Fox sightings in a city are related to certain land use classes and sociodemographics: results from a citizen science project

Urban areas are increasing worldwide [1], hence, they will become more important for wildlife, especially for small and middle-sized carnivores [2]. For wildlife various potential habitats exist in urban areas including buildings, streets, squares, gardens, parks and other areas all associated with a wide variety of disturbances by humans. These land use classes are used by wildlife in different ways e.g. gardens as resource of food, parks as hiding places or streets for migration. The red fox (Vulpes vulpes L., 1758) is one of the globally most adaptive and widely distributed carnivore species [2]. Until the 1980s, foxes in urban areas were mainly reported from the United Kingdom, however, since 1985 they are frequently noticed in many other cities in Canada, Australia, Switzerland, Germany, Japan and Austria [2–4, 25, 62]. There are several reasons why urban areas are attractive to foxes [2]. First, there is a high and constant food availability in urban areas. Second, cities provide safety from interspecific competition. Third, urban areas can provide more shelter or den sites compared to rural areas. Fourth, in cities foxes are safe from being hunted as legal regulations usually restrict hunting near houses [5].

Studying fox ecology in urban areas with focus on occurrence, distribution and use of different land use classes is especially important for public authorities when identifying possible risks of human–wildlife conflicts or disease outbreaks. However, despite the increasing trend in urban fox populations there is still little known about associations between foxes and surrounding land use classes. The challenges to study urban foxes with common non-invasive monitoring methods like camera trapping, transect sampling or hair sampling are great due to (i) omnipresence of people and dogs, (ii) risk of theft of the exposed research equipment in the public space [6–8] and (iii) private property and therefore no access for scientists to many urban habitats frequently favoured by animals (e.g., private gardens, industrial areas) [6, 7, 9]. Findings from studies dealing with urban foxes show that home ranges of urban foxes tend to be smaller than home ranges of rural foxes [10] and urban foxes have been shown to being less territorial and more living in family groups than rural foxes due to more stable food abundances (e.g., Bristol 37 individuals/km² [11], Melbourne 16 individuals/km² [12]; in comparison rural Britain 0.16 to 2.62 individuals/km² [13], rural Germany 0.7 to 2.7 individuals/km² [14]). For the city of Zurich, Switzerland, analysis of fox stomach contents showed that more than 50% of an average stomach content was from anthropogenic food resources [15].

The objective of the current study was to test whether a citizen science approach might be suitable to address the above-mentioned challenges regarding research on urban foxes [6, 9, 16]. We define citizen science as scientific research carried out with the aid of interested volunteers [17]. We are aware of the concerns regarding potential biases of citizen science data regarding geographical coverage and data quality. However, in wildlife research, citizen science has a long-standing tradition and is in the meanwhile also scientifically acknowledged [6, 8, 17]. Despite potential bias in geographic coverage due to unbalanced numbers of participants in some regions of a project or uncertainties in data quality when appropriate quality controls are missing when using citizen science, citizen science can be cost-effective when conducting long term monitoring [18]. Only a few studies address human–carnivore encounters in urban areas [9, 19, 20]. These studies found that wildlife sightings depend on habitat use and activity patterns of wildlife, but also on the use and accessibility of different land use classes by humans, and the visibility of wildlife in different habitats [19, 20]. It was shown by comparing radio-telemetry data and public sightings of urban coyotes that public sightings overestimated the use of more open vegetation as habitat compared to forests with short sight distances. Public sightings were biased towards habitats where people concentrated and daylight when people are more active, although coyotes were moving greater distances at night [19]. The positive association of coyote encounters with building densities in another study was due to more people being present in these areas and not due to coyotes using these areas more frequently [20]. Additionally, both humans and wildlife show certain activity patterns in their daily life, which may influence wildlife sightings [21].

The aim of the current study was to assess to what extent sightings of urban foxes by citizens are influenced by the surrounding land use and/or sociodemographic parameters. In this study, we define sightings as human–fox encounters which are reported via our citizen science project website. Additionally, we investigated temporal changes over years, months and daytime in urban fox sightings. To the best of our knowledge, our study is the first to use citizen science as a non-invasive method to study urban fox occurrence and distribution on a large scale and to include sociodemographic data as an explanatory variable [20]. Results should (i) help researchers to establish a large-scale monitoring system for urban areas by using a citizen science method, (ii) inform wildlife managers in urban areas on human–fox-encounters and therefore (iii) lay the foundation for future systems to prevent human–wildlife conflicts as well as spreading of fox-related diseases.

Research in urban wildlife ecology and human–wildlife interactions in cities becomes more important as more and more people live in urban areas [1, 2]. This is the first study analysing occurrence and distribution of urban fox sightings in relation to land use and sociodemographics in a European city using a citizen science approach. Fox sightings were not equally distributed across the year and over months, 51% of fox sightings were made between May and August. This could be explained by a fox population peak with many young foxes present and gradually starting to explore greater areas in these months [23]. It should also be considered that the internet platform “StadtWildTiere” was launched and promoted in the public at the end of May and citizens in general are more active outdoors in summer months. Fox sightings were also reported for every hour of the day (Fig. 1). Between 6 p.m. and midnight about 45% of all reported human–fox encounters took place, whereas between midnight and 6 a.m. only about 22% of the sightings were reported. This distribution of sightings is most likely a consequence of human behaviour and activity patterns, rather than due to fox activity patterns, since foxes are considered to be mainly active at night [24]. Gloor found that urban foxes in Zurich preferred public parks and other areas closed for humans during the first half of the night and used residential areas more, when human activity was low in these areas in the second half of the night [25]. For foxes in Bristol (UK) it was even shown that they crossed less roads before midnight than after, therefore supposedly adapting their activity patterns to reduce mortality risks by roads and avoiding human activity [26]. Fox sightings were reported throughout all districts of the city. Our analyses that fox sightings are affected by land use classes suggest that foxes prefer certain land use classes [20]. High conditional probabilities were calculated for different types of gardens and areas with a low building density, as well as for parks and squares. Our citizen science data are thus in line with telemetry studies on urban foxes [10, 25]. People also had good access to these land use classes and foxes are well visible, although during the day they tend to rest in vegetative structures [25]. Low conditional probabilities for fox sightings were calculated for agricultural areas, a diverse range of small other green areas, as well as factory premises and industrial areas and also forests. Based on several studies, one would assume that the number of sightings of foxes on these land use classes in Vienna was as high as that of gardens and parks, however several aspects should be considered [27–29]. First, a sighting of a fox on an agricultural field or in the forest within the city borders may not be as special for people, as a sighting in their own garden. Therefore, foxes seen on those land use classes might not be reported as often as foxes seen in gardens or in parks within the city. These results seem to be consistent with other research which found that sampling effort can bias results of citizen science projects [30–32]. Second, visibility of foxes in a forest is likely to be worse than in gardens or parks. Third, access to industrial areas and factory premises was limited to operating hours and to people who have access. There is the possibility, that in our project we might not have had enough citizen scientists with access to these land use classes, thus resulting in low conditional probability values. A special land use class category was the zoo, situated in the Schönbrunn castle grounds: while accounting for only 0.04% of the study area, foxes were reported in two out of three grid cells containing the zoo as land use class, resulting in the highest conditional probability value of all land use classes. Reported fox sightings from the zoo are sightings of a fox family, which is quite famous among Viennese people, roaming the premises of the zoo, and are not animals held in captivity. The last aspect to consider is of course the possibility that no foxes were present in areas with no reported sightings.

Similarly, to the conditional probability results, analyses with GLMs indicated that fox sightings increased with increasing area of private gardens, public parks and squares. This again mirrored habitat use by urban foxes on a large scale like found in various studies on smaller scales using other methods [10, 25, 33, 34]. These land use classes provide easy access for foxes to food resources as well as shelter. Additionally, these land use classes are also preferred by humans, which makes a human–fox-encounter more likely. As mentioned above, these results certainly do not indicate that land use classes with no fox sightings, inhabit no foxes. It might also be that non-reports from these land use classes originated in human perception of the land use classes as ‘not urban’, therefore not worthy to report a fox sighting to a project on urban wildlife. This is similar to other studies which refer to ‘reporting bias’ in citizen science projects [35, 36]. The GLM containing only sociodemographic predictor variables showed that education level is highly significant, which indicates that people with a university degree reported fox sightings more often than people with only a compulsory education. Since citizen science in general is not restricted to higher educated people [37], this result can be interpreted in a way that our project promotion was focused on the target group of people interested in wildlife. However, the result is also in line with previous findings showing that some citizen science projects seem to be more attractive to people with higher education (e.g. [38]). This challenge of reaching a broad target audience to avoid bias in data collection in citizen science projects could be addressed by training observers with different educational backgrounds or reaching a broader audience through different public relation activities. However, the problem of reaching a broad audience is existing in science communication as well, is multi-facetted and can only be solved by many parallel activities by scientists, communicators, politicians and NGOs. As expected, district area showed no significant influence on explaining fox sightings and human population density did not remain within the model after a stepwise AIC was performed.

Data quality is a core issue of every scientific research project, especially when citizen scientists are involved [39–44]. More than 60% of the sightings were submitted by citizen scientists without a photo for proof. Nevertheless, we considered submissions without a photo of the reported fox sighting to be sufficient, as the red fox is a well-known species and not easily confused with any other wildlife species living in Vienna. In citizen science projects, variation in observer quality and variation in sampling effort over time and space often pose challenges for data analysis [41, 45–47]. Including observer characteristics in statistical analysis can account for variation in data sets gathered by citizen scientists [21, 48]. In our study the combination of land use classes and sociodemographic data lead to a better model than just land use classes. When lacking information on the knowledge of every single observer, sociodemographic census data have been shown to be an important source of variation in citizen science data [20]. Additionally, 20% of the fox sightings were made in private gardens and other forms of private properties, which would be hard to access for researchers [9]. Therefore, citizen science proofed to be a feasible method to research urban foxes.

Citizen science adds different research possibilities to mammal monitoring in urban areas compared to more traditional monitoring methods like camera trapping and transect monitoring. A citizen science approach to wildlife monitoring is appropriate when interactions with wildlife are central to the research question [9]. This can be of high interest when working in urban areas, as human–wildlife contact is increased in certain areas of cities [49]. When researching urban wildlife, the success of a citizen science project can be affected by the species studied. The red fox is a charismatic well-known species and therefore a suited study model. However, even for urban rats, a species not liked by many people, citizen science is nowadays considered as a research method [50]. Additionally, a new possibility of comparing data from different cities arises, when data on wildlife sightings is gathered through the same project design as it is currently done within the project “StadtWildTiere” in Zurich (Switzerland), Berlin (Germany) and Vienna (Austria). Our findings could also have implications for wildlife management in cities or public health issues. For red foxes in Central Europe, infection with and zoonotic transmission of the Fox tapeworm (Echinococcus multilocularis) is already of interest for urban areas [51–54]. Human–wildlife interactions affect red fox populations as well as predation rate of the infected intermediate hosts of E. multilocularis and should therefore be considered in management strategies of this disease [55].

Despite common reservations against citizen science as a method, our study demonstrated that these can be partly overcome by including sociodemographic factors in the analyses. Taking the results of this study as basis for future citizen science projects in urban areas, we recommend to develop advanced citizen science projects with a broad focus on various target groups to foster the reporting of fox sightings on a large scale. The mostly positive feeling associated with a personal observation of wildlife in the city during a citizen science project could be followed by a more relaxed coexistence of humans and animals in the cities in general. Additionally, such projects would have the potential to predict the likelihood of human–fox encounters in different places of the city to inform public authorities on possible wildlife conflict areas and public health issues.