Predicting Hot Spots of Herpetofauna Road Mortality Along Highway Networks

ABSTRACT Road mortality is often spatially aggregated, and there is a need for models that accurately and efficiently predict hot spots within a road network for mitigation. We surveyed 145 points throughout a 353-km highway network in New York State, USA, for roadkill of reptiles and amphibians. We used land cover, wetland configuration, and traffic volume data to identify features that best predicted hot spots of herpetofauna road mortality. We resampled 40 points an additional 4 times over 4 years to evaluate temporal repeatability. Both amphibian and reptile road mortality were spatially clustered, and road-kill hot spots of the 2 taxa overlapped. One survey provided a valid snapshot of spatial patterns of road mortality, and spatial patterns remained stable across time. Road-kill hot spots were located where wetlands approached within 100 m of the road, and the best predictor was a causeway configuration of wetlands (wetlands on both sides of the road). We validated causeways as predictors of road mortality by surveying 180 causeways and 180 random points across 5 regions (17,823 km²) of northeastern New York. Causeways were 3 times more likely than random locations to have amphibian and 12 times more likely to have reptile mortality present, and causeways had a 4 times higher total number of amphibian roadkill and 9 times higher reptile roadkill than did random points. We conclude it is possible to identify valid predictors of hot spots of amphibian and reptile road mortality for use when planning roads or when conducting surveys on existing roads to locate priority areas for mitigation.     

Do Roads Reduce Painted Turtle (Chrysemys picta) Populations?

Road mortality is thought to be a leading cause of turtle population decline. However, empirical evidence of the direct negative effects of road mortality on turtle population abundance is lacking. The purpose of this study was to provide a strong test of the prediction that roads reduce turtle population abundance. While controlling for potentially confounding variables, we compared relative abundance of painted turtles (Chrysemys picta) in 20 ponds in Eastern Ontario, 10 as close as possible to high traffic roads (Road sites) and 10 as far as possible from any major roads (No Road sites). There was no significant effect of roads on painted turtle relative abundance. Furthermore, our data do not support other predictions of the road mortality hypothesis; we observed neither a higher relative frequency of males to females at Road sites than at No Road sites, nor a lower average body size of turtles at Road than at No Road sites. We speculate that, although roads can cause substantial adult mortality in turtles, other factors, such as release from predation on adults and/or nests close to roads counter the negative effect of road mortality in some populations. We suggest that road mitigation for painted turtles can be limited to locations where turtles are forced to migrate across high traffic roads due, for example, to destruction of local nesting habitat or seasonal drying of ponds. This conclusion should not be extrapolated to other species of turtles, where road mortality could have a larger population-level effect than on painted turtles.     

How Long Do the Dead Survive on the Road? Carcass Persistence Probability and Implications for Road-Kill Monitoring Surveys

Background

Road mortality is probably the best-known and visible impact of roads upon wildlife. Although several factors influence road-kill counts, carcass persistence time is considered the most important determinant underlying underestimates of road mortality. The present study aims to describe and model carcass persistence variability on the road for different taxonomic groups under different environmental conditions throughout the year; and also to assess the effect of sampling frequency on the relative variation in road-kill estimates registered within a survey.

Methodology/Principal Findings

Daily surveys of road-killed vertebrates were conducted over one year along four road sections with different traffic volumes. Survival analysis was then used to i) describe carcass persistence timings for overall and for specific animal groups; ii) assess optimal sampling designs according to research objectives; and iii) model the influence of road, animal and weather factors on carcass persistence probabilities. Most animal carcasses persisted on the road for the first day only, with some groups disappearing at very high rates. The advisable periodicity of road monitoring that minimizes bias in road mortality estimates is daily monitoring for bats (in the morning) and lizards (in the afternoon), daily monitoring for toads, small birds, small mammals, snakes, salamanders, and lagomorphs; 1 day-interval (alternate days) for large birds, birds of prey, hedgehogs, and freshwater turtles; and 2 day-interval for carnivores. Multiple factors influenced the persistence probabilities of vertebrate carcasses on the road. Overall, the persistence was much lower for small animals, on roads with lower traffic volumes, for carcasses located on road lanes, and during humid conditions and high temperatures during the wet season and dry seasons, respectively.

Conclusion/Significance

The guidance given here on monitoring frequencies is particularly relevant to provide conservation and transportation agencies with accurate numbers of road-kills, realistic mitigation measures, and detailed designs for road monitoring programs.     

Roadkills of vertebrate species on two highways through the Atlantic Forest Biosphere Reserve,southern Brazil

We assessed the magnitude, composition, and spatial and temporal patterns of road mortality of native vertebrates on two highways in southern Brazil from 18 January 2003 to 26 January 2004. The highways cross remnants of the Atlantic Rainforest, a global biodiversity hotspot, and differ in vehicle traffic and surrounding landscape. We compared the road-kill magnitude and composition of birds, mammals, and reptiles between roads and seasons. We used a modified K statistic to depict the spatial patterns of roadkills of these groups and tested the association between vehicle traffic and road mortality through linear regression. We recorded 869 kills of 92 species. The two roads differed regarding the abundance and composition of roadkills. Reptile road mortality was higher in summer than winter, but all other groups did not show significant difference in the magnitude of mortality between seasons. The composition of killed assemblages differed significantly for some of the taxonomic groups among seasons. We found only one positive association between roadkills and vehicle traffic (reptiles on one of the roads), suggesting that vehicle flow does not explain the road-kill temporal variation on these roads. Total vertebrate, bird, and mammal roadkills showed significant spatial aggregations possibly due to variation in vehicle traffic, highway design, and local landscape condition and arrangement. With expected expansion of the road network, mitigation measures for multi-species assemblages should include habitat protection, soil use regulation, road crossing structures, speed reducers, and campaigns to raise people’s awareness about road impacts on wildlife.     

Road density and wetland context alter population structure of a freshwater turtle

Roads are detrimental to wildlife populations that require contiguous networks of terrestrial and aquatic habitats. Many species of freshwater turtles are sensitive to habitat fragmentation caused by roads, and are susceptible to road mortality during overland migrations. The common long‐necked turtle (Chelodina longicollis) is an Australian freshwater turtle that frequently moves between wetlands, and so populations may incur negative impacts from road effects. Here, we assessed the relationship between C. longicollis and road density and landscape variables within populations inhabiting 20 wetlands distributed throughout greater Melbourne, Australia. The size frequency distribution of C. longicollis at sites surrounded by high road densities was skewed towards larger individuals, but there was no difference in the frequency of juveniles between high and low road density sites. Regression modelling revealed a clear positive relationship between road density and carapace length (CL) of C. longicollis; the mean CL at a site with the highest road density was predicted to be 23% greater than mean CL at a site surrounded by no roads. Female CL was also positively related to road density. There was a clear positive relationship between wetland age and CL, although this relationship was not as strong. While there was no relationship evident between road density and the proportion of female C. longicollis at a site, more females were captured at smaller ephemeral sites surrounded by a high proportion of green open space and located near drainage lines. We did not find evidence of sex‐related differences in road effects. These results suggest that roads may be affecting C. longicollis in the study area, but the direct cause of any effects is difficult to identify.     

Amphibian road mortality in Europe: a meta-analysis with new data from Poland

While the increasing vehicular traffic is widely suspected to play a role in the worldwide amphibian population decline, the research of amphibian road mortality is scarce, fragmented, fraught with methodological problems, and largely inconclusive. As the first attempt at a synthesis, we analyzed all available data on amphibian mortality in Europe and combined them with four previously unpublished surveys conducted by us. Based on our recalculation of road-kill counts in terms of species-specific road-kill recordability, we conclude that, in lowland Central Europe, the common toads, Bufo bufo, are the most common victims of vehicular traffic in suburban landscapes, while the common frogs, Rana temporaria and Triturus newts, prevail in rural landscapes. The green frogs also tend to be more frequent in rural areas. Common tree frogs, Hyla arborea, are unexpectedly rare in the road-kill record despite their terrestrial and migratory habits. In consideration of problems with obtaining accurate amphibian population estimates, we further propose the road kills-to-spawners ratio (R/S) as a working measure of the impact of road mortality on a local population. While the R/S ratio may not reflect the losses to an entire local amphibian population, it is free of the errors of whole-population estimates, which are notoriously difficult for amphibians. When corrected for species-specific road-kill recordability, most results suggest that the impact of roads on newts may have been underestimated and that the impact on common frog populations may be higher than on those of common toads. Electronic supplementary material  The online version of this article (doi:) contains supplementary material, which is available to authorized users. Communicated by H. Kierdorf     

Can road mortality limit populations of pool-breeding amphibians?

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, 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 >2.5 km per km² of landscape and traffic volumes >250 vehicles/lane/day within the migration range of a breeding population of spotted salamanders.     

Spatial distribution of road-kills and factors influencing road mortality for mammals in Northern New York State

One of the most obvious impacts of roads on wildlife is vehicle-induced mortality. The aims of this study were to examine the spatial pattern of mammal–vehicle collisions (MVCs), identify and examine factors that contribute to MVCs, and determine whether the factors that increase the odds of MVCs are similar between species. On 103 road surveys that covered 7,094 total km I recorded the location of each MVC along the survey route. I measured landscape and roadway features associated with each MVC and used kernel density and network analysis tools to identify road mortality hotspots and measure spatial clustering of MVCs. I used logistic regression to model the likelihood of MVCs for all mammal data and separately for Porcupine (Erethizon dorsatum), Raccoon (Procyon lotor), Skunk (Mephitis mephitis), Muskrat (Ondatra zibethicus) and Cottontail (Sylvilagus floridanus) data sets. I identified 51 MVC hotspots and found spatial clustering of MVCs for Porcupines, Raccoons and Skunks. Two landscape variables, distance to cover and the presence of an ecotone, as well as one road variable, road width, appeared as broadly important predictors of mammalian road mortality, though there was also species-specific variation in factors that increased the risk of MVCs. Field-measured variables were more important than remotely-measured variables in predicting the odds of MVCs. Conservation implications are that mitigation of landscape features associated with higher risk of vehicle-collisions may reduce the number of MVCs in general, but species-specific research is required to more carefully tailor mitigation efforts for particular species.     

Effects of road proximity on genetic diversity and reproductive success of the painted turtle (Chrysemys picta)

Roads have a severe impact on wildlife. Reptiles are particularly susceptible due to their attraction to roads and their low car-avoidance capacity. For example, a high number of road killed freshwater turtles resulted from females selecting the unpaved side of roads as nesting sites. However, roads are harmful not only for adults, but are also expected to affect egg survival and recruitment. In this work, we indirectly determined whether the proximity to roads affects the reproductive success of freshwater turtles. The painted turtle (Chrysemys picta) was chosen for its population density, which is higher than most turtle species considered endangered. Locations near roads (<100 m) and in natural areas (>500 m) were sampled in three geographically distant ecoregions. We estimated the diversity of microsatellite loci from nuclear and mitochondrial genomes to assess the size of the kin groups as a proxy of the reproductive success of females. Similar diversity at nuclear markers suggested a comparable historical and demographic background among populations. However, lower mitochondrial diversity, higher mean and variance in the size of kin groups as well as a lower number of kin groups were strongly associated with the proximity to roads. Results indicated that a lower proportion of females participated in the recruitment of populations close to the roads than in natural areas, resulting in fewer but larger families near roads. We expect similar results for species nesting on the roadside. Barriers or fences that prevent individuals from reaching the road may help reduce their impacts on these populations.     

Inter-Individual Variability of Stone Marten Behavioral Responses to a Highway

Efforts to reduce the negative impacts of roads on wildlife may be hindered if individuals within the population vary widely in their responses to roads and mitigation strategies ignore this variability. This knowledge is particularly important for medium-sized carnivores as they are vulnerable to road mortality, while also known to use available road passages (e.g., drainage culverts) for safely crossing highways. Our goal in this study was to assess whether this apparently contradictory pattern of high road-kill numbers associated with a regular use of road passages is attributable to the variation in behavioral responses toward the highway between individuals. We investigated the responses of seven radio-tracked stone martens (Martes foina) to a highway by measuring their utilization distribution, response turning angles and highway crossing patterns. We compared the observed responses to simulated movement parameterized by the observed space use and movement characteristics of each individual, but naïve to the presence of the highway. Our results suggested that martens demonstrate a diversity of responses to the highway, including attraction, indifference, or avoidance. Martens also varied in their highway crossing patterns, with some crossing repeatedly at the same location (often coincident with highway passages). We suspect that the response variability derives from the individual''s familiarity of the landscape, including their awareness of highway passage locations. Because of these variable yet potentially attributable responses, we support the use of exclusionary fencing to guide transient (e.g., dispersers) individuals to existing passages to reduce the road-kill risk.     

Identification methods and deterministic factors of owl roadkill hotspot locations in Mediterranean landscapes

Road fatalities are among the major causes of mortality for Strigiformes species and may affect the population’s survival. The use of mitigation strategies must be considered to overcome this problem. However, because mitigation along the total length of all roads is not financially feasible, the locations where Strigiformes roadkills are more frequent (i.e., road fatality hotspots) must be identified. In addition to hotspot identification, factors that influence the occurrence of such fatalities should be recognized to allow mitigation measures to be delineated. We used road fatality data collected from 311 km of southern Portugal roads over a 2-year period to compare the performance of five hotspot identification methods: binary logistic regression (BLR), ecological niche factor analysis (ENFA), Kernel density estimation, nearest neighbor hierarchical clustering (NNHC), and Malo’s method. BLR and ENFA modelling were also used for recognizing roadkill deterministic factors. Our results suggest that Malo’s method should be preferred for hotspot identification. The main factors driving owl roadkillings are those associated with good habitat conditions for species occurrence and specific conditions that promote hunting behavior near roads. Based on these factors, several mitigation measures are recommended.     

Estimating the probability of illegal desert tortoise collection in the Sonoran Desert

The expansion of road networks in desert tortoise (Gopherus agassizii) habitat in the Sonoran Desert has raised questions concerning appropriate mitigation to reduce impacts at the population level. Although some effects, namely road-kill and habitat loss, have been well documented, illegal tortoise collection has been insufficiently addressed. It has become increasingly important for wildlife and land-use managers to understand the cumulative impacts of roads on tortoises and the effect that those impacts have on population persistence. We estimated the probability of desert tortoise detection and collection along 2-lane paved, maintained gravel, and non-maintained gravel roads to evaluate whether collection probabilities were related to road type. Although collection probability did not vary by road type, the probability of desert tortoise detection by passing motorists was greatest on maintained gravel roads and fewest on non-maintained gravel and paved roads. These results have implications for effectively mitigating the impacts of roads on desert tortoises. Published 2011. This article is a U.S. Government work and is in the public domain in the USA.     

The influence of body-size and diet on road-kill trends in mammals

Roads negatively impact animals in a variety of ways. One of the most obvious impacts is vehicle-induced mortality. We have little understanding of the patterns and rates of road mortality for mammals. The aims of this study were to determine representative road-kill rates for local mammal species, compare road-kill rates between mammals of different sizes and diets, and compare patterns of road mortality between seasons. We hypothesized that carnivores would be killed less often than herbivores or omnivores and that medium sized mammals (1.0–10.0 kg) would be killed more often than expected based on their frequency in the area. At least 50% of the mammal species in the study area were impacted by road mortality, representing 21 species from 5 mammalian orders. Carnivores were found less often and medium-sized mammals more often than predicted. Porcupines (Erithizon dorsatum), raccoons (Procyon lotor), cottontails (Sylvilagus floridanus) and striped skunks (Mephitis mephitis) were the most frequent species to be killed. On average, 3.8 mammals were killed per 100 km of road. The road-kill rate varied between seasons and species and peaked in mid June. We also looked across studies to determine general rates of vehicle-induced mortality for mammals and found that an average of 8.8 mammals are killed per 100 km. Given the rapid decline of many mammal species and the consistent increase in vehicle miles traveled, efforts to preserve roadless areas and develop strategies to mitigate the effects of roads on mammals are warranted.     

Assessing Brazilian reptiles' road-kill risks using trait-based models

Reptiles are an understudied group in road ecology, despite evidence of their high vulnerability to road mortality. Recently, trait-based models have been demonstrated to be valuable tools for explaining and predicting road mortality risks for birds and mammals. The present study aimed to apply such models to reptiles for the first time. We fitted eight random forest regression models, controlling for different survey design variables, to explain 782 empirical road-kill rates for Brazilian reptiles and selected the best-performing model to predict road mortality risks for 572 continental species. The results showed that species that are habitat generalists, omnivorous, viviparous, cathemeral, and have intermediate clutch or litter sizes are at a higher risk of being road-killed. The relationships for other traits included in our models were uncertain, but our findings suggest that population density and species-specific behavioural responses to roads and traffic may play an important role in road mortality risks. Geographical location and survey design variables (especially sampling speed and sampling time) were more important in explaining the variance of the empirical road-kill rates than any of the tested ecological and functional traits. Besides adding evidence of the vulnerability of the Amazon region to vertebrate road-kills, this study highlights some similarities between the relationships identified here and those found for birds and mammals (such as with body mass and habitat breadth). We also corroborate that trait-based models are useful tools to aid in conservation efforts but indicate that they can be biased by the methodologies used to collect empirical data. Future road-kill surveys should therefore use methods specifically designed for reptiles and estimate both observer efficiency and carcass removal rates.     

Wetland complexes and upland–wetland linkages: landscape effects on the distribution of rare and common wetland reptiles

We used landscape ecology concepts to test the importance of upland–wetland linkages on the distribution of two common wetland species, the northern watersnake Nerodia sipedon sipedon and midland painted turtle Chrysemys picta marginata , and two rare wetland species, the copper-bellied watersnake Nerodia erythrogaster neglecta and Blanding's turtle Emydoidea blandingii . We tested if connectivity (proximity to other wetlands), connectivity quality (wetland distance to roads and forest area within 30, 125, 250, 500 and 1000 m of the wetland), and patch size (wetland size and shoreline length) affected the distribution of these four species. Our results show that both common species were more likely to occur in larger, less isolated wetlands, but their distribution were not influenced by proximity to roads or the amount of adjacent forest area surrounding the wetland. Both rare species were more likely to occur in wetlands that were farther away from roads and that had more surrounding forest. Proximity to other wetlands was not a significant predictor of either rare species' distribution. Our results suggest that management practices should focus on protecting wetland complexes and maintaining upland–wetland linkages by improving landscape connectivity, increasing forest area surrounding wetlands and reducing road effects.     

Identifying Hot Moments in Road-Mortality Risk for Freshwater Turtles

ABSTRACT Risk assessments can be used to identify threats, which vary both in space and time, to declining species. Just as hot spots describe locations where threat processes operate at a higher rate than in surrounding areas, hot moments refer to periods when threat rates are highest. However, the identification of hot moments can be challenging because the temporal complexity of some threat processes makes their effects on population viability difficult to predict. Declining throughout much of their range, Blanding's turtle (Emydoidea blandingii) populations are potentially most vulnerable to road mortality where road densities and traffic volumes are high. The temporal variations in road-mortality risk faced by these and other semiaquatic turtles at the population level are a consequence of several factors, including sex-specific movement characteristics and seasonal changes in traffic volume. We examined these risk factors for Blanding's turtle populations in Maine, USA, by integrating temporally explicit roadkill probabilities with demographic parameters informed by local and range-wide studies. Specifically, we used population simulations to estimate the relative risk for 14 2-week periods during the turtles’ active season. Our analysis clearly identified early summer as a period of elevated risk, with June through mid-July signaling a road-mortality hot moment for Blanding's turtles (for both M and F). These findings provide guidance for the implementation of temporally explicit conservation measures such as cautionary road signage, traffic management, and public outreach that, if timed strategically, could help to mitigate population impacts from road mortality.     

Habitat suitability modelling for species at risk is sensitive to algorithm and scale: A case study of Blanding's turtle,Emydoidea blandingii,in Ontario,Canada

Species distribution modelling (SDM) can help conservation by providing information on the ecological requirements of species at risk. We developed habitat suitability models at multiple spatial scales for a threatened freshwater turtle, Emydoidea blandingii, in Ontario as a case study. We also explored the effect of background data selection and modelling algorithm selection on habitat suitability predictions. We used sighting records, high-resolution land cover data (25 m), and two SDM techniques: boosted regression trees; and maximum entropy modelling. The area under the receiver characteristic operating curve (AUC) for habitat suitability models tested on independent data ranged from 0.878 to 0.912 when using random background and from 0.727 to 0.741 with target-group background. E. blandingii habitat suitability was best predicted by air temperature, wetland area, open water area, road density, and cropland area. Habitat suitability increased with increasing air temperature and wetland area, and decreased with increasing cropland area. Low road density and open water increased habitat suitability, while high levels of either variable decreased habitat suitability. Robust habitat suitability maps for species at risk require using a multi-scale and multi-algorithm approach. If well used, SDM can offer insight on the habitat requirements of species at risk and help guide the development of management plans. Our results suggest that E. blandingii management plans should promote the protection of terrestrial habitat surrounding residential wetlands, halt the building of roads within and adjacent to currently occupied habitat, and identify movement corridors for isolated populations.     

Integrated risk factors for vertebrate roadkill in southern Ontario

Road mortality of animals (roadkill) threatens public safety and wildlife populations. As mitigation tools, predictive models of roadkill are becoming more common in the published literature; however, few models generalize across multiple taxa, and thus are less useful for management scenarios that account for multiple target species. Using a dataset of 653 vertebrate roadkills collected from 2 parks in southern Ontario, we constructed generalized linear mixed models to determine the simultaneous risk factors for bird, frog, mammal, five-lined skink (Eumeces fasciatus), snake, toad, and turtle hatchling roadkills from among a set of 8 potential predictor variables. Posted road speed limit was the dominant roadkill predictor (positive coefficient), followed by maximum daily temperature (positive), habitat diversity (positive), and distance from wetlands (negative). All else being equal, as road speed limits increase from 20 km/hr to 50 km/hr, the model predicted the season's mean roadkill probability for a given location to increase from less than 0.1 to 0.75. Conversely, roadkill probability declined from 0.55 to 0.29 as distance from wetland edges increases from 0 km to 1 km. Model diagnostics calculated from randomly resampled cross-validation datasets indicated that the best model formulation had an averaged predictive accuracy of 67.5% and an area under the curve (AUC) of 0.867. The best model also reflected seasonal patterns of animal behavior, including late-summer frog movements and fall turtle hatching events. The best model also compared favorably to single-taxon equivalent models. To reduce the incidence of vertebrate roadkill, we recommend that managers lower road speed limits, especially on roads near diverse habitats and near wetlands, and on warmer days if temporary signage is used. © 2012 The Wildlife Society.     

Vehicle–wild vertebrate collision mortality on the highways of Tigray,Ethiopia, implications for conservation

In view of the upcoming road network improvement and expansion in Ethiopia, specifically in Tigray, it is also necessary to understand the potential impacts of road accidents with wildlife vertebrate animals. Road mortality detection surveys were conducted from March 2013 to June 2014. We surveyed around 530 km using vehicle with special emphasis given to roads surrounded by wetlands, forests, rocky areas and rivers each month. Additional information was also collected using a standardized questionnaire. A total of twenty species, 143 individuals of amphibians, reptiles, birds and mammals were recorded as road vehicular accidents in the surveyed area. Of all, mammals showed the highest species richness (80 individuals belonging to ten species) followed by birds (49 individuals belonging to eight species). The survey revealed most accidents happened during the early morning and late evening. This might be due to relatively high traffic and continued activity of wild animals at that time, besides the driver's inability to avoid accidents when it is dark. Wildlife underpasses during road construction, location of crossing structures, rules of wildlife conservation, improving driver's awareness are relevant in this context and may be the most important mitigation measures to reduce mortality of wildlife on the roads.     

The presence of rabbits adjacent to roads increases polecat road mortality

Road mortality is an increasing problem for terrestrial vertebrate conservation due to the increase of both road numbers and vehicle flow. We hypothesize that the probability of a predator being killed on the road is related to the presence of its prey adjacent to the road, which is likely to be related to the use that these predators make of road verges. We aim to identify the features of specific stretches of road where road-kills of a predator (European polecat) occur in Mediterranean landscapes, including the presence of its main prey (European rabbit) and landscape and road features. We compared 85 100 m long stretches of road where at least one road-kill was recorded with 104 stretches without any road-kill in a dry agricultural landscape in central Spain. We used regression analysis to investigate the relationship between road-kill occurrence and the features in the 67% of the cases. Road-kill stretches were characterised by greater numbers of rabbit burrows in the road verges and by higher traffic flow and speed (i.e. higher speed limit, lower proportion of heavy vehicles, wider road and lower proportion of unbroken central lines). Road-kill stretches also had more metres built over bridges and lower densities of people. We validated our best model with a dataset (the 33% of the cases) not included in its development, which correctly classified 82% of road-kill stretches and 89% of non-road kill stretches. Our results highlight the need for taking into account food resource distribution when studying causes of animal road-kills.