OUP user menu

Predation Risk as a Factor Affecting Sexual Segregation in Alpine Ibex

Stefano Grignolio , Iva Rossi , Bruno Bassano , Marco Apollonio
DOI: http://dx.doi.org/10.1644/06-MAMM-A-351R.1 1488-1497 First published online: 1 December 2007


Alpine ibex (Capra ibex) are polygynous ungulates that exhibit extreme sexual dimorphism and segregation. To test the hypothesis that risk of predation plays a key role in the development of sexual segregation (habitat segregation) in this species, compositions and locations of groups of ibex were recorded from February 2003 to October 2004. Individual fixes of marked adult males and females in the Gran Paradiso National Park in Italy were collected monthly over a 4-year period (January 2000-December 2003). Distances were measured between each location and the nearest refuge area (rocky slopes), and between each location and the nearest source of disturbance (roads and hiking trails). Group size was not related to distances from refuge areas or from sources of disturbance, but sex, age, and weaning played a key role in spatial segregation. During the period of rut, females kept nearer to rocky slopes and further from hiking trails than males. The presence of young also influenced the spatial behavior of females: mothers made use of refuge areas more frequently and of areas near hiking trails less frequently than other females. In contrast, no difference between the spatial behaviors of pregnant and non-pregnant females was recorded in spring. Age played an important role in space use by males, but not by females. Young males (2–5 years) stayed closer to rocks than did middle-aged males (6–11 years), and both stayed closer to rocks than did adult males (>11 years). Adult males stayed closer to hiking trails than middle-aged males; likewise, the latter stayed closer to hiking trails than young males. Males stayed near areas with human presence, likely because of good foraging opportunities. Although our findings do not exclude other explanations, they support the reproductive strategy-predation risk hypothesis. Moreover, predation risk was shown as an important factor leading to both sexual and age segregation in males.

Key words
  • alpine ibex
  • antipredator behavior
  • Capra ibex
  • predation risk
  • sexual segregation

Sexual segregation has been reported among fish, birds, and mammals (Bleich et al. 1997; Ruckstuhl and Neuhaus 2000); yet, of all the studies on different aspects of sexual segregation, those on mammals are dominant in literature (Bowyer 2004). In polygamous ungulates, sexual segregation is widespread and it is likely to be influenced by social, spatial, and temporal factors such as population density, resource distribution, environmental conditions, and periodicity of mating opportunities (Main et al. 1996). Sexual dimorphism also plays a key role. Ruckstuhl and Neuhaus (2002) showed that sexual dimorphism was correlated with behavioral differences in 19 ungulate species. Sexual segregation has been reported in both heterogeneous and homogeneous habitats. In heterogeneous environments the resources are partitioned and the animals use different areas, thus forming different groups (Festa-Bianchet 1988). In homogeneous environments, where the resources and the different kinds of habitat are evenly distributed, several ungulate species still show pronounced sexual segregation (Main et al. 1996; McCullough et al. 1989; Weckerly 1993). Mating strategies may considerably affect sexual segregation: for example, African buffalo (Syncerus caffer) are characterized by a prolonged mating period and, as a consequence, groups are mixed throughout the year (Prins 1989); in contrast, moose (Alces alces), whose rut is limited, show the most pronounced sexual segregation just at the end of winter (Miquelle et al. 1992). Barboza and Bowyer (2000) noted that sexual segregation had been traditionally defined in terms of such features as the different use of space, habitat, or forage by sexes. Additional features (e.g., social segregation) recently have been used to define sexual segregation (Bowyer 2004). In order to explain sexual segregation in ungulates, researchers suggested a plethora of hypotheses, which are based on different energetic requirements and reproductive strategies (reviewed in Main et al. 1996; Mooring et al. 2003; Ruckstuhl and Neuhaus 2000). These hypotheses are not necessarily mutually exclusive, and more factors are likely to be core-sponsible for the evolution of sexual segregation (Mooring et al. 2003).

The reproductive strategy-predation risk hypothesis postulates that males and females pursue different strategies to maximize reproductive success, with males maximizing their body condition and females maximizing survival of offspring. Females use habitats of increased security, even though containing less sources of food for offspring, whereas males are likely to use areas with more forage even though of greater predation risk, so as to increase their body size and be more competitive during the period of rut (Mooring et al. 2003).

The reproductive strategy-predation risk hypothesis presumes different habitat use by the sexes as a consequence of a different trade-off between forage quality and safe areas (Bleich et al. 1997; Ciuti et al. 2004; Corti and Shackleton 2002; Mooring et al. 2003). Accordingly, sexual body size dimorphism, which is associated with polygynous mating systems in ungulates (Weckerly 1998), is likely to be a key factor favoring sexual segregation (Mysterud 2000), and body size plays a main role in the reproductive strategy-predation risk hypothesis (Mooring et al. 2003). Body size generally influences predation risk; larger individuals are less vulnerable to predation than smaller ones (Ruckstuhl and Neuhaus 2000). Alpine ibex (Capra ibex) are characterized by extreme sexual dimorphism in body mass (about 100%—Giacometti et al. 1997), and males display a slow increase of body mass with considerable weight difference between age classes (Bassano et al. 2003). Recently, Grignolio et al. (2007) showed that antipredator behavior modified habitat selection of lactating female ibex, driving mothers to use more rugged escape terrain. Predation risk varies according to different environmental conditions and can modify sexual segregation. Kie and Bowyer (1999) reported that sexual segregation in a population of white-tailed deer (Odocoileus virginianus) decreased after reduction in the number of predators. Human activities in general are likely to produce disturbance (Creel et al. 2002; Szemkus et al. 1998) and responses can vary by sex. Both predation risk and human harassment may evoke different behavioral responses between and within sexes. In fallow deer (Dama dama), females have been found to avoid areas characterized by human disturbance, whereas males tend to use disturbed areas for their better forage, thus living closer to humans (Apollonio et al. 2005; Ciuti et al. 2004).

Alpine ibex mainly lived above tree line, in homogeneous habitats, and they selected alpine meadows and rocky slopes (Grignolio et al. 2003, 2007). It is likely that the flat meadows contained a higher quantity of forage compared to rugged terrains because of the high percentage of rock and the friability of the soil in the latter. Steep slopes and cliffs provide mountain ungulates with escape terrains; encounters and attacks are less likely to occur in these areas because mountain ungulates can move more easily in them than their predators (Bleich 1999; Bleich et al. 1997; Corti and Shackleton 2002; Frid 1997). Moreover, Grignolio et al. (2007) proved that females with young gradually used these areas less as the offspring grew.

Ibex showed minimal adaptability to snow cover, and in snowy winters they reduced their home ranges and selected rocky slopes without snow. In particular, larger males had greater mobility problems on frozen terrain and snow (Grignolio et al. 2003; Parrini et al. 2003; Villaret et al. 1997).

Although several authors showed sexual segregation in alpine ibex (spatial, habitat, and social), their explanations (Bon et al. 2001; Ruckstuhl and Neuhaus 2001; Villaret and Bon 1995; Villaret et al. 1997) did not include the reproductive strategy-predation risk hypothesis. The aim of our study was to investigate whether predation risk could account for spatial sexual segregation in alpine ibex. In particular we tested if male ibex used flat alpine meadows near human disturbance more than females because the meadows are assumed to have higher quality and a greater quantity of forage; females selected habitat closer to rocky slopes than males because rugged terrains are safer, although providing less forage; females with young stayed closer to rocky slopes than females without young; younger (i.e., smaller) males stayed closer to rocky slopes than older (larger) males; and in winter, males were forced to use rocky slopes because of snow cover. During this season, sexual segregation disappeared and both sexes used areas near rugged terrains.

Materials and Methods

Study area.—The study area was situated in the center of the Gran Paradiso National Park (GPNP; 45°35′N, 7°12′E), in the Levionaz Valley in the northwestern Italian Alps. The ibex habitat ranged between approximately 1,500 and 3,300 m elevation and consisted mainly of cliffs, slopes, and alpine meadows (Carex curvula and Festuca); conifer woods (Picea abies, Larix decidua, and Pinus cembra) and bushes (Rhododendron ferrugineum, Vaccinium myrtillus, and Juniperus communis) were not common in the areas used by ibex (Grignolio et al. 2003,2007). The GPNP has been free from the most relevant predators (e.g., lynx [Lynx lynx] and wolf [Canis lupus]) for about 1 century. Domestic ungulates have not been present in the Levionaz Valley for about 15 years. Roads are present and open to cars only at the bottom of the valley. Over the whole study area there are scattered hiking trails, and the trails used most frequently by people are those linking the bottom of the valley to the summer pasture alpine huts (Fig. 1).

Fig. 1

Study area map (the arrow indicates Gran Paradiso National Park, Italy) showing hiking trails and roads (continuous line), rocky slopes (squared areas), villages (black areas) at the bottom of the valley and summer pasture alpine huts (black circle). Dashed line A shows an area where the hiking trails are far from a refuge area, dashed line B shows an area where the hiking trails are closer to a refuge area. For more details, see the text.

Population and sample size.—The population at GPNP is the only surviving natural population of alpine ibex: this species risked extinction in the 19th century because of hunting and poaching. Many populations that were distributed in the Alps were extirpated and only a few hundred ibex survived in the territories that are now located within the GPNP. The last population included only 400–500 ibex after the Second World War. Legal trophy hunting was stopped during the 1960s but poaching continued to be a key factor for several years. Rearing of livestock in summer in alpine meadows was widespread in the GPNP and drastically decreased only in the last 20 years (Passerin d'Entrëves 2000). In the GPNP, the principle predators of ibex are virtually absent. During our study, scarce signs of transient and scattered individual wolves were found, but no stable presence was evident. The more than 140 ibex that were marked in the last 15 years did not suffer any predation events, nor were predation events reported in surveys made by park rangers the whole population under study. Predation of ibex by red fox (Vulpes vulpes) is only a supposition. GPNP staff (particularly rangers) never reported predation by red fox of ibex, either of young or adults (B. Bassano, pers. comm.). Predation by golden eagles is limited and focused only on young ibex a few months of age. Moreover, given that the hunting technique of golden eagles consists of pushing young ibex down rocky slopes to cause their death, it is likely that this kind of predation did not influence antipredator behavior. This would be consistent with the findings of Grignolio et al. (2007), who showed that lactating females used rugged escape terrain as refuge areas, and also is supported by another study of ours that entailed more than 700 h of observation of females with young during which no attack was seen (S. Grignolio et al., in litt.). Accordingly, we assume that the GPNP is free from predators of adult ibex. Censuses of alpine ibex in the study area ranged from 114 males, 113 females, and 61 young and yearlings in 2000, to 108 males, 119 females, and 30 young and yearlings in 2003. Data on individually recognizable ibex (83 males and 33 females) were analyzed.

In order to capture ibex we followed the methods of Bassano et al. (2004). We used chemical immobilization by means of a mixture of xylazine (about 70 mg per ibex; Rompun; Bayer Italia, Milano, Italy) and ketamine (about 50 mg per ibex; Ketavet Gellini International, Aprilia, Italy). After capture, we collected biometric data and biological samples; then we marked the ibex and used an antagonist of xylazine (Atipamezol, 1 ml per ibex; Antisedan, Priser Italia, Roma, Italy) to accelerate the recovery of the animal. We used radiocollars (285 g, corresponding to 0.7% of the mean weight of females; VHF Televilt; Lindesberg, Sweden) and livestock (sheep and goats) ear tags (5.5×6.8 cm; Allflex Europe, Vitré, France) to mark the animal. All aspects of animal capture and data collection were approved by the Italian National Wildlife Institute, and conformed to guidelines established by the American Society of Mammalogists (Gannon et al. 2007). In order to estimate the age of the ibex, we counted the clearly separated annuli in their horns (Ratti and Habermehl 1977; von Hardenberg et al. 2004). The age of marked females ranged from 2 to 14 years, whereas that of marked males ranged from 2 to 17 years. The birth period for alpine ibex occurred from June to early July; females were seen lactating in summer and, sporadically, in autumn. A female was considered to be lactating when she exhibited maternal behavior toward a young (suckling, leading, or lying by a young); young usually followed their mother until the following spring. In order to make a preliminary analysis of the influence of pregnancy on sexual segregation, we considered to be pregnant those females that were seen with young at least once, even though we risked classifying as nonpregnant those females that had lost their young just after its birth. The mating season began in December and ended around the 2nd week of January, and peaked in the last 2 weeks of December.

Data collection and analysis.—Any set of ibex whose individuals stayed within 50 m of each other (Frid 1997) was defined as a group. We used the group classification of Gross et al. (1995), which had been elaborated with reference to previous work on ibex (Aronson 1982). According to our classification, all groups were 1st subdivided into 3 main categories: male groups—when at least 75% of the members were males; female groups—when at least 75% of the members were females; and mixed groups—when the ratio of both sexes was lower than 75%. These criteria enabled us to classify each group in an unequivocal way, and using a composition of 75% of individuals of the same sex to define monosexual groups made us sure that male and female groups were unmistakably defined by the prevalence of 1 sex. Nevertheless, mixed groups comprised only about 8% of the total. Male groups were then subdivided into 2 unequivocal categories: adult groups (more than 50% of the individuals being >5 years of age) and young groups (more than 50% of the individuals being ≤5 years of age). Similarly, we also distinguished 2 subtypes of female groups: female with young groups (at least 1 young was present) and female without young groups (no young present).

From February 2003 to October 2004 we walked throughout the whole study area covering transects 10 times per month. We collected locations and compositions of each group by using binoculars and spotting scopes. Given that groups of ibex move slowly and infrequently, we avoided recounting the same animal or the same group by moving rapidly. Moreover, we used marked animals to identify the corresponding groups. Data were collected uniformly over the daylight hours. Group sizes were calculated by arithmetic mean for comparison with other studies, and by Jarman's (1974) typical group size (TGS). If compared to the arithmetic mean, TGS is a better representation of the social environment experienced by the average animal in the population, and is defined as:

Embedded Image

where n is the group size and N is the total number of animals in all groups.

The locations (fixes) of individual ibex were recorded from January 2000 to December 2003 (26,002 fixes). Every month at least 15 fixes were collected for each radiocollared ibex over the daylight hours. Moreover, we collected the locations of each observation of ear-tagged ibex (never more than twice a day).

We subdivided the study area into a grid of squares of 125×125 m, and we located each group and each fix in the center of the corresponding square. We measured the distance between each location and the nearest rocky slope, that is, the nearest refuge area, by means of geographic information system software. Furthermore, the distance was measured from each group and each fix to the nearest road or hiking trail that could be perceived as a source of disturbance. We decided to consider both distances because ibex could select areas that were either far (Fig. 1, segment A) or close (Fig. 1, segment B) to rocks and hiking trails. Whenever 1 of these 2 habitats prevailed, to employ just 1 distance could have biased the results.

To detect eventual differences in habitat quality between areas near disturbance and areas near refuge, we used both a land use map and a pastoral value map of meadows. The pastoral value map of meadows, as evaluated by the Staff of the Department of Agronomy, Forest and Land Management of the University of Turin, assessed the pastoral value (PV) of each homogeneous spot, whereby both the occurrence and the quality of each forage species were taken into consideration (Daget and Poissonet 1969). Researchers covered line transects homogeneously distributed in the meadows of the study area. Vegetation data were analyzed by computing percent contribution of each species to the total vegetation composition of each transect. Then, surveys were classified by means of a cluster analysis. Each cluster partition identified a unique vegetation type at a 1st level, which could be split into subtypes (facies) at a lower level. The validity of the classification was checked with a bootstrap validation procedure (Wishart 1999) that was performed starting from the original data set.

To emphasize the ecological difference between vegetation types, data were ordered according to a gradient analysis based on ecological indices (Whittaker 1967), and using the Landolt index (Landolt 1977). Starting from the vegetation composition of each survey, PVs were computed as described in Daget and Poissonet (1972). In addition, to assess the quality and quantity of the habitats used by ibex, we analyzed the habitat composition of the area covering 492 ha around hiking trails. This area was selected by establishing a border that ran 200 m from the hiking trails, parallel to the trails. We selected this distance because it was the mean distance we observed between males and hiking trails. We also evaluated the habitat composition of the area covering 572 ha around rocky slopes. Again, we established a border that ran 100 m around the perimeter of the rocky slopes. This distance was selected because it was approximately the distance of a line that separated the mean distances of males and females to refuge areas. The habitat compositions within these areas were different. Meadows (alpine meadows and pastures: 46.7%) were prevalent near the hiking trails (Fig. 2), whereas refuge areas near the rocks were characterized by unproductive habitats (rocks and stone ravines: 76.3%) and a low percentage of meadows (16.7%; Fig. 2). The 2 areas contained 60.7% of the fixes of males and 91.4% of the fixes of females, respectively. To evaluate differences in forage quality, we overlapped the 2 areas with a pastoral value map of meadows. This map usually takes into consideration the meadows only, so that the area around refuge areas overlapped only 16.5% of the map (mean PV = 2.8). In contrast, the area around the disturbance area overlapped 31.8% of the map and had a better PV (mean PV = 5.2).

Fig. 2

Habitat composition within refuge areas (572 ha near rocky slopes [white bars]) and disturbance areas (491 ha near hiking trails [black bars]). For more details, see the text.

Each month from May to October we also measured the forage quality in the principal meadows of the study site following Carranza and Valencia (1999) in 6 sampling areas, 3 in the meadows near rocky slopes and 3 in the flat areas. All observations in each area were made along the same fixed linear transect, at 5 sampling points that were 10 m from each other. Within approximately 1 m of each sampling point, we dropped a 30×30-cm sampling square, divided into 4 quarters, at 4 randomly selected locations. At the end of every month, we collected 400 samples of grass quality. At each sampling point we measured 3 parameters. The 1st was grass cover—a visual estimate of percent area covered by grass within the 30×30-cm sampling square. The 2nd was green index—the 4 corner tips of each square touched a leaf blade, which could be either green or brown. The green index was the number of tips at green leaves (0–4) against the number of tips touching any blade (0—4). If no tip touched a grass blade despite there being some grass cover, we recorded the greenness of the nearest leaf to any of the 4 tips, as suggested by Carranza and Valencia (1999). The 3rd was grass length—at the point with the most cover within the sampling square, we measured the length of the longest aerial part of the grass. In agreement with the findings of Carranza and Valencia (1999), these measurements of grass proved to be strongly correlated with biomass (dry mass), and they were considered a useful means of evaluating meadow productivity.

We used SPSS 12.0 (SPSS Inc., Chicago, Illinois) for statistical analysis. Data on distances of groups from refuge areas and on group size were transformed into a natural logarithm (ln(x + 1)) for statistical analysis to meet assumption of normality (Sokal and Rohlf 1995). To analyze data on group size, we used analysis of variance. Distance from groups to both refuge and disturbance areas were tested by means of multivariate analysis of variance (MANOVA), where season, group type, and group size were considered as fixed factors in the model. Our sampling design included repeated measures of the same ibexes. To avoid pseudoreplication (Machlis et al. 1985), the distances from the different areas were analyzed using linear mixed effect (LME) models considering ibex identity as a random factor and sex, age, season, and presence of young as fixed factors. Duncan post hoc test was used to detect differences between group types and age classes of males.


We defined the activity center of each home range by means of Ranges 6 software (Kenward et al. 2003), and for each radiocollared animal we measured the distance from its activity center to the activity centers of each other ibex. Male ibex were segregated from females. The mean distance between males (X̄ ± SE; 650.5 ± 36.26 m) was lower than the distance from the activity center of males to that of females (859.0 ± 53.62 m; paired-sample t-test: t = 3.153, d.f = 20, P = 0.005).

We observed 2,140 groups, and we classified them as adult males (n = 840), young males (n = 97), mixed (n = 167), females without young (n = 795), and females with young (n = 241). The size of the groups varied according to the group type. Mean group sizes (± SE) and TGS were, respectively: females without young (2.37 ± 0.09; TGS = 5.16), young males (4.37 ± 0.38; TGS = 7.67), females with young (5.85 ± 0.41; TGS = 12.78), adult males (6.49 ± 0.29; TGS = 17.71), and mixed (7.04 ± 0.44; TGS = 11.60). Group size differed significantly among types (F = 103.06, d.f. = 4, 2,135, P < 0.001); Duncan post hoc tests showed that only adult and young male groups had comparable sizes (P > 0.05).

The distance from refuge areas (MANOVA, F = 14.74, d.f. = 4, 2,135, P < 0.001) and from disturbance areas (MANOVA, F = 75.94, d.f = 4, 2,135, P < 0.001) varied according to the group type. Adult male groups were the farthest from refuge areas and the nearest to disturbance areas, whereas female groups kept closer to rocky slopes (Figs. 3A and 3B). Therefore, although group type influenced the location of the group, unlike in other ungulate species, group size was not influenced by the location (Table 1). Distances from refuge and disturbance areas to groups did not change over the seasons (Table 1). When we considered the distance from rocky slopes, Duncan post hoc tests grouped the 5 group types into 2 subsamples on the grounds of 2 behavioral patterns: male (adult and young) groups, and mixed and female (with and without young) groups (Fig. 3). In contrast, each type of group differed from the others in their distance from disturbance areas, except for mixed and female without young groups (Duncan post hoc tests; Fig. 3).

View this table:
Table 1

Results of MANOVA on distance of types of groups of alpine ibex (Capra ibex) from rocky slopes (refuge areas, r2 = 0.236, P < 0.001) and from hiking trails and roads (disturbance areas, r2 = 0.467, P < 0.001).

Dependent variabled.f.FP-value
InterceptIn distance from refuge areas1534.84>0.001
Distance from disturbance areas1545.52>0.001
SeasonIn distance from refuge areas30.970.408
Distance from disturbance areas30.970.406
Group typeIn distance from refuge areas48.09>0.001
Distance from disturbance areas435.85>0.001
Group sizeIn distance from refuge areas430.910.641
Distance from disturbance areas430.990.500
Season×group type×group sizeIn distance from refuge areas891.280.035
Distance from disturbance areas891.240.071
Fig. 3

Mean distance from different types of groups of alpine ibex (Capra ibex) to A) refuge areas and B) disturbance areas. Broken vertical lines subdivide groups into subsamples, as classified by Duncan post hoc tests. Differences among subsamples are significant (P < 0.05), whereas groups within the same subsample are not significantly different (P-value in graphs).

Sex influenced the spatial behavior of individual ibex. Females kept closer to rocks than all age classes of males. The mean distance from rocks (± SE) was 59.8 ± 1.1 m for females and 122.4 ± 1.8 m for males (LME: F = 158.5, d.f. = 1, 25,998, P < 0.0001; Fig. 4A). Mean distance from hiking trails differed between sexes even more (females = 351.6 ± 1.8 m, males = 178.2 ± 1.2; LME: F = 233.7, d.f. = 1, 25,998, P < 0.0001; Fig. 4B).

Fig. 4

Mean distance (with SE) from male and female alpine ibex (Capra ibex) to A) refuge areas and B) disturbance areas. Data are from individual ibex (Duncan post hoc tests: * P < 0.05).

Age influenced the use of refuge areas and disturbed areas: when 3 different age classes were taken into account, age of males proved to be an important factor. Young males (2–5 years) kept closer to rocks (104.3 ± 3.1 m) than did middle-aged males (6–11 years; 116.0 ± 2.4 m; Duncan post hoc test, P < 0.05), and both were closer than adult males (> 11 years; 146.1 ± 4.0 m; Duncan post hoc tests, P < 0.05). In contrast, adult males kept closer to hiking trails (167.2 ± 5.1 m) than did middle-aged males (211.7 ± 4.7 m; Duncan post hoc test, P < 0.05), and both were closer than young males (239.7 ± 7.7 m; Duncan post hoc tests, P < 0.05).

Individual spatial behavior varied according to the season (distance from rocks, LME: F = 46.5, d.f. = 3, 17,454, P < 0.0001; distance from hiking trails, LME: F = 327.6, d.f. = 3, 17,454, P < 0.0001). Season and age class were found to affect the distance from refuge areas (LME: F = 52.6, d.f. = 6, 17,454, P < 0.0001) and from disturbance areas (LME: F = 38.2, d.f = 6, 17,454, P < 0.0001). Season and presence of young were crucial factors in use of space by females. Spatial behavior of females varied according to the season (distance from refuge areas LME: F = 128.1, d.f. = 3, 8,420, P < 0.0001; distance from hiking trails LME: F = 88.5, d.f. = 3, 8,420, P < 0.0001). Lactating females kept closer to refuge areas (54.2 ± 1.6 m) than did females without young (62.1 ± 1.4 m; LME: F = 33.6, d.f. = 1, 8,420, P < 0.0001) and their locations were farther away from hiking trails (372.3 ± 8.9 m) compared to the locations of females without young (334.9 ± 5.1 m; LME: F = 36.1, df = 1, 8,420, P < 0.0001; Fig. 5). The interaction between such variables as season and presence of young also was significant (distance from refuge areas LME: F = 7.7, d.f. = 3, 8,420, P < 0.0001; distance from trails LME: F = 48.3, d.f. = 3, 8,420, P < 0.0001). In spring, no difference was found between pregnant females and nonpregnant ones (independent sample Mest: t = 0.622, d.f = 118, P = 0.535).

Fig. 5

Mean distance (with SE) from individual female alpine ibex (Capra ibex) with and without young to A) refuge areas and B) disturbance areas.

Differences between sexes were evident when data for individuals were examined as well: mean distances between locations and refuge areas for males and females were very different throughout the year (Fig. 6). We analyzed the data for November, December, and January (period of rut) and found significant differences in November (independent sample Mest: t = 4.489, d.f. = 162, P < 0.001) and January (independent sample t-test: t = 3.444, d.f. = 137, P = 0.001). Mean distance to disturbance areas did not differ between sexes in December and January, but a difference was evident in November (independent sample Mest: t = −2.606, d.f. = 162, P = 0.01).

Fig. 6

Monthly mean distance (with SE) from male (continuous line) and female (broken line) alpine ibex (Capra ibex) to A) refuge areas and B) disturbance areas.

Following the method of Carranza and Valencia (1999), grass cover in the monitored meadows in the flat areas (80.5% ± 4.31%) was higher than grass cover in the sampled areas near rocky slopes (60.3% ± 3.42%; independent sample Mest: t = 3.656, d.f = 32.3, P = 0.001; Fig. 7). Green indices of the 2 areas were not different (disturbance areas: 62.9% ± 5.45%; refuge areas: 59.7% ± 5.68%; independent sample Mest: t = 0.406, d.f. = 34, P = 0.688), but the number of tips touching any blade was different (disturbance areas: 3.4 ± 0.13; refuge areas 2.6 ± 0.14; independent sample Mest: t = 4.08, d.f. = 34, P > 0.001). Grass length did not differ between the sampling areas located in the flat meadows (122.4 ± 4.29 mm) and those near rocky slopes (118.3 ± 5.92 mm; independent sample Mest: t = 0.554, d.f. = 30.9, P = 0.583).

Fig. 7

Monthly mean grass cover (with SE) near refuge areas and disturbance areas. For more details about methodology see the “Materials and Methods.”


Alpine ibex showed sexual segregation and different use of habitats according to sex and age classes in an area that was free from predators of adult ibex and where golden eagles exerted only a scarce impact on young. Despite this, a different use of habitat considered refuge from predation was revealed between the sexes. Bon et al. (1995) proposed 2 nonexclusive hypotheses to explain why mouflons (Ovis aries) showed an antipredator strategy during the birth period in areas without predators: antipredator behavior has been selected for and it is still exhibited even without pressure from predators, and domestic ungulates and shepherds on mountain pastures may represent a source of disturbance in summer. Villaret et al. (1997) pointed out these 2 hypotheses to explain habitat segregation in alpine ibex during birth and lactation periods. Our findings exclude the 2nd interpretation. The study area has been free from domestic ungulates for about 15 years and therefore the antipredator behavior must have played a key role. Byers (1997) defined the presence of an antipredator behavior—that is, high vigilance or use of “predator-safe” habitats—in the absence of predators as the “ghost of predators past.” The explanation could be that these behaviors coevolved in ungulates and their predators for thousands of years and the absence of predators for a few centuries is not sufficient to remove them. Anthropic activities are not only a disturbance. Humans also can be perceived as a predator. Hunting and poaching of ibex used to be relevant factors affecting their status. Toward the end of the 19th century, hunting and poaching had reduced this species to a few hundred animals in the area that now constitutes the GPNP. After a recovery due to legal protection, during and after the Second World War the population at GPNP was still markedly reduced and comprised only 400–500 ibex. Undoubtedly, human influence was considerable and comparable to that of predators for centuries, at least until 6 or 7 ibex generations ago, and for this reason the “ghost of predators past” still seems to be present in ibex.

High mountain environments are characterized by a small number of habitat types. As a consequence, it was possible to make a clear distinction between refuge and disturbance areas and to show different behaviors according to sex. Male ibex made more frequent use than females of the areas near roads, villages, or hiking trails; these areas were characterized by higher human presence but also by a greater quantity of forage compared to steep rocky slopes. The amount of habitats characterized by good quality and quantity of forage (meadows) near risky areas was not the same as the corresponding amount near refuge areas (Fig. 2). Moreover, the density of grass near rugged terrains was lower compared to the density in flat areas. Males used disturbance areas more frequently, that is, they used areas with high forage quantity. Males used the pastures near villages and roads, particularly in spring, because they are free from snow and rich in new forage of high quality (Grignolio et al. 2003). Moreover, we also found a remarkable percentage of woodlands near trails (22.3%; Fig. 2): open larch wood can provide a large amount of forage, particularly in spring, when trees do not have needles and grass can grow without shading. However, the visibility in woodlands is more limited than in meadows and this may contribute in turning it into a riskier area for ibex.

In contrast, females kept farther from risky areas and closer to refuge areas; indeed, they lived all year above the tree line (Grignolio et al. 2004), in an environment characterized only by 3 habitat types (rocks and steep slopes, alpine meadows, and stone ravines). Like several other ungulate species such as Aepyceros melampus (Mysterud 2000), Giraffa camelopardalis tippelskirchi (Ginnett and Demment 1997; Young and Isbell 1991), Odocoileus hemionus (Main and Coblenz 1996), Ovis dalli dalli (Rachlow and Bowyer 1998), Cervus elaphus (Bonenfant et al. 2004), and D. dama (Apollonio et al. 2005; Ciuti et al. 2004), male ibex lived primarily in areas richer in food and with higher predation risk, whereas females selected safer habitats where the quantity of forage was lower. More precisely, male ibex used more areas with a higher percentage of meadows, and these meadows had higher density of grass and higher PV. Our results are consistent with those obtained on bighorn sheep by several researchers (Bleich et al. 1997; Mooring et al. 2003). In particular, Mooring et al. (2003) showed that groups of rams used sites with more abundant grass compared to the sites used by groups of ewes, and that females minimized predation risk for themselves and their offspring by choosing rugged escape terrains. Our findings about alpine ibex are similar even though the environmental conditions in our study area and those of Mooring et al. (2003) were very different. The main differences are that GPNP is free from predators and water sources are not a key factor because high mountains are rich in water throughout the year.

A positive and significant correlation between sexual dimorphism in body mass and behavioral differences between sexes has been shown for different species (Ruckstuhl and Neuhaus 2002). The behavioral difference increases along with the increase in difference of body mass. Moreover, body size is negatively associated with predation risk, with smaller individuals being more at risk because of natural predators. Our findings confirmed that within the same species and the same sex, weight and therefore also age influence sensitivity to predation risk and as a consequence also segregation. Young males (about 5 years of age) weigh as much as adult females and one-half as much as adult males (Bassano et al. 2003). Size and weight of male alpine ibex increases until they are 9–10 years of age, when they reach their maximum weight. Because young males preferred to keep closer to refuge areas and farther from disturbed areas than adult males, young males showed an intermediate behavior between that of females and adult males (Fig. 2). Our findings are consistent with those of Bon et al. (2001), who showed not only sexual segregation, but also intrasexual social segregation, thus supporting the predictions of the reproductive strategy-predation risk hypothesis.

Toigo (1999) found that lactating female ibex spent more time in antipredator behavior (vigilance behavior). In addition to confirming that finding, we also found that females with young were more segregated from males than females without young. Thus, the key role of antipredator behavior is further emphasized here than it was in an earlier study where we took into account only the females as a whole and their use of refuge areas (Grignolio et al. 2007). Once more, the presence of young proved crucial in determining different behaviors. In contrast, pregnant females did not show a more frequent use of refuge areas than other females in spring; despite the increase of energetic requirements during the last stage of pregnancy (Robbins 1993), pregnant females did not seem to spend more time further from refuge areas than other females. Similarly, difficulties in mobility due to pregnancy did not prevent pregnant females from using the same slopes and trails used by other females. Therefore, neither increased energetic requirements nor mobility problems forced pregnant females to change their antipredator behavior (Grignolio et al. 2007). Predation risk varies with reproductive status and body size, with those at greatest risk being the females with young, followed by females without young, young (small) males, and adult (large) males.

As a result of the sexual dimorphism that characterizes this species (Bassano et al. 2003) and the difficulty for males to move in deep snow cover (Parrini et al. 2003), we assumed that in winter males kept closer to the slopes, where the snow cover is shallower, and, as a consequence, sexual segregation would be markedly reduced. In contrast, our results showed that sexual segregation took place during winter too, and this suggests that the need to feed after the period of rut, as shown in moose (Miquelle et al. 1992), as well as the difficulty of moving among the rocks could force males to use areas that are farther from the slopes. Obviously, sexual segregation was absent during the period of rut (December), whereas in November, even though the rutting urge began to emerge, males and females lived separately. The fact that males employed a roving mating system (Clutton-Brock 1989) did not result in the sexes living together just before the rut.

In conclusion, the reproductive strategy-predation risk hypothesis seems to explain at least part of sexual segregation in GPNP. Our findings showed a different use of refuge and disturbance areas between sexes, between females with and without young, and between males of different ages, confirming our predictions. In contrast, deep snow cover did not modify sexual segregation. Moreover, other variables, such as prerut and postrut behavior and pregnancy, did not modify antipredator behavior.


This work was supported by a grant from the GPNP board. We are grateful to wardens of GPNP for capturing ibex and logistic support.

We thank A. Meriggi, E. Merli, and A. von Hardenberg for statistical advice. SG owes a particular debt of gratitude to S. Ciuti for his kind assistance and brilliant advice about sexual segregation.


  • Associate Editor was Martin B. Main.

Literature Cited

View Abstract