Introduction

Urbanisation is a rapidly increasing threat for global biodiversity (Seto et al. 2012; McDonald et al. 2019; Simkin et al. 2022) and urban wildlife have to tolerate multiple stressors. Yet some species have successfully established dense populations in urban habitats (Luniak 2004; Hubert et al. 2011; Francis and Chadwick 2012). Urban stressors are most intense in the city centre (Grimm et al. 2008; Bateman and Fleming 2012). However, cities are composed of a heterogeneous mosaic of habitat patches and spatial scales can have considerable influence on wildlife responses (de Satgé et al. 2019; Alberti et al. 2020; Uchida et al. 2020). An important driver of high urban population densities is the higher availability of food (Shochat et al. 2006; McCleery 2010). Natural food sources in urban environments are strongly modified by humans due to vegetation cover being reduced and distributed in patches; primarily concentrated in green spaces such as parks and private gardens (Singh et al. 2018). However, urban habitats and vegetation can also offer benefits for wildlife. For example, planting of ornamental plants and neophyts as well as the heat island effect lead to higher plant diversity, earlier flowering and extended growing seasons. These processes can result in a more continuous natural food availability (Shochat et al. 2006; Pickett et al. 2011; Singh et al. 2018).

Urban habitats also provide access to various intentional or unintentional food sources from humans, further contributing to a more stable food availability throughout the year (Shochat et al. 2006). Supplemental feeding can have diverse impacts on wildlife ecology and biology – both on the individual and the community level (Robb et al. 2008; Birnie-Gauvin et al. 2016), which are not necessarily advantageous. Abundant food sources in urban systems might not meet the nutritional needs of a species or even be detrimental (Birnie-Gauvin et al. 2017). Vertebrate species that receive supplementary food show higher population densities and body masses in general (Boutin 1990). The former is likely a result from increased birth rates, litter size and survival of juveniles and adults, affecting not only density, but also structure of populations (Dantzer et al. 2020). However, possible negative effects range from increased disease transmission and loss of natural behaviour to reductions in gut microbiome or community restructuring (Robb et al. 2008; Galbraith et al. 2015; Gillman et al. 2021). Artificially increased food abundance thus might act as false cue of habitat quality, thereby creating an ecological trap for urban species (Reynolds et al. 2017; Zuñiga-Palacios et al. 2021).

Given the local heterogeneity of urban green spaces and the high proportion of non-native and ornamental plants in parks and gardens, the availability of suitable natural energy rich and nutritionally adequate food can vary substantially between patches (Pickett et al. 2011; Seress and Liker 2015; Lepczyk et al. 2017). Moreover, utilization of urban green spaces by humans and their willingness to feed wildlife can also differ between urban patches, e.g., in relation to socio-economic status (Fuller et al. 2008; Hassell et al. 2021). An increasing number of studies investigate urban wildlife, however deeper insights into how green spaces within the same city core area differ in availability and composition of food sources in combination with mammalian population parameters, remain scarce.

The Eurasian red squirrel (Sciurus vulgaris, henceforth named “squirrel”) is a synurban species; able to successfully inhibit urban habitats (Luniak 2004; Jokimäki et al. 2017; Fingland et al. 2021). This arboreal small mammal is naturally forest dependent and favours seeds and nuts in its diet. However, as a food generalist, the species can successfully exploit urban resources (Reher et al. 2016; Fingland et al. 2021; Wist et al. 2022). Food availability is known to influence population demography in squirrels, as well as several other traits, such as home range, body mass and reproduction (Wauters et al. 2007, 2008; Di Pierro et al. 2010; Reher et al. 2016; Selonen et al. 2016). Urban squirrels also show behavioural adaptations to the urban environment and the ever-abundant presence of humans, for example, by altering their anti-predator behaviour or activity patterns (Uchida et al. 2019; Beliniak et al. 2021; Fingland et al. 2021). Urban green spaces, such as parks, are expected to act as suitable refuges for squirrels (Rezouki et al. 2014; Kopij 2015; Beliniak et al. 2022). However, most of these results were obtained from comparably large parks or small cities. We therefore focused on three small urban parks located in the centre of a big city approaching 2 million inhabitants. We deliberately chose parks that were similar in many environmental conditions and investigated the availability of natural and anthropogenically provided food sources. Furthermore, we examined population parameters of Eurasian red squirrels within the parks to explore whether they were influenced by heterogeneity in food abundance. Our results help to elucidate the interplay of urbanisation and local urban food abundance with mammalian population parameters.

Methods

Study sites

We conducted field work in northern Germany across three urban parks in the centre of Hamburg (Wohlerspark, Schanzenpark and Jacobipark; Fig. 1; Table 1). With around 1.9 million inhabitants (2,455 per km2) and an area of 755 km2, Hamburg is the second largest city in Germany. All parks were similar in several environmental factors. They were all small, with a size ranging from 4.6 to 7.4 ha, located in area with very high human population density and surrounded by a similar type of urban matrix (Table 1). In detail, we chose these parks using following criteria: (1) located in the city centre (6–9 km from the closest city limits), (2) area < 10 ha, (3) surrounded by a dense urban matrix (at least 2 large streets closer than 250 m plus large residential or commercial buildings directly adjacent to the park), and (4) high degree of diverse human utilization, such as sports, dog walking and various other outdoor activities (own observations prior to this study). In order to better assess regional conditions and population parameters, we also captured squirrels in a mixed forest about 30 km away (Hahnheide) for comparison purposes (Fig. 1; Table 1). The forest site was a nature conservation area with a high proportion of conifers and old trees (Bundesamt für Naturschutz 2019).

Fig. 1
figure 1

Map of the city of Hamburg and the four study sites: Wohlerspark, Schanzenpark and Jacobipark located in the city centre and the forest site Hahnheide outside the city

(source: OpenStreetMap 2023 and Natural Earth 2023)

Full size image
Table 1 Overview of study sites with total size (ha), human population density (including a ~ 1 km2 catchment area around the siteA) and trap days (product of number of traps by number of capture days) per site and season
Full size table

Natural food abundance

We compared tree species abundance and diversity per study site by using tree data obtained from Hamburg authorities (tree cadastres 2017, Freie und Hansestadt Hamburg), and the forestry office of Schleswig-Holstein (tree ledgers 2012, Schleswig-Holsteinische Landesforsten) respectively. Furthermore, we counted and classified tree seeds within two 1 × 1 m random seed plots at each trap location (20 × 20 m around the trap) at each study site in the summer season. This included fallen seeds as well as seed remains, as a measure of relative seed abundance per site (Wauters and Lens 1995). We further used literature values from the species or genus (Suppl. Table 1) for assessment of available energy-content in kJ per seed.

Supplemental food sources

For assessment of supplemental food sources, we counted and weighed anthropogenically provided food items per trapping period at each study site. We used mean values from three different days per season whenever possible as human activities varied with weather conditions and weekday. To collect these samples, we walked all paths of each park, observing the shrubbery on both sides and inspected open lawns. In the forest, we walked the whole trapping grid, but never observed supplemental food. Food was weighed to the nearest 0.5 g with a spring balance (KERN & SOHN GmbH, Balingen-Frommern, Germany). If weighing was not feasible (e.g. due to non-accessible position), the approximate weight was estimated by either comparing the amount with already measured food items (e.g. ~half fat ball) or by reproducing the set in the lab and then weighing it (e.g. a thin layer of oats 20 × 30 cm in size). We included all kinds of anthropogenically provided foods, since initial observations and a food choice experiment revealed that urban squirrels forage on a broad range of supplemental foods (Wist et al. 2022). We assigned the different food items to broader categories (e.g. bakery products or vegetables) for further analysis. The bird food category included all kinds of bird foods, such as seed mix, fat balls or scattered oats. Peanuts as high energy food were included into the “nuts or peanuts” category. We used literature values (Suppl. Table 2) to estimate the kJ content of the anthropogenic food items.

Trapping and animal handling

We trapped squirrels in all sites in summer 2017 or 2018 respectively. Furthermore, we trapped squirrels in the two most contrasting sites Wohlerspark and Hahnheide from winter 2016/2017 to autumn 2017/2018. We baited cage traps (20 × 20 × 50 cm; Tomahawk Live Trap, Hazelhurst, Wisconsin, USA) at 07:30 h with a mixture of seeds and nuts and kept them open for approximately 7 h. A longer trapping effort per day was not feasible due to the high degree of human activities in the later afternoons in the parks. Traps were placed around 100 m apart (Magris and Gurnell 2002), at approximately 1.2–1.5 m height in the park shrubbery or on trees and checked regularly. Park traps were prebaited for 1–2 days prior to each trapping period. Forest traps were prebaited every few days for one month as forest individuals are not as accustomed to search for foreign food sources.

We used a cloth cone for handling non-anesthetized trapped squirrels (Koprowski 2002). Individuals were weighed with a spring balance to the nearest 5 g (KERN & SOHN GmbH, Balingen-Frommern, Germany), sexed and aged (adult, subadult, juvenile) depending on their body mass, length and inspection of reproductive organs (Magris and Gurnell 2002; Wauters et al. 2007). The nose-anus length was determined to the nearest 0.5 cm by using a measuring tape (Wauters and Dhondt 1989a, b; Magris and Gurnell 2002; Wauters et al. 2007). Finally, the individuals were marked via subcutaneous implantation of a passive integrated transponder (ID-100B; Trovan Ltd., East Yorkshire, UK) between the shoulder blades. Animals were released immediately at the capture site after around 10 min of handling time.

Population density

We used the minimum number of individuals known to be alive (MNA) per site in one season as estimate of population density, a common method for squirrels (Magris and Gurnell 2002, Gurnell et al. 2004, Wauters et al. 2008). Furthermore, we calculated an index of relative abundance (MNAind) per site and season for adequate visualization and comparison, as exact estimates were not possible (Jokimäki et al. 2017):

$${MNA}_{ind}\frac{MNA}{trapping days*number of traps*ha*100}$$

Statistics

Statistics were performed in R 4.3.1 (R Core Team 2023) within the “RStudio” environment (RStudio Team 2023). Graphical data exploration was performed with the packages ‘lattice’ (Sarkar 2008) and ‘ggplot2’ (Wickham 2016). We used the ‘psych’ package (Revelle 2023) for descriptive statistics and Shapiro-Wilk test or Levene’s test respectively for testing normality and equality of variance.

We compared tree species diversity at the four study sites by using the relative abundance in percent per tree species and site since absolute tree numbers were not available for the forest. We then calculated the Shannon-Wiener Index (H’) per site (package ‘vegan’, Oksanen et al. 2013). Despite data transformation, kJ content from fallen seeds in the counted seed plots (nJakobipark = 18, nSchanzenpark = 12, nWohlerspark = 19, nHahnheide = 38) did not follow a normal distribution or equality of variance (Shapiro-Wilk test and Levene’s test, both p < 0.05). We therefore used the non-parametric Kruskal Wallis test with kJ content of fallen seeds per seed plot as dependent variable to investigate differences in natural food availability between sites. We used Dunn’s test with adjusted p values for post-hoc comparisons (package ‘FSA’) (Ogle et al. 2023). Unless otherwise stated n represents the number of individuals / seed plots per site or group. A level of p < 0.05 represents statistical significance.

Results

Natural food sources

Number of tree species was highest in the smallest park (Wohlerspark) with 45 species in total (10 per ha), followed by Schanzenpark with 36 species (5 per ha) and Jacobipark with 32 species (5 per ha). The forest study site had 11 tree species (0.3 per ha). Among the urban sites, Wohlerspark also had the highest number of trees per ha (136), which was around 1.5-times the trees per ha of Schanzenpark and 3 times that of Jacobipark. All parks were higher in tree diversity when compared to the forest site (Table 2). However, among the parks, Wohlerspark had the lowest tree diversity (Shannon-Wiener-Index H’ 2.1 vs. 2.7 in the other two parks). See Table 2 and Suppl. Table 3 for tree details. With regard to tree composition, Wohlerspark was dominated by lime (Tilia spec.) which accounted for around 55% of the park trees, whereas total proportion of conifers was only 8%. Schanzenpark was dominated by maple (Acer spec., 27%) and lime trees (Tilia spec., 26%), with only 3% conifers in the park. The most common trees in Jacobipark were European white birch (Betula pendula, 20%), maple trees (Acer spec., 20%) and oaks (18%, mostly Quercus robur). Conifers accounted for 9% of the Jacobipark trees. In the forest, European beech (Fagus sylvatica, 36%) and spruce (34%, mostly Picea abies) were the most abundant trees and the proportion of conifer trees was 59%.

Table 2 Overview of trees at the four different study sites
Full size table

kJ availability from seeds in summer differed significantly between sites (Chi2 = 41.45, df = 3, p < 0.001). Pairwise post-hoc Dunn’s test revealed higher kJ availability in the forest vs. Schanzenpark (p < 0.001), but also vs. Wohlerspark (p < 0.001). Furthermore, among the urban sites, Jacobipark had higher kJ availability from fallen seeds than Schanzenpark (p < 0.001) or Wohlerspark (p < 0.001, Fig. 2).

Fig. 2
figure 2

Available kJ from fallen seeds in the seed plots (nJakobipark = 18, nSchanzenpark = 12, nWohlerspark = 19, nHahnheide = 38) at the four different study sites. Boxes show the upper and lower quartiles (box limits) with the median as centre line. Whiskers indicate the 5 and 95% percentiles and outliers are represented as circles. Different letters indicate significant differences between sites (Dunn’s test:p ≤ 0.001)

Full size image

Supplemental food sources

Anthropogenically provided food sources at the four study sites in summer ranged from zero supplementation in the forest to 1,683 g found in total on an average day in the highest supplemented park, corresponding to an available energy of 6,537 kJ (Table 3). In total, we found 13 different food items in Wohlerspark, three in Jacobipark and two in Schanzenpark (Suppl. Table 4). The availability of supplemental food sources in Wohlerspark, which was sampled throughout the year, clearly peaked in winter and was lowest in spring (Suppl. Table 5). Additionally, we observed squirrels from Wohlerspark repeatedly feeding on various foods, such as fat balls, watermelon, grapes, cupcakes, bread, rice crackers and BBQ leftovers during field work. We also observed hand feeding with walnuts, hazelnuts and peanuts in Wohlerspark on most trapping days, but not in the other two parks. This manifested itself in active begging behavior from the squirrels at Wohlerspark.

Table 3 Overview of available anthropogenic food items in g and kJ at the four different study sites in the summer season; given are mean values from three days of observation. Peanuts have been included in a common category with nuts
Full size table

Squirrel population density and structure

In total, we trapped 67 squirrels at the urban sites (314 trap days) and 10 squirrels (428 trap days) in the forest. Within the urban sites we trapped 46 individuals in Wohlerspark (230 trap days), 12 in Jacobipark (48 trap days) and 9 in Schanzenpark (36 trap days). Summer population densities calculated as MNAind were substantially higher in the three urban parks compared to the forest. The smallest park (Wohlerspark) housed a 42-fold higher squirrel density than the forest. The density in this park was also higher than in the other two parks: 2.2 times higher compared to Jacobipark and 2.6 times higher compared to Schanzenpark. Compared to the forest, Jacobipark had a 20-fold higher and Schanzenpark a 16-fold higher population density. Specifically, we found a MNAind of 0.21 individuals per ha in summer in the forest, 4.10 in Jacobipark, 3.38 in Schanzenpark and 8.83 in Wohlerspark. We did not trap any individuals in the forest in winter.

We never trapped subadults or juveniles in the forest, whereas they were trapped in all seasons in Wohlerspark (Table 4). In summer, 23% of the trapped individuals in Wohlerspark were juvenile or subadult and in Jacobipark they accounted for 25% of the individuals. We trapped more juveniles or subadults than adults in Schanzenpark (56%), but absolute trapping numbers were low. Since all juveniles were trapped in the same corner of the park, they might have been siblings from the same litter. Apart from one exemption (Wohlerspark in winter), we always trapped more males than females, however, the proportions were neither statistically different between the four sites in summer (p = 0.40) nor between the forest and Wohlerspark in spring (p = 1.00, odds ratio 0.51, CI 0.01–6.63) or autumn (p = 0.49, odds ratio 0.00, CI 0.00-6.75). We did not recapture any of the forest individuals from one season to another during this study, but three males were recaptured in the years 2018/2019 during another study. In Wohlerspark, recapture rate was fairly high with 20 (54%) out of the 37 adult individuals being recaptured at least once in another season, thereof five individuals in two seasons, nine in three seasons and six in all four trapping seasons. Ten (53%) of the subadults were recaptured as adults in following seasons. See Table 4 as well as Figs. 3, 4 and 5 for full details on population structures and trapping success.

Table 4 Overview of trap days (product of no. of traps by no. of capture days per site and season) and population structure per study site and season. MNA: minimum number alive per ha; MNAind: Index of MNA, calculated to account for differences in trapping effort (i.e., varying no. of trapping days and used traps)
Full size table
Fig. 3
figure 3

Squirrel density in summer at the four different study sites; shown as index of MNA (MNAind; note the logarithmic scale)

Full size image
Fig. 4
figure 4

Squirrel density across the year for the forest site (Hahnheide, open bars) and the most contrasting urban site with regard to size and food supplementation (Wohlerspark, grey bars); shown as index of MNA (MNAind; note the logarithmic scale). We did not trap any individuals in winter in the forest despite extensive pre-baiting, 10 trapping days and squirrel feeding signs (stripped spruce cones)

Full size image
Fig. 5
figure 5

Proportions (%) of female (white) and male (grey) squirrels trapped in summer at all four study sites. Proportions were not statistically different (Fisher’s exact test: p = 0.40)

Full size image

Discussion

Our study highlights the heterogeneity of urban habitat patches that appear to be very similar. Urban stressors peak at urban centres, yet the spatial heterogeneity of cities can lead to distinct environmental conditions for wildlife in different green urban fragments even when located closely together (Alberti et al. 2020). The urban parks we studied showed pronounced differences in natural and supplemental food availability, though being located in the core area of the same city and sharing similar environmental conditions in terms of human population density, park size or type of urban matrix. Tree species richness as well as number of trees in total and per ha were highest in the smallest park (Wohlerspark), which might indicate higher availability of seeds and shelter for wildlife, compared to the other two parks. However, this park had the lowest tree diversity among the parks and was strongly dominated by lime trees. Lime tree seeds are not favored by squirrels, but represent secondary food items (Bertolino et al. 2004; Krauze-Gryz and Gryz 2015). Conifers, an important primary food source for squirrels, were scarce in all parks, accounting for only 3 to 9% of the local trees, thus suitability as permanent habitat could be disputable. Availability of kJ from fallen tree seeds was lower in Wohlerspark and Schanzenpark compared to the third park and the forest, but this could change throughout the year with different types of food ripening at different times. Natural seed availability might have been influenced by gardening activities in the parks where measures such as cutting of trees and cleaning of paths etc. occurred regularly during the trapping seasons. These activities might also impact the possibilities to retrieve hoarded food items.

Besides natural food sources, anthropogenically provided foods also differed between the parks in amount and variety in the summer season and were highest in Wohlerspark. Throughout the year, we observed multiple feeding spots in this park with intentionally provided wildlife food, for example, bird food or nuts. However, supplemental foods in the parks also included leftovers, e.g., processed foods or bakery products, both inappropriate food for squirrels with regard to nutrients. We received reports and witnessed first-hand that squirrels in the park foraged on various food items, e.g., fat balls, rice crackers, cookies and even cupcakes. Moreover, peanuts – a non-native food – were frequently offered to squirrels in Wohlerspark. The chemical composition of peanuts, such as a high phosphorus content, is likely to restrict the peanut intake or can lead to nutrient deficiencies in squirrels (Shuttleworth 2000). We did not take any data on trash bin content in this study, but squirrels often inspected them and food intake from trash bins was documented for grey squirrels (Rimbach et al. 2022).

Supplemental foods can make up a significant part of the diet of urban wildlife and induce substantial shifts in the feeding repertoire (Dasgupta et al. 2021; Wist et al. 2022). For example, they accounted for more than half of the stomach content of urban foxes (Contesse et al. 2004). Consequently, abundant supplemental food sources in urban habitats can increase body mass or condition (relation between body mass and a measure of body length) in species that are able to exploit them successfully, as it is the case for, e.g., gulls, chipmunks and many carnivores (Auman et al. 2008; Bateman and Fleming 2012; Lyons et al. 2017). On the other hand, effects on health and fitness of wildlife can be ambiguous and an urbanized diet can have multiple side-effects (Andersson et al. 2015; Murray et al. 2015; Pollock et al. 2017). This depends on further parameters, such as supplemental food quality and the simultaneous availability and quality of natural foods sources (Shuttleworth 2000; Birnie-Gauvin et al. 2017). We have indication for lower body condition in the highly supplemented Wohlerspark (Suppl. Table 6, Suppl. Figure 1), possibly due to unsuitable food types and a lower availability of natural food sources. Body mass in squirrels can be age dependent, but usually varies with season and food availability (Wauters and Dhondt 1989a). We do not have indications for seasonal food shortages, i.e., body mass changes over the year in the Wohlerspark population (Suppl. Table 7, Suppl. Figure 2).

A study from Poland also documented lower body condition in urban squirrels when compared to forest individuals (Beliniak et al. 2022), though another recent study found higher body masses in urban squirrels (Tranquillo et al. 2023). These ambiguous results comply with studies on other species. House sparrows (Passer domesticus), for example, ranged in studies from no difference in body condition between urban and rural individuals to a significantly decreased body condition in urban individuals (Bókony et al. 2010; Meillère et al. 2017; Jiménez-Peñuela et al. 2019). We suggest that the heterogeneity of study sites, particularly in local food abundance and quality might have contributed to these findings, as we were able to show that even closely located patches can already differ substantially. Differences in city size, park area, human population density as well as socio-economical or cultural factors can influence natural food availability and also the amount of refuse or active wildlife feeding (Fuller et al. 2008; Aronson et al. 2017; Hassell et al. 2021). Furthermore, urban wildlife might also face different levels of intra- and interspecific competition for food sources. We frequently observed other species in high numbers in the parks, such as crows, jays and doves, but also rats. Most of these species interacted with the squirrels, particularly at feeding spots, and pillaged squirrel hoards.

Increased population density is a common feature in successful urban species, including squirrels, and strongly driven by the availability of supplemental food sources (Luniak 2004; McCleery 2010; Shochat et al. 2010; Jokimäki et al. 2017). Interestingly, species densities even seem to be highest in green patches surrounded by the densest urban infrastructure. Probably because the matrix does not offer enough other habitat options, supplemental feeding occurs and numbers of natural predators are often reduced (Luniak 2004; Parker and Nilon 2012). Local squirrel densities and habitat use at rural sites usually vary with forest stand structure, tree diversity and the proportion of conifers, representing food availability (Wauters et al. 2008; Rima et al. 2010; Flaherty et al. 2012; Dylewski et al. 2021). As expected, squirrel densities in this study were higher at all urban sites when compared to the forest, but with distinct variation between the parks. The park with the highest amounts and diversity of supplemental foods also had highest squirrel population density, despite natural food sources might have been limited.

So far, studies in rural habitats, forest fragments, or larger urban parks, have found that densities of S. vulgaris rarely exceed 2.0 individuals/ha (Wauters et al. 1994, 1997; Magris and Gurnell 2002; Bosch and Lurz 2011; Rezouki et al. 2014; Haigh et al. 2015; Beliniak et al. 2021, 2022). Yet densities of grey squirrels (Sciurus carolinensis) and other synurban species can be up to 100 times higher in urban than in rural habitats (Rodewald and Gehrt 2014; Merrick et al. 2016). Our results for the Wohlerspark population even exceeded this level in the autumn trapping session. However, this might have been influenced by hoarding behaviour of squirrels in autumn, where foraging distances increase and thus individuals are more likely to be caught (Bertolino et al. 2004), together with a lower number of autumn trapping days in the forest location were trapping success was always low or even absent in winter.

Reliable conclusions about differences in population structures or breeding individuals were hampered by low absolute squirrel numbers in most sites. Generally, mammal species successful in urban habitats have larger litters (Santini et al. 2019). Yet, urbanisation can decrease the reproductive success within a species (de Satgé et al. 2019). However, the juvenile to adult ratio in all parks of our study was high and we assume that these individuals were locally born. Year-round access to abundant supplemental foods might enable squirrels to allocate more resources into reproduction (Wauters et al. 1995, 1997), although this possibly might contribute to the observed lower investment into body condition in highly supplemented parks. Higher proportions of juveniles within urban populations were also found in other studies on urban squirrels (Beliniak et al. 2022). The high recapture rate of juveniles and subadults as adults in following seasons in Wohlerspark might indicate high offspring survival and successful establishment in the parks, again possibly supported by the high amounts of supplemental foods and higher tree numbers, i.e. higher shelter availability. Furthermore, urban fragmentation can limit dispersal, and dispersal distances in urban squirrels are rather short (Selonen et al. 2018).

Our results provide insight into differences between populations living in close proximity in the same urban core area, though we acknowledge the imbalance in trapping effort and success among the sites and seasons in our study. Living in an urban park with high access to supplemental food seems to support high population densities, however might not necessarily improve other fitness relevant traits. Urban grey squirrels, for example, showed poorer physiological condition represented by higher blood glucose levels and lower levels of, e.g., potassium resulting from an anthropogenic diet (Schmidt et al. 2019). We predict that there might be some kind of threshold where positive effects of urban conditions are outweighed by the costs of other urban factors, such as malnutrition of the urban diet or negative side-effects of high population densities.

Future studies should address long-term fitness consequences, since reproductive success and survival are influenced by food availability and body mass (Wauters and Dhondt 1989b; Wauters et al. 1995) and there might be different effects on different life stages. Urban planning and conservation actions should focus on ensuring sufficient natural food availability for urban wildlife such as squirrels, e.g., by increasing tree diversity and particularly conifer numbers. It is also crucial to inform the human population about suitable supplemental feeding and possible threats to urban wildlife. Reducing littering and limiting the access to trash bins might be beneficial to prevent animals from foraging on refuse and non-natural food items. These measures might also assist in preventing the spread of wildlife diseases at clumped feeding spots and onto humans.

Conclusions

Our study highlights the importance of studying the response of wildlife to urban conditions not only on a large, but also on a small scale. Differences in population parameters between different urban habitats within the same city can be substantial, even between parks with similar area and similar type of surrounding urban matrix. Small parks can hold surprisingly high population densities, likely dependent on vegetation structure and the availability of supplemental food sources and shelters. However, we cannot exclude that these habitats might also prove to be ecological traps on a long-term basis. Further studies are necessary investigating underlying mechanisms and consequences on long-term survival at those sites. It is also crucial to consider the urban heterogeneity when drawing conclusions from urban studies and deciding about appropriate conservation actions for a certain site.