CSIRO PUBLISHING

www.publish.csiro.au/journals/wr

Wildlife Research, 2005, 32, 723–731

Rabbit warren distribution in relation to pasture communities in Mediterranean habitats: consequences for management of rabbit populations G. Gea-IzquierdoA,B, J. Muñoz-IgualadaA and A. San Miguel-AyanzA A

Departamento de Silvopascicultura, ETSI Montes, Universidad Politécnica de Madrid, Ciudad Universitaria s/n, 28040 Madrid, Spain. B Corresponding author. Email: [email protected]

Abstract. Iberian wild rabbit numbers have decreased in the last decades. The management implemented to recover rabbit populations includes several techniques, one of the most common being the construction of artificial rabbit warrens. To optimally distribute the artificial warrens in the field it is essential to understand natural warren microhabitat. Few studies have investigated the relationship between rabbits and grassland communities. In this work we study the spatial distribution and characteristics of rabbit warrens as well as their relation to grasslands in Mediterranean woodlands of central Spain. During the summer of 2001, three 12.5-ha study plots, including the most representative grassland communities of the area, were selected. All rabbit warrens were surveyed and the active and total entrances, shrub cover, grassland community and warren cover type were characterised. A grassland community selection index was calculated and the warren spatial distribution analysed. Ploughed lands and shallow soils were unsuitable for warren establishment. The mean number of burrow entrances per warren was 5.8 (4.4 active), and warren clustering occurred only in ploughed plots. However, pasture communities composed of annual and perennial species growing on unploughed deep sandy soils were preferentially selected. Most warrens (81.4%) were dug under some kind of protection such as shrub roots and rocks. According to our results, when designing rabbit restocking programs that include the provision of artificial warrens, unploughed deep soils with plenty of shrubs and rocks should be preferentially selected to locate the artificial warrens, which should be spaced so there are ~10 warrens per hectare and ~5–6 burrow entrances per warren.

Introduction Conservation of endangered species is a priority worldwide. The endemic and extremely threatened Iberian lynx (Lynx pardinus Temminck) and Iberian imperial eagle (Aquila adalberti Brehm) rely completely on the European rabbit as prey and have been negatively influenced by rabbit population decline (González et al. 1990; DGCN 2002). In addition to their great ecological importance, rabbits are a valuable game species for hunters. The decrease in rabbit numbers has also led to increased negative interactions between hunters and predators (Villafuerte et al. 1998; Angulo and Villafuerte 2003). Currently, several practices are used to enhance rabbit numbers. Among them, the offering of artificial warrens (most designs are made of concrete, plastic, wood or stones) to rabbits is one of the most widespread (CBD-Hábitat 2001). The great importance of warrens to rabbits and their social behaviour have been highlighted by many authors (Wheeler et al. 1981; Cowan 1987; Parer et al. 1987; Kolb 1991a; Surridge et al. 1999). The most prosperous rabbit populations are usually associated with high warren densities (Williams et al. 1995; Palomares 2001a; Lombardi et al. © CSIRO 2005

2003) and breeding success is greater in deep warrens than in shallow breeding stops (Mykytowycz and Gambale 1965; Parer 1977; Webb 1993; Surridge et al. 1999). Warrens are preferably dug in sandy soils (Parker et al. 1976; Rogers and Myers 1979; Parer and Libke 1985), although, according to several authors (Kolb 1985; Cowan 1987, 1991; Parer et al. 1987; Myers et al. 1994), warrens in sandy soils are usually smaller than warrens in heavy soils. Additionally, Parer and Libke (1985) reported that warrens are larger in soils with high carbonate content. Other factors that seem to determine the structure and location of warrens are the weather, the use of landscape by humans and the presence of predators capable of digging (Palomares 2003a). The importance of warrens is greater in open habitats with no shrub protection (Gibb and Fitzgerald 1995; Palomares and Delibes 1997; Lombardi et al. 2003). Therefore, the characterisation and ecology of warrens is one of the most important issues when designing measures to aid recovery of rabbit populations. Most studies related to warrens have been conducted in countries (UK, Australia, etc.) where the species, introduced by humans centuries ago, is considered an agricultural pest (Rogers et al. 1994). Far 10.1071/WR04129

1035-3712/05/080723

724

Wildlife Research

less is known about rabbit warrens in the Iberian Peninsula despite the great variability in habitats used by rabbits and the historical changes in agricultural practices in rabbit habitats (Rogers et al. 1994; Branco et al. 2002). For example, the pasture communities in those areas of the Iberian Peninsula under Mediterranean climates are different from pasture growing in areas with temperate climates. In West Iberia the rabbits, and hence their immediate predators, are frequently associated with a particular type of open woodlands, so-called ‘dehesas’. In this habitat, the patchy distribution of the pasture communities within the landscape depends on weather, topography, animal activities and human influence (Pérez Corona et al. 1998; Vázquez de Aldana et al. 2000). As a consequence they are particularly complex and diverse, both in species diversity (Marañón 1986; Figueroa and Davy 1991) and in their dynamics and phenology (San Miguel 2001). No studies have analysed the relation of rabbits to the different pasture communities and the use they make of them, despite their potential importance both for rabbit feeding and warren digging. Additionally, owing to the development of GPS technologies, only recently has it been possible to perform large-scale studies dealing with microhabitat selection and the spatial relationships of warrens (Palomares 2003a). None of these studies have been performed in central Spain. In this work we analyse the relationship between rabbit warrens and the different pasture communities found in a typical west Iberian Mediterranean habitat. We also study the spatial distribution of rabbit warrens and characterise the associated habitat variables.

G. Gea-Izquierdo et al.

2001; Rivas-Martínez and Penas 2004). The current vegetation, however, is a typical Iberian Mediterranean mosaic of open holm oak woodlands, shrublands, xerophytic grasslands and crops (‘dehesa’). The estate area is extensively managed for multiple purposes: large and small game, extensive raising of livestock, agriculture and other activities such as birdwatching. Its biological importance comes from the endangered populations of predators that survive there. It contains five nesting pairs of Iberian imperial eagle, Iberian lynx, and black vulture. Additionally, the rabbit population sustains a significant game activity: ~5000 rabbits are taken by hunters every year, though 25000 was the usual average before the arrival of rabbit haemorrhagic disease. Experimental plots Three 500 × 250 m plots (P1, P2 and P3) were selected to represent the most important biotopes inhabited by rabbits within this habitat (Fig. 1). The plot size (12.5 ha) could be considered the home range of monitored rabbits in southern Spain (Lombardi et al. 2003). The experimental plots were within the general area where rabbit-restocking and rabbit-enhancement management techniques have been performed (CBD-Hábitat 2001). The major features of the plots are described below (Fig. 2). More complete descriptions of the pasture communities can be found in San Miguel (2001) and Rivas-Martínez and Penas (2004). The topography was almost flat within the three plots (Fig. 1).

Methods The study area The study was conducted in a 5000-ha private estate located within the Toledo province of central Spain. The climate is continental Mediterranean, characterised by hot and dry summers and cold winters, with spring and autumn the usual rainy seasons. The annual rainfall varies greatly both between and within years (from 200 to more than 1300 mm, average 480 mm). The mean annual temperature is mild (15°C), but the range between maximum and minimum annual temperatures is from 42°C to –10°C (figures also variable between years). Hence, the vegetative growth period of plant communities varies greatly between years (Figueroa and Davy 1991; Pérez Corona et al. 1998; Vázquez de Aldana et al. 2000). Soils are acidic and nutrient poor, derived from granite and cuarcite parent rock materials. In this study, we use the concept of ‘vegetation communities’, as described in Phytosociology by Braun-Blanquet (1928). Phytosociology is based on the theoretical existence of ‘climax’ vegetation communities, similarly to American Clementsian’s statements (Clements 1916), with a particular approach and a specific systematic nomenclature based on indicator species and succession. However, the existence of real vegetation communities is controversial (Gleason 1926); we just use it here as a tool for stratifying vegetation. In the study site, the potential vegetation is a sclerophyllous evergreen holm oak forest: Pyro bourgaeanae–Quercetum rotundifoliae community in which evergreen holm oak (Quercus rotundifolia) is the dominant forest tree species and Pyrus bourgaeana is the species that best reflects the particular condition of the Q. rotundifolia forest (Rivas-Martínez et al.

Fig. 1.

Location of the study plots within the study area.

Distribution and ecology of rabbit warrens in central Spain

Plot P1 was located on the area showing the apparent highest rabbit abundance within the study site. Perennial, though withering in late summer, grassland communities were abundant. Those growing on deep and well drained soils, dominated by Stipa gigantea, Stipa lagascae and other rough perennial grasses, included in the Agrostio–Stipion plant community (alliance), were widely distributed throughout the plot. On the other hand, those growing on pseudogley (seasonally flooded) soils were dominated by Agrostis castellana (alliance: Agrostion castellanae) and were restricted to valley bottoms. Additionally, oligotrophic annual communities growing on unploughed land (phytosociological order: Helianthemetalia guttati) were also abundant, whereas subnitrophyllous annual grasslands growing on fallow land (phytosociological order: Sisymbrietalia officinalis) were scarce. The only tree species was holm oak (Quercus rotundifolia Lam.). The dominant shrub species was shrub-like coppiced holm oak.

Fig. 2.

Wildlife Research

725

Retama sphaerocarpa (L.) Boiss, Cytisus multiflorus (L’Hér), Cistus ladanifer L., Lavandula stoechas L. and Thymus mastichina (L.) L., were other shrub species present. Field data were collected in July 2001. Plot P2 was placed near a seasonal stream and included arable lands mixed with shrubland thriving on shallow granitic soils with outcrops. The dominant pasture community was Helianthemetalia guttati. However, there were other pasture communities: terophitic subnitrophyllous grasslands growing on fallow land and dominated by annual Bromus species (phytosociological order: Thero–Brometalia) and perennial grassland growing on seasonally flooded soils: phytosociological alliance Agrostion castellanae. Scattered ash trees (Fraxinus angustifolia Vahl.) grew on stream banks, usually mixed with holm oaks. Mastic (Pistacia terebinthus L.), bramble (Rubus Group ulmifolius Schott) and hawthorn (Crataegus monogyna Jacq.) were the shrub species present in this plot. Field data were also collected in July 2001.

Warren and pasture community distribution in the three study plots.

726

Wildlife Research

Plot P3 included large areas of arable land with randomly distributed shrub patches. Dense and tall Mediterranean round-headed club-rush meadows (phytosociological order: Holoschoenetalia) grew on seasonal stream banks, where rabbits were usually observed during dusk and dawn. P3 was mostly covered by subnitrophyllous annual grasslands, which were abundant owing to the periodic cultivation of its stony soil. The tree and shrub layers were similar to those described for the previous plots. Field data were collected in September 2001. Field techniques Field data were collected with the help of GPS devices. Within each plot all warrens were surveyed, registering their coordinates as well as the following variables. Pasture community: to a plant association level of phytosociological order or alliance, following Rivas-Martínez and Penas (2004). Shrub cover: estimated within an imaginary 25-m-radius circle around each warren, categorised 0–5%, 6–25%, 26–50%, 51–75% and 76–100%. Total number of burrow entrances in the warren: all entrances, active and inactive. Number of active burrow entrances in the warren: a burrow entrance was regarded as active if rabbit footprints were visible or when it did not present a rough edge and grass or cobwebs across it (Parer and Wood 1986; Kolb 1991a). Warren protection type: divided into four categories depending on the structures under which the warrens were dug: (a) unprotected warrens, (b) below shrub roots (holm oaks), (c) under rocks, and (d) below rocks and shrub roots. Distance between warrens: nearest-neighbour warren distances were calculated from the GPS warren points registered with the aid of a GIS. Finally, after grouping the values of the three plots we calculated a pasture community selection index (wi = oi /pi) following Manly et al. (1993), to analyse the possible preferential use by rabbits of the different pasture communities, where wi was the selection index of pasture community i (range ≥ 0; values over 1 meaning preference, under 1 avoidance), oi was the relative number of burrows over the total of pasture community i, and pi: was the relative area over the total of pasture community i. Data analysis As our warren population was completely mapped, clustering was studied using the Donnelly index (RD) for the nearest neighbour without including any boundary strip. The Thompson and Campbell and Clark (ZT) indexes were calculated for the second, third and fourth nearest neighbours, either when sample size was smaller (P2 and P3) or larger (P1) than 50 (Krebs 1999). We employed those indexes to determine whether there was any pattern in the warren distribution and, hence, the random null hypothesis could be rejected. Non-parametric tests were performed in all cases as normality could not be achieved by any transformation of the data. We used the Kruskal–Wallis variance analysis (K–W, hereafter) and the Mann–Whitney U test (M–W, hereafter) to check for differences between more than two and just two groups respectively (Quinn and Keough 2002). To check for differences in wi we used the G-test. Significant differences were considered with a confidence level α = 0.05.

Results

G. Gea-Izquierdo et al.

P = 0.089), whereas they were clustered in both P2 and P3 (see Table 1 and Fig. 2). The largest warren had 29 burrows, 27 of them active. Significant differences were found between the three plots both for the total number of burrow entrances per warren (Table 1) and the active burrows per warren (K–W, χ2 = 6.37, P = 0.041). These differences seemed to come from a smaller warren size in P2 than in P1 and P3 (M–W between P1 and P3: Z = 1877.5, P = 0.799 for total burrows; Z = 1673.5, P = 0.246 for active burrows). Active burrows averaged 5.0 ± 4.7 (mean ± s.d.) in P1, 2.9 ± 2.6 in P2, and 3.8 ± 4.2 in P3, corresponding to densities of 51.4 active burrows per hectare in P1, 8.2 in P2, and 9.2 in P3. The most common warren size, both total and active, was 2 burrow entrances in P1 and P2 and 4 burrow entrances in P3. Within the three plots, both warrens without activity (0% of active burrows) and completely active (all burrows with evidence of rabbit use) were found. The average activity levels were similar in P1 and P2, 77.2% and 74.3% respectively, but significantly different from P3, which averaged 58.3% (Table 1). Warren–habitat relationships The mode shrub cover around warrens was 25–50% in the three plots (Table 1). The total burrows in each pasture community along with the respective selection indexes (wi) are shown in Table 2. According to the G-test (χ2 = 843.57, P < 0.001) we could reject the null hypothesis of equal number of burrows in each pasture community. Periodically ploughed lands (Thero– Brometalia, wi = 0.137) were a major pasture type, occupying about one-third of the grassland area, but seemed to be unsuitable for warren establishment, whereas Agrostio– Stipion (wi = 2.500) communities were less widespread but highly preferred. Helianthemetalia (wi = 1.247) communities were the most common pasture and occupied an intermediate position. Communities of Agrostion castellanae (wi = 0.125) and Holoschoenetalia (wi = 0.281) occupied a small portion (8%) of the total area and supported few burrows and Sisymbrietalia (wi = 1.667) communities occupied such a small area (0.1 ha) that their suitability for warren establishment remains unclear. Most warrens were dug beneath the shelter of shrubs, predominantly holm oaks: warrens beneath shrub roots accounted for 42.8% of the total, while 29.9% were beneath shrubs and rocks. A further 8.7% were located under rocks and only 18.6% were unprotected (Table 3). No differences were found between the four groups, neither in the total number of burrows per warren (K–W, χ2 = 0.13, P = 0.987), nor in the number of active burrows (K–W, χ2 = 2.03, P = 0.566).

Warren distribution and characteristics The average nearest-neighbour warren distance was significantly lower in P1 (18.6 m) than in P2 (28.6 m) and P3 (24.7 m), which were statistically homogeneous (Table 1). Warrens were randomly distributed in P1 (RD = 1.09,

Discussion Warren distribution and characteristics In our study area, rabbits are abundant in some areas while absent from other apparently similar areas. This is common

Distribution and ecology of rabbit warrens in central Spain

Wildlife Research

in Iberia, where rabbit populations and warrens tend to be patchily distributed (Parker et al. 1976; Soriguer and Rogers 1981; Parer and Libke 1985). The horizontal precision estimated using GPS was 2.983 ± 0.121 m (mean ± s.d.), hence the spatial results are particularly accurate. Warrens were clustered in P2 and P3, possibly owing to the lack of optimum places for digging, but were not clustered in P1. Clustering in P3 was probably related to periodical ploughing, which destroys warrens and prevents the construction of new ones. This is a common practice to combat rabbits in regions where they are considered a pest (Cooke 1981; Wheeler et al. 1981; Kolb 1985). In the same way, warren clustering in P2 was probably the result of both ploughing and soil shallowness. In Australia, most warrens are found at a depth of over 45 cm, and even that is probably too shallow for digging big warrens (Parer and Libke 1985). Although our warrens were clustered, they seem to be less clustered than in other populations (Parer et al. 1987; Cowan 1991; Mutze 1991), maybe owing to the overall sandy structure of the soil. In a 10-ha plot of chalkland, mean nearest-neighbour distances of 28 m and randomly distributed warrens have been reported (Cowan 1987). Although variable, our results are similar. Warrens were slightly bigger in P1 and P3 (respective means: 6.2 and 6.0 burrow entrances) than in P2 (4.2 burrow entrances), most probably because the soils were shallower in the latter. However warren size is usually very variable

and our results are in accordance with those of other authors. In Doñana (south-west Spain), on sandier soils, Rogers and Myers (1979) reported 1–20 burrow entrances (mean 3.9) in two different areas including many shrub warrens, and slightly smaller sizes (1–10, mean 3.4) in another area where no warrens were dug under shrubs. For the same Spanish region, in a three-year study, Palomares (2003a) observed variations in warren sizes between years. His results were higher, with values of 3.8–6.8 entrances per warren per year (range: 1–126) for Mediterranean shrubland, an ecosystem similar to our study site, and from 8.1 to 11.7 (ranging from one to a peak of 218 entrances per warren) for grasslands. In the same area, Lombardi et al. (2003) found variations in the number of entrances, from 2.3 for scrubland to 7.5 for grassland warrens. In France, mean warren sizes range from 1.0 up to 24.0, for a variety of landscapes in parkland (Rogers et al. 1994). Outside of the rabbits’ endemic range, in Britain, Cowan (1991) reported an average warren size of 11.5 burrow entrances per warren. These results, clearly higher than ours, were recorded in chalkland, hence, in heavier soils, where the higher stability of the soil would enable larger warrens (Kolb 1985). Finally, in Australia, the average warren size is reported to usually lie between 3 and 15 burrow entrances per warren, but to vary greatly according to soil type, habitat and the level of human disturbance (Williams et al. 1995). For example, Parer (1977) reported a modal size of 1.0

Table 1. Summary statistics for P1, P2 and P3 RD (for first nearest neighbour) and ZT (for the second, third and fourth): Donnelly and Thompson or Campbell and Clark aggregation indexes to test the hypothesis of random spatial pattern (Krebs 1999). χ2 = chi-square value associated with the Kruskal–Wallis test; P = probability level. Different letters mean differences between medians by Mann–Whitney test. An asterisk denotes statistical significance (α = 0.05)

No. of warrens Warren density (warrens per hectare) Burrow density (burrows per hectare) No. of total burrow entrances per warren Mean (s.d.) Minimum Maximum Percentage of active burrow entrances Mean (s.d.) Minimum Maximum Distance to nearest warren neighbour (m) Mean (s.d.) Minimum Maximum First nearest neighbour RD Second nearest neighbour ZT Third nearest neighbour ZT Fourth nearest neighbour ZT Shrub cover (%) Mode Minimum Maximum

Plot 1

Plot 2

Plot 3

χ2

P

131 10.5 64.4

35 2.8 11.8

30 2.4 14.4

– – –

– – –

6.2 (5.0) a 1 29

4.2 (4.3) b 1 24

6.0 (4.9) a 1 26

8.38

0.015

77.2 (30.7) a 74.3 (34.1) a 58.3 (30.6) b 0 0 0 100 100 100

11.55

0.003

18.6 (9.0) a 28.6 (18.0) b 24.7 (18.6) b 4.9 11.8 10.8 60.8 76.7 110.5 1.09 0.77* 0.70* – 0.01 –2.72* – – –3.20* – – –2.75*

11.97

0.003

– – – –

– – – –

11.35

0.003

25–50 a 0–5 50–75

727

25–50 b 5–25 50–75

25–50 b 0–5 50–75

728

Wildlife Research

Table 2.

G. Gea-Izquierdo et al.

Habitat selection for burrow establishment by rabbits in Mediterranean grassland communities (grouping the values of the three plots) Habitat selection index (Manly et al. 1993): wi > 1 indicates preference and wi < 1 indicates avoidance

Grassland community

Area (P1+P2+P3) (ha) (% of total)

Sisymbrietalia officinalis Thero–Brometalia Helianthemetalia guttati Agrostio–Stipion giganteae Agrostion castellanae Holoschoenetalia Total

0.1 12.3 16.1 6.0 1.8 1.2 37.5

0.3 32.8 42.9 16.0 4.8 3.2 100.0

(mean 3.9 in the peak of activity), with a maximum of 37 burrow entrances per warren for the Canberra district. However, for the same district, Parer et al. (1987) recorded warrens with up to 99 entrances, referencing from other authors surveys a mean of 2.5–7.0 burrow entrances per warren and no warrens with more than 26 entrances. In arid central Australia, huge warren complexes are often found along intermittent water courses, with average warren size as high as 33 entrances per warren and some warrens exceeding 100 entrances (Mutze 1991). Warren densities were different between our plots. P1, with 10.5 warrens per hectare (64.4 burrow entrances per hectare), would represent the best habitat in our study area. In Britain, Kolb (1991a) reported the highest densities we have found, with 496 total burrow entrances per hectare, 293 of them active, on an area of sand dunes (a soft, easy-to-dig substrate), and 200 burrows per hectare in a similar area. In another study, average densities of 40 burrows per hectare were recorded on chalkland (Cowan 1987). In France, in a variety of landscapes and, hence, soils, densities of 0.4–2.6 warrens per hectare are reported (Rogers et al. 1994). Warrens do not seem to change much in size and location between years but they exhibit a seasonal variation in active burrow entrances. The percentage of active burrows within warrens decreased through the summer (from 72.7% and Table 3.

Burrow entrances (P1+P2+P3) (n) (% of total) 6 50 602 450 7 10 1125

0.5 4.5 53.5 40.0 0.6 0.9 100.0

Selection index (wi) 1.67 0.14 1.25 2.50 0.13 0.28 –

74.3% in July for P1 and P2 to 58.3% for P3) in line with the usual seasonal decline in rabbit numbers in Mediterranean ecosystems (Soriguer 1981). This seasonal variation in active burrow entrances is to be expected as there is usually a positive correlation between active burrow entrances and rabbit abundance (Parker et al. 1976; Cooke 1981; Parer and Wood 1986; Cowan 1991). Myers et al. (1994) observed stable populations to have no more than 1.3 adult rabbits per warren burrow entrance, whereas Parker et al. (1976) recorded 0.72 rabbits per active burrow entrance Warren–habitat relationships The great importance of shrubs, especially for surfacedwelling rabbits, has been shown in many studies (Wheeler et al. 1981; King et al. 1984; Kolb 1991b; Moreno et al. 1996; Banks et al. 1999; Palomares 2001b; Lombardi et al. 2003). Rabbits prefer habitats with low tree cover and medium shrub cover mixed with grasslands to feed on (Rogers and Myers 1979; Soriguer and Rogers 1981; Rogers et al. 1994; Moreno and Villafuerte 1995; Fa et al. 1999; Palomares et al. 2001). Our results agree with these findings: warrens were preferentially located in medium shrub cover areas (25–50% cover), somewhat less than Jaksic and Soriguer’s (1981) suggested optimum shrub densities of 70–80%, which we consider to be too high.

Distribution of warrens according to warren protection Warren protection Shrub roots Shrub roots and rocks Rocks

No. of warrens No. of total burrow entrances Mean (s.d.) Mode Minimum Maximum No. of active burrow entrances Mean (s.d.) Mode Minimum Maximum

Unprotected

84

59

17

36

5.9 (5.2) 3.0 1 29

5.7 (4.5) 2.0 1 18

5.4 (4.0) 4.0 2 18

6.3 (5.5) 2.0 1 24

4.2 (4.6) 3.0 0 27

4.7 (4.5) 2.0 0 18

4.9 (4.0) 4.0 1 17

4.3 (4.2) 2.0 0 19

Distribution and ecology of rabbit warrens in central Spain

We do not know whether the botanical composition of grasslands is directly related to warren digging. It might be presumed that it is not very important, considering the great spectrum of habitats where rabbits are found around the world. However, as plant communities are closely linked with soil properties and traditional human management, they provide helpful complementary information that enables us to better comprehend empirical data. Thero–Brometalia habitat had very few rabbit warrens relative to its large area because of the disturbance caused by periodic ploughing. Furthermore, it helps us to understand the great loss of rabbit habitat caused by the expansion of arable land that took place in Spain from the 1960s, which resulted in very large increases in periodically ploughed arable lands and left them without permanent vegetation cover (Palomares et al. 2001; Virgós et al. 2003; Calvete et al. 2004). On the other hand, Agrostio–Stipion communities were strongly favoured by rabbits for warren establishment. Agrostio–Stipion communities thrive on deep, well drained unploughed lands with frequent stone outcrops that favour establishment of deep, dry stable warrens. Additionally, they are composed of a mix of annual and perennial grasses that provide valuable green feed for rabbits during summer, in contrast to Thero–Brometalia, Helianthemetalia and Sysymbrietalia communities, which are composed predominantly of annuals that dry off during summer. Helianthemetalia communities (comprising 43% of the plot area and having 53% of all burrows), which thrive in unploughed but shallower soils, were also positively selected, though not as strongly as Agrostio–Stipion. Communities of Holoschoenetalia and Agrostion castellanae, comprising 8% of the total sampled area, which is representative of their usual prevalence in those types of habitats, seem to be less selected. That could be related to the risk of warren collapse because of flooding during rainy seasons that would kill kittens inside the nest chambers (Copson et al. 1981; Parer and Libke 1985; Trout et al. 2000; Palomares 2003b). However, as those communities grow on damp soils, they have longer vegetative periods than their neighbours, hence offering some green pasture in the dry season. Therefore Holoschoenetalia and Agrostion castellanae communities seem to play an important role in the distribution and organisation of warrens (Bell and Webb 1991; Mutze 1991). More than 81% of warrens were located beneath some kind of protection, mostly shrub-like holm oaks. This phenomenon seems to be a characteristic of Mediterranean habitats (Rogers and Myers 1979; Palomares 2003a) and could be a result of high predation pressure (Jaksic and Soriguer 1981). Other factors such as lower risk of floods, higher stability and better temperature-buffering capacity can also contribute to the preference for these locations (Palomares 2003a). Following previous studies (Palomares et al. 2001; Palomares 2003a), we expected unprotected

Wildlife Research

729

warrens to be larger than protected warrens. However, despite the mean size of unprotected warrens being larger, the differences were not statistically significant. Consequences to management of rabbit populations The effects of human activities changing landscape features and the consequences for rabbit populations should be taken into account when managing Mediterranean rangelands. For example, low livestock stocking rates homogenise the habitat towards a closed woodland and are likely to have negative consequences for rabbit populations (González and San Miguel 2004). The pasture communities that thrive on soils with little human disturbance (e.g. ploughing) were preferred by rabbits. This also has consequences for rabbit population management, because selecting pasture communities with plant associations that are favoured by rabbits as the sites for restocking programs is likely to enhance the success of restocking. Additionally, high densities of warrens seem usually to be associated with prosperous rabbit populations (Williams et al. 1995; Palomares 2001a; Lombardi et al. 2003). Consequently, we suggest artificial warrens to have 5–6 burrow entrances, be spaced with an equivalent density of 10 units per hectare and should be located in unploughed, deep soils with plenty of shrubs and rocks. Acknowledgments We acknowledge the Finat family for their altruism in allowing us to work in their estate and the CBD-Habitat Foundation for their support. Special thanks go to Dr Greg Mutze, Dr C. Calvete and Dr J. San Miguel for their extremely valuable comments on a previous version of the manuscript, and Dr J. Solana for statistical assessment. We are indebted to S. Mancebo and the Department of Topography of the ETSI Montes for lending us the GPS devices. References Angulo, E., and Villafuerte, R. (2003). Modelling hunting strategies for the conservation of wild rabbit populations. Biological Conservation 115, 291–301. doi:10.1016/S0006-3207(03)00148-4 Banks, B., Hume, I. D., and Crowe, O. (1999). Behavioural, morphological and dietary response of rabbits to predation risk from foxes. Oikos 85, 247–256. Bell, D. J., and Webb, N. J. (1991). Effects of climate on reproduction in the European wild rabbit (Oryctolagus cuniculus). Journal of Zoology 224, 639–648. Branco, M., Monnerot, M., Ferrand, N., and Templeton, A. R. (2002). Postglacial dispersal of the European rabbit (Oryctolagus cuniculus) on the Iberian Peninsula reconstructed from nested clade and mismatch analyses of mitochondrial DNA genetic variation. Evolution 56, 792–803. Braun-Blanquet, J. (1928). ‘Pflanzen Sociologie.’ 1st edn. (SpringerVerlag: Berlin.) Calvete, C., Estrada, R., Angulo, E., and Cabezas-Ruiz, S. (2004). Habitat factors related to wild rabbit conservation in an agricultural landscape. Landscape Ecology 19, 531–542.

730

Wildlife Research

CBD-Hábitat (2001). Fundación para la Conservación de la Biodiversidad y su Hábitat. Memoria final del Proyecto LIFE “Gestión del Monte mediterráneo para la conservación de Águila imperial, Buitre negro, Cigüeña negra y Lince ibérico 99/NAT/E-006336” de la Comisión Europea. Documentación interna no publicada. Madrid, España. Clements, F. E. (1916). ‘Plant Succession: An Analysis of the Development of Vegetation.’ (Carnegie Institution of Washington: Washington, DC.) Cooke, B. D. (1981). Rabbit control and the conservation of native mallee vegetation on roadsides in South Australia. Australian Wildlife Research 8, 627–636. doi:10.1071/WR9810627 Copson, G. R., Brothers, N. P., and Skira, I. J. (1981). Distribution and abundance of the rabbit, Oryctolagus cuniculus (L.), at subantarctic Macquarie Island. Australian Wildlife Research 8, 597–611. doi:10.1071/WR9810597 Cowan, D. P. (1987). Group living in the European rabbit (Oryctolagus cuniculus): mutual benefit or resource localization? Journal of Animal Ecology 56, 779–795. Cowan, D. P. (1991). The availability of burrows in relation to dispersal in the wild rabbit Oryctolagus cuniculus. Symposia of the Zoological Society London 63, 213–230. DGCN (2002). Censo diagnóstico de las poblaciones de Lince ibérico (Lynx pardinus Temminck) en España. Dirección General de Conservación de la Naturaleza. Ministerio de Medio Ambiente. Madrid. Fa, J. E., Sharples, C. M., and Bell, D. J. (1999). Habitat correlates of European rabbit (Oryctolagus cuniculus) distribution after the spread of RVHD in Cadiz Province, Spain. Journal of Zoology 249, 83–96. doi:10.1017/S0952836999009085 Figueroa, M. E., and Davy, A. J. (1991). Response of Mediterranean grassland species to changing rainfall. Journal of Ecology 79, 925–941. Gibb, J. A., and Fitzgerald, B. M. (1995). Dynamics of sparse rabbits (Oryctolagus cuniculus), in Orongorongo Valley, New Zealand. New Zealand Journal of Zoology 22, 115–121. Gleason, H. A. (1926). The individualistic concept of the plant association. Bulletin of the Torrey Botanical Club 53, 7–26. González, L. M., and San Miguel, A. (Coords) (2004). Manual de buenas practicas de gestión en fincas de monte mediterráneo de la Red Natura 2000. Dirección General para la Biodiversidad. Ministerio de Medio Ambiente. Madrid (España). González, L. M., Bustamante, J., and Hiraldo, F. (1990). Factors influencing the present distribution of the Spanish imperial eagle Aquila adalbeti. Biological Conservation 51, 311–319. doi:10.1016/00063207(90)90116-7 Jaksic, F. M., and Soriguer, R. C. (1981). Predation upon the European rabbit (Oryctolagus cuniculus) in Mediterranean habitats of Chile and Spain: a comparative analysis. Journal of Animal Ecology 50, 269–281. King, D. R., Wheeler, S. H., and Robinson, M. H. (1984). Daytime locations of European rabbits at three localities in south-western Australia. Australian Wildlife Research 11, 89–92. doi:10.1071/ WR9840089 Kolb, H. H. (1985). The burrow structure of the European rabbit (Oryctolagus cuniculus L.). Journal of Zoology 206, 253–262. Kolb, H. H. (1991a). Use of burrows and movements by wild rabbits (Oryctolagus cuniculus) on an area of sand dunes. Journal of Applied Ecology 28, 879–891. Kolb, H. H. (1991b). Use of burrows and movements of wild rabbits (Oryctolagus cuniculus) in an area of hill grazing and forestry. Journal of Applied Ecology 28, 892–905. Krebs, C. J. (1999). ‘Ecological Methodology.’ 2nd edn. (Addison Wesley Longman: Menlo Park, CA.)

G. Gea-Izquierdo et al.

Lombardi, L., Fernández, N., Moreno, S., and Villafuerte, R. (2003). Habitat-related differences in rabbit (Oryctolagus cuniculus) abundance, distribution, and activity. Journal of Mammalogy 8, 26–36. doi:10.1644/1545-1542(2003)084<0026:HRDIRO>2.0.CO;2 Manly, B. F. J., MacDonald, L. L., and Thomas, D. L. (1993). ‘Resource Selection by Animals: Statistical Desing and Analysis for Field Studies.’ (Chapman and Hall: London.) Marañón, T. (1986). Plant species richness and canopy effect in the savanna-like ‘dehesa’ of S.W. Spain. Ecologia Mediterranea XII, 131–141. Moreno, S., and Villafuerte, R. (1995). Traditional management of scrubland for the conservation of rabbits Oryctolagus cuniculus and their predators in Doñana National Park, Spain. Biological Conservation 73, 81–85. doi:10.1016/0006-3207(95)90069-1 Moreno, S., Villafuerte, R., and Delibes, M. (1996). Cover is safe during the day but dangerous at night: the use of vegetation by European wild rabbits. Canadian Journal of Zoology 74, 1656–1660. Mutze, G. J. (1991). Long-term effects of warren ripping for rabbit control in semi-arid South Australia. Australian Rangelands Journal 13, 96–106. Myers, K., Parer, I., Wood, D., and Cooke, D. (1994). The rabbit in Australia. In ‘The European Rabbit: The History and Biology of a Succesful Colonizer’. (Eds H. V. Thompson and C. M. King.) pp. 108–157. (Oxford University Press: New York) Mykytowycz, R., and Gambale, S (1965). A study of the inter-warren activities and dispersal of wild rabbits, Oryctolagus cuniculus (L.), living in a 45-ac paddock. CSIRO Wildlife Research 10, 111–123. Palomares, F. (2001a). Comparison of three methods to estimate rabbit abundance in a Mediterranean environment. Wildlife Society Bulletin 29, 578–585. Palomares, F. (2001b). Vegetation structure and prey abundance requirements of the Iberian lynx: implications for the design of reserves and corridors. Journal of Applied Ecology 38, 9–18. doi:10.1046/j.1365-2664.2001.00565.x Palomares, F. (2003a). Warren building by European rabbit (Oryctolagus cuniculus) in relation to cover availability in a sandy area. Journal of Zoology 259, 63–67. doi:10.1017/S09528369 02002960 Palomares, F. (2003b). The negative impact of heavy rains on the abundance of a Mediterranean population of European rabbits. Mammalian Biology 68, 224–234. doi:10.1078/1616-5047-00088 Palomares, F., and Delibes, M. (1997). Predation upon European rabbits and their use of open and closed patches in Mediterranean habitats. Oikos 80, 407–410. Palomares, F., Delibes, M., Revilla, E., Calzada, J., and Fedriani, J. M. (2001). Spatial ecology of Iberian lynx and abundance of European rabbits in southwestern Spain. Wildlife Monographs 65. Parer, I. (1977). The population ecology of the wild rabbit, Oryctolagus cuniculus (L.), in a Mediterranean-type climate in New South Wales. Australian Wildlife Research 4, 171–205. doi:10.1071/ WR9770171 Parer, I., and Libke, J. A. (1985). Distribution of rabbit, Oryctolagus cuniculus, warrens in relation to soil type. Australian Wildlife Research 12, 387–405. doi:10.1071/WR9850387 Parer, I., and Wood, D. H. (1986). Further observations of the use of warren entrances as an index of the number of rabbits, Oryctolagus cuniculus. Australian Wildlife Research 13, 331–332. doi:10.1071/ WR9860331 Parer, I., Fullagar, P. J., and Malafant, K. W. (1987). The history and structure of a large warren of the rabbit, Oryctolagus cuniculus, at Canberra, A.C.T. Australian Wildlife Research 14, 505–513. doi:10.1071/WR9870505

Distribution and ecology of rabbit warrens in central Spain

Parker, B. S., Hall, L. S., Myers, K., and Fullagar, P. J. (1976). The distribution of rabbit warrens at Mitchell, Queensland, in relation to soil and vegetation characteristics. Australian Wildlife Research 3, 129–148. doi:10.1071/WR9760129 Pérez Corona, M. E., Vázquez de Aldana, B. R., García Criado, B., and García Ciudad, A. (1998). Variations in nutritional quality and biomass production of semiarid grasslands. Journal of Range Management 51, 570–576. Quinn, G. P., and Keough, M. J. (2002). ‘Experimental Design and Data Analysis for Biologists.’ (Cambridge University Press: Cambridge.) Rivas-Martínez, S., and Penas, A. (Coords) (2004). Atlas y Manual de los Hábitat de España. Ministerio de Medio Ambiente, Madrid. Rivas-Martínez, S., Fernández González, F., Loidi, J., Lousa, M., and Penas, A. (2001). Syntaxonomical checklist of vascular plant communities of Spain and Portugal to association level. Itinera Geobotanica 14, 5–341. Rogers, P. M., and Myers, K. (1979). Ecology of the European wild rabbit, Oryctolagus cuniculus (L.), in Mediterranean habitats. Journal of Applied Ecology 16, 691–703. Rogers, P. M., Arthurs, C. P., and Soriguer, R. C. (1994). The rabbit in continental Europe. In ‘The European Rabbit: The History and Biology of a Succesful Colonizer’. (Eds H. V. Thompson and C. M. King.) pp. 22–63. (Oxford University Press: New York.) San Miguel, A. (2001). ‘Pastos Naturales Españoles.’ (Mundi-Prensa: Madrid.) Soriguer, R. C. (1981). Biología y dinámica de una población de conejos (Oryctolagus cuniculus L.) en Andalucía occidental. Doñana Acta Vertebrata 8, 1–379. Soriguer, R. C., and Rogers, P. M. (1981). The European wild rabbit in Mediterranean Spain. In ‘Proceedings of the World Lagomorph Conference’. (Eds K. Myers and C. D. MacInness.) pp. 600–613. (University of Guelph: Canada.) Surridge, A. K., Bell, D. J., and Hewitt, G. M. (1999). From population structure to individual behaviour: genetic analysis of social structure in the European wild rabbit (Oryctolagus cuniculus). Biological Journal of the Linnean Society 68, 57–71. doi:10.1006/ bijl.1999.0330

Wildlife Research

731

Trout, R. C., Langton, S., Smith, G. C., and Haines-Young, R. H. (2000). Factors affecting the abundance of rabbits (Oryctolagus cuniculus) in England and Wales. Journal of Zoology 252, 227–238. doi:10.1017/S095283690000995X Vázquez de Aldana, B. R., García Ciudad, A., Pérez Corona, M. E., and García Criado, B. (2000). Nutritional quality of semi-arid grassland in western Spain over a 10 year period: changes in chemical composition of grasses, legumes and forbs. Grass and Forage Science 55, 209–220. doi:10.1046/j.1365-2494.2000.00217.x Villafuerte, R., Viñuela, J., and Blanco, J. C. (1998). Extensive predator persecution caused by population crash in a game species: the case of red kites and rabbits in Spain. Biological Conservation 84, 181–188. doi:10.1016/S0006-3207(97)00094-3 Virgós, E., Cabezas-Díaz, S., Malo, A., Lozano, J., and López-Huertas, D. (2003). Factors shaping European rabbbit abundance in continuous and fragmented populations of central Spain. Acta Theriologica 48, 113–122. Webb, N. J. (1993). Growth mortality in juvenile European wild rabbits (Oryctolagus cuniculus). Journal of Zoology 230, 665–677. Wheeler, S. H., King, D. R., and Robinson, M. H. (1981). Habitat and warren utilization by the European rabbit, Oryctolagus cuniculus (L.), as determined by radio-tracking. Australian Wildlife Research 8, 581–588. doi:10.1071/WR9810581 Williams, K., Parer, I., Coman, B., Burley, J., and Braysher, M. (1995). ‘Managing Vertebrate Pests: Rabbits.’ (Australian Government Publishing Service: Canberra.)

Manuscript received 8 December 2004, accepted 22 September 2005

http://www.publish.csiro.au/journals/wr