Can road mortality limit populations of pool-breeding amphibians? Wetlands Ecology and Management

Abstract

We integrated road maps, traffic volume data, and pool locations in a modeling study to estimate the potential effects of road mortality on populations of pool-breeding spotted salamanders (Ambystoma maculatum Shaw). Population projections based on spotted salamander life tables imply that an annual risk of road mortality for adults of >10% can lead to local population extirpation; mitigation efforts (tunnels, road closures, and other measures) should seek to reduce road mortality rates to below this threshold. For central and western Massachusetts, we estimated that salamanders would be exposed to at least this threshold level of risk at 22–73% of populations (assuming a 100 vs. 500m migration distance, respectively). We conclude that road mortality can be a significant source of additive mortality for individual spotted salamanders in many parts of the species’ range. Efforts to prevent such mortality by transportation planners are likely warranted strictly on a biological basis in areas with road densities >2.5km per km2 of landscape and traffic volumes >250 vehicles/lane/day within the migration range of a breeding population of spotted salamanders.

Full text

-1
Can road mortality limit populations of pool-breeding amphibians?
James P. Gibbs
1,
* and W. Gregory Shriver
2
1
College of Environmental Science and Forestry, State University of New York, 350 Illick Hall, 1 Forestry
Drive, Syracuse, NewYork 13210, USA;
2
U.S. National Park Service, Marsh-Billings-Rockefeller National
Historic Park 54 Elm Street, Woodstock, VT 05091, USA; *Author for correspondence (e-mail:
jpgibbs@syr.edu; phone: +1-315-470-6764; fax: +1-315-470-6934)
Received 9 September 2003; accepted in revised form 9 July 2004
Key words: Ambystoma maculatum, Conservation, Demography, Pool-breeding amphibian, Road mor-
tality, Spotted salamander
Abstract
We integrated road maps, traffic volume data, and pool locations in a modeling study to estimate the
potential effects of road mortality on populations of pool-breeding spotted salamanders (Ambystoma
maculatum Shaw). Population projections based on spotted salamander life tables imply that an annual risk
of road mortality for adults of >10% can lead to local population extirpation; mitigation efforts (tunnels,
road closures, and other measures) should seek to reduce road mortality rates to below this threshold. For
central and western Massachusetts, we estimated that salamanders would be exposed to at least this
threshold level of risk at 22–73% of populations (assuming a 100 vs. 500 m migration distance, respec-
tively). We conclude that road mortality can be a significant source of additive mortality for individual
spotted salamanders in many parts of the species’ range. Efforts to prevent such mortality by transportation
planners are likely warranted strictly on a biological basis in areas with road densities >2.5 km per km
2
of
landscape and traffic volumes >250 vehicles/lane/day within the migration range of a breeding population
of spotted salamanders.
Introduction
Road mortality of amphibians is a worldw ide
conservation concern (e.g., W yman 1991; Ashley
and Robinson 1996; Hels and Buchwald 2001),
and the United States is no exception. About one-
fifth of the land area of the conterminous United
States experiences some ecological effect of roads
and traffic (Forman 2000), an effect that is
increasing in concert with the expansion of the
road network and increase in traffic volumes
(National Research Council 1997).
Amphibians are one taxonomic group likely
sensitive to this expansion because they frequently
are killed on roads, particularly during the highly
synchronized overland migrations many pool-
breeding species undertake annually (Wyman
1991). Mortality of dispersing juveniles, as well as
adults, during the post-breeding period also occurs
and could be significant, but is less well recognized,
in large part because the phenomenon is more
diffuse in time and space and therefore more
difficult to study.
Amphibian migration is primarily nocturnal,
when traffic vo lumes are highly reduced relative to
their daytime rates (about 80% of traffic volume
occurs during the day: Festin 1996). Moreover,
amphibian demography, particularly that of anurans
Wetlands Ecology and Management (2005) 13: 281–289 Ó Springer 2005
DOI 10.1007/s11273-004-7522-9

(frogs and toads), is characterized by extraordi-
nary potential for population increase (e.g., Ber-
ven 1990). Thus, it is possible that amphibian road
mortality is an unfortunate yet demographically
insignificant phenomenon. The sparse research
conducted on the topic to date, however, suggests
otherwise.
Most research on the effects of road mortality on
amphibians has entailed simple tallies of numbers of
individuals killed at particular road crossing sites, in
some cases yielding surprisingly large counts. For
example, road-kill counts along a 3.6 km section of a
two-lane paved causeway in Ontario, Canada over
two seasons yielded >32,000 individual amphibians
(Ashley and Robinson 1996). Wyman (1991)
reported average mortality rates of 50.3 to 100% for
hundreds of salamanders attempting to cross a
paved rural road in New York State, USA.
Attempts to translate such counts of road-killed
animals into population-level assessments of the
consequences of road mortality are few. Hels and
Buchwald (2001) estimated that 10% of amphibian
populations near roads were killed on roads
annually at their study area in Denmark. Carr and
Fahrig (2001) concluded that road mortality exerts
more negative effects on populations of vagile
species of frogs than on sedentary (largely aquatic)
species in Ontario, Canada. In the same region,
Fahrig et a1. (1995) demonstrated that local frog
populations decreased with increasing traffic vol-
ume. Only Gittins (1983), studying co mmon toads
(Bufo bufo) in Great Britain, concluded that road
mortality had a minimal effect on populations.
Transportation policy makers are becoming
increasingly aware of the issue of wildlife mortality
and predisposed to addressing it (Forman et al.
2002). Unfortunately, we do not yet know whether
road mortality is sufficient to pose a significant risk
to the viability of local or perhaps even regiona l
amphibian populations. Moreover, the few, site-
specific studies of amphibian road mortality con-
ducted to date have not provided the type of
information needed to advance planning efforts.
Particularly useful would be studies that identify
threshold combinations of road density and traffic
volume that lead to unacceptably high levels of
road mortality in amphibian populations.
We implemented a modeling approach first
described by Gibbs and Shriver (2002) to examine
the magnitude of road mortality on pool-breeding
amphibians. The modeling approach integrates
maps of actual road networks, data on traffic
volumes, and simulated animal movements to
estimate the effects of road density and traffic
volume on animal demography. We used spotted
salamanders (Ambyst oma maculatum Shaw) as a
model organism because (1) they are wi despread in
the eastern United States, (2) they are frequently a
target of efforts to mitigate amphibian road mor-
tality, and (3) their demography, characterized by
extended generations times, and slow locomotion
suggest that they would be among the amphibians
most sensitive to additive sources of adult mor-
tality, such as that generated by roads (Wyman
1991; Petranka 1998).
Methods
Estimating mortality in relation to traffic volume
The probability that a salamander would be killed
as it attempted to cross a road, p
killed
, was esti-
mated based on an equation adapted from Hels
and Buchwald (2001):
p
killed
¼ 1  e
Na=v
where N = traffic volume in vehicles/min,
a = width of the kill-zone on a road (m), and
v = velocity (m/min) of the salamander moving
through the kill-zone. The kill-zone was estimated as
two times tire width (0.5 m total) per lane plus two
times total salamander body length (snout-to-tail
tip, see Hels and Buchwald 2001; data on mean adult
length obtained from Petranka 1998). A conserva-
tive value was used for v, salamander velocity, that
is, 1 m/min, based on our estimates of 12 adult A.
maculatum crossing a road (see next paragraph);
these animals averaged 1.3 m/min (±0.63 SD).
Assessing the validity of mortality estimates
We assessed the validity of our mortality estimates
in relation to traffic volume by measuring actual
salamander mort ality caused by fatal encounters
with vehicles on roads between March 15 and May
1 in 2003. We conducted surveys on New York
State Route 91 at Labrador Hollow (UTM 18
414312E, 4737775N [NAD27]) in Apulia, New
York. At this site, salamanders descend from the
forested slopes of Morgan Hill State Forest and
282

migrate across the highway en route to breeding
pools in the Labrador Hollow Unique Area. We
constructed 21, 3-m- long drift fences, spaced
approximately 47 m apart, on the breeding-pool
side of a 1 km length of highway (Figure 1). Drift
fences wer e located 1 m from the road and angled
10° toward the road. Each drift fence was outfitted
with one pitfall trap located in the center of the
drift fence. Pitfall traps were metal cans 0.5 m
deep and 7.5 cm in circumference. On six nights
with sufficient precipi tation to facilitate salaman-
der movement, we monitored salamander cross-
ings and vehicle passage. Observers repeatedly
walked along the highway from fence 1 to 21 (a
‘survey’ of ca. 30 min) and counted and removed
live and dead (or injur ed and presumably dying)
animals. Live and dead salamanders were tallied
(1) along segments of the highway between fences,
and (2) from the segments of the highway facing
drift fences where live animals captured in the drift
fence pitfalls were included in the count. This
sampling design provided two road crossing mor-
tality estimates for salamanders (calculated as
dead animals captured/total animals captured).
We estimated ‘tr ue’ mortality using salamanders
captured on each survey in segments of the road
facing drift fences and in drift fence pitfalls only
(‘true’ because all salamanders would either be
killed on the highway or captured in drift fences).
We also estimated an ‘inflated’ mortality using
road captures only (‘inflated’ mortality because all
animals killed were cou nted, but some live animals
escaped unnoticed between passes). Vehicle pas-
sage during surveys was recorded to the nearest
second to later estimate traffic volume during a
given survey period.
Estimating mortality in relation to road density
Annual road-associated mortality (d
road
) in spot-
ted salamanders was estimated as
d
road
¼ 1 ð1  p
killed
Þ
n
crossings
where p
killed
is the probability a salamander is
struck by a vehicle on any given road crossing (see
above) and n
crossings
is the number of road cross-
ings an individual salamander undertakes annu-
ally. The equation is derived from binomial
probabilities (Zar 1984 : 370–375) and estimates
the likelihood that a salamander is struck (and
presumably killed) during its cumulative road
crossings each year.
Figure 1. Sampling design used to estimate rates of road mortality of adult spotted salamanders in relation to traffic volume at
Labrador Hollow, Apulia, New York, March–May 2003.
283

We used spatially explicit sim ulations of move-
ments of individual salamanders on round-trip
movements between upland territories and breed-
ing pool areas to estimate the relationship between
n
crossings
and road density based on actual road
networks (Figure 2). Salamander movements were
simulated in landscapes within a 7224 km
2
region
of central and western Massachusetts (Figure 3),
which encompassed the Connecticut River Valley
as well as portions of the Berkshire Mountains. All
geographic data were obtained from MassGIS
(http://www.state.ma.us/mgis/massgis.htm) and
analyzed with ArcView 3.2 software and associ-
ated extensions. We randomly chose 500 breeding
pools within this area (from the data layer created
based on the survey conducted by Burne [2001]),
and counted the average number of times a single
salamander would have to cross a road on eight,
roundtrip, linear movements (to the N, NE, E, SE,
S, SW, W and NW) oriented upon the breeding
pool (Figure 2). Two dispersal distances were
evaluated – 100 and 500 m away from the pool –
to bracket the low and high estimates of average
adult salamander movement distances (125 m)
reported by Semlitsch (1998). Road density in a
corresponding circle of radius 100 or 500 m of
each breeding pool was also calculated by dividing
the total length of roads present by the area of the
circle (as km roads/km
2
of landscape). From these
data (n = 500 pairs of road density vs. salaman-
der road crossing frequency for each dispersal
distance) the relationship between road density
and average n
crossings
was estimated via least-
squares linear regression analysis (with intercepts
forced through zero).
Demographic significance of road mortality
We estimated mortality for each of the 500
breeding pools assessed in Massachusetts, using
the observed road density surrounding each site as
well as a statew ide, night-adjusted traffic volume
estimate. This value was estimated to be 160
vehicles/lane/day, based on the number of all
motor vehicles registered in Massachusetts multi-
plied by the annual miles traveled by an average
vehicle in the United States in 1998 divided by the
lane ‘mileage ’ in the stat e (FHA 1999). Traffic
volumes were reduce by 80% to conform to the
Figure 2. Example of simulated patterns of movement by an individual adult spotted salamander moving on 8 round-trips (straight
lines) from a breeding pool (open dot) and crossing (filled dots) roads (solid lines). In this case the average frequency of road crossing
was 1.75 (average of 0, 0, 0, 2, 2, 2, 4 and 4). Road density was subsequently estimated within a circular area of radius = migration
distance and centered upon the same pool. Simulated movements were subsequently replicated 500 times for two movement distances
(100 and 500 m away from pools).
284

proportion of overall traffic volume that occurs
between 1800 and 0600 h (Festin 1996), the peak
period of the salamanders’ nocturnal movements
(Petranka 1998).
We evaluated the potential significance of
additive road mortality to spotted salamander
population viability by integrating published esti-
mates of vital rates into a population projection
model that estimated the number of adults (N
a,t
)in
a given year (t) as:
N
a;tþ1
¼ N
a;t
 r
a
þ N
e;t
 r
m
ðK
l
 N
e;t
Þ=K
l
 r
j
where (N
e,t
) is the number of eggs produced in a
given year (t):
N
e;tþ1
¼ N
a;t
 r
a
u and
r
a
= adult annual survival rate, u
m
= average
eggs produced per individual, r
m
= survival rate
from egg to metamorphosis, r
j
= survival rate of
juveniles through their first winter, K
l
= the car-
rying capacity of larval habitat.
The mod el reflects limitation of adult popula-
tions through density dependent process in the
larval stage, typical of most pool-breeding
amphibians (see Vonesh and de la Cruz 2002).
Based on Petranka’s (1998) synopsis of the
scientific literature on spotted salamander biol-
ogy, annual adult survival rate, r
a
, was estimated
at 0.7, survival rate of egg to metamorphosis was
estimated at 0.04, and r
j
= survival rate of
juveniles through their first winter was estimated
at 0.6. The clutch size parameter, u, incorporates
number of eggs per mass (ca.80), number of egg
masses laid per female per year (ca.2), and annual
breeding probability (ca.0.38), all halved to reflect
contributions by both males and females. Last,
starting populations were set at 10,000 eggs and
100 breeding adults (an approximately stable age
distribution) and K
l
, carrying capacity of the
larval habitat, was set at 10,000. Under this sce-
nario, a spotted salamander population was
projected over 25 years under varying levels of
adult survival (r
a
= 0.4, 0.5 and 0.6), each
reflecting varying levels of additive road mortality
in relation to the scenario of no road mortality
(r
a
= 0.7).
Results
Estimating mortality in relation to traffic volume
Mortality estimates were highly and positively
correlated with volume of traffic that passed dur-
ing the same interval: p
killed
= 1.537*vehicles/min
0.0048, R
2
= 0.927, Figure 4).
Assessing the validity of mortality estimates
We co unted 330 live an d dead spotted sala-
manders on road crossing surveys. Mortality
rate at drift fences (‘true’ mortality rate) was
estimated at 19% (16 dead/70 live) whereas
mortality estimated from captures on the road
between drift fences (‘inflated’ mortality rate)
was 25% (61 dead/183 live). We used road
mortality estimates from 5 surveys along the
road during which we collected >10 animals
between drift fences, reduced these estimates by
6% (the inflation factor between true and
inflated estimates), and found that field esti-
mates of mortality were nearly identical
(Figure 4) to those predicted by the equation of
Hels and Buchwald (2001).
Figure 3. Area of central and western Massachusetts, USA,
where salamander movements were simulated in relation to
roads (upper). Circles represent the 500 randomly chosen
breeding pools mapped by Burne (2001) used in the analysis.
Arrow on map of the United States (lower) indicates location of
Massachusetts.
285

Estimating cross ing frequency in relation to road
density
The relationship between road density (RD) and
expected road-crossing frequency (n
crossings
)
(Figure 5) was linea r and positive for two migr a-
tion distances, with slope increasing but model fit
declining as dispersal distance increased: 100 m,
n
crossings
= 0.0658*RD (R
2
= 0.81) and 500 m,
n
crossings
= 0.356*RD (R
2
= 0.54).
Demographic significance of road mortality
Based on our probability model that integrated
traffic volume and road density, we estimated that
for the 500 breeding pools assessed in Massachu-
setts, the median road mortality rate was 17%
annually for animals moving 100 m and 37% for
animals moving 500 m. Population projections
indicated that a spotted salamander population
with a starting number of eggs = 10,000 (at a
breeding pool’s carrying capacity) and starting
number of adults = 100 would (after achieving a
stable age distribution) attain an equilibrium adult
population size of about 200 after 20 years
(Figure 6). Additive road-associated mortality of
10% woul d lead to a stable adult population,
albeit at a lower equilibrium level, of about 80
individuals (Figure 6). Road mortality levels of 20
and 30% would lead to population extirpation
(Figure 6) within 25 years. Thus, a threshold rate
Figure 4. Relationship between road mortality predicted on the basis of traffic volume, salamander rate of locomotion, and sala-
mander length (based on Hels and Buchwald 2001) vs. that estimated in the field during five surveys at Labrador Hollow, Apulia, New
York, in 2003.
Figure 5. Relationship between road density and estimated road crossing frequency for spotted salamanders moving at two distances
(100 and 500 m).
286

of somewhere between 10 and 20% likely occurs
above which population declines owing to road
mortality would be expected to ensue.
Discussion
Can road mortality limit populations of pool-
breeding amphibians? These analyses imply that
road mortality could be a significant source of
additive mortali ty for spotted salamanders in
central and western Massachusetts. Moreover, our
population projections, which integrate these road
mortality estimates with the demography of the
spotted salamander, indicate that road mortality
alone could lead to local population extirpation in
many areas. Our estimates are somewhat conser-
vative insofar as they consider only mortality
during breeding migrations whereas some move-
ments do occur outside the breeding season and
may further expose adults and migrating juveniles
to road mortality.
Notably, our population projections suggest
that the ‘dose-response’ curves between sizes of
salamander populations and levels of road mor-
tality are not a simple linear relationship. Scenar-
ios for patterns of local population decline with
increasing exposure to road mortality are likely
quite complex in amphibians and low levels of
road mortality may have little effect on salaman-
der populations. This is because many population
processes in amphibians are the result of density-
dependent interactions in the larval stage (e.g.,
Berven 1990). Thus, initial road-ca used declines in
the adult population may be partially compen -
sated for by increased juvenile recruitment, as
reduced input of eggs results in less larval crowd-
ing, and hence higher rates of metamorphosis (e.g.,
Vonesh and de la Cruz 2002). Once egg input of
adults drops below that needed for saturation
of larval habitats, however, the number of meta-
morphs produced and the number of breeding
adults will be linearly and positively related. Thus,
we expect the relationship between the intensity of
road mortality and populations of spotted sala-
manders is one in which salamander populations
remain stable in the face of increasing road mor-
tality and then go into rapid decline after a
threshold has been crossed, a pattern largely borne
out by the outputs of the population model
(Figure 6). Further research is needed to under-
stand these processes and estimate critical thresh-
olds of road mortality on salamander population
persistence.
Although we have validated a critical compo-
nent of the model, that is, road mortality in rela-
tion to traffic volume (Figure 4), substa ntial
uncertainties remain with the relationship between
road crossing frequency and road density. In
particular, our analysis assumed that salamanders
traverse the landscape in a linear fashion and cross
any road encountered. That salamanders navigate
completely independently of roads, however, is
naive. In fact, roadbeds and ditches may physically
Figure 6. Projections of a single spotted salamander population over 25 years under scenarios of adult survival = 0.7 (typical estimate
of adult survival in areas without heavy road mortality) vs. road-mortality reduced levels of adult survival = 0.6, 0.5 and 0.4. The
model reflects limitation of adult populations through density dependent processes in the larval stage, that is 10,000 eggs, with starting
population of 100 adults and 10,000 eggs, with the population achieving a stable age distribution between years 1 and 5.
287

prevent road crossings and the salamanders
themselves may avoid crossing roads. Some sala-
manders may remain on road surfa ces for
extended periods, either arrested physiologically
by debilitating compounds associated with salts
used to de-ice road surfaces or to linger upon the
warmer surfaces for purposes of thermoregulation.
Detailed behavioral studies of movements of
individual salamanders in relation to roads would
greatly clarify these issues and suggest ways to
model them more accurately. W e expect, for
example, that high levels of road mortality such as
those estimated do indeed occur where salaman-
ders must cross roads to reach vital habitats (e.g.,
isolated breeding ponds that limit population
recruitment in an area such as our study site in
Labrador Hollow) that are separated by roads
from adult habitat on uplands (e.g., Wyman 1991).
However, aversions to road crossing, coupled with
options for habitat choices, would deflate our
mortality esti mates to an unknown degree.
It must be emphasized that this study was per-
formed at a regional spatial scale to examine the
question of whether road mortality could lead to
the extirpation of regional A. maculatum popula-
tions. Mortality at any particular site could be
dramatically greater or less than what we estimate
depending on local road densities and traffic vol-
umes. For example, daily traffic volumes on ‘main
roads’ and other arteries in this part of the United
States can be on the range of 1000–10,000 vehicles
per day. Some restricted access highways carry
10,000–100,000 vehicles/day (FHA 1999). Roads
with such traffic volumes are wholly lethal to any
migrating salamander.
The applicability of our findings to other
amphibians and regions is unclear. Frogs and
toads move more quickly across roads and are
less vulnerable to road mortality (Wyman 1991).
Frogs may be more mobile, however, particularly
during the post-breeding period (e.g., Carr and
Fahrig 2001) and hence may undertake road
crossings at greater frequency over an entire
season. Frogs are also more active diurnally,
when traffi c volumes are dramatically higher
(Festin 1996). Even for other salamanders, vari-
ation in life hist ory may make application of
these results tenuous. For example, the red-spot-
ted newt (Notophthalmus viridescens) wanders for
years prior to sexual maturation during which
time it is likely highly vulnerable to road mor-
tality. Notably this species is among the most
sensitive to habitat fragmentation (Gibbs 1998)
and chronic road mortality may contri bute to this
sensitivity. Conversely, plethodontid salamanders
may be more limited in their movements than
Ambystoma and hence less vulnerable. From a
geographical perspective, these results are likely
applicable to other regions that are urbanized to
a similar degree as the region studied in central
Massachusetts (for exampl e, much of the north-
eastern United States excluding northern New
England and Northern New York).
To convert our analyses into policy relevant
information, we identified the combinations of
traffic vo lume and road density that produce a
predicted level of 25% or more additive mortality
associated with roads (Figure 7). Based on this
analysis, we estimated that combinations road
densities >2.5 km per km
2
of landscape and
traffic volumes >250 vehicles/lane/day within the
dispersal and migrati on range of a particular
breeding population of spotted salamanders
could generate demographically signi ficant mor-
tality levels (Figure 6). Although interactions
among salamander movement, road mortality,
and population persistence remain unclear, it is
apparent that conservation planning should
accommodate local movements and breeding
migrations of salamanders if local and perhaps
even regional populations are to remain secure.
Moreover, if efforts are successful in limiting
rates of traffic-caused mortality to <10% (See
Figure 6) of all individuals attempting to cross
Figure 7. Combinations of road density and daily traffic vol-
ume that produce an estimated level of annual road-associated
mortality of 25% in adult spotted salamanders migrating to two
different distances (100 and 500 m).
288

roads during migration circuit to a particular
pond, e.g., by tunnel construction, road closure,
or physically transporting individuals, then those
efforts are likely warranted to stave off local
population extirpation.
Acknowledgments
We are grateful to R. Brooks and P. Paton for
inviting us to the Vernal Pools Symposium at The
Wildlife Society Conference in Burlington,
Vermont, in 2003, thereby motivating us to
undertake this analysis. We are also thank them
for insightful comments and helpful editing, which
was similarly provided by two anonymous
reviewers. Many undergraduate and graduat e
students at SUNY-ESF kindly assisted with road
surveys at Labrador Hollow, particularly Joel
Strong, Debra Endr iss, Nancy Karraker, Sara
Ashkannejhad, Stephanie Smith, and Matthew
Young. Mary Clements of the New York State
Department of Transportation arranged permis-
sion and material support for us to conduct the
road surveys, for which we are grateful.
References
Ashley E.P. and Robinson J.T. 1996. Road mortality of
amphibians, reptiles and other wildlife on the Long Point
Causeway, Lake Erie, Ontario. Can. Field-Nat. 110: 403–412.
Berven K.A. 1990. Factors affecting population fluctuations in
larval and adult stages of the wood frog (Rana sylvatica).
Ecology 71: 1599–1608.
Burne M.R. 2001. Massachusetts Aerial Photo Survey of Po-
tential Vernal Pools. Natural Heritage and Endangered
Species Program, Westborough, MA, USA.
Carr L.W. and Fahrig L. 2001. Impact of road traffic on two
amphibian species of differing vagility. Conserv. Biol. 15:
1071–1078.
Fahrig L., Pedlar J.H., Pope S.E., Taylor P.D. and Wagner J.F.
1995. Effect of road traffic on amphibian density. Biol.
Conserv. 73: 177–182.
FHA (Federal Highway Administration). 1999. Highway Sta-
tistics – 1998. Office of Highway Policy Information, FHA,
Washington, DC, Available from http://www.fhwa.dot.gov/
ohim/hs98/hs98page.htm.
Festin S.M. 1996. Summary of National and Regional Travel
Trends: 1970–1995. Office of Highway Information Man-
agement, Federal Highway Administration, Washington,
DC, Available from: http://www.fhwa.dot.gov/ohim/blue-
book.pdf.
Forman R.T.T. 2000. Estimate of the area affected ecologically
by the road system in the United States. Conserv. Biol. 14:
31–35.
Forman R.T.T., Sperling D. and Swanson F.J. 2002. Road
Ecology: Science and Solutions. Island Press, Covelo, Cali-
fornia, USA.
Hels T. and Buchwald E. 2001. The effect of road kills on
amphibian populations. Biol. Conserv. 99: 331–340.
Gibbs J.P. 1998. Distribution of woodland amphibians
along a forest fragmentation gradient. Landscape Ecol.
13: 263–268.
Gibbs J.P. and Shriver W.G. 2002. Estimating the effects of
road mortality on turtle populations. Conserv. Biol. 16:
1647–1652.
Gittins S.P. 1983. Population dynamics of the common toad
(Bufo bufo) at a lake in mid-Wales. J. Anim. Ecol. 52: 981–
988.
National Research Council. 1997. Towards a Sustainable Fu-
ture: Addressing the Long-term Effects of Motor Vehicle
Transportation on Climate and Ecology. National Academy
Press, Washington, DC, USA.
Petranka J.W. 1998. Salamanders of the United States and
Canada. Smithsonian Institution Press, Washington, DC,
USA.
Semlitsch R.D. 1998. Biological delineation of terrestrial buffer
zones for pond-breeding salamanders. Conserv. Biol. 12:
1113–1119.
Vonesh J. and de la Cruz O. 2002. Complex life cycles
and density dependence: assessing the contribution of
egg mortality to amphibian declines. Oecologia 133: 325–
333.
Wyman R.L. (ed.) 1991. Global Climate Change and Life on
Earth. Routledge, Chapman and Hall, New York, NY, USA.
Zar J.H. 1984. Biostatistical Analysis. 2nd edn. Prentice Hall,
NJ, USA.
289