Throughout its range, the red fox Vulpes vulpes is widely perceived as a major predator of game and livestock (e.g. Sargeant et al. 1984, Saunders et al. 1995, White et al. 2003). Although predation on pest species may be beneficial, large numbers of foxes are culled annually in an attempt to increase the abundance of economically important species (Tapper 1992, Reynolds & Tapper 1996). Culling is also increasingly viewed as an effective management tool for the conservation of rare or endangered species (Côté & Sutherland 1997). However, the evidence for the effectiveness of widespread culling is equivocal: whilst some studies have shown that reduced predator pressure from foxes, or foxes in combination with other predators, can increase prey abundance and/or productivity (e.g. Marcström et al. 1988e.g. Marcström et al. 1989, Pech et al. 1992, Tapper et al. 1996, Kinnear et al. 1998, Banks 2000), this is not always the case (e.g. Pech et al. 1992, Mayot et al. 1998, Kauhala et al. 1999). Furthermore, such studies have focussed on a relatively restricted number of species, mainly lagomorphs and tetraonids, with several prey populations being confined to islands with limited opportunity for immigration or emigration, or where the fox is an introduced species. Thus there are relatively few data on the impact of fox predation on a range of prey species in large-scale agricultural habitats.

For a given prey species, the impact of fox predation will, at the most basic level, be determined by the numbers of individuals killed by foxes and the relative importance of this rate of mortality in the overall dynamics of the prey. Consequently, a simple estimate of the effects of predation can be obtained from the simultaneous measurement of fox numbers, fox diet and prey dynamics. More complex assessments would include the impact of predation on different age and sex classes and the timing of predation relative to reproduction: however, such data are lacking for most prey species. Therefore, although such studies do not definitively indicate the effect of predation, they can be used to help identify species that may be adversely affected by a predator (Risbey et al. 1999) and serve as a useful precursor to more intensive and expensive investigations.

In Britain, the impact of predation by foxes is of fundamental importance in light of recent restrictions on the use of dogs to control fox numbers (Burns et al. 2000;  www.hmso.gov.uk). The diet of foxes in Britain has been investigated extensively (see review in Baker & Harris 2003). However, the data currently available are of limited use in identifying the possible impact of predation by foxes in rural Britain. First, previous studies span > 60 years, with most having been conducted between 1970 and 1980. In the last 50 years there have been significant changes in the abundance of several prey species in the UK, particularly rabbits Oryctolagus cuniculus, pheasants Phasianus colchicus and field voles Microtus agrestis (Tapper 1992, Harris et al. 1995, Harris et al. 2000). Second, the data collected in these studies have been analysed and presented in a number of different ways, thereby limiting their comparability. Lastly, only a single study has concurrently measured prey abundance and, therefore, been able to estimate the possible impact of fox predation on species of concern (Reynolds & Tapper 1995a).

Fox diet is known to vary substantially between sites (Reynolds & Tapper 1995a, Baker & Harris 2003), presumably in relation to prey availability. In agricultural landscapes, prey availability is likely to be influenced by patterns of agricultural production, but also by other land use interests. Consequently, to obtain a complete picture of the possible impact of foxes in Britain, it is necessary to obtain data from areas covering a wide range of agricultural and land use practices. For example, the study of Reynolds & Tapper (1995a) was conducted on an area dominated by cereal crop production and with game-rearing interests, including the control of predators (although fox control was suspended for part of that study). However, crops comprise only 25% of the total area of agricultural holdings in the United Kingdom (MAFF 2000), and the presence and degree of fox control varies among holdings; in two nation-wide surveys, 38–42% of farmers reported that they undertook no fox control (Vaughan et al. 2003, White et al. 2003).

The purpose of our study was to assess the possible impact of fox predation on a mixed organic farm in Wiltshire, UK, where no predator control was undertaken, by comparing the estimated mass of rabbits, pheasants, field voles, wood mice Apodemus sylvaticus and bank voles Clethrionomys glareolus consumed annually by the resident fox population with the measured spring biomass of these prey species. The total mass of each prey species consumed annually was estimated from field data on fox diet composition and fox density in conjunction with the approximate annual food requirements for juvenile and adult foxes derived from a published study of captive foxes (Sargeant 1978) corrected for the larger body size of foxes in Britain. To enhance comparability, we have adopted the approaches outlined by Reynolds & Tapper (1995a).

Material and methods

Our study was conducted during 1995–1996 on an organic farm at Castle Combe, Wiltshire, UK (51°20'N, 2°13'W; altitude 92 m a.s.l.). Temperature range was approximately −7.3°C to 30.0°C with an annual rainfall of 506 mm (data for 1996 from Lyneham weather station 51°30'N, 1°59'W; UK Meteorological Office). The principal farming practices were the production of cereal and fodder crops and the maintenance of beef cattle; several fields were leased annually for sheep grazing. The farm covered an area of 352 ha; this consisted of 213 ha open fields (of which approximately 35% was cereal crops and 65% pasture/fodder crops annually), 110 ha woodland, 23 ha set-aside (i.e. rough grassland) and 6 ha marshland managed as a nature reserve. Neighbouring land, to which we did not have access, comprised farmland, a golf course, a motor racing circuit and a village. Predators were not culled on the study farm, but neighbouring landowners did control foxes. As fox territories overlapped these neighbouring farms, the resident fox population was subject to an unknown level of culling.

Pheasants were reared and released annually on two neighbouring farms but not on the study farm. At the commencement of the study there was only a limited amount of rabbit control, conducted principally by the farm manager. During winter 1995/96 and winter 1996/ 97 more intensive control was undertaken because of grazing damage to cereal crops. In all instances, rabbits were taken by ferreting during autumn and winter. Rabbits on neighbouring farms were culled all year round, primarily by shooting and ferreting.

Other vertebrate prey present included: domestic chickens, available from a nearby intensive poultry unit (where dead birds were often discarded) and from households in the vicinity where small numbers were kept for private use; and ducks and geese (Anseriformes) and rails (Ralliformes), available principally from the private nature reserve on the study farm. Passeriformes, Columbiformes, grey squirrels Sciurus carolinensis, common rats Rattus norvegicus, roe deer Capreolus capreolus and muntjac deer Muntiacus reevesi were present across the study site. Brown hares Lepus europaeus were present within 2 km of the study farm but were only observed on the study farm and immediate neighbouring farms on one occasion. Consequently, we have considered them absent as prey. Four other mammalian carnivores (badger Meles meles, stoat Mustela erminea, weasel Mustela nivalis and American mink Mustela vison) and four predatory bird species (kestrel Falco tinnunculus, sparrowhawk Accipiter nisus, buzzard Buteo buteo, and tawny owl Strix aluco) were also present.

Fox scats were collected during 1995 and 1996 from across the study farm during other routine activities, e.g. trap checking and prey abundance surveys. Scats were stored at −5°C prior to examination. Seasons were defined as spring = March-May, summer = June-August, autumn = September-November, and winter = December-February. Scats of cubs collected at breeding dens (April-June) were analysed separately (Lindström 1994).

Scats were analysed according to the procedures outlined by Reynolds & Aebischer (1991). Faeces were oven dried at 80°C to constant mass, weighed to the nearest 0.01 g, and then soaked for 24 hours in water before being physically separated and strained through a 0.5-mm sieve. The strained liquid was then left to settle for a further 24 hours before decanting off most of the water, and air-dried. The remaining micro-fragment was then examined for earthworm chaetae. The total number of chaetae in the micro-fragment was converted to mass ingested as described by Reynolds & Aebischer (1991).

Each component of the macro-fragment was separated, dried and weighed. Fur types were identified to species using the keys of Day (1966) and Teerink (1991). Feather types were identified to order using Day (1966). Insect components were identified to order. All Apodemus fur was assumed to represent predation on wood mice; yellow-necked mice Apodemus flavicollis were present, but because of their more restricted distribution on the farm, we have assumed that wood mice were much more likely to be consumed. All lagomorph fur was assumed to represent predation on rabbits. The excreted mass of each macro-group was converted to ingested mass using conversion factors derived from captive feeding studies (Reynolds & Tapper 1995a).

For comparability with other studies, diet composition is presented as both the frequency of occurrence of each prey type in scats and percentage mass ingested. For the calculation of percentage mass ingested, only items considered to be of nutritive value were utilised. Mammalian prey was classified into three categories based on size: small mammals (< 0.1 kg), medium-sized mammals (0.1–5.0 kg) and large mammals (> 5.0 kg). Unidentified remains of mammals and birds were assumed to be in direct proportion to the distribution of identified prey and were re-allocated accordingly. Limits for the contribution of each prey type to the diet were estimated using the bootstrapping procedure outlined by Reynolds & Aebischer (1991).

Results

Fox diet

Medium-sized mammals were the most frequently recorded prey in the faeces of adult foxes, followed by non-edible vegetation (e.g. grass, small twigs and pine needles), earthworms, insects, birds and small mammals, respectively (Table 1). Other prey types were typically found in < 10% of scats in each season. There were only minor seasonal variations in the frequency of occurrence of the different prey groups. The most frequently occurring items in the diet of dependent cubs were non-edible vegetation and medium-sized mammals, with a greater occurrence of insects and birds and less reliance on earthworms (see Table 1).

Table 1.

Frequency of occurrence of prey types in fox faecal samples. Figures give the percentage of faeces containing each prey group. Because of multiple occurrences of prey groupings in the same scats, subtotals and totals are not the sum of the individual components. Figures in parentheses give the number of scats analysed. T denotes that the prey occurred in < 0.5% of scats.

The single most important prey in terms of mass ingested was rabbit, which accounted for 74% of the diet of adult foxes (range: 72–79% across seasons; Table 2). Three other prey groups each comprised ≥ 5% of the prey intake of adult foxes; small mammals (mean 7%, range: 5–8% seasonally), large mammals (mean 6%, range: 2–8%), and wild galliformes (mean 5%, range: 4–6%). Dependent cubs were less reliant on rabbits (58%) and consumed larger amounts of wild galliformes (17%), passeriformes (14%) and large mammals (10%).

Table 2.

Percentage mass of prey ingested. Figures in parentheses give 95 percentile ranges obtained from bootstrapping (Reynolds & Aebischer 1991). Bootstrapping limits for mean adult diet was calculated from random samples of 150 scats from each season to avoid overdependence on any single season. T denotes that the prey accounted for < 0.5% of prey ingested.

Rabbit abundance

Approximately 940 adult rabbits were killed during the rabbit control programme; of these, 827 rabbits from 24 warren units provided usable data for quantifying the relationship with warren size. For three warrens ferreted in both years, we have used an average value to calculate the relationship between adult numbers and warren size. The number of active holes recorded was significantly correlated with the number of adult individuals ferreted and killed: y = 0.33x + 13.2, where y is the number of adult rabbits killed and x is the number of active entrance holes (Pearson correlation coefficient: r = 0.81, N = 24, P < 0.001). We identified 79 warren units across the study site, with a total of 2,895 active entrance holes. This would equate to a ferretable population of approximately 969 adult rabbits and a standing crop of 2,548 adult rabbits. Assuming an average weight of 1.5 kg (Reynolds & Tapper 1995a), the pre-breeding biomass on the study site was 3,822 kg (~ 1,086 kg/km²).

Combining the seasonal pattern of consumption for adults, juveniles at the breeding den and juveniles away from the breeding den, and the seasonal diet composition of adults and cubs (i.e. juveniles at the breeding den; see Table 2), each fox group would consume approximately 443 kg of rabbit over the course of one year. At minimum and maximum fox densities of 0.40–0.85 groups/km², this is equivalent to the consumption of 624–1,329 kg of rabbit on the study site as a whole (~177–378 kg/km²; Fig. 1A).

Figure 1.

The pattern of consumption of A) rabbits, B) wood mice, C) field voles, D) bank voles and E) pheasants at Castle Combe, Wiltshire, UK, during 1995–1996. Columns denote the mass of each prey group in the corresponding season. Rabbit biomass was estimated from ferreting and a warren survey; wood mouse and field vole biomass was estimated from trapping grids in woodland and grassland habitats (see Table 3); bank vole biomass was estimated as 0.57 times wood mouse density (Harris et al. 1995); pheasant biomass from census data. The solid and dashed lines indicate, respectively, the maximum and minimum cumulative mass of wood mice consumed as estimated from the analysis of faecal samples (see Table 2), the annual requirements of a group of foxes assuming that all juveniles disperse and maximum and minimum fox density.

Wood mouse, bank vole and field vole abundance

Multiplying the spring densities of each species (Table 3) by the amount of available habitat (110 ha of woodland and 23 ha of grassland), the small mammal population in spring comprised approximately 5,832 wood mice and 483 field voles. In addition to these two habitat types, there was an additional 75 ha of cereal crops that would also have supported wood mice; field voles are rarely found in this habitat. Assuming a density of 1/ha for wood mice in cereal crops (Harris et al. 1995), this would indicate a total population of 5,907 wood mice and 483 field voles. Assuming a standard weight of 20 g for an adult of each species (Yalden & Morris 1990), the standing crop of wood mice and field voles was 118 kg (~34 kg/km²) and 10 kg (~3 kg/km²), respectively. The corresponding standing crop for bank voles, based on the abundance of wood mice, was 3,367 individuals, with a total biomass of 67 kg (~19 kg/km²).

Table 3.

Seasonal small mammal density in woodland (N = 4 trapping grids) and grassland (N = 6 trapping grids) habitats. Figures give number per hectare (mean ± SD) and are based on the minimum number trapped in each trapping session.

Each fox group would consume approximately 5 kg of wood mice, 5 kg of bank voles and 19 kg of field voles annually. At minimum and maximum fox densities, these figures are equivalent to the consumption of 7–15 kg of wood mice (~2–4 kg/km²), 27–57 kg of field voles (~8–16 kg/km²) and 7–15 kg of bank voles (~2–4 kg/km²) on the study site as a whole (Fig. 1B–1D).

Discussion

In our study, adult foxes were most reliant on rabbits throughout the year; on average, rabbits comprised 74% of the diet (see Table 2). Other prey groups consumed, in decreasing order of importance, were small mammals, large mammals and wild Galliformes. The occurrence of large mammals probably reflected scavenging on road casualties (e.g. badgers and deer) or illegally poached individuals (deer), although foxes will predate deer fawns. The diet of juveniles at breeding dens was markedly different, with a greater reliance on birds, in particular wild Galliformes (see also Lindström 1994). This probably reflects the optimum allocation of food resources by adults during central place foraging (Lovari & Parigi 1995).

To assess the potential impact of fox predation on a given prey species it is necessary to consider the mass of prey consumed in the context of the likely recruitment of a comparable biomass over the course of one year. The maximum estimated mass of rabbits consumed on the study site was 1,329 kg per annum, which is equivalent to the consumption of 886 adult individuals (1.5 kg each). In comparison, assuming an equal adult sex ratio, the estimated biomass of the pre-breeding population (3,822 kg) is equivalent to a reproductive population of approximately 1,274 females. Maximum adult mortality for a high-density population in the UK is likely to be in the order of 65% (Cowan 1991b), indicating a loss of 1,656 adult individuals (2,484 kg) in the course of one year. Assuming that fox predation was wholly additive to this mortality, the total biomass lost (1,329 + 2,484 = 3,813 kg = 2,542 adult individuals) would be fully compensated by the recruitment of just 2.0 adult offspring for each female within the pre-breeding population. In reality, predation by foxes was likely to be the major form of mortality in this population, such that the assumed mortality rate is likely to wholly incorporate any estimate of predation; this would have been fully compensated by the recruitment of 1.3 adult offspring per female (i.e. 1,656/1,274). These recruitment rates are entirely plausible given that productivity (at birth) in agricultural landscapes in England averages 14–22 young per female per year (Trout & Smith 1995), although there are few specific data on the rate at which young are recruited into breeding populations at different densities. Therefore, despite the heavy reliance on rabbits, it is unlikely that fox predation would have regulated the size of the rabbit population (see Pech et al. 1992, Banks 2000).

This is also likely to be the case considering predation in addition to the mortality exerted by the rabbit control programme. During the control programme undertaken in two winter seasons approximately 940 rabbits were killed. As this mortality was exerted at around the time of the onset of breeding in the rabbit population, it would have had the effect of reducing the size of the pre-breeding rabbit population. With approximately equal numbers taken each year and an equal sex ratio, the pre-breeding population would have been reduced to approximately 1,039 females. To recover to pre-control levels (2,548 adults) with an additive 65% adult mortality rate would necessitate a recruitment rate of 2.6 adult offspring per female in the reduced population, which is in the likely range for rabbits in agricultural landscapes (Trout & Smith 1995). At this density, therefore, the level of control exerted is unlikely to have exerted any economic benefit in reducing crop damage.

Foxes are generally unlikely to exert a limiting effect on generalist small mammal species in lowland agricultural landscapes, as these typically constitute only a minor component of the diet (Baker & Harris 2003), and foxes occur at substantially lower densities. In this study, maximum consumption rates by foxes accounted for 13% of the spring biomass of wood mice (see Fig. 1B) and 22% of the spring biomass of bank voles (see Fig. 1D). In conjunction with a high intrinsic rate of increase (wood mice, r = 12.43; bank vole, r = 6.22; Bright 1993), it is improbable that foxes exerted any influence on the wood mouse or bank vole populations.

In contrast, minor prey species with specialised requirements may be more vulnerable to the effects of fox predation. For example, field voles have been shown to be susceptible to the effects of habitat fragmentation (Crone et al. 2001) and, in this study, they were confined to small, isolated areas of rough grassland. Despite this limited distribution, they comprised the major portion of the small mammal component in the diet of foxes. As a result, the mass consumed by foxes exceeded the spring biomass of voles by a factor of 2.7–5.7 times (see Fig. 1C) and this species is also likely to have been consumed by a range of other predators (Dyczkowski & Yalden 1998). However, field voles have a high intrinsic rate of increase (r = 8.21; Bright 1993), such that even substantial losses to predators may be offset by reproduction. In addition, the trapping regime utilised may have underestimated field vole density, e.g., Gurnell & Flowerdew (1994) recommend using a small inter-trap interval and pre-baiting traps to maximise capture rates. As a result, the impact of predation on field voles is equivocal.

Fox predation did appear to exert an effect on the pheasant population. The mass of pheasant consumed annually (56–120 kg) was equivalent to 38–81% of the spring standing crop (see Fig. 1E). The majority of the mass consumed was taken during spring and summer (41–87 kg), when pheasants were breeding, and prior to the likely immigration of released birds from neighbouring farms. Consequently this provides an indication of the potential impact on the breeding population. Most predation during these seasons was likely to have involved females, as they are particularly vulnerable during nesting and incubation (Hill & Robertson 1988). Assuming (probably conservatively) that 60% of the mass of prey consumed during spring and summer related to predation on females, the consumed mass of 41–87 kg would be equivalent to the loss of 22–47 females from a minimum estimated pre-breeding population of 67 (i.e. 57/0.85), indicating that a minimum of 20–45 females survived the breeding season. These data would indicate a survival rate of approximately 0.30–0.67 for the period 1 March - 31 August; these are at the low end of the scale for rates calculated in other studies (Robertson 1991, Riley & Schulz 2001).

We do not have figures for the productivity of the females in this population, as autumn counts would exclude females that had successfully reared broods to independence. However, data from the British Trust for Ornithology's nest recording scheme suggest that approximately 29% of females surviving the breeding season would have successfully produced a brood (Robertson 1991); we have elected to use this figure rather than the 66% for the Pheasant Nest Recording Scheme run by the Game Conservancy Trust (Robertson 1991), as this latter scheme is focussed only on shooting estates where predator control is typically undertaken. Assuming an average brood size of 3.2 young (Robertson 1991), these figures indicate the production of 19–42 fledged young, approximately half of which would be females. This is below the level of productivity required to sustain the population and excludes losses of breeding females to other causes during the breeding season (e.g. agricultural machinery, road traffic accidents) and subsequent autumn and winter mortality of both juvenile and adult pheasants. Consequently, predation by foxes would have been sufficient to eradicate the pheasant population within a few years. Additional factors that may also have reduced productivity in our study include reduced availability of optimal habitats for mating and reproduction (Robertson et al. 1993) and low chick survival as a consequence of poor invertebrate availability (Hill 1985), although pesticides had not been used for a number of years.

Yet despite this level of predation, there was little difference in pheasant population size between the two years surveyed, indicating that the population appeared to be maintained by the immigration of captive-reared birds from neighbouring farms. Therefore, the results of this study and others (Robertson 1988, Reynolds & Tapper 1995a, Mayot et al. 1998, Riley & Schulz 2001) suggest that in Britain the current abundance of this introduced species is heavily dependent on the release of millions of birds annually (Tapper 1992), and that maintaining populations at current levels is reliant on practices such as habitat management and predator control (Mayot et al. 1998). Such practices are likely to have beneficial (e.g. Clarke & Robertson 1993) and detrimental (e.g. Tompkins et al. 2002) effects on other species in agricultural communities. Therefore, this system represents an unusual example of the repeated facilitated introduction of an alien species. However, it has been proposed that released birds may have a detrimental effect on the productivity of wild-reared (individuals reared in the wild by females that were themselves hand-reared; Game Conservancy Trust 1997) or wild birds (birds reared in the wild by females that were themselves reared in the wild; Game Conservancy Trust 1997), such that free-living populations that do not consist of large numbers of hand-reared birds may be more robust to the impacts of fox predation.