Talk Using citizen science to survey roadkill over large spatiotemporal scales: the example of South Africa
Abstract

Roads affect wildlife through a range of mechanisms from habitat loss and degradation by decreasing landscape connectivity to direct mortality from collisions with vehicles (roadkill). It is challenging to estimate the direct impact of wildlife-vehicle collisions on wildlife populations, but this cause of mortality has been rated amongst the highest modern risks to our fauna. Repeated road surveys conducted by trained personnel are the ideal way of monitoring impacts of roadkill on wildlife, but are impractical over large areas. However, with the development of “citizen science” projects, in which members of the public participate in data collection, it is now possible to monitor this phenomenon over scales far beyond the limit of traditional field studies. This tool has far-reaching implications for ecologists, but has not yet been rigorously tested.

In this study, we used social media and a simple smartphone app (“Road Watch”) to mobilize citizen scientists to collect data on mammalian roadkill across South Africa (with an estimated road network length of 1.2 million km). From 1999 to 2015, a total of 2680 roadkill occurrences were reported by citizen scientists and stored in the Endangered Wildlife Trust’s roadkill database. We identified two main types of reporters: those that conducted regular road patrols (each person reporting >50 roadkill) and those that submitted opportunistic data. Throughout this study, we compared data from regular and opportunistic reporters, and found few significant differences between these data sets.

Based on regular reporter data, carnivores (44%), very small species (≤10kg, 25%) and nocturnal species (43%) were most often found killed on South African roads. The five species reported most frequently were scrub hares (Lepus saxatilis, 10%), various rabbit sp. (9%), bat-eared foxes (Otocyon megalotis, 8%), black-backed jackals (Canis mesomelas, 7%) and aardwolves (Proteles cristatus, 5%). Unidentified mammals represented 8% of the dataset. Mammalian roadkill were most often reported on national roads (32%) and paved roads in general (96%) compared to smaller and unpaved roads. We identified three major roadkill hotspots representing high-risk areas for wildlife, two located in the Northern Cape province and one in the Free State. They were all centered on heavily used national highways linking Johannesburg to two of the country’s biggest cities: Cape Town and Durban.

This study is the first to provide a nationwide survey of mammalian roadkill in South Africa. Some groups of mammals are reported more frequently as roadkill than others e.g. carnivore and nocturnal species. While this pattern may be partly due to observer bias, it confirms what has been described in smaller studies elsewhere in Africa. Our analysis demonstrates that citizen science surveys can be used to provide solid roadkill data, in determining trends across very large areas difficult for scientists to monitor. These trends can determine potential roadkill hotspots and species at risk, which in turn can become the focus of more detailed monitoring, ultimately leading to the implementation of roadkill-reduction-measures.

Keywords
roadkill, wildlife-vehicle collisions, traffic mortality, mammals, cluster analysis, citizen science