Study area

The study was carried out in the Alpes-de-Haute-Provence in the southeastern French Alps (~UTM coordinates: 285 000 E, 4908 000 N, 32 T), in an area of 6,460 ha situated at 1,200–1,700 m a.s.l. (Fig. 1). The climate in the area is somewhat Mediterranean, with an average annual precipitation of 950 mm, in which snow from December through March is included, and a mean annual temperature of 7°C. The hazel grouse ‘population’ in the study area was not isolated, but was connected to surrounding populations through movements. The study area was situated ca 50 km from the southern limit of the expanding hazel grouse range. Interviews with local hunters and poachers suggest that hazel grouse first colonised the study area in the 1950s, and that by the 1970s the species was quite common. Large inter-connected tracks of forests cover 76% of the study area (see Fig. 1). Open land includes pastures, cultivated valleys, alpine meadows, and at higher elevations, rocky ridges. At a larger scale, i.e. within the 40-km diameter circular area (shown in Fig. 2) centred in the study area, the forest cover made up only 46%, and the habitat was more fragmented (see Fig. 2). High elevations exceeding 2,000 m (indicated in Fig. 2) formed natural barriers to hazel grouse movements.

Most forests are the results of natural regeneration. About half of the forested area was covered by old stands (> 100 years), the remainder being young stands originating from either clear-cutting 40–50 years ago or from natural reforestation of abandoned pastures. The vegetation of the four separate forests Fissac, Peyronnière, Sansenu and Pinée in the study area was described either in 10-m radius plots regularly spaced at a density of four per ha (for Fissac = 154 ha, Peyronnière = 46 ha and Sansenu = 83 ha) or in 18 1-ha squares in the Pinée forest (180 ha). In these plots we recorded: number of trees and number of species of all mature trees (i.e. with a diameter at breast height (DBH) of > 17.5 cm), number and species of deciduous shrubs and small trees, canopy cover of deciduous and coniferous trees at three layers, cover of field layer and an index of horizontal cover of the understorey using a 60 × 60 cm checkerboard mounted on a 1.2 m pole (Table 1).

Figure 2.

Forest cover at a landscape scale in the southeastern French Alps, with indication of the forest cover of the study area (), the forest cover outside the study area (), the natural barrieres > 2,000 m a.s.1. (---) and the circular area of 40 km diameter.

In the study area, > 80% of the stands were dominated by conifers such as spruce Picea abies, Scots pine Finns sylvestris or white fir Abies alba, and it was especially characteristic for the forests Fissac and Peyronniere. Mature deciduous trees were scarce in these stands. However, a sparse but diversified broad-leaved understorey was usually present and included the following plant species: mountain ash Sorbus aucuparia, whitebean S. aria, wild cherry Prunus avium, hazel Corylus avellana, hawthorn Crataegus monogyna, willow Salix spp., silver birch Betula pendula, wayfaring tree Viburnum lantana, snowy mespilus Amelanchier ovalis, wild cotoneaster Cotoneaster integerrimus, fly honeysuckle Lonicera xylosteum, dog rose Rosa canina and bumet rose R.pimpinellifolia. The remaining 20% of the stands were either beech Fagus silvatica mixed with a few conifers (in the Pinée forest) or mixed stands rich in both coniferous and deciduous trees (in the Sansenu forest; see location in Fig. 1).

Table 1.

Vegetation characteristics of the four forests Sansenu, Fissac, Peyronnière and Pinée in the study area in the southeastern French Alps. Diameters at breast height (DBH) of < 7.5 cm, 7.5–17.5 cm and > 17.5 cm are indicated by *, ** and ***, respectively.

The Fissac and Peyronniere forests were grazed in summer and autumn by cattle. Forest management was usually extensive. Private forests often remained unexploited for decades, and when logging did occur tree species of no economic value were left behind. The communal forests were managed by the National Forestry Service which made use of selective cutting that resulted in a fine-grained mosaic of tree patches of different ages.

The community of predators is diversified and the raptors likely to capture young or adult hazel grouse are sparrow hawk Accipiter nisus, goshawk Accipiter gentilis, buzzard Buteo buteo, golden eagle Aquila chrysaetos, kestrel Falco tinnunculus, eagle owl Bubo bubo and tawny owl Strix aluco. The main carnivores are red fox Vulpes vulpes, pine marten Martes martes and beech marten M. foina. Stoat Mustela erminea and polecat Mustela putorius are rare. Shooting of hazel grouse is prohibited in the Alpes-de-Haute-Provence.

Capture, radio-tracking and census

Between 1998 and 2001, we captured hazel grouse in spring (March-April) and in autumn (September-October) by luring them into nylon fishing nets with whistles that imitate the song of either the male or the female (Swenson 1991). Capture was biased towards males because they responded better to the whistle.

To obtain a representative sample of birds for estimating survival rates, we captured birds throughout the study area and in most forest types (see Fig. 1). The dispersion of juveniles of up to 25 km from the study area (M. Montadert, unpubl. data) further increased the variety of habitats sampled to study survival.

Birds were weighed and equipped with a radio-necklace tag weighing 7 or 11 g that included a mortality sensor. The life expectancy of the tags was 12 months, but some functioned two years. Adults were distinguished from juveniles (< one year old) by aspect of the clear band of the first primary (Stenman & Helminen 1974, Bonczar & Swenson 1992). Because all juveniles were captured in autumn after broods break up, we have no data on juvenile survival from hatching to late August.

We relocated radio-tagged birds by approaching them to ≤ 100 m, once per week in winter, and 2–3 times per week during the other seasons. Locations were determined with an accuracy of ±25 m and coordinates stored in a GIS. Causes of deaths were classified by appearance of remains as being due to raptors, mammals, disease or accidents. Birds killed by raptors were recognised by a pile of intact feathers (rachis or shaft not sectioned) that included the tag. Mammals (fox or mustelids) did not pluck the bird completely on the capture site. The tag was generally found with the necklace cut open, lying near some feathers sectioned near the base of the rachis.

We evaluated reproductive success between 1999 and 2001 on a 140-ha area at the end of July (in the Fissac forest), using a drive-count method derived from Rajala (1966) and modified by Leclercq (1984). The results were expressed as the percentage of juveniles among all birds seen (age ratio). Sexes could not be distinguished. We compared these results with those obtained by the same method in two other study areas in the Jura mountains (northeastern France; Desbrosses 1997, B. Leclercq, unpubl. data, M. Montadert, pers. obs.). Just before break up of broods (in mid-August-beginning of September) we also recorded, partly for modelling purposes, the number of juveniles reared per radio-tagged hen that had been alive in mid-June.

Data analysis

We estimated survival rates using the Kaplan-Meier procedure modified by Pollock, Winterstein, Bunck & Curtis (1989), assuming that: 1) the probabilities of capture are independent, 2) newly marked individuals have the same subsequent survival probability as the birds previously marked, 3) survival times of different individuals are independent, 4) marking does not influence survival of the birds, and 5) loss of signal is independent of the fate of the animal.

Of the birds followed more than one year, eight males and one female were reintroduced into the calculations as new individuals. Juveniles entered the group of adults on 15 June, i.e. at about one year of age. Of four birds that died because their radio collar became entangled in branches, two had survived for more than one month, and we included them in the data set as being still alive at end of tracking. We considered one case of ‘premature signal loss' (i.e. after < 6 months duration) to be due to tag failure and not to mortality. For this individual, we ruled out emigration by performing an aircraft search and considered the bird to be alive at the time of signal loss.

We calculated an annual survival rate for adults. For juveniles we calculated survival over nine months, including the period from 15 September (shortly after brood break up) to 15 June. We assumed no between-year differences in annual and seasonal survival and combined data from different years to obtain acceptable precision of survival estimates. Variance and standard error were calculated by use of the method of Cox & Oakes (1984). A normal Z test was used to compare survival rates by sex and age (Pollock et al. 1989). We retained a significance level of α = 0.05.

Table 2.

Mean survival rates (S) of radio-tagged hazel grouse in the southeastern French Alps during 1998–2001. N is the number of radiotagged hazel grouse, N′ the endpoint sample sizes used to estimate both S and standard error (SE).

We compared survival rates during two 6-month periods which differed in vegetation cover, and which therefore might have affected the vulnerability to predators. The periods were: May-October (leaves on deciduous trees and shrubs) and November-April (leaves absent and snow often present).

The relative importance of the causes of deaths was evaluated by computing cause specific mortality rates with MICROMORT software (Heisey & Fuller 1985). With this method, the probability that an individual succumbs to a specific cause of mortality during a definite lapse of time can be estimated. Annual variation in reproductive indices was analysed by a r × k Fisher test.

To estimate the population growth rate, λ, we employed a deterministic matrix population model (Caswell 1989) with two age classes of females (juveniles and adults). Adult survival was calculated for 15 June-15 June, juvenile survival for 15 September-15 June. The mean reproductive rate of hens was calculate as the number of juveniles reared by hens into late August divided by the number of hens carrying radios and being alive in mid-June. Calculations were done using ULM software (Legendre & Clobert 1995).

Survival

We captured and radio-tagged 65 hazel grouse during March 1998-October 2001, of which 49 were males (26 adults, 23 juveniles) and 16 females (10 adults, six juveniles). The annual survival rate of adult males was 0.72 (95% C.I.: 0.59–0.85) and 0.60 (95% C.I.: 0.40–0.91) for adult hens (Table 2). Although the two rates were not significantly different (one-tailed Z = 0.683, P = 0.25), we did not calculate a common adult survival rate because the seasonal pattern of mortality differed (see below). The survival rate during 15 September-15 June of immature males was 0.71 (95% C.I.: 0.53–0.89) and was similar to that of adult males for the same period (0.76; 95% C.I.: 0.34–0.91; one-tailed Z = 0.427, P = 0.33). The number of adult and juvenile females marked was too small to make a meaningful comparison of their survival during 15 September-15 June. To obtain a conservative estimate of juvenile female survival during 15 September-15 June for the demographic model, we combined the data for the two sexes, which gave a survival rate of 0.74 (95% C.I.: 0.58–0.90).

The survival of adult males during November-April (0.76) was lower than during May-October (0.95; onetailed Z = 2.52, P = 0.006). Of the 16 male deaths, only three occurred during May-October. The survival of females (adults and juveniles pooled) during November-April (0.90) was probably higher than during May-October (0.70; one-tailed Z = 1.54, P = 0.06). Of the six female deaths, only one occurred during November-April. Of the five females that died during May-October, one died during laying, two were killed on the nest during incubation and two died while they were accompanying broods. In this period, no mortality occurred among hens who had lost their nests. The survival of females during May-October (0.70) was lower than that of adult males for the same period (0.95; onetailed Z = 2.08, P = 0.02).

Causes of death

We recorded 22 deaths of natural causes (i.e. disease and predation; Table 3). The only case of disease, a coliba-cillosis, occurred in an adult male in June. We believe that we correctly distinguished between raptor and mammal kills. Although identification of the species was rarely possible, we suspect that most raptor kills were made by goshawks. Among the mammals, red fox and pine marten were the most likely predators. We attributed two deaths to pine martens. In one case, we found a hazel grouse hidden in a squirrel nest 25 m up a fir tree. In the other case, we found the radio tag hidden under a large pile of branches. A red fox probably killed one radio-tagged male whose feathers we found next to a den. Lastly, a dog probably killed a hen incubating at the edge of a road commonly travelled by hikers. Small sample sizes precluded testing for differences in the importance of the various mortality factors given in Table 3.

Table 3.

Cause specific annual mortality rates of adult male and female, and for immature (of both sexes) hazel grouse in the southeastern French Alps computed after the method given by Heisey & Fuller (1985). The numbers of deaths recorded are given in parantheses.

Population growth

We retained the following values to implement the simple demographic model: annual survival of adult females = 0.60 (for 15 June-15 June), survival of juvenile females = 0.74 (i.e. the common survival of juveniles; for 15 September-15 June), and reproductive success of adult females (i.e. juvenile females in August/ adult females alive on 15 June) = 0.75 (average reproductive rate of 1.5 young per hen in August with an assumed equal sex ratio among the young). According to this model, the rate of increase, λ, was 1.15.

Reproductive rate

Our indices of reproduction varied significantly from 1999 to 2001. For age ratios they were 9, 60 and 23% for juveniles, respectively, and 1.14, 3.00 and 0.25 young per radio-marked hen (Fisher r × k test: for age ratios, P = 0.02, for juveniles per radio-marked hen, P = 0.06). Both methods indicate that reproduction was highest in 2000, when nest success was highest (all five radio-marked hens raised young). Our mean age ratio of 34% was similar to the means recorded in two populations from the Jura Mountains in northeastern France, but our mean density of 6.0 adults/100 ha was higher than in the Risol forest during 1995–2001, where the mean reproductive success was 34% and the mean density 3.1 adults/100 ha; and in the Risoux forest during 1983–2001, where the mean reproductive success was 37% and the mean density 2.1 adults/100 ha.

Annual survival

Our estimation of the annual survival rates for adult males (0.72) and for adult females (0.60) are higher than those reported hitherto on hazel grouse. In Finland, Gaidar (1973) obtained 0.23 from recaptures of ringed birds. These Finnish data likely underestimated the survival because all birds that emigrated from the study area were considered dead. In Sweden, the annual survival estimated by use of the method of Trent & Rongstad (1974) was 0.48 for males and 0.19 for females (Swenson 1991). The values obtained in Sweden were strongly influenced by the proportion of radio-tagged birds that lived in poor forest habitats characterised by a single-layered structure (Swenson 1991). In contrast, the coniferous or mixed forests occupied in the south eastern Alps were generally well structured with a mixture of young, medium and mature coniferous trees (see Table 1) that offered good protection against predators. We conclude that good cover may largely explain the relatively high survival recorded in France.

Seasonal variation in survival of the sexes

The seasonal distribution of mortality appeared to differ according to sex. Female hazel grouse suffered high predation during the period of reproduction (May-July). The predation risk is high in tetraonid hens at this time (Angelstam 1984, Willebrand 1988, Caizergues & Ellison 1997, Hewitt, Keppie & Stauffer 2001) owing to the energy cost of laying and incubation, and the reluctance of hens to leave their nests or young at the approach of a predator. Predation on males was highest during November-April. In this period the cover is reduced by loss of leaves on deciduous trees and shrubs, which may render singing cocks vulnerable to predators, particularly when the ground vegetation is covered by snow. Display and singing activities of cocks are most frequent in both March-April and September-October. In the latter period, cover is good and the mortality of males was low. Finally, when the cover is poor the conspicuous plumage of males may render them more vulnerable than females to predators (Gotmark & Hohlfalt 1995).

Juvenile versus adult survival in males

Juvenile tetraonids often survive over winter less well than adults (Lindstrom 1994), but there are exceptions, for example in spruce grouse Dendragapus canadensis (Keppie 1979, Beaudette & Keppie 1992). In our study, the survival of males during 15 September-15 June was similar in juveniles and adults. Thus the dispersal of juvenile males between September and March, with some settling 4–25 km from the study area (M. Montadert, unpubl. data), and their intensive erratic movements in autumn, did not result in higher mortality than in the more sedentary adults. Also in black grouse Tetrao tetrix dispersal did not appear to be costly in terms of survival (Caizergues & Ellison 2002).

Cause specific mortality

Like in other galliformes (Bergerud 1988, Marcstrom, Kenward & Engren 1988, Connelly, Apa, Smith & Reese 2000, Hewitt et al. 2001), predation was the main cause of mortality. The impact of raptors is often higher than that of mammals, especially in open landscapes. Examples are black grouse (Angelstam 1984, Willebrand 1988), willow grouse Lagopus lagopus lagopus (Smith & Willebrand 1999), red grouse L. lagopus scoticus (Hudson 1992, Thirgood, Redpath, Rothery & Aebischer 2000) and rock partridge Alextoris graeaca (Bernard-Laurent 1989, Bernard-Laurent 1990). Raptors were also more important predators than mammals on hazel grouse in Sweden (Swenson 1991) and on black grouse occupying a forest habitat in the Alps (Caizergues & Ellison 1997). In the hazel grouse population we studied, predation by raptors was not markedly higher than that by mammals, despite the fact that raptors are common in the area. We conclude that the discrete behaviour of the hazel grouse associated with a dense and well-structured forest habitat constrained the ability of raptors to detect and predate hazel grouse. Thus we agree with the suggestion put forward by Swenson (1991, 1993) that hazel grouse are more vulnerable to raptors in open forest habitats where the birds are more visible than in denser habitats.

Demographic performance and range extension

The simple demographic model suggests that a closed hazel grouse population with the survival and reproductive rates we estimated could potentially increase at an annual rate of 15%. We believe that in fact some net emigration was occurring from this and other nearby populations in the southeastern French Alps. Reproduction in our study area was no better than that of hazel grouse in the Jura, where low-density populations are stable or declining (Magnani, Cruveille, Huboux, Collard, Roche & Longchamp 1991, Bernard-Laurent & Magnani 1994, Deloche & Magnani 2002). We therefore suggest that the positive growth rate estimated in our study area, and the range expansion in the southeastern Alps, are largely due to high survival of birds > two months.

The forests in our study area are dominated by conifers and are relatively young. The current logging practices, involving selective cutting, result in the fact that these forests are multi-layered, and well structured horizontally in a fine-grained mosaic. Such habitat is found throughout the southern French Alps, including many thousands of hectares not yet colonised by hazel grouse. Conservation of current source populations, and of the unoccupied forests, is required to assure that the hazel grouse continue to expand its range.