How reliable are your data? Verifying species identification of road-killed mammals recorded by road maintenance personnel in São Paulo State, Brazil
Introduction
Across the world, many scientific studies rely on data collected by volunteers. The reasons are varied, but they include a) Project goals that are too ambitious in time or money to be conducted by paid professionals on their own, b) Making use of local knowledge and experience, c) Engaging the public and other stakeholder groups, and d) The public and organized stakeholders taking initiative to document an issue that they would like to see addressed (e.g. Hadj-Hammou et al., 2017; Newson et al., 2017; Steger et al., 2017). Despite the growing use of data collected by volunteers and non-experts in science there are often problems or perceived problems with the data quality (Jollymore et al., 2017; Mitchell et al., 2017; Tredick et al., 2017).
Many citizen science projects in the biological and ecological sciences require species identification by volunteers or non-experts. However, the ability of volunteers or non-experts to correctly identify the species varies widely between species, and rare species tend to be most difficult to identify correctly (Swanson et al., 2016; Rüdisser et al., 2017; Vantieghem et al., 2017). Nonetheless, in the field of road ecology volunteers or non-experts have been found to correctly identify 97% of the reported road-killed animals for which an image was available (Waetjen and Shilling, 2017). Species or species groups that were most commonly misidentified were squirrels, other small mammal species, and birds (Waetjen and Shilling, 2017). However, volunteers or non-experts, especially those that are only occasional contributors, are also known to report charismatic and easily identifiable species more than other species groups (Paul et al., 2014; Périquet et al., 2018). In this article, we further explore the reliability of species identification of road-killed animals by non-experts.
Wildlife-vehicle collisions are a concern around the world and it affects human safety, property and wildlife (Conover et al., 1995). Many roadkill studies focus on large common mammal species (usually mainly ungulates), as they pose the greatest threat to human safety in Europe (e.g. Groot Bruinderink and Hazebroek, 1996), Africa (e.g. Drews, 1995; Eloff and van Niekerk, 2005), North America (e.g. Huijser et al., 2009), South America (e.g. Bueno et al., 2013; Huijser et al., 2013; Medici et al., 2016), Asia (Vidya and Thuppil, 2010), and Oceania (e.g. Klöcker et al., 2006; Bond and Jones, 2013). However, if the focus is also on biological conservation or biological diversity in general, then rare, threatened, or endangered species, or all species are important, independent of their body size (e.g. Brokie et al., 2009; Braz and França, 2016). Since biodiversity in the tropics tends to be higher than in temperate regions, roadkill studies in the tropics tend to report a wide range of species including relatively rare and small species (Hobday and Minstrell, 2008; Baskaran and Boominathan, 2010; Wang et al., 2013; Mohammadi and Kaboli, 2016).
For non-experts, it is typically easier to correctly identify large mammals that are hit by vehicles than smaller sized species as it is more likely that at least some portion of a large animal has remained intact (Santos et al., 2016). Furthermore, in areas with high biodiversity, species identification is typically more challenging because there are often multiple species that have similar appearance, for example small wild Felids from genus Leopardus. For these reasons, it may be harder for non-experts to correctly identify road-killed mammals in the tropics than, for example, in North America or Europe. Non-experts also tend to use species descriptions based on common names which can be confusing and may result in unclear data that is excluded from further analyses (Ford and Fahrig, 2007; Swanson et al., 2016). Species descriptions based on common names can relate to more than one species, or the same common name may be associated with different species depending on the region. Yet, many roadkill studies, including those from the tropics, rely on data collected by non-experts (Huijser et al., 2013; Vercayie and Herremans, 2015; Bíl et al., 2017).
Non-experts include law enforcement personnel (e.g. crash reports), road maintenance personnel (carcass removal data), and volunteers (carcass observation data) (Lee et al., 2006; Huijser et al., 2007; Heigl et al., 2017). The most severe wildlife-vehicle crashes, e.g. those with at least an estimated US$1000 in vehicle repair costs and those that include human injuries and human fatalities, are reported by law enforcement personnel and are included in crash databases (Huijser et al., 2007). It is typically a standard task of road maintenance personnel to remove and report carcasses of large mammals on or near the highway as they present a safety hazard for the traveling public (Huijser et al., 2007). Large mammal-vehicle collision data collected by road maintenance personnel or law enforcement personnel are typically readily available, they tend to have (the potential for) similar search and reporting effort in time and space, and they often relate to large geographical areas. However, the data may still have varying spatial accuracy, especially when cross-roads or other landscape features are used as a location description rather than mile reference posts or coordinates based on a Global Positioning System (Huijser et al., 2007). In some cases, citizens (or volunteers) are asked to submit observations of road-killed wildlife. These observations are typically incidental and lack documented search or reporting effort. Such citizen science programs are sometimes targeted at rare or small species that are not well represented in the data collected by road maintenance crews or law enforcement personnel (e.g. McClintock et al., 2015; Heigl et al., 2017). Citizen science data can relate to specific highway sections, but they can also relate to much larger geographical areas (e.g. Lee et al., 2006 vs. Shilling and Waetjen, 2015). Though citizen science data typically have undocumented search and reporting effort, they can still have spatial similarity to data collected by researchers or others that used consistent search and reporting effort (Paul et al., 2014).
While many roadkill studies rely on data collected by non-experts, data quality control is often lacking (but note these exceptions: Paul et al., 2014; Dwyer et al., 2016; Swanson et al., 2016; Périquet et al., 2018). In this study, we investigated whether road maintenance personnel (non-experts) correctly identified the species of the road-killed wild mammals in a tropical region: São Paulo State, Brazil. We used two different and complementary methods to verify species identification: i) We investigated images of road-killed mammals that were associated with the data collected by road maintenance personnel, and ii) We presented images of alive and dead mammals to road maintenance personnel and asked them to identify the species. Finally, we formulated recommendations to improve the reliability of species identification of road-killed animals by non-experts. We believe the findings and recommendations do not only apply to our study area, but that they are also useful elsewhere throughout the world where wildlife data, specifically roadkill data, are collected by people who are not experts in species identification.
Section snippets
Datasets
We investigated the reliability of species identification by road maintenance personnel of road-killed animals. We used two different methods, each with their corresponding data sets and analyses:
Which Portuguese common names for road-killed animals relate to what species?
For data set 1, Portuguese common names for road-killed mammals that had the highest reliability (≥90%) with being associated with a particular species and that had a sample size of at least 15, were “lobo-guará” and “lobo” for maned wolf (Chrysocyon brachyurus), “cachorro-do-mato” for crab-eating fox (Cerdocyon thous), “coelho” and “lebre” for European hare (Lepus europaeus), “capivara” for capybara (Hydrochoeris hydrochaeris), and “tamanduá-mirim” for southern tamandua (Tamandua tetradactyla)
Discussion
In our study, the two methods used were complementary and strengthened our conclusions regarding the reliability of species identification of road-killed animals by non-experts. The data showed that non-experts usually correctly identified certain common, large, or highly recognizable species, similar to the findings of Waetjen and Shilling (2017) and Périquet et al. (2018). Rare or rarely seen species (e.g., striped hog-nosed skunk (Conepatus semistriatus), lesser grison (Galictis cuja), tayra
Conclusions
We conclude that road maintenance personnel typically correctly identified certain common, large, or highly recognizable species. However, rare or rarely seen species, species that resemble other species (e.g. small wild canids and felids), or species that are not highly recognizable are often misidentified, ambiguously described, or not identified at all. Interestingly, the ability of road maintenance personnel to correctly identify the most common road-killed small wild canids and felids is
Acknowledgements
We thank the Forest Science Department (Escola Superior de Agricultura “Luiz de Queiroz”, Universidade de São Paulo), the Interdisciplinary Program in Applied Ecology (PPGI-EA) and the Wildlife Ecology, Management and Conservation Lab (LEMaC). We are indebted to Conselho Nacional de Pesquisa e Desenvolvimento Científico e Tecnológico (CNPq) for the productivity fellowship to KMPMBF (grant #308503/2014-7) and we also thank to ARTESP (Agência Reguladora de Serviços Públicos Delegados de
References (44)
- et al.
Srazenazver.cz: a system for evidence of animal-vehicle collisions along transportation networks
Biol. Conserv.
(2017) - et al.
Diet and body size of North American mammal road mortalities
Transp. Res. D
(2007) - et al.
Citizen science for water quality monitoring: data implications of citizen perspectives
J. Environ. Manag.
(2017) - et al.
Large-scale citizen science improves assessment of risk posed by wind farms to bats in southern Scotland
Biol. Conserv.
(2017) - et al.
An evaluation of citizen science data for monitoring wildlife along a highway right-of-way
J. Environ. Manag.
(2014) - et al.
Immediate behavioral responses of humans and Asian elephants in the context of road traffic in southern India
Biol. Conserv.
(2010) - et al.
Road kill of animals by highway traffic in the tropical forests of Mudumalai Tiger Reserve, southern India
J. Threat. Taxa
(2010) - et al.
Mamíferos da Serra da Macaca São Miguel Arcanjo
- et al.
Roads and macropods: interactions and implications
Aust. Mammal.
(2013) - et al.
Wild vertebrate roadkill in the Chapada dos Veadeiros National Park, Central Brazil
Biota Neotropica
(2016)
Long-term wildlife road-kill counts in New Zealand
New Zeal. J. Zool.
Influence of landscape characteristics on capybara road-kill on highway BR-040, southeastern Brazil
Oecologia Australis
Review of human injuries, illnesses, and economic losses caused by wildlife in the United States
Wildl. Soc. Bull.
Roadkill of animals by public traffic in Mikumi National Park, Tanzania, with notes on Baboon mortality
Afr. J. Ecol.
Using citizen-collected wildlife sightings to predict traffic strike hot spots for threatened species: a case study on the southern cassowary
J. Appl. Ecol.
Game, fences and motor vehicle accidents: spatial patterns in the Eastern Cape
S. Afr. J. Wildl. Res.
Ungulate traffic collisions in Europe
Conserv. Biol.
Getting the full picture: assessing the complementarity of citizen science and agency monitoring data
PLoS One
Amphibian and reptile road-kills on tertiary roads in relation to landscape structure: using a citizen science approach with open-access land cover data
BMC Ecol.
Distribution and abundance of roadkill on Tasmanian highways: human management options
Wildl. Res.
Animal-vehicle collision data collection. A synthesis of highway practice
Cost–benefit analyses of mitigation measures aimed at reducing collisions with large ungulates in the United States and Canada; a decision support tool
Ecol. Soc.
Cited by (20)
The road less traveled: Addressing reproducibility and conservation priorities of wildlife-vehicle collision studies in tropical and subtropical regions
2021, Global Ecology and ConservationCitation Excerpt :The paucity of studies reporting detection probability makes it difficult to compare mortality rates across studies and sites. This issue is exacerbated by imperfect species identification due to the condition of decaying carcasses, as well as the increased species richness and presence of cryptic species within high biodiversity regions, which can distort our perception of biodiversity trends (Vieites et al., 2009; Abra et al., 2018). Most wildlife-vehicle collision studies also focused on medium- to large-sized mammals a priori, which exhibit higher detection probability and carcass persistence time than smaller carcasses (Teixeira et al., 2013; Ratton et al., 2014).
An estimate of wild mammal roadkill in São Paulo state, Brazil
2021, HeliyonCitation Excerpt :For the corrected database, we used the scientific name of species. However, for Portuguese species descriptions that were not encountered in the subset of the data used by Abra et al. (2018), we kept the Portuguese common name and added “unconfirmed” to the species name. To improve understanding for international readers, we also included the common English name for these unconfirmed species.
Wildlife-vehicle collisions - Influencing factors, data collection and research methods
2020, Biological ConservationCitation Excerpt :WVC research is based on historical events that derive from different survey methods and data sets, mainly from ecological surveys or police registrations. Data are also collected by non-experts (Abra et al., 2018), or by camera or LIDAR scanners integrated in the vehicle (Alden et al., 2016) or in the road infrastructure (Chen et al., 2019). Despite the variety of collection methods, the number of underreported WVCs is high, as Gkritza et al. (2014), for instance, regrets, although even most the animals die in such an accident.
Traffic in the sky: ranking the hazard bird species to aircraft-collision in Brazil
2024, Ornithology ResearchA dataset of road-killed vertebrates collected via citizen science from 2014–2020
2022, Scientific Data