Assessing large-scale wildlife responses to human infrastructure development

Abstract

Significance
Nature is increasingly threatened by rapid infrastructure expansion. For the first time, to our knowledge, we quantify the high pervasiveness of transportation infrastructure in all European countries. Unfortunately, spatial definition of the areas ecologically affected by infrastructure at large scales is complicated. Thus, we present a method for assessing the spatial extent of the impacts on birds and mammals at regional and national scales. As an illustration, its application to Spain shows that most of the country is affected, predicting moderate and severe declines for birds and mammals, respectively. The lack of areas that could be used as controls implies that scientists may no longer be able to measure the magnitude of road effects on wide-ranging mammals in most of Europe.

Full text

Assessing large-scale wildlife responses to human
infrastructure development
Aurora Torres
a,1
, Jochen A. G. Jaeger
b
, and Juan Carlos Alonso
a
a
Departamento de Ecología Evolutiva, Museo Nacional de Ciencias Naturales, Consejo Superior de Investigaciones Científicas (CSIC), E-28006 Madrid, Spain;
and
b
Department of Geography, Planning, and Environment, Concordia University, Montreal, QC, Canada H3G 1M8
Edited by Rodolfo Dirzo, Stanford University, Stanford, CA, and approved May 25, 2016 (received for review November 13, 2015)
Habitat loss and deterioration represent the main threats to wildlife
species, and are closely linked to the expansion of roads and human
settlements. Unfortunately, large-scale effects of these structures
remain generally overlooked. Here, we analyzed the European trans-
portation infrastructure network and found that 50% of the conti-
nent is within 1.5 km of transportation infrastructure. We present
a method for assessing the impacts from infrastructure on wildlife,
based on functional response curves describing density reductions in
birds and mammals (e.g., road-effect zones), and apply it to Spain as a
case study. The imprint of infrastructure extends over most of the
country (55.5% in the case of birds and 97.9% for mammals), with
moderate declines predicted for birds (22.6% of individuals) and
severe declines predicted for mammals (46.6%). Despite certain limi-
tations, we suggest the approach proposed is widely applicable to the
evaluation of effects of planned infrastructure developments under
multiple scenarios, and propose an internationally coordinated strat-
egy to update and improve it in the future.
anthropogenic development
|
birds
|
Europe
|
mammals
|
road-effect zone
Habitat loss and degradation are the primary drivers of the
decline and extinction of wildlife populations in terrestrial
ecosystems (1), with the main precursors of these impacts being
roads and human settlements (2). If current trends continue, by
2030, urban areas will increase by 1.2 million km
2
globally and,
by 2050, our planet will accommodate more paved-lane kilo-
meters than required to reach Mars (3, 4). The largest expected
infrastructural undertakings will occur in developing nations (3,
4), including many regions that sustain exceptional levels of
biodiversity and vital ecosystem services. These structures will
alter ecological conditions, cut through highly suitable habitat,
and further reduce the populations of many wildlife species (5–
7). However, large-scale consequences of these trends remain
unknown (8). Global and continental schemes for prioritizing
road building have recently been proposed to limit the envi-
ronmental costs of infrastructure expansion while maximizing its
benefits for human development (9, 10). The refinement of these
zoning plans would greatly benefit from more detailed estimates
of the imprint of infrastructure on wildlife populations. Human
footprint models combine spatial data regarding human activities
with assessments of their effects to estimate their overall impact
(11–13). The burgeoning availability of detailed geospatial layers
of infrastructure contrasts with the lack of quantification of their
effects, which still relies on expert knowledge and is mostly based
on single species or local studies (14). As a result, mapping of the
area of influence of infrastructure ranges from a few hundred
meters (15) up to 50 km (10, 11, 16, 17).
The main difficulty in quantifying the area of influence of
infrastructure on wildlife, that is, the area over which the eco-
logical effects extend into the adjacent landscape [e.g., “road-
effect zone”(2)] has been the lack of reliable distance thresholds
for these effects (18). Most effects on local species abundances
occur within a specific distance from the infrastructure and level
off as distance increases (19, 20). For instance, this decrease in
population density varies by taxonomic class, with mammals
being affected over larger distances than birds (21).
The objective of our work is to assess the spatial extent of the
impacts from infrastructure on wildlife populations at a large scale,
based on taxa-specific functional distance-decay curves (Fig. 1). We
first examine the pervasiveness of transportation infrastructure in
Europe, a continent with extensive data and broad variability in
both infrastructure development and wildlife distribution, and
then, using Spain as an example, we explore how the pervasiveness
of infrastructure translates into the distribution of six emblematic
species of the Iberian fauna, pointing out large-scale effects and
strengthening the evidentiary basis of impact assessments on
wildlife at regional or national scales. Finally, we present a method
to model the area of influence of infrastructure and apply it for
birds and mammals in Spain. Worldwide, the Mediterranean Basin
is the biodiversity hotspot most affected by urban expansion (4);
thus, our results for Spain may help predict the level of threat for
other biodiversity hotspots undergoing rapid development.
Our results reveal both the pervasiveness of human infrastructure
and its negative influence on wildlife populations, particularly
among wide-ranging mammals. Despite its limitations, our approach
may represent a useful tool for conservation and land management,
enabling (i) assessments of the human footprint of infrastructure
or wilderness mapping, (ii) the definition of roadless areas, and
(iii) projections of future human influence under alternative sce-
narios, as well as supporting strategic infrastructure planning.
Results
How Far to the Nearest Infrastructure? Almost a quarter of all land
area in Europe (22.4%) is located within 500 m of the nearest
transport infrastructure, and 50% is within 1.5 km (Table S1).
For the EU-28 (the 28 member states currently forming the
European Union), these numbers are almost identical (22.8%
and 1.5 km, respectively). Ninety-five percent of all Europe is
Significance
Nature is increasingly threatened by rapid infrastructure ex-
pansion. For the first time, to our knowledge, we quantify the
high pervasiveness of transportation infrastructure in all Eu-
ropean countries. Unfortunately, spatial definition of the areas
ecologically affected by infrastructure at large scales is com-
plicated. Thus, we present a method for assessing the spatial
extent of the impacts on birds and mammals at regional and
national scales. As an illustration, its application to Spain
shows that most of the country is affected, predicting moder-
ate and severe declines for birds and mammals, respectively.
The lack of areas that could be used as controls implies that
scientists may no longer be able to measure the magnitude of
road effects on wide-ranging mammals in most of Europe.
Author contributions: A.T. de signed research; A.T. performe d research; A.T. analyzed
data; and A.T., J.A.G.J., and J.C.A. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
1
To whom correspondence should be addressed. Email: aurora.torres@mncn.csic.es.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
1073/pnas.1522488113/-/DCSupplemental.
www.pnas.org/cgi/doi/10.1073/pnas.1522488113 PNAS Early Edition
|
1of6
ECOLOGYENVIRONMENTAL
SCIENCES

located within 9.2 km of a transport infrastructure (within 8 km in
the case of EU-28), with the farthest distances in Iceland (83.5 km).
The densest transport network is located in Central Europe,
particularly in the three Benelux countries (Belgium, The Neth-
erlands, and Luxembourg; Fig. 2), whereas landscapes with low
road density are located in northern latitudes and in areas with
large mountain ranges (Alps and Carpathians). Spain stands out as
the country with the highest median and average distances to
transport infrastructure (1.9 and 2.7 km, respectively), with the
exception of most of the northern countries, namely, Iceland,
Norway, Estonia, Finland, Latvia, and Lithuania, as well as the
Principality of Andorra. This median distance is almost halved
when using the more precise Base Cartográfica Nacional (BCN100;
centrodedescargas.cnig.es/CentroDescargas/index.jsp) (869 m) in-
stead of EuroGlobalMap (EGM; www.eurogeographics.org/content/
euroglobalmap), revealing the underrepresentation of transport in-
frastructure in the EGM. Aside from transportation infrastructure,
50% of all land area in Spain is located within 1.6 km of the nearest
built-up area and within 718 m from the nearest impervious in-
frastructure (Fig. 3). Most land is located near infrastructure, and the
proportion of land added to the accumulation curve rapidly becomes
smaller as the distance increases, so 99% of Spanish land is within
7.6, 6.4, and 5.2 km from a built-up area, transport corridor, and
impervious infrastructure, respectively, whereas the farthest locations
are at 15.4, 16.6, and 13.4 km, respectively.
Regarding the effects of proximity to infrastructure on em-
blematic species, the distribution maps of all six species show the
highest number of cells with positive presence data within the
second band (at 500–1,000 m from the infrastructure) (Fig. 4).
However, prevalence shows differences between taxa; higher
values at increasing distances to transport infrastructure in the
Spanish imperial eagle, Iberian lynx, and Brown bear; and no
clear pattern in the Tawny owl, Great bustard, and Gray wolf.
What Is the Area of Influence of Infrastructure on Birds and Mammals
in Spain? The area of influence of infrastructure, as reflected by
a mean species abundance (MSA)<0.95 compared with non-
disturbed distances, covers 55.5% [confidence interval (CI) =
48.3–64.4%] of the country in the case of birds, and extends over
almost all of Spain for mammals (97.9%, CI =95.1–99.2%). The
results for transportation infrastructure alone are very similar
(birds: 49.4%, CI =42.6–58.0%; mammals: 95.8%, CI =91.8–
98.2%). For birds, spatial clusters of low MSA values are clearly
observed, but many large unaffected areas remain available (Fig.
5A), whereas for mammals, low MSA values prevail across Spain
(Fig. 5B;MSA values for transport infrastructure alone are shown
in Fig. S1). These MSA values predict an average decline of 22.6%
(CI =16.7–29.7%; for transport infrastructure alone: 19.0%, CI =
9.6–25.6%) in bird numbers and 46.6% (CI =33.0–60.7%; for
transport infrastructure alone: 42.9%, CI =29.6–56.9%) in mam-
mal numbers compared with the undisturbed situation.
Are All Habitats Similarly Affected? Although all habitat types
showed similar patterns of proximity to human infrastructure,
some differences were observed (Fig. 6A). Farmland is most af-
fected by transport infrastructure and built-up areas, and the
lowest MSA values are found here (mean ±SD =0.729 ±0.277
and 0.496 ±0.168 for birds and mammals, respectively; Fig. 6B).
The second most affected habitat is wetlands (birds: mean ±SD =
0.790 ±0.254; mammals: 0.539 ±0.176,), due mostly to the in-
fluence of maritime wetlands (Table S2). Forests and scrublands
share similar effect values, whereas bare lands are the least af-
fected. In the remotest locations (beyond 10 km to impervious
areas), the differences among habitats are more evident. Those
locations mainly correspond to bare rocks (32.8%), natural grass-
lands (23.6%), and sclerophyllous vegetation (22.9%).
Discussion
In Europe, half of the continent’s surface is located within 1.5 km,
and almost all land within 10 km, from a paved road or a railway
line. Riitters and Wickham (22) reported shorter distances to the
nearest road in the United States, where 50% of the land was
within 382 m of a road (compared with 869 m in Spain). However,
the US road map at that time included unpaved and private roads.
Fig. 1. Relationships between MSA of birds and mammals and distance to
infrastructure obtained by Benítez-López et al. (21) through metaregressions
and used in the present study to model the area of influence of in-
frastructure in Spain. Solid lines represent the MSA curve estimated for birds
(gray) and mammals (black) as a function of distance to infrastructure.
Dashed lines represent the 95% confidence bands for the predictions.
0300 600 km
0
5
10
≥ 50
Distance to the nearest
transport corridor (km)
Fig. 2. Mapped dis tances to the nearest transport in frastructure (pave d roads
and railways; details are providedin Table S3) in Europe (36 countries; Table S1)
based on the small-scale pan-European topographic dataset EGM v7.0 (2014),
using a Lambert azimuthal equal area projection. Distances were quantified at
a resolution of 50 m for inland Europe and islands larger than 3,000 km
2
and
ranged from 0 to 83.5 km.
2of6
|
www.pnas.org/cgi/doi/10.1073/pnas.1522488113 Torres et al.

Another reason for the difference is that over much of the less
developed United States, the road system results from the original
subdivision of land into rectangular ownership parcels with roads
regularly spaced along owner boundaries (23). Thus, the US sys-
tem was really designed to minimize the distance to the nearest
road. Given that the more accurate input map of paved roads and
railway lines in Spain halved the estimated distance obtained from
the EGM, we consider the European estimates to be very con-
servative. However, the observed patterns are consistent with
previous measurements of landscape fragmentation, urban sprawl,
and wilderness areas (24–26).
Spain is one of the European countries less affected by road-
mediated effects and where many roadless areas can still be rec-
ognized; however, on the other hand, this country is under a high
human footprint from a global perspective (27). All of our example
species were more frequently distributed at relatively close dis-
tances to transportation infrastructure, because most of the land is
located at such short distances (Fig. 3), so wildlife does not have
many options to occupy remote areas. Even so, the first 500-m
band is systematically being “avoided”by four species with differ-
ent ecological requirements and functional traits (Great bustard,
Spanish imperial eagle, Iberian lynx, and Brown bear), even though
a high percentage of land is available within that band (Fig. S2).
Given that these analyses are based on occurrences, and that the
presence cells in the first 500-m band probably hold lower numbers
of individuals than presence cells in subsequent bands, these four
species would not only be found farther from infrastructure if land
at such distances were available but could also be less abundant in
cells that are closer to infrastructure. Also, the increasing preva-
lence of some species with higher distances to transport infra-
structure (Spanish imperial eagle, Iberian lynx, and Brown bear)
suggests that they prefer remote sites or that they were better able
to persist there in past times of strong direct persecution. These
detrimental effects at large scale illustrate the high level of exposure
for wide-ranging carnivores, like the critically endangered Iberian
lynx, for which road casualties are a major mortality cause (20
road-kill mortalities in 2014 in a total population of ca. 320 in-
dividuals; www.iberlince.eu/index.php/port/). In contrast, the Tawny
owl and the Gray wolf are known to use areas next to roads (28, 29),
whereas the Great bustard is characteristic of cereal farmland, a
habitat strongly pervaded by infrastructure (Fig. 6).
Area of Influence of Human Infrastructure for Birds and Mammals.
Proximity to infrastructure contributes to average decreases by
25% and 50% compared with the undisturbed situation in birds
and mammals, respectively, based on data from Benítez-López
et al. (21). Moreover, in the case of mammals, there is almost no
area left unaffected from transport infrastructure. For road
ecology, this result implies that researchers may no longer be
able to measure the whole extent of road effects on wide-ranging
mammals as well as birds with large effect-distances, because
core areas of significant size that could be used as controls are
now almost inexistent, and this implication extends to most of
Europe and a sizeable part of the United States (30) (Fig. 2).
Farmland plays an important role in the conservation of bio-
diversity throughout Europe, with more than half of all species
depending on this habitat type (31). We found the effects from
impervious infrastructure to be more evident in farmlands, so
this threat may also be contributing to the biodiversity decline
Fig. 3. Accumulation curves for the proportion of total land area in Spain
located within a certain distance from the nearest built-up area, transport
infrastructure (paved roads and railways), and impervious infrastructure (in-
cluding built-up areas, transport infrastructure, and other sealed surfaces).
Fig. 4. Level of exposure to human infrastructure varies throughout a
species’distribution, which we illustrate by considering the distributions of
six emblematic species of the Mediterranean fauna. The bars (Left,yaxis)
indicate the proportions of each species’distribution found within each
500-m distance band to transport infrastructure (xaxis), whereas the blue
dots (Right,yaxis) indicate the prevalence for each band (i.e., the ratio
between the number of cells in which the species was present divided by the
total number of cells available at such distances in peninsular Spain).
Torres et al. PNAS Early Edition
|
3of6
ECOLOGYENVIRONMENTAL
SCIENCES

that has mostly been associated with the agricultural intensification
process (32, 33). Moreover, in farmlands and other open-habitat
types like bare lands, the infrastructure imprint is potentially wider
than what our results indicate because of the higher visibility of
infrastructure (14, 21). A future meta-analysis should determine
the specific distance decay functions for different types of habitat
once enough data are available.
Areas characterized by a low imprint of infrastructure may
clearly be priority sites for protecting roadless areas (17, 34).
However, some places still hosting important biodiversity are no
longer in remote areas, suggesting that extinction debts are likely.
In this regard, the reductions predicted for birds and mammals are
inherently based on how we have managed wildlife over the past
decades in the affected areas. Hence, areas with a high imprint of
infrastructure have become challenges for conservation planning,
where potential extinctions (which are most likely debts at pre-
sent) should be prevented by reinforcing remnant populations and
restoring vital ecological processes.
Applicability of the Approach and Next Steps. The approach explained
here for Spain provides the most detailed picture obtainable now-
adays of the magnitude and spatial distribution of infrastructure-
induced effects on birds and mammals, is readily transferable
to other places, and can contribute to future regional and national
infrastructure planning. However, it has certain limitations:
(i) geographic bias, (ii) undistinguished effects of different
infrastructure types, and (iii) low inferential strength of the studies
considered in the meta-analysis. There is a major geographic
bias in the research conducted about the impacts of roads on
wildlife, with vast areas of the globe being largely ignored (35).
This aspect is not a major problem for the present study be-
cause species from Europe are well represented in Benítez-
López et al.’s meta-analysis (21), but the applicability of this
approach beyond Europe and North America may be limited.
As for the second limitation cited above, previous studies have
found different effect distances for different road types or
traffic levels (36), which would affect the accuracy of estimates.
However,thereisstillasubstantial debate around this topic;
thus, we decided to ignore differences between infrastructure
types to retain consistency with Benítez-López et al. (21), who
did not find a significant difference. Finally, most studies used in
the meta-analysis followed a control-impact study design, by
comparing bird and mammal numbers in the impacted area with a
reference state. Although this design is widely used to quantify
impacts from a variety of pressures (e.g., 37), it has lower in-
ferential strength than a before-after-control-impact (BACI) de-
sign (38). Unfortunately, due to time and logistical constraints, the
proportion of BACI-designed studies is still very small (39).
Most of the urban development and more than one-third of the
transportation infrastructure expected to exist by 2050 are not yet
built (3, 4). Nine-tenths of all road construction in the coming 40 y
is expected to occur in developing nations (3, 4) and to be aimed at
improving the conditions of large human populations with low
average incomes. Infrastructure-mediated impacts are expected to
be most damaging in species-rich ecosystems, such as tropical
forests, where few roads currently exist (9, 40). Our approach can
be used in those areas for regulating the expansion of new in-
frastructure, supporting regional planning and road development
schemes, and increasing the efforts to mitigate their detrimental
effects. As infrastructure building progresses, it will be increasingly
difficult to quantify its effects, because the core areas that can be
used as control sites will be rare and more isolated. Therefore,
there is a trade-off between the uncertainty of using effect mea-
sures from studies with low inferential strength and the urgent need
to respond to rapid development using the evidence available to-
day, in consideration of the precautionary principle. We propose to
overcome, at least partially, the weaknesses of our approach
through regular updates of the wildlife-response meta-analysis
(21). The addition of new species’datasets would allow fine-tuning
of the parameters of the response functions, as well as revealing the
differences among habitat types. Moreover, the investigation of
groups of species with similar functional traits that may provide
new response functions would be a useful means of developing the
applicability of this study further, when conservation needs to be
focused on particular taxa or wildlife communities or where there
are fewer data available. In general, large-sized mammals with
lower reproductive rates and larger home ranges are more sus-
ceptible to negative road effects (41), but for tropical areas, we
would expect larger effect distances on apex predators, large-sized
mammals and birds, and forest specialists because of their marked
avoidance behavior (40). As a first step, we have conducted a re-
view of five major traits, namely, body mass, home range size, re-
productive rate, longevity, and trophic level, of the 232 species
included in the study of Benítez-López et al. (21). By creating this
database (available in Dataset S1) we intend to ease the way for
broader application of the insights derived from this study and give
impetus to further applied research in developing regions, which
are in great need of solutions and increased representation (7, 42).
In moving forward, we are making a call to scientists and practi-
tioners to coordinate a database and network of studies about
infrastructure-mediated impacts on wildlife populations across
A
B
Fig. 5. Predicted MSA of birds (A) and mammals (B) across Spain (Left;two
large maps) according to proximity to human infrastructure, based on the
effect distance-decay curves fitted for empirical data by Benítez-López et al.
(21). (Right) Adjacent smaller maps represent the upper (Top) and lower
(Bottom) CIs. MSA layers were reclassified into six effect intensity zones for
representation. (Upper Right) Small map showing the location of five major
cities is included for reference.
4of6
|
www.pnas.org/cgi/doi/10.1073/pnas.1522488113 Torres et al.

ecosystems and geographical areas (43) and to make use of this
approach as a powerful conservation planning tool.
Materials and Methods
Distance Analysis. We measured proximity to transportation infrastructure in
inland Europe (and islands larger than 3,000 km
2
, as well as Malta) based on
the EGM v7.0 (1:1,000,000 scale; EuroGeographics, 2014), a pan-European
open dataset containing seamless and harmonized geographic information.
We exclusively considered paved roads and railway lines, excluding aban-
doned and underground sections (Table S3). We then calculated Euclidean
distances to the nearest transport infrastructure for 36 countries, at a res-
olution of 50 m.
The consistency of the EGM database was assessed against the most recent
and precise geographical information system (GIS) database of transportation
infrastructure for Spain (BCN100, 1:100,000 scale; National Geographic Institute
of Spain, 2014; Table S4). In addition, we measured the pervasiveness of built-
up areas and all infrastructure combined. We used the Spanish Land Cover and
Use Information System (1:25,000 scale; National Geographic Institute of Spain,
2005; www.siose.es) to create the map of built-up areas (Table S5) and other
impervious infrastructure (e.g., parking lots, irrigation ponds; Table S5). All
maps were converted to raster format (15 m). For each cell, we calculated the
Euclidean distance to the nearest transport infrastructure, built-up area, and all
impervious infrastructure combined. We were not able to calculate distances
for Europe and Spain for even higher resolution because of computational
limitations for smaller pixel sizes.
Effects of Proximity to Transportation Infrastructure on Species Distribution.
We overlaid distance maps to transportation infrastructure with distribution
maps (10 ×10-km cells) (44) of six emblematic species of the Iberian fauna
known to be neg atively affected by roads at local scales: Strix aluco (Tawny
owl), Otis tarda (Great bustard), Aquila adalberti (Spanish imperial eagle),
Canis lupus (Gray wolf), Lynx pardinus (Iberian lynx), and Ursus arctos (Brown
bear) (28, 29, 38, 45–47). For each species, we quantified the median distance
to transport infrastructure in presence cells and classified resulting dis-
tances by bands of 500 m from the nearest infrastructure for graphical
representation as a normalized histogram. Most wildlife species affected
by human development have escape distances on this order of magnitude or
higher and home ranges of many hectares to several square kilometers, so this
bandwidth seemed appropriate. A more detailed, continuous distribution of
each species in relation to the nearest transport infrastructure and considering all
pixels in each distribution cell is shown in Fig. S2. Counting how many presence
cells fell into each 500-m band, we calculated both the relative proportion of
the species distribution that each band represented and their prevalence (i.e.,
the presence cells divided by the total number of cells available in each band).
Modeling the Area of Influence of Infrastructure. We estimated the overall
effect of the Spanish transportation, and other impervious infrastructure on
mean species abundances for birds (MSA
b
) and mammals (MSA
m
)andde-
termined the spatial distribution of the predicted effect zones. The MSA in-
dicator expresses the difference between the averaged mean abundance for
various species in the proximity of an infrastructure relative to their abun-
dance in a control location free of infrastructure (48). MSA values range from
no individuals remaining (0) to no effect on species abundance (1). Using a
meta-analytical approach, Benítez-López et al. (21), within the framework of
the Global Biodiversity model GLOBIO assessments, tested the relationship
between MSA and distance to infrastructure through generalized linear mixed
models (GLMM), and provided functional distance-decay response curves for
birds and mammals (Fig. 1). This study was undertaken using 49 studies and 90
datasets, which included 201 bird species (52% present in Spain) and 33 mammal
species (12% present in Spain), but it shows a substantial geographic bias be-
cause 88% of the studies came from Europe and North America. In addition, the
mammal datasets were biased toward ungulates (representing 58.1% of the
datasets considered, whereas carnivores, rodents, proboscideans, and lagomorphs
represented, respectively, 16.3%. 18.1%, 4.7%, and 2.3% of the datasets). How-
ever, because ungulates are species with usually very large home ranges and
many large carnivores worldwide have also been shown to be severely affected
by the presence of roads, the findings are likely to be applicable to many other
places worldwide. These functions have been previously applied only once, to
assess the impacts on roads in areas of high diversity value in Sweden (49).
Based on the statistics from the metaregressions, we generated two spatial
datasets about the predicted infrastructure effects on birds and mammals
and four spatial datasets showing the associated upper and lower 95% CIs at
a resolution of 15 m by applying a logit transformation
AB
Fig. 6. Variations through habitat types in the exposure to human infrastructure and in predicted detrimental effects on birds and mammals in Spain. (A)Box
plots of the distances to the nearest built-up area, transport infrastructure, and all impervious infrastructure combined for the five habitat typesconsidered.
(B) Proportion of land inside each intensity zone (Fig. 2) for birdsand mammals per habitat type, based on proximity to impervious infrastructure (outside circle) or
transport infrastructure alone (inside circle) (colors correspond to MSA legend in Fig. 2). Habitat illustrations courtesy of Marina Pinilla (Valencia, Spain).
Torres et al. PNAS Early Edition
|
5of6
ECOLOGYENVIRONMENTAL
SCIENCES

MSAðestimatedÞ=
eu
1+ eu ,
where MSA
(estimated)
is the predicted MSA at the observed distance from the
infrastructure and uis the linear equation describing the log-transformed
probability of the presence of a species at a certain distance xfrom the
infrastructure
u=ln Pi
1− Pi=β0+β1x,
where β
0
is the intercept (β
0
-birds =−0.863; β
0
-mammals =−0.607) and β
1
is the regression coefficient for the distance (β
1
-birds =0.00447 m
−1
;
β
1
-mammals =0.00083 m
−1
). The coefficients were obtained from the authors
of the meta-analysis. The distance variable xcould take the value of each cell
in the raster containing the Euclidean distance from an infrastructure. Given
that 61.1% of the datasets considered by Benítez-López et al. (21) corresponded
to road effects and the rest to other infrastructure, we used both a raster of
distances to transportation infrastructure alone (as a conservative measure) and
another with all impervious infrastructure combined to explore the sensitivity
of our estimates.
Finally, we analyzed the overall effect of the infrastructure by habitat
types on a national scale, by overlaying distance and MSA layers on a land
cover map [European Commission Program to Coordinate Information on
the Environment (Corine) land cover 2006; www.eea.europa.eu/data-and-
maps/data/clc-2006-vector-data-version-3] and calculating statistics for each
habitat. We report the results for five major classes in Results, namely,
wetland, bare land (open space with little or no vegetation), farmland,
scrubland, and forest, but the results for land cover classes at finer thematic
resolution are available in Table S2.
ACKNOWLEDGMENTS. We thank A. Benítez, R. Alkemade, and P. Verweij
for sharing the statistics from their meta-analysis; R. Early and F. Ferri-
Yañez for comments on an earlier version of this paper; and E. T. Game,
M. D. Madhusudan, and two anonymous reviewers for useful comments
that greatly improved the manuscript. The Spanish Ministry for Science
and Innovation provided funding for this study (Project CGL2008-02567).
A.T.’s work was funded through a FPU (Formación de Profesorado Univer-
sitario) PhD grant from the Spanish Ministry of Education.
1. WWF (2014) Living Planet Report 2014: Species and Spaces, People and Places, eds
McLellan R, Iyengar L, Jeffries B, Oerlemans N (WWF International, Gland, Switzer-
land).
2. Forman RT, et al. (2003) Road Ecology: Science and Solutions (Island Press, Wash-
ington, DC).
3. Dulac J (2013) Global Land Transport Infrastructure Requirements: Estimating Road
and Railway Infrastructure Capacity and Costs to 2050 (IEA, Paris).
4. Seto KC, Güneralp B, Hutyra LR (2012) Global forecasts of urban expansion to 2030
and direct impacts on biodiversity and carbon pools. Proc Natl Acad Sci USA 109(40):
16083–16088.
5. Fahrig L, Rytwinski T (2009) Effects of roads on animal abundance: an empirical re-
view and synthesis. Ecol Soc 14(1):21.
6. Clarke RT, Liley D, Sharp JM, Green RE (2013) Building development and roads: Im-
plications for the distribution of stone curlews across the Brecks. PLoS One 8(8):
e72984.
7. Clements GR, et al. (2014) Where and how are roads endangering mammals in
Southeast Asia’s forests? PLoS One 9(12):e115376.
8. van der Ree R, Jaeger JAG, van der Grift EA, Clevenger AP (2011) Effects of roads and
traffic on wildlife populations and landscape function: Road ecology is moving to-
ward larger scales. Ecol Soc 16(1):48.
9. Laurance WF, et al. (2014) A global strategy for road building. Nature 513(7517):
229–232.
10. Laurance WF, Sloan S, Weng L, Sayer JA (2015) Estimating the Environmental Costs of
Africa’s Massive “Development Corridors”.Curr Biol 25(24):3202–3208.
11. Sanderson EW, et al. (2002) The human footprint and the last of the wild. Bioscience
52(10):891–904.
12. Woolmer G, et al. (2008) Rescaling the human footprint: A tool for conservation
planning at an ecoregional scale. Landsc Urban Plan 87(1):42–53.
13. Theobald DM, Reed SE, Fields K, Soulé M (2012) Connecting natural landscapes using
a landscape permeability model to prioritize conservation activities in the United
States. Conserv Lett 5(2):123–133.
14. Forman RTT, Deblinger RD (2000) The ecological road-effect zone of a Massachusetts
(U.S.A.) suburban highway. Conserv Biol 14(1):36–46.
15. González-Abraham C, et al. (2015) The human footprint in Mexico: Physical geogra-
phy and historical legacies. PLoS One 10(3):e0121203.
16. Carver S, Comber A, McMorran R, Nutter S (2012) A GIS model for mapping spatial
patterns and distribution of wild land in Scotland. Landsc Urban Plan 104(3-4):
395–409.
17. Selva N, Switalski A, Kreft S, Ibisch PL (2015) Why keep areas road-free? The impor-
tance of roadless areas. Handbook of Road Ecology, eds van der Ree R, Smith DJ,
Grilo C (Wiley Blackwell, London), pp 16–26.
18. Leu M, Hanser SE, Knick ST (2008) The human footprint in the west: A large-scale
analysis of anthropogenic impacts. Ecol Appl 18(5):1119–1139.
19. Palomino D, Carrascal LM (2007) Threshold distances to nearby cities and roads in-
fluence the bird community of a mosaic landscape. Biol Conserv 140(1-2):100–109.
20. Eigenbrod F, Hecnar SJ, Fahrig L (2009) Quantifying the road-effect zone: Threshold
effects of a motorway on Anuran populations in Ontario, Canada. Ecol Soc 14(1):24.
21. Benítez-López A, Alkemade R, Verweij PA (2010) The impacts of roads and other
infrastructure on mammal and bird populations: A meta-analysis. Biol Conserv 143(6):
1307–1316.
22. Riitters KH, Wickham JD (2003) How far to the nearest road? Front Ecol Environ 1(3):
125–129.
23. Thrower N (1966) Original Survey and Land Subdivision: A Comparative Study of the
Form and Effect of Contrasting Cadastral Surveys (Rand McNally, Chicago).
24. Hennig EI, et al. (2015) Multi-scale analysis of urban sprawl in Europe: Towards a
European de-sprawling strategy. Land Use Policy 49:483–498.
25. Jaeger JAG, Soukup T, Madriñán LF, Schwick C, Kienast F (2011) Landscape
Fragmentation in Europe. Joint EEA-FOEN Report. (European Environment Agency
& Swiss Federal Office for the Environment, Luxembourg City, Luxembourg).
26. Ceaus
¸u S, et al. (2015) European wilderness in a time of farmland abandonment.
Rewilding European Landscapes, eds Pereira HM, Navarro LM (Springer, Cham,
Switzerland), pp 25–46.
27. Kareiva P, Watts S, McDonald R, Boucher T (2007) Domesticated nature: Shaping
landscapes and ecosystems for human welfare. Science 316(5833):1866–1869.
28. Grilo C, Reto D, Filipe J, Ascensão F, Revilla E (2014) Understanding the mechanisms
behind road effects: Linking occurrence with road mortality in owls. Anim Conserv
17(6):555–564.
29. Colino-Rabanal V, Lizana M, Peris S (2011) Factors influencing wolf Canis lupus
roadkills in Northwest Spain. Eur J Wildl Res 57(3):399–409.
30. Watts RD, et al. (2007) Roadless space of the conterminous United States. Science
316(5825):736–738.
31. Stoate C, et al. (2009) Ecological impacts of early 21st century agricultural change in
Europe–a review. J Environ Manage 91(1):22–46.
32. Donal PF, Gree RE, Heath MF (2001) Agricultural intensification and the collapse of
Europe’s farmland bird populations. Proc Biol Sci 268(1462):25–29.
33. Benton TG, Vickery JA, Wilson JD (2003) Farmland biodiversity: Is habitat heteroge-
neity the key? Trends Ecol Evol 18(4):182–188.
34. Dickson BG, Zachmann LJ, Albano CM (2014) Systematic identification of potential
conservation priority areas on roadless Bureau of Land Management lands in the
western United States. Biol Conserv 178:117–127.
35. Taylor BD, Goldingay RL (2010) Roads and wildlife: Impacts, mitigation and implica-
tions for wildlife management in Australia. Wildl Res 37(4):320–331.
36. Reijnen R, Foppen R, Meeuwsen H (1996) The effects of traffic on the density of
breeding birds in Dutch agricultural grasslands. Biol Conserv 75(3):255–260.
37. Chaplin-Kramer R, et al. (2015) Spatial patterns of agricultural expansion determine
impacts on biodiversity and carbon storage. Proc Natl Acad Sci USA 112(24):
7402–7407.
38. Torres A, Palacín C, Seoane J, Alonso JC (2011) Assessing the effects of a highway on a
threatened species using before-during-after and before-during-after control-impact
designs. Biol Conserv 144(9):2223–2232.
39. Lesbarrères D, Fahrig L (2012) Measures to reduce population fragmentation by
roads: What has worked and how do we know? Trends Ecol Evol 27(7):374–380.
40. Laurance WF, Goosem M, Laurance SGW (2009) Impacts of roads and linear clearings
on tropical forests. Trends Ecol Evol 24(12):659–669.
41. Rytwinski T, Fahrig L (2012) Do species life history traits explain population responses
to roads? A meta-analysis. Biol Conserv 147(1):87–98.
42. Edwards DP, et al. (2014) Mining and the African Environment. Conserv Lett 7(3):
302–311.
43. van der Ree R, Jaeger JAG, Rytwinski T, van der Grift EA (2015) Good science and
experimentation are needed in road ecology. Handbook of Road Ecology, eds van der
Ree R, Smith DJ, Grilo C (Wiley Blackwell, London), pp 71–81.
44. MAGRAMA (2012) Inventario Español de Especies Terrestres. Inventario Español del
Patrimonio Natural y de la Biodiversidad (Ministerio de Agricultura, Alimentación y
Medio Ambiente, Madrid). Spanish.
45. Bautista LM, et al. (2004) Effect of weekend road traffic on the use of space by
raptors. Conserv Biol 18(3):726–732.
46. Ferreras P, Aldama JJ, Beltrán JF, Delibes M (1992) Rates and causes of mortality in a
fragmented population of Iberian lynx Felis pardina Temminck, 1824. Biol Conserv
61(3):197–202.
47. Mateo-Sánchez MC, Cushman SA, Saura S (2013) Scale dependence in habitat selec-
tion: The case of the endangered brown bear (Ursus arctos) in the Cantabrian Range
(NW Spain). Int J Geogr Inf Sci 28(8):1531–1546.
48. Alkemade R, et al. (2009) GLOBIO3: A framework to investigate options for reducing
global terrestrial biodiversity loss. Ecosystems (N Y) 12(3):374–390.
49. Karlson M, Mörtberg U (2015) A spatial ecological assessment of fragmentation and
disturbance effects of the Swedish road network. Landsc Urban Plan 134:53–65.
6of6
|
www.pnas.org/cgi/doi/10.1073/pnas.1522488113 Torres et al.