Study Area

Our study area is situated in the Pannonian ecoregion, southwestern Hungary (46°24′N, 17°27′E) in the Boronka Nature Conservation Area (BNCA; 78.3 km²). The BNCA district is characterised by extensive forests of hornbeam and oak Fraxino pannonicae-Carpinetum, which contain island-like formations of beech Leucojo verno-Fagetum, Austrian oak Quercetum petraeae-cerris and Scots pine Pinus silvestris within a mixed oak stock. The forests surround eutrophic fish ponds. Detailed information on the vegetation and the water habitats of the BNCA is given by Lanszki et al. (2001). All interventional forestry procedures are performed in an environmentally sound manner, because the area was declared a protected zone in 1991 and part of the forest is a strictly protected core area. Thus, no logging and only a minimum of fish pond management was carried out, and consequently the degree of human influence in the area was kept at a minimum. The nearest plough-lands are at least 1.5 km away and the closest villages 4-5 km away from the study area. Between 1996/97 and 2000/01 the population densities of game species (individuals/km², mean ± SE) were the following: red deer Cervus elaphus 3.0 ± 0.35, fallow deer Dama dama 0.2 ± 0.02, roe deer Capreolus capreolus 2.4 ± 0.25, wild boar Sus scrofa 2.9 ± 0.35, brown hare Lepus europaeus 0.5 ± 0.11 and pheasant Phasianus colchicus 1.9 ± 0.49 (Csányi 1999, 2000, 2001). The climate is continental and during our study the mean (± SE) winter temperature was 1.7 ± 0.6°C (range: -0.4 - +3.2°C). Duration of snow cover was 33.4 ± 11.9 days (range: 7-71 days) and snow depth was 6.5 ± 1.1 cm (range: 3.1-8.8 cm). Summer temperature was 20.7 ± 0.5°C (range: 20.0-22.2°C), and mean annual precipitation was 711 ± 104 mm (range: 563-943 mm).

Rodent Community

During April 1998-February 2001, 26 small mammal trapping sessions were carried out (one period in winter and two or three periods each season from spring to autumn). In each session, trapping was conducted at two stations for four consecutive nights using glass-doored wooden live traps. The traps (180 ×70 × 70 mm) were distributed in a grid 10 × 10 (at the first site: 100 traps) and 7 × 7 (at the second site: 49 traps). Quadrate grid points were marked for every 10 metres and a given trap was placed on the same point in each period during the study. The first station was situated in hornbeam and oak forests, characteristic of the BNCA; the second in a mixed forest (oak and pine). Walnuts, maize and ham were used as bait. The traps were checked twice daily: at 06:00 and 20:00 (with eight consecutive checks per session). For individual identification of captured animals we removed the terminal knuckle of the toes (Begon 1979), and recorded the sex, age and weight of the animal. Minimum number alive (MNA) was determined from capture-mark-recapture data (Krebs 1989). Biomass of small mammals living in the forest (in kg/ha) was calculated from summarised capture data (from individual weights and MNA) in each season. Data obtained from small mammal trapping, performed over three years, were averaged according to season for the preference calculations. All small mammal examination was permitted by the Directorship of the Danube-Drava National Park.

Scat Collection and Diet Analysis

Diet composition of the red fox and pine marten was studied by analysis of scats collected during December 1996-February 2001. Scats were collected twice a month, on a standard route (approximately 5 km long). Pine marten and fox scats were distinguished on the basis of size, shape and smell characteristics. Scats of stone martens Martes foina and pine martens can be misclassified, as they are very similar and both species can inhabit the same habitat (Herrmann 1994, Pedrini et al. 1995a,b, Genovesi et al. 1996). Stone martens, however, in general tend to prefer agricultural land and small forest patches close to villages, whereas pine martens tend to select larger forest complexes. For example, stone marten sightings were rarely made far (≥ 10 km) away from urban areas in the central Italian Alps (Pedrini et al. 1995b). In larger forested areas of central Europe, like the Białowieża Forest (which is similar to the area examined in our study), stone martens mostly occur in villages and seem to avoid large forest complexes. Radio-tracked stone martens in the Białowieża Forest mostly lived in villages, while movements between villages were very uncommon (1-2 nights per year; A. Zalewski, unpubl. data). Our study area in BNCA is distant from any human settlement, and only pine martens were recorded in the area during our study period both by direct observation and by snow tracking. On the basis of the above considerations and data, we assume that stone martens did not occur in our study area during the period of sample collection; if some stone martens did occur in the area, we assumed that it was an occasional presence with little or no effect on the results of our study.

A total of 1,010 fox and 332 pine marten scats were analysed using a standard procedure (Jędrzejewska & Jędrzejewski 1998). Scats were soaked in water, washed through a sieve (0.5 mm mesh) and dried. All food remains were separated and identified with the aid of keys from Teerink (1991), März (1972), Brown et al. (1993) and our own reference collection. Diet composition of the predators was expressed in two ways: relative frequency of occurrence (%Occ) and percentage of biomass consumed (%Bio). To calculate the relative frequency of occurrence the number of occurrences recorded for the given food source was multiplied by 100 and then divided by the total number of food types identified. All dry food remains were weighed and multiplied by coefficients of digestibility (insectivores and small rodents 23, medium-sized mammals 50, wild boar 118, deer 15, birds 35, amphibians and reptiles 18, fish 25, insects, crayfish and molluscs 5, fruit, seed and other plant material 14) to obtain an estimate of the percentage of fresh weight (biomass) of food consumed (Jędrzejewska & Jędrzejewski 1998). For wild boar and cervids we used various coefficients of digestibility as was suggested by Jędrzejewski & Jędrzejewska (1992). Wild boar meat eaten by fox and marten were from whole carcasses (mortality caused by disease), and medium-sized predators mostly consume meat and less frequently bones or skin with hairs. In contrast to wild boar, remains of cervids were mainly left by hunters (limbs and internal organs), and thus corresponded to what would have been left over from wolf kills where the prey is utilised almost completely. Therefore, scavengers often eat bones and skin, and coefficients of digestibility are likely to be lower (see Jędrzejewski & Jędrzejewska 1992). The prey species were classified according to weight (Clevenger 1993) and on their characteristic zone of occurrence recorded for predatory species (Gittleman 1985). We selected three prey zonation categories: 1) terrestrial and mainly terrestrial but sometimes arboreal, 2) arboreal and mainly arboreal but sometimes terrestrial, and 3) aquatic or water-linked (Gittleman 1985; for more detail see Appendix I). We used log-linear likelihood tests on frequency of occurrence data to test for dietary differences among seasons and years. Owing to the large number of comparisons (eight dietary categories), we adjusted the level of significance to 0.0064 with a Bonferroni correction. χ²-test was applied for distribution analysis of prey consumption on the basis of weight and characteristic zonation.

Trophic niche breadth was calculated in accordance with Levins (Krebs 1989): B  =  1/Σp_i², where p_i  =  the relative frequency of the i^th food item; and standardised across food items: B_A  =  (B-1)/(n-1), rating from 0 to 1. The following food categories were used in the calculations related to trophic niche and the comparative analysis of diet composition for predator species: 1) small mammals, 2) medium-sized mammals, 3) carcasses, 4) birds, 5) other vertebrates (reptiles, amphibians and fish), 6) invertebrates, and 7) fruits, seeds and other plant matter. Trophic niche overlap was calculated by means of the Renkonen index: P_jk  =  [Sn(minimum p_ij, p_ik)]100, where P_jk  =  percentage overlap between species j and species k; p_ij and p_ik  =  proportion of resource i represented within the total resources used by species j and species k; n  =  total number of food items (Krebs 1989). The standardised food niche breadths were compared with general linear models (GLM procedure in SPSS). One-way analysis of variance was used for seasonal niche-overlap calculation. Ivlev's index (E_i) of preference according to small mammal dominance was applied as follows: E_i  =  (r_i - n_i)/(r_i + n_i), where r_i  =  percentage biomass of the given (i^th) food category in the diet and n_i  =  percentage of biomass of the given (i^th) taxon in the environment (Krebs 1989). Electivity varies from -1.0 to +1.0. T-test was applied to compare the Ivlev's indices in two, bank vole or Apodemus mice dominated, periods. We used the SPSS 10 for Windows (1999) statistical package to process data.

Density and Biomass of Rodents

Apodemus mice species in BNCA were the yellow-necked mouse Apodemus flavicollis (51.4%), common field mouse A. sylvaticus (46.3%) and striped field mouse A. agrarius (2.3%). The various Apodemus species were grouped together because it was not possible in every case to identify the species on the basis of the hair or teeth found in predator scats. The trapping results showed that bank vole and Apodemus mice were the dominant rodent species in the forests of BNCA (Fig. 1), and they comprised on average 99.2 ± 0.4% (mean ± SE) of the small mammal community. Their number and dominance in the community varied both annually and seasonally. Based on the dominance of the small mammal species two periods were distinguished: 1) dominance of bank voles and 2) dominance of Apodemus mice (see Fig. 1). The density of small mammals increased from 19.9 ± 6.9 individuals/ha in spring to 69.3 ± 12.5 individuals/ha in autumn and decreased during winter. Small mammal numbers were highest in 1999, and, due to the mild winter weather of 1999/2000, a large proportion of them survived the winter (see Fig. 1). Despite this, no conspicuously high small-mammal density was observed in the autumn of 2000 (see Fig. 1).

Figure 1

Availability of small mammals in the Boronka Nature Conservation Area, Hungary, expressed as abundance (A), dominance (B) and biomass (C). Data are based on minimum number alive (MNA) obtained using the mark-recapture technique. •  =  bank vole; ○  =  Apodemus spp.; +  =  shrews; -  =  dormice.

Predator Diets

Two main food items were most important in predator diets: small rodents and plant material; together they comprised 69-93% of the food biomass of foxes and 55-79% of the food biomass of pine martens. The predators supplemented their diets with birds, ungulate carcasses and invertebrates. Generally, pine martens preyed more often on reptiles, amphibians and fish (χ²  =  32.62, df  =  1, P < 0.0001) and consumed more plant material (χ²  =  19.27, df  =  1, P < 0.0001), whereas red foxes preyed more often on small mammals (χ²  =  36.80, df  =  1, P < 0.0001) and fed more often on carcasses (χ²  =  14.21, df  =  1, P < 0.0002).

Prey weighing < 50 g comprised up to 83% of fox diet and 87% of marten diet (Fig. 2) and both predators hunted prey weighing 51-300 g. The highest prey weight category (> 300 g) proved to play an important role in fox diets (13%). Prey weight did not differ significantly between fox and marten diets (χ²  =  9.27, df  =  4, P  =  0.055). The majority (88%) of prey species consumed by the terrestrial fox lived at ground level (see Fig. 2). Foxes seldom ate arboreal species, or species associated with aquatic habitats (7 and 5%, respectively). The pine marten, despite its outstanding capacity for climbing, also preyed primarily on terrestrial species (frequency: 81%). In total, 14% of pine marten prey species were typically arboreal and only 5% were associated with aquatic habitats. The diets of the two predators differed significantly (χ²  =  33.27, df  =  2, P < 0.001) on the basis of zonation of prey.

Figure 2

Distribution of the frequency of prey in the diet of red foxes (▪) and pine martens (□) on the basis of weight (A) and zonation (B), in the Boronka Nature Conservation Area, Hungary.

Seasonal and Inter-year Variation in the Predator Diets

Predator diet compositions showed significant differences between seasons (Tables 1 and 2). Percent occurrence of rodents in both predators' diets was higher in winter and autumn and lower in summer (see Table 1). Both predators hunted more bank voles than Apodemus mice. The proportion of both bank voles and Apodemus mice in fox diets did not vary between seasons (see Table 2). Pine martens preyed upon bank voles significantly more in winter than in summer. In both predators' diets, carcasses, birds, invertebrates and plant material varied significantly between seasons. Invertebrates and plant materials (mainly fruits and maize Zea mays) were more often consumed in summer and autumn, carrion in winter and spring, and birds in spring and summer (see Tables 1 and 2). Pine martens preyed upon reptiles, amphibians and fish significantly more in winter and spring than in summer and autumn.

Table 1

The diet of red fox and pine marten in the Boronka Nature Conservation Area, Hungary. Data were collected during December 1996-February 2001. %Occ  =  percent relative frequency of occurrence; %Bio  =  percent of biomass consumed; -  =  occurring in proportions of < 0.05%.

Table 2

Summary of log-linear analysis of the variation in red fox and pine marten diets during four seasons and four years (1997-2000) from Boronka Nature Conservation Area, Hungary. Numbers in italics indicate significant values.

Diets of both predators changed between years (see Table 2). Occurrence of bank voles in fox diet decreased with decreasing density of this rodent (see Table 2 and Fig. 1). However, the season*year interaction was significant, indicating that fox diet did vary between seasons in various years (see Table 2): the occurrence of bank voles was higher in summer and autumn in the first two years (high vole density), and in the next two years (low vole density) foxes hunted more voles in winter and spring. The frequency of occurrence of Apodemus mice showed an opposite trend. These rodents were often eaten by foxes in 1999 and less often in 1997. Therefore, the occurrence of small mammals in total was relatively stable between years in the fox diet (see Table 2). However, red foxes ate more small rodents in the autumns and winters of 1997 and 1998 than in the winters and springs of 1999 and 2000. Carcasses, birds and other vertebrates (frogs, reptiles and fish) occurred equally often in scats from different years. Invertebrates were eaten by foxes more often in 1997 and 1998 than in the following two years. The occurrence of invertebrates in 1998 and 1999 decreased in summer and increased in winter, whereas in 1997 and 2000 the opposite occurred. Plant materials were eaten more by foxes in 1999 and 2000 than in the first two years. A significant three-way interaction (see Table 2) showed that the degree of differences varied among years and seasons. In years of lower consumption of plants (i.e. 1997 and 1998), this type of food was consumed more often in autumn, whereas in years of higher consumption (i.e. 1999 and 2000) plants were found more often in summer.

Diets of pine martens varied much less between years than did fox diets. The frequency of occurrence of bank voles and Apodemus mice in pine marten diets varied only slightly between years (see Table 2). However, the total of small mammals consumed was higher in the first two years than in the last two years. Other types of food did not vary significantly between years.

Throughout the years, rodents were the most important prey resource for both red foxes and pine martens. However, in periods of lower rodent consumption both predators supplemented their diet with other food items. In summer and autumn, with an increase in the percentage biomass of rodents in fox and marten diets, there was a decrease in consumption of plant material and birds (see Table 1). In winter and spring, birds and carcasses were the most important alternative food, and their consumption decreased with the predators' consumption of rodents.

Small Mammal Abundance and Preference by Predators

In summer and autumn, percentage biomass of bank voles in pine marten diet significantly correlated with the number of voles in the local populations (Fig. 3), but the correlation was non-significant for foxes. In winter and spring bank vole density varied to a minor degree (0.9-1.5 MNA/100 trap nights). Furthermore, percent biomass of Apodemus mice in fox diets was positively correlated with numbers of mice in the population in summer and autumn, but not in winter and spring (see Fig. 3). In pine marten diets this relationship was not significant in either period.

Figure 3

Relationship between bank voles (A,B) and Apodemus mice (C,D) densities (MNA/100 trap nights) and percent biomass of these rodents in red fox (A,C) and pine marten (B,D) diets during 1997-2000.

The red fox and pine marten ate bank voles primarily in winter, while in summer both predators generally consumed less of the various species of mice than expected by availability (Fig. 4). In 1998-1999, when bank vole numbers were greater or equal to the number of mice in the rodent community (see Fig. 1), both predators preyed on bank voles in proportion to their total biomass in the rodent community; the electivity index (E_i) was 0.12 ± 0.04 (range: 0.04-0.26) for fox and E_i  =  -0.12 ± 0.12 (range: -0.60 - +0.11) for pine marten (see Fig. 4). In this period, red fox ate less Apodemus mice (E_i  =  -0.30 ± 0.05; range: -0.39 - -0.15), but pine marten consumed this rodent in proportion to its abundance in the rodent community (E_i  =  -0.06 ± 0.06; range: -0.31 - -0.03). In the second period (from autumn 1999; see Fig. 1), when Apodemus mice outnumbered the other species, foxes and martens preferred bank voles (E_i  =  0.35 ± 0.08 for fox and 0.25 ± 0.18 for marten) and ate less Apodemus mice (E_i  =  -0.28 ± 0.05 for fox and E_i  =  -0.27 ± 0.10 for marten). The difference in bank vole preference between the first and second periods was significant for both the red fox (t-test: P  =  0.027) and pine marten (t-test: P  =  0.015) and for Apodemus mice in respect of pine marten (t-test: P  =  0.035). Red foxes and pine martens similarly consumed fewer shrews (E_i on average -0.12 for both predators), and various species of dormice (E_i  =  -0.06 for fox and -0.10 for marten) than expected by availability.

Figure 4

Preference by red foxes (A) and pine martens (B) for bank voles (▪) and Apodemus mice (□) in spring (Sp), summer (Su), autumn (Au) and winter (Wi) during 1998-2000 in the Boronka Nature Conservation Area, Hungary.

Food-niche Breadth and Food Niche Overlap

The number of food items in the diet of foxes living in the forest was 66 prey and 15 plant taxa, and for pine martens it was 62 and 11, respectively. The general linear model for the standardised trophic niche breadth, calculated with predator species, seasons and years as independent variables, was not significant (F_7,30  =  1.52, P  =  0.21). Pine martens had a standardised trophic niche breadth similar to that of red foxes (F₁  =  1.12, P  =  0.30) and both predators' food niche breadth did not vary significantly between seasons (F₃  =  1.55, P  =  0.22; Table 3) and years (F₃  =  1.65, P  =  0.21). Trophic niche overlap between the red fox and pine marten, proved to be of a high degree (71.6 ± 3.3%), but differed non-significantly between seasons (one-way ANOVA: F₃  =  1.14, P  =  0.38; see Table 3).

Table 3

Seasonal standardised trophic niche breadths and overlaps (± SE) of the red fox and pine marten in the Boronka Nature Conservation Area. Calculated from the percent biomass of food types in the scats.