Introduction

The raccoon dog (Nyctereutes procyonoides, Grey 1834), a medium-sized omnivorous canid, indigenous to East Asia, was introduced to the former USSR in the mid-1900s, where it rapidly established and spread throughout Central and Northern Europe at a rate of approximately 40 km per year, and occasionally up to 120 km per year (Lavrov 1971). In Denmark, the first raccoon dog was recorded in 1980 on the Jutland peninsula (Baagøe and Ujvári 2007), and since then the raccoon dog has established populations throughout Jutland but not yet on the remaining island regions of Eastern Denmark (Miljøstyrelsen 2020), although observations have been made on Funen (Pers. Comm. Mariann Chriél).

Following an increased presence of the raccoon dog across EU member states, the European Union classified the species as invasive in 2017, due to its estimated threat to native biodiversity (European Union 2017). Primary concerns regarding their threat to biodiversity revolve around the negative impacts that the raccoon dog may have on native mesopredator and prey species; particularly regarding the predation of threatened bird and amphibian populations. In addition, the potential for disease transmission exists, such as the rabies virus, tapeworm (Echinococcus multilocularis), roundworm (Trichinella spiralis) and sarcoptic mange (Sarcoptes scabiei) (Kauhala and Kowalczyk 2011), however, there are no known pathogens that have been newly introduced by the raccoon dog to Europe (Pagh and Chriél 2017). The raccoon dog may locally predate ground nesting bird colonies (Koshev et al. 2020; Salewski and Schmidt 2019; Viksne et al. 2011), and may successfully scare away nesting birds as large as graylag geese (Anser anser) (Dahl and Åhlén 2019). This is particularly concerning as ground-nesting bird populations are in severe decline across Europe (McMahon et al. 2020). Amphibian species, particularly those with vulnerable isolated populations are another topic of concern (Miljøstyrelsen 2017; Mulder 2013) as amphibians and their spawn are a common occurrence in the raccoon dog diet, especially during the spring and summer months (Elmeros et al. 2018; Sutor et al. 2010). Nevertheless, there is no evidence that these organism groups are doing worse in parts of Europe long occupied by raccoon dogs than elsewhere. Various studies have examined the niche overlap between the raccoon dog and native mesopredators in Europe, such as the red fox (Vulpes vulpes), badger (Meles meles), pole cat (Mustela putorius), and pine marten (Martes martes) (Baltrunaité 2010; Drygala and Zoller 2013a, c; Elmeros et al. 2018; Kauhala et al. 1998b; Sidorovich et al. 2000a). They were found to indeed have niche overlap, particularly dietary, however, differences in diet were variable enough that only minor competition may occur when resources are plentiful (Drygala and Zoller 2013b; Elmeros et al. 2018; Jędrzejewski et al. 1989; Kauhala et al. 1998a; Sidorovich et al. 2000b). In times of limited resources, and in areas where the raccoon dog does not display winter passivity, competition with native predators could potentially occur (Watabe and Saito 2022). Raccoon dogs may utilize badger setts and fox dens for shelter and pup rearing, and have been observed to peacefully cohabitate with badgers and foxes (Kowalczyk et al. 2008). Although evidence of localized influence on prey species and niche overlap with other mesopredators exists, the general impact that the raccoon dog has on Europe’s native biodiversity remains unclear.

The raccoon dog is an opportunistic generalist, having a wide dietary niche (Castelló 2018; Elmeros et al. 2018), a high reproductive capacity (Helle and Kauhala 1995; Kauhala 1996b; Kowalczyk et al. 2009; Pagh et al. 2020), effective dispersal (Drygala et al. 2010; Herfindal et al. 2016), and an ability to hibernate (Kauhala and Saeki 2004; Mustonen et al. 2007; Ward and Wurster-Hill 1990); all of which collectively attribute to their successful establishment of novel areas. In general, they prefer habitats such as meadows, moist deciduous and mixed forests with abundant understory, river valleys, lakeshores, marshes, and moist heath. However, they are also known to occupy other habitats such as woodlands and agricultural areas (Castelló 2018). Habitat use by raccoon dogs can be affected by the availability of food (Castelló, 2018), shelter and suitable den sites. Raccoon dogs are largely nocturnal; however, they may also be quite active during the day but typically only if they have thick vegetation as refuge (Ikeda et al. 2016; Kauhala et al. 2007; Schwemmer et al. 2021; Zoller and Drygala 2013). In Finland, periods of highest seasonal activity are between March and April, potentially due to the mating season and in response to an increase in food availability. Periods of lowest activity are between November and February (Kauhala et al. 2007), associated with a period of winter dormancy or intermittent wintertime passivity (Mustonen et al. 2007; Süld et al. 2017). During the autumn months, raccoon dogs accumulate extensive fat stores (Korhonen et al. 1991), in preparation for winter dormancy (Kauhala and Saeki 2004; Ward and Wurster-Hill 1990).

Scientific research examining raccoon dog ecology in Denmark is limited, and ecological assessments in different ecoregions will allow for more accurate and effective risk assessments regarding the species’ impact on native biodiversity. Therefore, this study aimed to examine relevant ecological traits of the raccoon dog in Denmark, including habitat preferences and movement patterns such as travel speed, using GPS-based telemetry data, to provide a better understanding of the role that the raccoon dog plays in the ecosystem. Habitat preferences were examined according to the biological seasons, and travel speed was examined according to time of day, time of year and habitat type. Further examination of travel speed was made according to the time of day across biological seasons. Referring to results from similar studies in Europe, we hypothesized that the raccoon dog in Denmark would exhibit distinct shifts in their:

  1. (i)

    Selection of habitat types according to the biological seasons, with an overall preference for high-coverage forests and habitats closely associated with water,

  2. (ii)

    Travel speed according to the time of day between the biological seasons, by exhibiting a unimodal nocturnal activity pattern, and higher speeds during the spring and summer months, and lower speeds during the winter denning months, and

  3. (iii)

    Travel speed according to habitat type, by exhibiting lower speeds in preferred habitats and higher speeds in avoided habitats.

Methods

Study area

The study area includes the Danish peninsula of Jutland as well as the neighboring island region of Funen (56.2639°N, 9.5018°E) (Appendix A – Fig. 6). This is the region of Denmark where raccoon dogs are known to be established. The study area spans approximately 32,800 km2 and shares a border with Germany. The land is heavily cultivated with 65.8% of the land area being dedicated to agricultural practices (Bank 2018), and the remaining proportion of natural land primarily consisting of coniferous and deciduous forests, wetlands, coastal dunes, grasslands, and moors and heathlands. Denmark is low in elevation with average elevation being 31 m above sea level, and has a temperate climate with an average annual temperature of 9.1 °C and extreme temperatures of -23.1 °C and 34 °C (2011–2020) (Danmarks Meteorologiske Institut 2021).

Telemetry data preparation

The telemetry data used in this study was collected via GPS-based satellite transmitters in the form of radio-collars which were placed on feral raccoon dogs that were trapped and released back into Danish nature on both Jutland and Funen for monitoring purposes over the course of 10 years (2011–2020). All trapping, collaring, and handling of raccoon dogs occurred prior to the commencement of this study. The collared raccoon dogs were released into areas that were deliberately chosen by the monitoring program and were also manually moved from place to place on occasion. This monitoring of raccoon dogs is part of a national program run by the Environmental Protection Agency—Ministry of Environment of Denmark (Naturstyrelsen 2010). The primary objective of this monitoring effort was to utilize the knowledge of raccoon dog monogamous pairing behavior to locate others to facilitate more effective eradication efforts. Individuals that were collared were adults and the sex of the individual was only recorded if they were used on a long term basis. The geospatial telemetry data was collected via an enterprise known as “FollowIt” (Followit 2018).

The geographic location of each collar was recorded at various time intervals with the majority being recorded at 10-min, 2-h and 3-h time intervals. The coordinates were registered in decimal degrees formatting, in addition to the date and time of point registration. The coordinates were converted to UTM format prior to analysis and mapped in qGIS (version 3.1). Before using the raw coordinate data, coordinate duplicates, points registered outside Jutland and Funen (except those registered in northern Germany, indicating movement between Denmark and Germany), locations where raccoon dogs were known to be handled, and if the interfix speed exceeded 20 km per hour (indicating car travel), were removed.

Habitat selection

Telemetry and land cover data preparation

For the habitat selection analysis, telemetry data registered within Denmark only was used, and the first three days of each individual raccoon dog’s telemetry data were removed from the dataset to minimize bias in habitat choice associated with the deliberate human placement of the individual. To categorize habitat types, a land cover raster layer of Denmark was acquired from the Sentinel-2 Global Land Cover project (Malinowski et al. 2020). In Denmark, 10 land cover classes were present: artificial surfaces and constructions, cultivated areas, broadleaf tree cover, coniferous tree cover, herbaceous vegetation, moors and heathland, marshes, peatbogs, natural material surfaces (f.ex. river pebble banks, sand dunes, beaches), and water bodies (full descriptions found in Appendix A - Fig. 5). The overall accuracy of this land cover estimation was estimated at 94% (Malinowski et al. 2020). To limit the raster layer to only include areas that raccoon dogs can physically access, the layer was cropped further to the size of the Danish land mass with a 3 km coastal buffer, which included portions of the ocean and fjords (see Appendix A - Fig. 6). The buffer range of 3 km was chosen because there were a few telemetry points located within this distance from the coastline, specifically within the fjords.

Analysis

Habitat selection was analyzed across four seasons, which were defined according to previous literature (Drygala et al. 2008; Zoller and Drygala 2013). The seasons include oestrus and gestation (OG, March – April), birth and cub-rearing (BCR, May—July), extensive foraging and fat accumulation (EF, August – October) and reduced activity (RA, November – February). To determine if any particular habitat type was being either selected or avoided, an electivity index (Jacob’s D) was calculated for each individual within each habitat type (Manly et al. 1993). The formula used for calculating this index is as follows:

$$D=\frac{\left(r-p\right)}{r+p-2rp}$$

where r represents the proportion of habitat used, and p represents the proportion of habitat available. The electivity index (D) may range from -1 to + 1, where negative values indicate avoidance and positive values indicate selection. A value of 0 indicates neutrality, with neither selection nor avoidance. The proportion of habitat available (p) was calculated uniquely to each individual based on the proportion of each habitat type found within their unique utilization distribution (UD). The UD was calculated using kernel density estimation, using a 95% isopleth level and reference bandwidth. To ensure an accurate UD estimation, only individuals with a minimum of 40 telemetry points were included in this analysis (Kauhala et al. 2010; Seaman et al. 1999), resulting in a total of 129 individuals with the mean number of telemetry points per individual being 1,900, the minimum being 41 and the maximum being 5,992. To ensure that the UD represented areas raccoon dogs could realistically access, the UD was cropped to have the same boundaries as the land cover raster with a 3 km coastal buffer (see Appendix A - Fig. 7b). A land cover raster layer was created for each individual that corresponded to their UD size, and the proportion of each habitat type available was determined by extracting the raster values from their unique raster layers and dividing the number of each habitat’s raster cells by the total number of raster cells. This method assumes that all geographic areas within the individual’s UD are equally accessible to the individual. Each individual’s telemetry data was subset according to season and the UD was estimated according to this data (for an example, see Appendix A - Fig. 8). For individuals monitored over multiple years, the seasonal data was pooled together for all years.

The proportion of habitat used (r) was defined as the number of telemetry points for each individual found within each habitat type, divided by the total number of telemetry points of that individual. Only the telemetry points found within the individual’s UD were used. Once the proportion of habitat used and available were calculated, the electivity index was calculated for each habitat and individual separately.

Statistical analysis

To determine if habitat use was selective or random, the selection index value for each habitat type across four seasons was tested for being significantly different from the value of 0. The Anderson–Darling test was used to test for normality, and a One Sample T-test was used for normally distributed data, and a Wilcoxon Signed Rank test was used for non-normal data.

To determine if the selection pattern for each habitat type varied according to season, linear mixed effect models were built for each habitat type separately (see Appendix A - Table 3 and Statistical modelling for model details and methodology). To test if the fixed effect Season had a significant influence on the response variable Selection Index, Wald Chi-Square tests were performed when normality assumptions were met, whereas the Kruskal Wallis Rank Sum Test was used when normality assumptions were not met. If significant, Multiple Comparisons of Means with Tukey Contrasts was used for the normal data, and the Dunn test for the non-normal data, to determine which seasons differed from one another.

Travel speed

Telemetry data preparation

For the analysis of travel speed the first three days of data collection and the data acquired from individuals that traveled between Denmark and Germany’s border were included. To examine each individual’s path of movement, a ‘track’ was built using the make_track function of the amt package using R software. This track consisted of all registered fixes placed in consecutive order by time per individual. The track for each individual was converted into a ‘step’ format, using the steps function of the amt package which compiles two consecutive telemetry points into one ‘step’ using ‘step length (straight-line distance)’ and ‘step time duration’ (hereafter referred to as ‘interfix time’ (IT).

Analysis

We used the animal’s rate of travel (speed, in meters/hour) to infer overall activity patterns according to (Ensing et al. 2014; Merrill and Mech 2003; Owen-Smith and Goodall 2014; Palomares and Delibes 1993). Through the examination of movement intensity, we can generally infer the type of activity that the animal is executing, with slower speeds indicating rest, or intense foraging and higher speeds indicating directional movement activities, such as dispersal. For each individual, the rate of travel was calculated for each step of their track. The rate of travel for each step was calculated as follows:

$$Speed=\frac{step\,length\,(m)}{interfix\,time\,(hr)}$$

The IT was chosen based on prioritizing the lowest possible IT to optimize accuracy of speed calculations while still maintaining a high enough sample size for further analysis. The majority of the telemetry data was collected using ITs of 3 h, 1 h, 6 h and 10 min (in that order). There was a 2-min buffer around each IT, therefore the ITs included in this analysis are of a 2-min range (see Appendix B - Fig. 9 and Table 7). Only individuals with a minimum of 50 ‘steps’ were used in the analysis.

Travel speed was examined according to two main predictors: the time of day between seasons (defined by 24 h across four seasons) and habitat type (defined by 10 habitat types). The seasons were defined according to the raccoon dog’s biological seasons. Due to sample size constraints, the seasons for this analysis were redefined as OG (April), BCR (July), EF (October) and RA (January).

To optimize the analysis in response to sample size constraints, different ITs were used according to the hypothesis being tested. To choose the most appropriate IT for each hypothesis, priorities were given to choosing that i) which was lowest, ii) provided a high enough overall sample size, iii) and provided a relatively well-balanced design. For examining Speed according to time of day between seasons, an IT of 10 min was used (Appendix B - Tables 8 and 9), according to habitat type an IT of 3 h was used (Appendix B - Table 10). To understand the methodology behind these choices please refer to Appendix B (Methodology).

The habitat type variable represented the same ten land cover categories as in the habitat selection analysis and was developed by extracting the raster values from the land cover raster for each corresponding ‘step’ that was used to calculate each speed value. Since each ‘step’ used to estimate speed is comprised of two consecutive telemetry points, one of the telemetry points was chosen to extract the raster values to. It was determined that using either the first or last telemetry point of the ‘step’ did not make a difference in the percent variation explained of the Speed variable, so the habitat type of the first telemetry point of the ‘step’ was used to create the variable.

Statistical analysis

All models used in this analysis were linear mixed effects models (see Appendix B - Table 11 and Statistical modelling for model details and methodology). The residuals of all models were not normally distributed; therefore, response variables were log10 transformed to meet the assumptions of a linear mixed effects model. Post hoc pairwise comparisons for the model terms of interest were performed by estimating the least square means (otherwise known as estimated marginal means) for the fixed effect term of interest and then performing pairwise comparisons on these means (using the ls_means function of the lmerTest package (Kuznetsova 2020)). Least square means were used due to the unbalanced nature of the experimental groups and to account for the presence of other influential variables within the model (Mangiafico 2016).

Results

Habitat selection

Availability and usage of habitat types

The largest proportion of land area was cultivated areas (37.3%), water bodies (19.6%) and herbaceous vegetation (19.3%). While the most raccoon dog telemetry points were found in peatbogs (24.5%), marshes (21.2%), cultivated areas (17.6%), herbaceous vegetation (14.4%), and broadleaf tree cover (11.3%) (Fig. 1, Appendix A – Table 1).

Fig. 1
figure 1

Proportions for each habitat type that is available and used by the raccoon dog according to the entire study region in Denmark and entire telemetry dataset. Habitat available represents the proportion of the land cover raster that is represented by each habitat type and the habitat used represents the proportion of telemetry points found within each habitat type from the entire telemetry dataset

Full size image

Selection according to season

Throughout the year, the raccoon dog consistently preferred peatbogs, marshes, and broadleaf tree cover and consistently avoided water bodies, natural material surfaces, cultivated areas, and artificial surfaces and constructions. The degree of preference or avoidance of these habitat types varied only slightly between the seasons. In contrast, the raccoon dog’s selection patterns in coniferous tree cover and herbaceous vegetation varied between preference and avoidance throughout the year, with these habitats being avoided during the summer months and preferred or neutral during the winter months (Fig. 2, Appendix A – Tables 2, 3, 4 and 5).

Fig. 2
figure 2

Median habitat selection index of the raccoon dog in Denmark observed in 10 habitat types across four seasons. Seasons are defined by four main distinctions in raccoon dog ecological behaviors: OG = oestrus and gestation, BCR = birth and cub-rearing, EF = extensive foraging and fat accumulation, and RA = reduced activity. Index values above 0 indicate preference of the habitat, whereas values below 0 indicate avoidance, and values at 0 indicate neutrality. Error bars indicate 95% confidence intervals. Significance is indicated by asterisks: * p < 0.05, ** p < 0.01, *** p < 0.001, and indicates if the index value significantly differs from 0

Full size image

Overall variation in the median selection or avoidance of habitats observed between the seasons was highest in coniferous tree cover (ΔSI 0.62), followed by moors and heathland (ΔSI 0.57), cultivated areas (ΔSI 0.43), herbaceous vegetation (ΔSI 0.35), broadleaf tree cover (ΔSI 0.20), artificial surfaces and constructions (ΔSI 0.13), peatbogs (ΔSI 0.11), marshes (ΔSI 0.08), water bodies (ΔSI 0.02), and lastly natural material surfaces (ΔSI 0.0). Overall selection patterns across all habitats were most similar between the BCR and EF seasons, and most dissimilar between BCR and RA (Appendix A - Table 6).

Travel speed

Raccoon dogs traveled with an overall mean speed of 402 m/h ± 799 SD (IT: 10 min, N = 30), 229 m/h ± 467 SD (IT: 1 h, N = 115), and 280 m/h ± 500 SD (IT: 3 h, N = 111) (Appendix B – Table 7).

Time of day

The time of day played a significant role in explaining variation in travel speed (Appendix B – Table 12), with there being a unimodal nocturnal trend. Nocturnal behaviour was observed consistently throughout each biological season with minor variations; the most notable occurring during the peak hours of movement. Daily travel speeds were lowest during RA, followed by EF, then BCR, and highest during OG. Higher speed levels persisted for longer during EF, and significantly higher speeds occurred during the peak hours of BCR and OG, particularly when compared to RA (Fig. 3, Appendix B – Tables 13, 14 and 15).

Fig. 3
figure 3

Mean speed (least square) of raccoon dog movement in Denmark in meters per hour according to time of day across four seasons. Seasons are defined by four main distinctions in raccoon dog ecological behaviors: OG = oestrus and gestation, BCR = birth and cub-rearing, EF = extensive foraging and fat accumulation, and RA = reduced activity. Data was collected using an interfix time of 10 min. Least square mean values are in the log-10 scale. Error bars indicate standard error values

Full size image

Habitat type

Raccoon dog travel speed varied significantly between habitat types, with the highest mean speeds being in artificial surfaces and constructions, water bodies, natural material surfaces and cultivated areas, and the lowest speeds being in peatbogs and marshes. (Fig. 4, Appendix B – Tables 16, 17, 18 and 19).

Fig. 4
figure 4

Mean speed (least square) of raccoon dog movement in Denmark in meters per hour according to habitat type. Data was collected using an interfix time of 3 h. Least square mean values are in the log-10 scale. Error bars indicate standard error values. Letters indicate significant differences between habitat types

Full size image

Discussion

Reliability of the method

Habitat selection

Autocorrelation of the telemetry points may be a challenge in telemetry data, however, studies have found that the benefits associated with controlling for autocorrelation, do not outweigh the costs of not controlling for autocorrelation, as it can severely reduce the sample size and also limit the biological significance of the analysis (De Solla et al. 1999; Fieberg 2007). Lastly, the inability to analyze differences in habitat selection patterns according to sex, was another limitation, as there was a limited number of individuals that were sexed within the dataset, and there was also a strong bias of males being monitored as opposed to females.

Travel speed

The complications associated with interpreting speed values estimated at different IT´s lie with making accurate distance traveled estimates between the telemetry points used to estimate speed. The distance traveled between the two points is measured as a straight-line distance, which often is not representative of the actual trajectory that the animal travels. Common biases associated with straight-line distance estimation is the sampling frequency of the points, and tortuosity of the path traveled – which can vary incredibly according to various factors. Animals with more directional movements (lower tortuosity), may be in the process of dispersing, whereas an animal with nondirectional movements (higher tortuosity), may be considered established (Herfindal et al. 2016). The straight-line distance traveled can vary substantially according to the sampling frequency used where distance traveled is increasingly underestimated, and that several fixes per minute are required to achieve tolerably accurate estimates (Noonan et al. 2019; Rowcliffe et al. 2012; Sennhenn-Reulen et al. 2017). Research has shown that with a one minute sampling interval, distance was underestimated by 6.3% on average, and with a 2 h sampling interval, distance was underestimated by 32.3% (Sennhenn-Reulen et al. 2017). It was therefore a priority to use the lowest IT possible if sample size allowed.

Habitat selection

The preference for peatbogs, marshes and broadleaf tree cover corresponds to various other studies within the raccoon dogs introduced range (Baltrunaité 2010; Kauhala and Auttila 2010; Melis et al. 2015; Mustonen et al. 2012). Preference for these habitats was observed year-round, indicating their importance throughout the year as habitats that are providing their main source of food. The overall avoidance of artificial surfaces and constructions, and water bodies corresponds to other research (Drygala et al. 2008; Sutor and Schwarz 2013). Overall selection patterns across all habitats were most similar between the birth and cub rearing and extended foraging seasons, and most dissimilar between birth and cub rearing and reduced activity seasons. This dissimilarity was expected, as in Denmark—resource availability is highest during the months of birth and cub rearing and lowest during the months of reduced activity. During periods of lower resource availability, it would be most advantageous for the raccoon dog to broaden its habitat selection patterns to optimize sufficient resource and shelter acquisition.

During the reduced activity season, raccoon dogs showed markedly different selection patterns compared to the other seasons, specifically within four habitats: broadleaf tree cover, coniferous tree cover, herbaceous vegetation, and moors and heathland. They used these habitat types more during the winter than during the other three seasons. Similar preferences for coniferous and deciduous forests during winter were found in Germany, Finland, and Lithuania. The higher preference for broadleaf and coniferous tree cover during winter in Denmark may indicate that they are utilizing this habitat type for denning, shelter and resting purposes as well as for food, considering these forests offer berries year round, which are known to be a reliable food source prior to and throughout winter (Elmeros et al. 2018; Kauhala 1993; Sutor et al. 2010). In Sweden, the raccoon dog selected coniferous forests during winter but selected against the habitat during the rest of the year (Melis et al. 2015). In Germany, raccoon dogs were observed utilizing abandoned badger setts in coniferous forests all year round, for the purpose of daytime resting places, shelter from bad weather, winter resting, and for cub-rearing (Sutor and Schwarz 2012), and in Poland, an increased use of burrows and hollow trees (which are primarily found in forests) as den sites occurred as mean daily temperature decreased (Kowalczyk and Zalewski 2011). In Denmark, used badger setts are more likely to be found in areas of high forest coverage (Jepsen et al. 2005).

The shift from avoidance to an almost neutral selection pattern for herbaceous vegetation and moors and heathland during the winter indicates that during times of lower resource availability, these habitat types become useful to the raccoon dog. In Finland, raccoon dogs preferred moist heaths during the winter months due to the availability of berries (Kauhala 1996a), and in Sweden the raccoon dog showed a neutral selection pattern for heathlands and meadows during the winter but selected against these habitats during the rest of the year (Melis et al. 2015). In Germany, it was found that grasslands were a preferred habitat throughout the year, as these conditions are suitable for various prey species of the raccoon dog, such as small mammals, birds and invertebrates (Sutor and Schwarz 2013). Certain insectivores (primarily shrews) and birds are frequently consumed prey items during the winter in Denmark (Elmeros et al. 2018), which may explain the shift to neutral selection pattern of herbaceous vegetation in our study. This shift in preference may indicate an increased use of areas located closer to human settlements, such as mowed yards and gardens. Artificial surfaces, which was classified largely as human settlement areas and major highway systems, were also utilized slightly more during the winter months. This slight increase in usage of artificial surfaces during the winter may be due to the opportunistic foraging of carrion occurring near human settlement areas. In Finland, (Mustonen et al. 2012) found that anthropogenic areas such as gardens, roads, and railroads were all preferred habitats of the raccoon dog during winter. The preference for gardens was supported by their winter diet analysis which indicated consumption of anthropogenic food sources, which has been observed in other areas of Europe (Mustonen et al. 2012; Süld et al. 2017; Sutor et al. 2010).

During the birth/cub-rearing season, four habitats; broadleaf tree cover, coniferous tree cover, herbaceous vegetation and moors and heathland, saw an overall decrease in usage. The opposite trend was observed within cultivated areas where usage during birth and cub rearing increased, however it was still avoided. In Germany, a similar result was found, where raccoon dogs avoided farmland areas except during the cub-rearing season where there was a neutral selection pattern for the habitat, and a slight difference between the sexes was found with selection of cereal fields where females preferred this habitat and males did not (Drygala et al. 2008). Cultivated areas may become a more useful habitat for females during the cub-rearing season, while she is out foraging to maintain her milk supply, or it may simply be a matrix habitat that may be located near the den-site. During cub-rearing, habitat preference may not only depend on rich habitats in relation to basic diet, but also the need for a place of refuge for the cubs. The need for a safe denning place may limit habitat use to be more local around the den.

Travel speed

We predicted that the raccoon dog in Denmark would display a unimodal nocturnal activity pattern with minor shifts throughout the year, with lower speeds during the winter and higher speeds during the spring/summer. Our results showed that the raccoon dog indeed displayed such patterns, with peak speeds occurring at night, and minimal movement occurring during the day, across all four seasons. This has been observed in a variety of studies across the raccoon dog’s introduced range (Drygala et al. 2001; Kauhala et al. 2007; Schwemmer et al. 2021; Zoller and Drygala 2013), and native range (Ikeda et al. 2016; Seki and Koganezawa 2011, N. viverrinus subspecies). Diurnal activity was observed as well, although much less frequently. Changes in diurnal activity has been observed between the seasons in other studies, where the proportion of daytime movements increased as mean daily air temperatures decreased (Seki and Koganezawa 2011), and where diurnal activity was more prevalent during the cub-rearing season and correlated positively with increasing day length (Zoller and Drygala 2013). We observed an increase in diurnal activity during the birth/cub-rearing season where speed was significantly higher than all the other seasons during hours 06, 07 and 12, which corresponds to other studies (Ikeda et al. 2016; Ogurtsov et al. 2018; Zoller and Drygala 2013). This increased level of diurnal activity is likely due to the 24-h care associated with rearing cubs. Other minor variations in daily movement patterns were observed seasonally, with the lowest speed levels occurring during the reduced activity season, and highest occurring during oestrus/gestation and birth/cub-rearing seasons. Mating can occur between February – March (Helle and Kauhala 1995; Miljøstyrelsen 2020), when male spermatogenic cells (Qiang et al. 2003) and testosterone (Rudert et al. 2011) reach their peak levels. This increased travel speed during the month of April relative to the other months may be a result of mating behavior. Raccoon dogs may also exhibit an increase in foraging activity following an extended period of reduced activity over winter. During the extensive foraging/fat accumulation season higher speed levels were maintained for a longer period, which is likely due to an increase in time spent foraging in a 24-h period, in preparation for winter.

We predicted that the raccoon dog would travel slower in habitats that were preferred and faster in habitats that were avoided, when compared to the habitat selection analysis. According to ‘area restricted search’ behavior, an animal will remain in an area longer where resources are high as opposed to areas where resources are low, and will also move more slowly and tortuously in such high resource areas (Turchin 1998). Our results correspond well with this theory and our prediction, with raccoon dogs moving faster within their avoided habitats and slower within their preferred habitats. In Japan, a similar pattern was observed, where the raccoon dog displayed slower speeds in habitats that were favored (Saeki et al. 2007). Our results provide indirect evidence of raccoon dog habitat preferences and supports the results from our habitat selection analysis.

Conclusion

In conclusion, the raccoon dog in Denmark possesses ecological traits that are similar to their conspecifics found elsewhere in their introduced range. They prefer moist habitats and high coverage forests year-round and will additionally favor coniferous forests and heathlands during winter. They exhibit a unimodal nocturnal activity pattern throughout the year with highest speeds occurring during spring and summer and lowest during winter. They are capable of shifting their habitat use and movement intensity according to season in response to environmental stimulus, and with a gradually warming climate, these patterns may shift during the milder winters to come. With their overall level of ecological plasticity, the raccoon dog is likely capable of adapting to the ongoing climatic changes locally in much of Europe and expand their current distribution even further. Having a detailed understanding of how the raccoon dog interacts with its environment on a local scale can inform assessments regarding their likely future presence and space use as well as any associated ecological effects in the study region and potentially also more broadly.