Introduction

Urbanization has been widely reported as one of the main threats to biodiversity, representing the major determinant of habitat loss for wild species (Antrop 2004; Garden et al. 2006; McKinney 2006). In spite of habitat degradation, urban areas offer a range of habitat niches and may host a high species diversity (Niemelä 1999; Collins et al. 2000; Chace and Walsh 2004; Magle et al. 2012). Wild species adapt to live in these new conditions, expanding or modifying their natural ecological niches (Luniak 2004; Russo and Ancillotto 2014; Mori and Bertolino 2015). Native (Harris 1986; Beringer et al. 2002; DeStefano and DeGraaf 2003; Parker and Nilon 2008; Herr et al. 2009) and nonnative (Bertolino et al. 2004; Clergeau and Vergnes 2011; Walther et al. 2011) species may benefit from urbanization, in terms of food availability, human protection and shelter sites. Nevertheless, urbanization increases interactions between animals and humans, enhancing opportunities for conflict (Conover et al. 1995; Heimlich and Anderson 2001; DeStefano and DeGraaf 2003).

Attitude towards urban/suburban wildlife is mainly positive (Coluccy et al. 2001; König 2008; Menchetti and Mori 2014), although it may range from “wildlife as a resource” to “wildlife as pests” (Conover et al. 1995). Wildlife influences human activities through (1) transmission of zoonoses (König 2008; Mackenstedt et al. 2015); (2) predation on livestock, poultry and scavenging of urban waste (Doncaster et al. 1990; Prange et al. 2003; Abay et al. 2011); (3) attacks to humans (Timm et al. 2004; Abay et al. 2011; Bateman and Fleming 2012); (4) religion and popular beliefs (Daigle et al. 2002) and (5) crop damages (Connelly et al. 1987; Conover et al. 1995). Human-induced stress may also alter animal behaviour in urban ecosystems (Ditchkoff et al. 2006). As a consequence, the urbanization of wildlife may pose a challenge to resource managers, emphasizing the importance to analyse the behaviour of problematic species in urban environments.

The crested porcupine Hystrix cristata is a large monogamous rodent of African origin, probably introduced to Italy in the early medieval times (Bertolino et al. 2015). The members of a pair share the same home range and show overlapping activity rhythms, but, during reproductive periods, partners alternate in the den with newborns (Mori et al. 2016). This species recently underwent a substantial range expansion in Italy, reaching areas where it was historically absent (Mori et al. 2013). The crested porcupine is listed in the Berne Convention (All. II) and in the Habitat Directive (All. IV). It has been protected by the Italian law since 1977 (National Law 968/1977) and listed among the “especially protected species” in the National Law 157/1992. Therefore, killing porcupines in Europe is severely banned. Despite this level of legal protection, the species is still extensively poached (Amori et al. 2008; our own data, for Southern Tuscany, 23.07%, N = 60 radio-tagged porcupines). Porcupines are sensitive to disturbance, e.g. in Western Terai (Nepal), this rodent tended to be active when its main predator, the common leopard Panthera pardus, was not (Fattorini and Pokheral 2012). It is unknown whether poaching influences ranging behaviour of porcupines in areas free of natural predators.

Throughout the world, porcupines are considered as agricultural pests (Hystrix indica: Alkon and Saltz 1985a; Khan et al. 2000; Hystrix africaeaustralis: Gaigher and Currie 1979; Hystrix brachyura: Chuan 1969; Greaves and Khan 1978; Linkie et al. 2007). As to the crested porcupine, crop damages lamented by farmers are actually negligible in Central Italy (Laurenzi et al. 2016). Major damage seems to occur in private vegetable gardens in the surroundings of human settlements (Laurenzi et al. 2016), thus increasing the conflict with humans. Intolerance against this rodent is locally increased because of wounds provoked to hounds during hunting trips targeting other den-living species, e.g. the red fox Vulpes vulpes (Mori et al. 2014a), as well as because of damage to riverbanks through burrow digging.

So far, the only published study on the diet of the crested porcupine throughout the year has been carried out in a rural area, mainly composed by fallow and woodland (Bruno and Riccardi 1995). According to Bruno and Riccardi (1995) and Mohamed (2011), porcupines appear to be generalist rodents who feed on a variety of plant parts in relation to seasonal availability.

Porcupines have also been observed in suburban and urban areas (Ramat Aviv, Israel: Sever and Mendelssohn 1989; Siena, Italy: Lovari et al. 2013; Rome, Italy: Petrozzi et al. 2012; Grosseto, Italy: E. Mori, personal observation), but no data are available on their ecology, except for their ranging behaviour (Lovari et al. 2013) and activity rhythms (Corsini et al. 1995). In this work, we have aimed at a holistic approach (i.e. habitat selection, activity rhythms, food habits and selection) to the study of some ecological factors shaping up the ranging behaviour of crested porcupines in a suburban area, where the probability of encroaching human activities is the highest.

We predicted that (1) if crested porcupines are generalist herbivores, food selection would reflect local environmental availability of each food resource, including cultivated vegetables; (2) ranging movements would change according to the seasonal distribution of food resources and (3) in presence of diurnal human harassment, crested porcupines would develop more strictly nocturnal activity rhythms to avoid direct encounters with man.

Materials and methods

Study area

Our study area (about 120 ha, 270–345 m a.s.l., 43° 20′ N, 11° 20′ E) was included in a suburban area located in the northern outskirts of Siena, in the close surroundings of the New General Hospital “Le Scotte” in the suburb of San Miniato (Municipality of Siena). The area was heavily used for human activities, in particular small-scale agriculture and construction works. The climate was relatively mild, with average monthly temperatures of >0 °C in cold months (i.e. October 1990–March 1991, temperature range: 4.4–6.8 °C) and <30 °C in warm months (i.e. April–September 1991: 11.7–24.3 °C). Monthly average rainfall, concentrated in autumn and winter, was c. 63.8 ± 8.4 mm (Lovari et al. 2013). About 42% of the study site included farmlands (Fig. 1). Proportion of deciduous woodland (Quercus pubescens, Quercus cerris, Quercus ilex, Acer campestre, Ulmus minor, Castanea sativa and introduced Robinia pseudoacacia) was 14.8%, together with buildings and other human settlements (17.4%) as well as irrigation ditches (4.0%) and fallow (14.8%). Shrubwood (Cornus mas, Crataegus monogyna, Clematis vitalba, Rubus ulmifolius, Prunus spinosa, Spartium junceum) covered about 4.8% of the area, and the remaining 2.2% was characterized by coniferous woodland (Pinus pinea).

Fig. 1
figure 1

Habitat composition of the study area and location of den setts, province of Siena (©Google Earth). Asterisks mark the setts used by radio-tagged porcupines; D represents the location of all the known den setts in the study area

Full size image

Captures and radio-tracking

Crested porcupines were caught in home-made metal box traps (40 × 50 × 140 cm), baited with fruits and vegetables (plums, apples, carrots, potatoes) and kept active for at least seven nights a month (Lovari et al. 2013; Mori et al. 2014b). Trapped porcupines were sedated according to the protocol in Massolo et al. (2003). If adult (Van Aarde 1985; Mori and Lovari 2014), they were equipped with a radio collar (VHF radio collars: AVM P2, Livermore, CA).

Each crested porcupine was radio-tracked for one full night every 4 days and one full day (24 h) every 2 weeks. Porcupines were trapped all at the same time of the year (late May–early June) and, thus, all monitored for 14–20 h/week/individual for 1 year. Only biologically independent fixes (Lair 1987) were used for our analysis.

The mean location error was determined by positioning radio-collars in 100 known locations, at ground level, and by calculating the difference in metres between actual and estimated locations. Radio-tracking procedures included a mixture of distance (1 fix/70 min, mean location error = 32 m) and homing-in locations (1 fix/15 min, mean location error = 15 m), with a mean value of 41 fixes/month/individual. Seasonal home range (hereafter, HR) sizes were estimated through the 95% minimum convex polygon (MCP 95%) and the 95% fixed kernel (ker 95%: Lovari et al. 2013), by using the statistical software R 3.1.3, packages ade4 (Dray and Dufour 2007) and adehabitat (Calenge 2006). The software RANGES V (Kenward and Hodder 1992) was used to assess HR overlap (ker 95%) between members of the same pair.

Habitat selection

All the fix locations were ascribed to a category of habitat type. The Ivlev’s electivity index (Ivlev 1961) was computed to assess the selection of crested porcupine for habitat types, through the statistical software R 3.1.1, package Gplots (Warnes et al. 2014). This index ranges from −1 (total avoidance) to +1 (total preference): selection/avoidance occurs at values >+0.3 or <−0.3, respectively (Laurenzi et al. 2016).

Activity rhythms

Activity rhythms were assessed seasonally (Corsini et al. 1995) to evaluate the temporal behaviour of crested porcupines in relation to moon phases and round-the-clock. The activity of crested porcupines was determined through the variation of signal intensity within 60 s (Garshelis and Pelton 1980; Corsini et al. 1995). Porcupines were considered as being “inactive” when in the den: thus, duration of activity was measured as the difference between the times of onset and termination of surface activity (Corsini et al. 1995).

Nocturnal fixes were divided according to four moon phases: phase 1, from new moon to ¼ (532 fixes collected in 28 nights); phase 2, from ¼ to ½ (483 fixes collected in 23 nights); phase 3, from ½ to ¾ (494 fixes collected in 32 nights); and phase 4, over ¾ (456 fixes collected in 27 nights). Variation in moonlight intensity has been reported also over the 29.5 days of lunar cycle (day 0, new moon; day 15, full moon: Mori et al. 2014c). The dependence of activity rhythms on moon phases was assessed through circular statistics for each season (Mori et al. 2014c). We computed the Rayleigh test for uniformity to the circular distributions to test for the concentration of fixes around the mean (Batschelet 1981; Mori et al. 2014c), through the statistical software R 3.1.1, package circular (Lund and Agostinelli 2009). The main statistics were calculated to test if inactive fixes were randomly distributed along a lunar cycle or not. The density curves indicating the concentration of inactive fixes along the lunar cycle were kernel smoothed and double plotted on Cartesian axes to better show circularity (Mori et al. 2014c).

Feeding ecology

Four seasons have been astronomically defined. Droppings were collected once a month throughout 1 year, along the routes used by our radio-tagged porcupines, identified through radio tracking in the preceding days. A total of 8–10 droppings per month were analysed, as this number was considered to be representative to assess the proportion of species included in the diet of this rodent (Bruno and Riccardi 1995). Droppings were analysed following the protocol by Bruno and Riccardi (1995). A solution of NaOH 0.059 M was used to dissolve the mucous film which covered the droppings (10–20 min at 40 °C: Zyznar and Urness 1969). Then, the sample was washed through a 1-mm square mesh sieve. Fragments from the washed sample were separated by hand according to the food categories they belonged to; then, they were identified either by eye or through a stereomicroscope (Wild M3C, Heerbrugg: 128–320×).

Food categories were (1) monocotyledonous herbs (leaves and stems), (2) dicotyledonous herbs (leaves and stems), (3) roots and rhizomes, (4) bulbs and tubers, (5) fruits and (6) cultivated vegetables. “Bark” was not included among the categories as debarking by porcupines was never observed in the course of this study. Each category was identified at the family to species levels through comparison with private reference collections and herbarium specimens. Volumes of each food category were measured by water displacement (Jackson 1980; Bruno and Riccardi 1995). Relative frequencies and volumes were then plotted in a graph (Kruuk 1989). The availability of trophic resources was measured through the point intercept method (Goodall 1953; Jonasson 1988). This method measures the relative cover of vegetal categories which builds up the diet of the crested porcupine (monocotyledonous herbs, dicotyledonous herbs, plants with bulbs and tubers, plants with rhizomes, fruit plants, vegetables). It is designed to sample within-area variation and quantify changes of plant species cover over time. A total of 90 linear transects of 10 m were placed randomly along the study site throughout the year, using a stratified sampling designs, and throughout all the identified habitat types. On each transect, 50 plant hits were carried out, one every 20 cm. Assessment points were accomplished with a 30-cm pin, which was sharpened to create a point and placed perpendicularly to the ground. Detected plants were then reported on a field tab together with the date, the number of the transect and the habitat type. The relative frequency of each category was calculated. Food selection by porcupines was then evaluated through the Ivlev’s electivity index, by using the R 3.1.1 package Gplots (Warnes et al. 2014). Selection/avoidance occurs at values >+0.3 or <−0.3, respectively (Laurenzi et al. 2016).

Results

We caught a total of 18 crested porcupines, 11 of which were adults and were equipped with radio collars. Among our research individuals, 44.4% were poached within the first 3 months from radio tagging (i.e. unequivocal signs of poaching activities were found: a severed radio collar and a used snare, often in combination with a number of bloodied quills) and 33.4% disappeared so that just 22.2% were radio tracked for at least one full year (two pairs: M1-F1 and M2-F2). We are aware of the snags related to our small sample size; on the other hand, with a much greater number of experimental individuals, a holistic approach (as the one we have used in our study) would have been particularly challenging. All poached/disappeared porcupines showed the tendency to range in the close neighbourhoods of human settlements (59.81% fixes, N = 1428), differently from the others.

Of our radio-tagged pairs, one had its den in a bramble thicket, whereas the other one changed three setts in 1 year, eventually locating the last one in a bramble thicket. Seasonal HRs are shown in Table 1; home ranges were larger in the warm months with respect to the cold ones (Wilcoxon matched pairs signed-rank test: W = 56; P = 0.01).

Table 1 HR size (MCP 95% and ker 95%) of radio-tracked porcupines (hectares)
Full size table

No significant differences in seasonal HR sizes occurred between sexes (Mann-Whitney U test: U = 27, P = 0.65), with extensive spatial overlap between HRs of members of the same pair (median = 78.85%, Q1 = 75.42%, Q3 = 87.43%, throughout the year: Table 2). HR overlap of different pairs was negligible (4.2–16.8%).

Table 2 Percentage of HR (ker 95%) and activity overlap for two pairs in 1 year of radio tracking
Full size table

Temporal overlap was high between pair members, with a decrease in winter for pair M1-F1 and in autumn for pair M2-F2, when cubs were present (median = 80.55%, Q1 = 57.08%, Q3 = 89.18%, throughout the year: Table 2).

Temporal activity of radio-tagged porcupines (median value per night 7 h 40 min, Q1 = 7 h 32 min, Q3 = 8 h 6 min, throughout the year: Table 3) did not vary significantly between warm (spring–summer) and cold months (autumn–winter: Wilcoxon matched pairs signed-rank test W = 286, P = 0.19). Porcupines completely avoided daylight hours and were only active at night.

Table 3 Mean temporal activity (±SD) of each individual in the four seasons
Full size table

The highest peaks of inactivity were recorded always around the full moon (winter: day 13.6 ± 2.6, Rayleigh test Z = 7.37, P < 0.001; autumn: day 13.9 ± 2.8, Rayleigh test Z = 7.37, P = 0.003; spring: day 14.8 ± 1.4, Rayleigh test Z = 31.43, P < 0.001; summer: day 15.8 ± 0.9, Rayleigh test: Z = 46.03, P < 0.001: Fig. 2).

Fig. 2
figure 2

Density estimates of nocturnal inactive fixes of crested porcupines throughout the lunar cycle, double plotted to show circularity, for all the seasons (autumn, N = 492 fixes; winter, N = 478 fixes; spring, N = 483 fixes; summer, N = 512 fixes)

Full size image

Irrigation ditches, human settlement and coniferous woodland were avoided throughout the year; fallow in winter, spring and summer; and shrubwoods only in autumn. Deciduous woodland was positively selected throughout the year and shrubwoods in winter, spring and summer. Farmlands were used proportionally to their local availability (Table 4).

Table 4 Ivlev’s electivity index (E) for habitat selection by crested porcupine
Full size table

A total of 113 faecal samples collected throughout the year (winter, N = 31; spring, N = 19; summer, N = 34; autumn, N = 29) were analysed. Fruits were the staple of the diet mainly in autumn and winter, while roots and rhizomes prevailed in spring and dicotyledons, followed by fruits (especially figs Ficus carica: 33.3% relative frequency and 68.2% relative volume on the total of fruits) in summer (Fig. 3).

Fig. 3
figure 3

Volume of main food categories in total diet of crested porcupines, i.e. volume of each dietary component versus its relative frequency of occurrence (autumn, N = 29 droppings; winter, N = 31 droppings; spring, N = 19 droppings; summer, N = 34 droppings). 1, dicotyledons; 2, monocotyledons; 3, roots and rhizomes; 4, bulbs and tubers; 5, fruits; 6, vegetables

Full size image

All the food categories were used proportionally to their local availability, with few exceptions. Roots and rhizomes were selected in winter and spring (especially Rumex crispus: 93.3% relative frequency and 95.7% relative volume, on the total of roots and rhizomes); bulbs and tubers (especially Cyclamen hederifolium: 78.0% relative frequency and 53.8% relative volume, on the total of bulbs and tubers) were preferred in autumn and underused in winter and spring. Fruits were positively selected in autumn. Cultivated vegetables were underused in autumn and spring, used in proportion to their availability in summer and not available in winter (Table 5).

Table 5 Ivlev’s electivity index (E) for food selection by crested porcupine
Full size table

Discussion

Differently from other areas (Alkon and Saltz 1988; Corsini et al. 1995; Mori et al. 2014c), activity rhythms of crested porcupines were strictly nocturnal in our study, with a sharp moonlight avoidance (especially in summer, when nights are short and cloudless), comparable to what was observed in areas where potential predators occur (Negev desert: Alkon and Saltz 1988; Western Terai: Fattorini and Pokheral 2012). Despite this, in summer, Indian porcupines foraged also during bright nights, to fulfil their water requirements (Alkon and Saltz 1988). In the absence of natural predators, light avoidance is not so strict (Corsini et al. 1995; Mori et al. 2014c), but poaching pressure may act as a force vicariant that of natural predators to maintain this behaviour. In our study area, poaching was especially common and at least 72.7% of artificially marked porcupines (N = 11, including 64% of radio-tagged ones) were killed in 12 months. Whereas the pair M1-F1 remained in the same den throughout the whole year, as commonly observed in crested porcupines (Monetti et al. 2005; Lovari et al. 2013; Mori et al. 2015), the pair formed by M2 and F2 changed their den three times in 1 year, progressively using more closed and inaccessible habitats (from open woodland to a S. junceum shrubwood, to a bramble thicket). In fact, digging dens in thick thorny bushes (cf. Fig. 1) could be a ploy to avoid poaching.

A high HR overlap between members of the same pair has been confirmed (Mori et al. 2014b). Pair partners also showed a high overlap of activity rhythms (Mori et al. 2014c), which only decreased in winter for pair 1, with cubs in the den (Mori et al. 2016). In closely related species, the availability of food resources has been reported as the main determinant of habitat selection (H. indica: Sharma and Prasad 1992; H. africaeaustralis: De Villiers et al. 1994). In the Mediterranean area, summer provides porcupines with a high amount of food categories (i.e. epigeal plants, fruits, vegetables: Bruno and Riccardi 1995), but underground plant storage organs may be difficult to reach because of sun-baked soil, thus determining an expansion of HR sizes (Mori et al. 2014b). As to habitat selection in suburban areas, crested porcupines avoided coniferous woodlands, fallows and human settlements, as they do not provide rodents with enough food resources, are unsuitable for denning and/or expose porcupines to poaching. As to avoidance of irrigation ditches, most likely the presence of water or mud made them difficult habitats for porcupine movements. Differently from what has been observed in other studies (Fattorini and Pokheral 2012; Mori et al. 2014b), in our study area, grass in fallows was kept short, thus providing little cover to porcupines. A positive selection occurred for deciduous woodlands, which were overused throughout the year with respect to their local availability. Shrubwoods were also selected throughout the year but avoided in autumn, when a high availability/consumption of fruits (mainly chestnuts and acorns) occurred in woodland. Furthermore, all the porcupine den setts detected in this study were located in closed habitats, i.e. deciduous woodland and shrubwood. While Mori et al. (2014b) showed a positive selection of cultivations in summer, farmlands were used proportionally to their availability in our work, which may partly contradict their findings. In our study area, tin fences (50-cm high from the ground) were used as a prevention to crop damage in house gardens, thus removing a particularly attractive food source, e.g. flower bulbs, melons, courgettes and onions (Laurenzi et al. 2016).

The crested porcupine has been confirmed to be a “generalist” herbivore, which feeds on a high variety of plants and organs, i.e. roots, stems and fruits (Bruno and Riccardi 1995; Mohamed 2011), in relation to their local seasonal availability. Fruits built up the staple of the diet in autumn and winter, when chestnuts were abundant. The presence of fruit with hard, thick epicarp was greater in the cold months with respect to warm ones. These fruits are a long-lasting, rich reserve of carbohydrates and have a clumped distribution in woodlands (Pignatti 1982), confirming results of our analysis of habitat selection. In summer, as to fruits, figs appeared to be a particularly attractive food resource, although concentrated around human inhabited, underused areas. One may predict that animals will tend to simultaneously maximise net energy intake and minimise risk of predation, to maximise fitness (Dill 1987). This prediction may explain why the usage of a sought after, clumped food resource, such as figs, contrasted with the short time spent in human inhabited areas: porcupines must have traded safety for food intake, i.e. quickly feeding and going back to safer habitats to avoid being poached (cf. Cavallini and Lovari 1991, for V. vulpes; De Villiers et al. 1994, for H. africaeaustralis).

The positive selection of roots (especially R. crispus) in spring could depend on field ploughing, which enhanced availability of underground storage organs in arable lands in that season. A positive selection for bulbs and tubers occurred early in the autumn, when the tuber of the ivy-leaved cyclamen C. hederifolium was overused with respect to its local availability. This may be due to the fact that secondary metabolites accumulated in underground storage organs are eliminated when the plant is flowering, thus reducing their toxicity (Fiori 1969). Furthermore, the palatability of the cyclamen is reduced by the increased concentration of the toxic glycoside outside the flowering period (Fiori 1969).

Despite being described as a species whose diet is composed mainly by vegetal underground storage organs (Santini 1980; Pigozzi and Patterson 1990), our work revealed that porcupines preferred fruits and stems. The greater use of epigeal plant parts (about 71.1% of total frequency, in our work) with respect to roots may represent a strategy to reduce poaching risk because of the longer time and higher amount of energy requested to search, dig and extract underground storage organs (Alkon and Saltz 1985b; Downs et al. 2015). Furthermore, water content of ripe fruits may allow crested porcupines not to move in moonlight nights, differently from what observed in an arid environment by Alkon and Saltz (1988). The use of hypogeal parts of the plant has been confirmed, although they were mainly consumed when fruits were poorly available and when the plough brought up roots, bulbs and rhizomes. Consumption of cultivated vegetables was very low and occurred only in summer (mainly on courgettes and tomatoes), proportionally to their local availability. Vegetables represent a source of rich and clumped food for the crested porcupine. Their usage may have been limited by the tin fences previously described and, most likely, also by the availability of other food resources present in less risky environments (Laurenzi et al. 2016).

Although the genus Hystrix originated in Asia (Nowak 1991, but see van Weers 2005), the species H. cristata evolved in Africa (Mohr 1965; Trucchi and Sbordoni 2009; Trucchi et al. 2016), where most of its present range occurs. During the Pleistocene, in the open habitats of Asia and Africa, e.g. savanna, the herbivore guild was particularly rich in species, e.g. 32 species of bovids and 9 species of equids in Africa (Kappelman et al. 1997; Steele 2007; Bernor et al. 2010; Gentry 2010), as well as several tens of potential porcupine predators (Werdelin and Peignè 2010). The presence of the latter may have determined the evolution of quills, whereas the former may have provided superior competitors (e.g. ruminants) for above ground food resources. In fact, earlier porcupines (e.g. Hystrix refossa: van Weers 1994) showed a hypsodont pattern of dentition, adapted to chew grass. As a consequence of competitor and predator presence, porcupines may have adapted to suboptimal habitats on time (e.g. subdeserts: Gutterman 1982; Alkon and Saltz 1985b; Alkon and Olsvig-Whittaker 1989; Bragg et al. 2005) and/or to feed on underground plant organs, which requires time and energy to dig them out, but it exposes them to predation risk. Quills, preference for thick thorny cover and nocturnal habits could be the results of this adaptation. In Italy, in near absence of predators and in presence of very few potentially competing wild ruminants, one could expect porcupines to have expanded again their ecological niche to the food-rich open habitats, which might explain the unusually high use of epigeal wild plants we recorded.