Selecting the Spatio-Temporal Scale
The rationale for selecting the spatio-temporal scales depends on the species to be
monitored (Roedenbeck et al. 2007; van der Grift et al. 2013). Large animals with
large home ranges and daily movements require the survey of a railway stretch
large enough (e.g., 30 km) to reflect their spatial requirements (Iuell et al. 2003;
Seiler and Helldin 2006). By contrast, if we aim to study a local amphibian population during seasonal migrations, a 1 km railway stretch can be reasonable (Hels
and Buchwald 2001). Another important issue is the habitat preference. If the target
species is forest-dwelling, we must focus on stretches crossing forests (Iuell et al.
2003). For more general species, the railway stretch selection should cover and
reflect a hierarchical distribution and abundance of the existing habitats (van der
Grift et al. 2013) or, alternatively, we may choose different railway stretches, each
representing a different habitat (Roedenbeck et al. 2007). If a species shows clear
seasonal movements with a high probability of railway crossings at certain periods
of the year (e.g., amphibian breeding season migrations), then wildlife mortality
studies should focus on those periods.
Estimating the Number of Casualties on Railways
After assessing the baseline situation in a given region, the following guidelines are
important to achieve an appropriate monitoring plan. In electrified railways, besides
the mortality due to train-collisions (almost all vertebrate groups), we also expect
mortality due to overhead electric line collisions (birds and bats) and electrocution
at cables and pylons (mainly birds). Once the sources of mortality are established,
we should adopt the most practical prospecting methods (e.g., on foot, on foot with
a search dog, or using a motor vehicle) according to the railway features (width,
train speed and volume, vegetation on verges, topography, etc.) (Dorsey et al. 2015;
SCV 1996).
Counting dead animals along railways is more challenging than on roads
because railways often cross remote areas and their accessibility is often difficult
(tunnels, steep topography, etc.) (Dorsey et al. 2015; Wells et al. 1999). Therefore,
most studies report counts obtained by the transportation agency personnel, such as,
train drivers and maintenance workers, who often lack wildlife experience, leading
to inaccurate identifications and underestimation of the mortality (Wells et al.
1999).
An important issue is the sampling effort required in each situation to effectively
detect patterns of causalities (Costa et al. 2015; Santos et al. 2011, 2015). As
recommended for roads, railway surveys should be carried out early in the morning
(to reduce scavenging, but see also sources of bias below), preferably by two
experienced observers walking at specified railway stretches, one on each side of
the rail (Peña and Llama 1997), and covering a 10 m sight strip whenever logistically possible. The use of a vehicle could be a better choice if parallel dirty roads
exist (at least on one side), and in cases where the surveyed stretch is too long
26
F. Carvalho et al.
The rationale for selecting the spatio-temporal scales depends on the species to be
monitored (Roedenbeck et al. 2007; van der Grift et al. 2013). Large animals with
large home ranges and daily movements require the survey of a railway stretch
large enough (e.g., 30 km) to reflect their spatial requirements (Iuell et al. 2003;
Seiler and Helldin 2006). By contrast, if we aim to study a local amphibian population during seasonal migrations, a 1 km railway stretch can be reasonable (Hels
and Buchwald 2001). Another important issue is the habitat preference. If the target
species is forest-dwelling, we must focus on stretches crossing forests (Iuell et al.
2003). For more general species, the railway stretch selection should cover and
reflect a hierarchical distribution and abundance of the existing habitats (van der
Grift et al. 2013) or, alternatively, we may choose different railway stretches, each
representing a different habitat (Roedenbeck et al. 2007). If a species shows clear
seasonal movements with a high probability of railway crossings at certain periods
of the year (e.g., amphibian breeding season migrations), then wildlife mortality
studies should focus on those periods.
Estimating the Number of Casualties on Railways
After assessing the baseline situation in a given region, the following guidelines are
important to achieve an appropriate monitoring plan. In electrified railways, besides
the mortality due to train-collisions (almost all vertebrate groups), we also expect
mortality due to overhead electric line collisions (birds and bats) and electrocution
at cables and pylons (mainly birds). Once the sources of mortality are established,
we should adopt the most practical prospecting methods (e.g., on foot, on foot with
a search dog, or using a motor vehicle) according to the railway features (width,
train speed and volume, vegetation on verges, topography, etc.) (Dorsey et al. 2015;
SCV 1996).
Counting dead animals along railways is more challenging than on roads
because railways often cross remote areas and their accessibility is often difficult
(tunnels, steep topography, etc.) (Dorsey et al. 2015; Wells et al. 1999). Therefore,
most studies report counts obtained by the transportation agency personnel, such as,
train drivers and maintenance workers, who often lack wildlife experience, leading
to inaccurate identifications and underestimation of the mortality (Wells et al.
1999).
An important issue is the sampling effort required in each situation to effectively
detect patterns of causalities (Costa et al. 2015; Santos et al. 2011, 2015). As
recommended for roads, railway surveys should be carried out early in the morning
(to reduce scavenging, but see also sources of bias below), preferably by two
experienced observers walking at specified railway stretches, one on each side of
the rail (Peña and Llama 1997), and covering a 10 m sight strip whenever logistically possible. The use of a vehicle could be a better choice if parallel dirty roads
exist (at least on one side), and in cases where the surveyed stretch is too long
26
F. Carvalho et al.