely that true mortality rates due to train strikes are higher than reported. For
example, as few as 50% of strikes with large mammals were reported by standard
observers (train engineers) along the Canadian Pacific Railroad (CPR) during a
six-year period (Wells et al. 1999). In other cases, strikes may not be reported
except when large groups (>450) of wildlife are killed (Chaney 2011) or when
strikes occur within protected areas (Waller and Servheen 2005). Long-term data of
train strikes along the CPR exist because strikes have been reported to Parks
Canada for at least 30 years. However, other railroads may not report or record
strikes with such consistency.
Studies on roads have analyzed the spatial pattern of road-kills, which showed
that these occurred in clusters (Finder et al. 1999; Clevenger et al. 2003; Malo et al.
2004). The spatial pattern of road-kills has been explained by landscape, environmental and infrastructure variables (Finder et al. 1999; Hubbard and Danielson
2000; Gunson et al. 2006; Kassar 2005). These studies have helped inform management actions targeted at reducing road-kills (Clevenger et al. 2001; Grilo et al.
2009). At least five general factors are thought to affect the spatial pattern of
road-kills and train strikes (Seiler and Helldin 2006). Along railroads, these include:
animal (e.g., wildlife abundance and behavior); train (e.g., train speed and frequency); railroad design (e.g., curvature or alignment); and landscape (e.g., vegetation type) variables (Huber et al. 1998; Bashore et al. 1985; Finder et al. 1999;
Seiler and Helldin 2006). Driver behavior variables are largely removed from train
strike analyses because trains generally cannot stop or swerve to avoid animals.
Landscape variables derived from land cover data have been found to be the best
predictors of road-kill rates (Bashore et al. 1985; Finder et al. 1999; Roger and Ramp
2009). We suggest that estimates of relative abundance be used to assess the spatial
pattern of road-kills and train strikes. The importance of including wildlife abundance in the analysis of factors contributing to wildlife strikes was demonstrated for
moose: strikes coincided with locations of high moose abundance in wintering areas
and on migration routes (Gundersen et al. 1998; Ito et al. 2008). Train variables and
railroad design, as well as animal abundance, vary along the CPR and may also affect
strike rates. If train or railroad variables altered the probability of a strike, the rate of
136
B.P. Dorsey et al.
Like roads, railroads affect wildlife through direct mortality, habitat loss and habitat
fragmentation (van der Grift 1999; Forman et al. 2003; Davenport and Davenport
2006). Direct mortalities result when trains strike wildlife. Strikes can be a significant
source of mortality for some wildlife populations and have been reported for decades
(Child 1983; Gundersen et al. 1998; Bertch and Gibeau 2010a). Studies have reported
strike rates for large mammals such as grizzly bears (Ursus arctos) (Bertch and
Gibeau 2010a, b) and moose (Alces alces) (Child 1983, 1991; Modafferi 1991).
In Canada’s Rocky Mountain National Parks, train strikes are a leading source of
mortality for grizzly bears (0.35 year
−1 ) and black bears (1.95 year
−1 ), and the
second largest source of mortality for deer (Odocoileus spp.), elk (Cervus elaphus)
and moose (Bertch and Gibeau 2010a, b) in the Banff and Yoho National Parks. It is
likely that true mortality rates due to train strikes are higher than reported. For
example, as few as 50% of strikes with large mammals were reported by standard
observers (train engineers) along the Canadian Pacific Railroad (CPR) during a
six-year period (Wells et al. 1999). In other cases, strikes may not be reported
except when large groups (>450) of wildlife are killed (Chaney 2011) or when
strikes occur within protected areas (Waller and Servheen 2005). Long-term data of
train strikes along the CPR exist because strikes have been reported to Parks
Canada for at least 30 years. However, other railroads may not report or record
strikes with such consistency.
Studies on roads have analyzed the spatial pattern of road-kills, which showed
that these occurred in clusters (Finder et al. 1999; Clevenger et al. 2003; Malo et al.
2004). The spatial pattern of road-kills has been explained by landscape, environmental and infrastructure variables (Finder et al. 1999; Hubbard and Danielson
2000; Gunson et al. 2006; Kassar 2005). These studies have helped inform management actions targeted at reducing road-kills (Clevenger et al. 2001; Grilo et al.
2009). At least five general factors are thought to affect the spatial pattern of
road-kills and train strikes (Seiler and Helldin 2006). Along railroads, these include:
animal (e.g., wildlife abundance and behavior); train (e.g., train speed and frequency); railroad design (e.g., curvature or alignment); and landscape (e.g., vegetation type) variables (Huber et al. 1998; Bashore et al. 1985; Finder et al. 1999;
Seiler and Helldin 2006). Driver behavior variables are largely removed from train
strike analyses because trains generally cannot stop or swerve to avoid animals.
Landscape variables derived from land cover data have been found to be the best
predictors of road-kill rates (Bashore et al. 1985; Finder et al. 1999; Roger and Ramp
2009). We suggest that estimates of relative abundance be used to assess the spatial
pattern of road-kills and train strikes. The importance of including wildlife abundance in the analysis of factors contributing to wildlife strikes was demonstrated for
moose: strikes coincided with locations of high moose abundance in wintering areas
and on migration routes (Gundersen et al. 1998; Ito et al. 2008). Train variables and
railroad design, as well as animal abundance, vary along the CPR and may also affect
strike rates. If train or railroad variables altered the probability of a strike, the rate of
136
B.P. Dorsey et al.