A forecasting map of avian roadkill-risk in Europe: A tool to identify potential hotspots

Highlights

• Using road casualties from 9 European countries we estimated the bird's roadkill probability.

• We calculated a cumulative risk of roadkill considering the species in the communities.

• We combined community cumulative risk of roadkill with road density across the continent.

• We elaborated a forecasting map of avian roadkill-risk in Europe to estimate hotspots.

Abstract

In this study, we propose a novel strategy for identifying potential hotspots of avian roadkills in Europe. The proposed approach combines information about the spatial distribution of bird species at a comparatively higher risk of roadkill with data on road density. First, using a large dataset collected from several European studies and reports, we extracted the frequency of occurrence of bird casualties for 209 breeding bird species recorded in roadkill events. We standardized the relative frequency of roadkill from 0 (species never recorded in bird casualties' reports) to 1 (species with the higher number of roadkill's), obtaining a continuous variable that indicates the potential risk of roadkill species by species. Second, using published data on the spatial distribution of breeding bird species in Europe, we calculated the cumulative risk of roadkill in each bird assemblages, by considering the sum of the values estimated for each species in the previous step. Third, we calculated the road density in each spatial unit. Finally, we elaborate a forecasting map of potential avian roadkill-risk across Europe, by combining the data on road density and cumulative roadkill risk of bird communities.

The tool proposed can help to identify potential hotspots at different spatial scales where the risk of avian roadkill is high, offering the possibility to improve conservation measures in road planning. Briefly, the prediction of where there is aligned convergence between communities with highly ranked species and landscapes with dense road networks can be used in procedures modelling wildlife-car collisions, for transportation mitigation projects.

Introduction

The urbanization process is characterized by the concentration and expansion of human settlements. Consequently, the development of interconnectivity tissue (the road network) guarantees the connectivity among urban centers (Marzluff et al., 2008). There is currently more than 36,000,000 km of roads on the global scale, following the estimates of the World Development Index of World Bank, World Road Statistics of the International Road Federation and World Factbook of the Central Intelligence Agency on the World. Road networks are a prominent feature of the capacity of humans for penetrating the last remaining natural areas and landscapes, causing severe habitat loss, fragmentation, and many other well-known disturbances (Bennett, 1991; Forman et al., 2003). Specifically, roads are commonly associated with a significant decline of biodiversity (Benítez-López et al., 2010; Kociolek and Clevenger, 2009; Loss et al., 2014a; van der Ree et al., 2011). The main adverse ecological effects of roads on vertebrate wildlife are related to habitat loss associated to the construction and spread of road networks (Davenport and Davenport, 2006; Kociolek et al., 2011); to traffic disturbance, as noise pollution (Kaseloo and Tyson, 2005; McClure et al., 2013; Zhou et al., 2020), light pollution (De Molenaar et al., 2006), and even fostering changes in species ranges, habitat use, local abundance and distribution (Cooke et al., 2020; Delgado et al., 2008). Roads can also represent ecological barriers for non-flying terrestrial animals, reducing the species' range or increasing the isolation of some populations (van der Ree et al., 2011). However, perhaps the most immediate and critical effect of roads is the direct mortality due to collisions with the traffic (Collinson et al., 2014; Reijnen et al., 2008), and at the same time the road barrier effect (Jaeger et al., 2005). In addition, roads and traffic may be interacting with other drivers, such as climate change, leading to the decline of bird diversity in Europe (Huntley et al., 2008; Morelli et al., 2020).

In Europe, several bird species use roads and roadsides for different types of activities, from hunting (e.g., carnivorous species as shrikes or raptors) to breeding or nesting activities (Helldin and Seiler, 2003; Morelli et al., 2015, Morelli et al., 2014). Scavenger species such as vultures and kites, or even strigids (nocturnal raptors), are likely to become road casualties when they are attracted to corpses on the road (Vidal-Vallés, 2018). For decades, birds were considered less affected by roadkill than mammals because of their ability to fly. However, there is increasing evidence about birds as road casualties indicating that vehicle collision is a significant cause of death (Kociolek et al., 2015). Some studies suggest that the number of birds killed by vehicle collisions in Europe is vast (up to 5–10% of bird mortality caused by roads and traffic for W Palearctic; Møller et al., 2011). In Germany, there is an estimated average of more than 9 million birds killed on the roads every year (Fuellhaas et al., 1989). Variable estimations of annual averages come from Spain (12 million) (PMVC, 2003), Bulgaria (7 million) (Nankinov and Todorov, 1983), the Netherlands (2 million) (Schrijver, 1993), England (27 million), Denmark (1.1 million) and Sweden (8.5 million) (Erritzøe et al., 2003). We do not know if there is an increasing or decreasing trend in road–caused mortality for European birds, since we lack long-term monitoring. However, these numbers are likely to increase if the road network continues to develop across the continent.

Numerous factors would explain bird susceptibility to become roadkill. Aside from road and traffic design variables (Clevenger et al., 2002; Husby, 2016), research has focused on bird responses, and functional traits have been analyzed as reasons for susceptibility to road mortality (Santos et al., 2015). Firstly, road mortality depends on bird escape and vehicle avoidance behavior (Husby and Husby, 2014). Though, some species seem to be more susceptible than others to the roadkill, showing a frequency as road casualties higher than expected (Erritzøe et al., 2003; Santos et al., 2015). It is assumed that the relative frequency of traffic casualties is associated with the eco-functional traits of the species. For example, a higher incidence of car-collisions is expected for birds that are abundant near to roads (Santos et al., 2016), for frequent roadside users (Morelli et al., 2014), for birds more prone to cross the roads (Johnson et al., 2017), but also merely for species that occupy landscapes where the road networks are more spread (e.g., lowland bird populations, agricultural, urban and periurban areas, etc.). Interspecific differences could also be related to different sensitivity to collision risk (Husby and Husby, 2014; Møller et al., 2011).

With this in mind, we can hypothesize that avian communities overall more prone to roadkill than others could characterize some geographical areas in Europe. For example, a bird community composed of five species often recorded in road casualties should have a cumulative roadkill risk higher than that of another community consisting of five species never recorded as road casualties. Then, combining areas with bird communities composed by species with a high frequency of roadkills and simultaneously characterized by a dense road network, especially with high-speed roads or highways, we can expect an increased overall risk of bird-car collisions. To predict zones of potentially elevated collision risk would help to identify areas posing a consistent threat for the conservation of bird diversity (Gunson et al., 2011).

The accurate detection of the potential hotspot of wildlife-vehicle collisions should be the first step to elaborate regional and local strategies aiming to reduce the problem and to identify areas (or in some cases, species) with higher mitigation needs. Different mitigation strategies and measures can be developed and applied, depending on the area and road characteristics, but to mitigate the ecological impacts of roads effectively, continued actions and monitoring are necessary (Karlson et al., 2014; Trombulak and Frissell, 2000). The use of forecasting models, which can combine different layers of information to predict an event, primarily used in ecology (Guisan and Thuiller, 2005), constitutes a valid tool also for transportation mitigation projects as well as urban and road planning (Gunson et al., 2011; Santos et al., 2013; Soldatini et al., 2011). A good example is constituted by the use of ecological modelling to create risk maps (Yañez-Arenas et al., 2014).

In this study, we combined data on the spatial distribution of bird species characterized by high incidence in roadkill events in Europe, with data on road density across the continent, by using a fixed spatial unit. We compared the spatial congruence/mismatch between the road network, the bird species richness, by considering the different types of environments. Then, we elaborated, on a large spatial scale, a forecasting map of the potential risk of collisions between cars and birds, across Europe. We tried to identify potential hotspots where the risk of avian roadkill is higher, to be considered as areas where it should be necessary to focus on mitigation strategies.