Bodil Elmhagen

Interference Competitionbetween

Arctic and Red Foxes

1

Summary of the DPhil thesis

Interference Competitionbetween

Arctic and Red Foxes

BODIL ELMHAGEN

Department of ZoologyStockholm University

2003

2

Doctoral dissertation 2003

Bodil ElmhagenDepartment of ZoologyStockholm UniversitySE – 106 91 StockholmSweden

Interference Competition between Arctic and Red Foxes

Abstract

In this thesis, I investigate the relationship between arctic foxes Alopex lagopus and red foxesVulpes vulpes in Swedish mountain tundra habitat (fjällen). The arctic fox population was severelyreduced by hunting in the early 20th century. It has not recovered despite protection since 1928 andit is endangered, while the red fox population increased in 1930-1960.

I found a high food niche overlap between arctic and red foxes and they responded similarlyto changes in the prey base, indicating similar prey preferences. Hence, arctic and red foxes shouldcompete for the same territories; more precisely the ones in low altitude areas close to the tree-linewhere prey abundance is relatively high. In the 19th century, arctic foxes bred in all tundra habitats.An analysis of present den use showed that arctic foxes have retreated to higher altitudes as theyrarely used the lower parts of their former range. Instead, red foxes did. Arctic foxes were highlydependent on the availability of Norwegian lemmings Lemmus lemmus for reproduction, while redfoxes at lower altitudes had better access to alternative prey.

Interference competition imply that there are behavioural interactions between competingspecies, e.g. fighting or predation, but interactions can also be more subtle and imply that inferiorspecies avoid encounters with stronger competitors by changing their habitat use. Red foxes arelarger than arctic foxes. Hence, they have an advantage in direct fights and arctic foxes may eitherbe driven away from their dens when red foxes establish in the vicinity, or they avoid habitatswhere they risk encounters with red foxes. I found that arctic foxes almost exclusively used denssituated farther than 8 km from inhabited red fox dens. In two out of three cases when they bredcloser to red foxes, there was red fox predation on arctic fox cubs. Further, simulations of arcticfox avoidance of areas surrounding inhabited red fox dens in a spatially explicit population model,indicated that relatively small numbers of red foxes might have a large impact on arctic foxpopulation size and distribution.

Thus, the results of this thesis indicate that interference competition with red foxes hashampered arctic fox recovery after the initial population decline, by causing a substantial reductionin arctic fox habitat. Further, red foxes have taken over the most productive areas and remainingarctic fox habitats is of such low quality that it is uncertain whether it can maintain even a smallarctic fox population.

ISBN 91-7265-632-82003 © Bodil ElmhagenCover paintings and layout by Bodil Elmhagen

3

Contents

Abstract 2Introduction 4

When do species compete? 4Competition, Predation and the Ecology of Fear 4Arctic and Red Foxes 5Arctic and Red Foxes in Sweden 8Questions 9

Methods 10Results and Discussion 13

Fox den use 13Fox diets and virtual food niches 14Specialists, generalists and numerical responses 16Interference competition 18Population effects 19

Conservation Implications 21Conclusions 22References 23Sammanfattning (Swedish summary) 28Acknowledgements 33

The complete version of this thesis includes all the above and the following papers, which are referred to by theirRoman numericals in the summary.

Papers I – V

I Dalerum F, Tannerfeldt M, Elmhagen B, Becker D, Angerbjörn A. 2002. Distribution,morphology and use of arctic fox dens in Sweden. Wildl. Biol. 8: 185-192.

II Elmhagen B, Tannerfeldt M, Verucci P, Angerbjörn A. 2000. The arctic fox (Alopexlagopus) – an opportunistic specialist. J. Zool. 251: 139-149.

III Elmhagen B, Tannerfeldt M, Angerbjörn A. 2002. Food-niche overlap between arctic andred foxes. Can. J. Zool. 80: 1274-1285.

IV Tannerfeldt M, Elmhagen B, Angerbjörn A. 2002. Exclusion by interference competition?The relationship between red and arctic foxes. Oecologia 132: 213-220.

V Shirley MDF, Elmhagen B, Lurz PWW, Rushton SP, Angerbjörn A. 2003. Modelling thespatial population dynamics of arctic foxes (Alopex lagopus): the effects of red foxes andmicrotine cycles. Submitted manuscript.

4

INTRODUCTION

The effects of competition between indivi-duals of different species may extend to pop-ulation levels where population densities anddistributions are affected. The arctic foxAlopex lagopus was abundant in Swedishmountain tundra areas in the late 19th cen-tury. At the turn of the century, excessivehunting was followed by a drastic populationdecline (Lönnberg 1927). Despite protectionsince 1928, there are now only about 70arctic foxes left. The red fox Vulpes vulpespopulation increased dramatically in the1930-1960’s (Lindström 1989a) and inter-specific competition may be a factor whichhas hampered arctic fox recovery (Herstein-sson et al. 1989). This thesis investigates therelationship between arctic foxes and redfoxes, especially what consequences red foxcompetition may have on the arctic foxpopulation.

When do species compete?

All species require specific environmentalconditions and resources to be viable. Theyrequire a niche (Hutchinson 1957). Inter-specific competition occurs when individualsof different species use the same resource,i.e. when there is an overlap in a nichedimension, and supply of that resource islimited. When all individuals are equallygood competitors, interspecific competitionleads to decreased reproductive success ordecreased survival in both species. However,it is more common with asymmetrical com-petition, where individuals of one species aremore affected than those of the other are(Connell 1983).

A high overlap in a niche dimension mayindicate severe competition, but there is nosimple relationship between the degree ofniche overlap and competition. Competitioncan cause an individual to change its use ofresources, and intense competition can there-fore be associated with low niche overlap

(Colwell and Futuyma 1970, Abrams 1980).Further, a high overlap in one niche dimen-sion is often associated with a low overlap inanother dimension (Schoener 1974). Forexample, if two individuals with similar foodniches also compete for territories, dietaryoverlap may remain high while spatial over-lap decreases.

It is useful to distinguish between thefundamental and the realised niches of aspecies. The fundamental niche is its broad-est possible niche, which comprises allenvironments where the species would beviable in the absence of interspecific compe-tition and predation. The realised niche is thepart of the fundamental niche where thespecies remains when competitors and pre-dators are present (Hutchinson 1957). A highoverlap between the fundamental niches oftwo species indicate that they will competewhen they are sympatric, while a relativelylower overlap between their realised nichesindicate that one or both species havechanged their use of resources as a responseto competition. If the distribution or densityof a superior competitor decrease, the inferiorspecies will experience a competitive releasewhich allows it to expand its realised nicheand increase in density or distribution. On thecontrary, when an inferior competitor nolonger has access to necessary resources, ithas lost its realised niche and will go extinctdue to competitive exclusion.

Fundamental and realised niches areconcepts which apply to an entire species, butsome species are found in many differentenvironments. In such cases, it can be usefulto identify also the niches of local popula-tions, using the concepts virtual and actualniches as equivalents of fundamental andrealised niches (Colwell and Futuyma 1970).

Competition, Predation & the Ecology of Fear

Competition is usually classified as eitherexploitation or interference competition (Park1962). Exploitation competition occurs when

5

one individual uses a resource and therebydeprives another individual of the oppor-tunity to do so. For example if food avail-ability for one individual is affected by otherindividuals’ consumption of that foodresource. Interference competition involvesbehavioural interactions between the individ-uals, such as fighting and predation. It mayalso be more subtle and individuals of aspecies may avoid stronger competitors ofother species rather than risk loosing a fightor being killed. Exploitation and interferencecompetition are both common among terres-trial animals (Schoener 1983).

A guild can be defined as a group ofspecies which uses similar resources andtherefore may compete. When such compe-tition involves predation, so-called intraguildpredation, the predator will benefit both byan immediate energetic gain and by de-creased exploitative competition (Polis et al.1989). Predators can kill more than a “doom-ed surplus” of their prey, thereby affectingprey abundance (e.g. Lindström et al. 1994).However, prey respond to the presence ofpredators by changing their feeding be-haviours and habitat selection. They prefersafe habitats, “predation refuges”, rather thanproductive ones. Hence, the fear of predatorsmay affect both predation rate and preyfitness. The behavioural response in fear-driven interactions may even be moreimportant for prey abundance than predationitself (Brown et al. 1999). Interspecifickilling is common among mammalian carni-vores and the combined effects of compe-tition and intraguild predation can causeinferior species to change their habitat use oractivity patterns (Palomares and Caro 1999).Thus, “competition refuges” may be import-ant for the co-existence of competitors justlike predation refuges are important for co-existence of predator and prey (e.g. Durant1998).

In some carnivores, intraguild predationhas a considerable impact on mortality rates.For example, predation by coyotes Canislatrans was the major cause of death for kit

foxes Vulpes macrotis and swift foxes Vulpesvelox (Ralls and White 1995, Sovada et al.1998). Despite this, there was no indicationof spatial segregation between coyotes andeither fox species (White et al. 1994, Kitchenet al. 1999). However, other carnivores showbehavioural responses to predator presence.Avoidance patterns has been found incoyotes which avoid wolves Canis lupus, redfoxes which avoid coyotes and lynx Lynxlynx, and cheetahs Acinonyx jubatus whichavoid lions Panthera leo and hyenas Crocutacrocuta (Harrison et al. 1989, Thurber et al.1992, Durant 1998, Fedriani et al. 1999).Intraguild predation on the inferior speciesoccurs in all these cases (Sargeant and Allen1989, Thurber et al. 1992, Palomares andCaro 1999, Sunde et al. 1999). However,predation appears to be rare, and avoidanceby the inferior species may be more import-ant for the overall effect on distributions andpopulation sizes. Coyote densities were high-er in areas where wolf densities were low andcoyote home ranges were found outside ornear wolf pack territory boarders (Fuller andKeith 1981, Thurber et al. 1992). Further, thereduction in wolf numbers in North Americawas followed by an increase in coyoteabundance (Mech 1970). Likewise, red foxesestablished home ranges outside or close tocoyote territory boarders, with minimalspatial and temporal overlap (Voigt and Earle1983, Major and Sherburne 1987, Sargeant etal. 1987, Harrison et al. 1989). Reductions incoyote or lynx populations have also beenfollowed by increased abundances of redfoxes (Linhart and Robinson 1972,Palomares 1996).

Arctic and Red Foxes

Evolution, morphology and distributionThe arctic fox is closely related to NorthAmerican swift and kit foxes. The speciesdiverged as glaciations became morecommon in the late Pleistocene and the arcticfox evolved in an arctic environment (Geffen

6

et al. 1992, Mercure et al. 1993). The currentdistribution of the arctic fox is circumpolarand includes tundra areas in the Arctic, NorthAmerica and Eurasia, and the mountaintundra in Fennoscandia (Audet et al. 2002).

The arctic fox has several adaptations totundra climate, such as a short, rounded bodywith short ears and legs. It has the bestinsulative winter fur of all mammals and itdoes not have to increase its metabolic ratesignificantly to maintain homeothermy undernormal temperature conditions (above -40°C)(Prestrud 1991). Further, its resting metabolicrate decreases in winter, possibly an adaptionto conserve energy (Fuglei and Øritsland1999). The arctic fox appears in two colourmorphs. The white morph is brownish-greyish dorsally and whitish ventrally insummer, but turns white in winter. The bluemorph is brownish all seasons. In general,both morphs are present in a population, butthere is a positive correlation between thefrequency of white foxes and snow cover.Due to the camouflage effect of their colour,the white morph probably has a selectiveadvantage over the blue in snowy inlandhabitats, while the opposite is true in coastalareas with less snow (Hersteinsson 1989).

The traditional placement of arctic foxesin a separate genus, Alopex, is due to themorphological differences between arcticfoxes and the fox species in the genusVulpes. However, genetic data suggest thatarctic foxes should not be genericallydistinguished from Vulpes (Geffen et al.1992).

The red fox evolved in Eurasia and latercolonised North America. Its current distri-bution covers North America, Europe andRussia, but it does not penetrate into the higharctic. It has also been introduced to otherareas (Larivière and Pasitschniak-Arts 1996).The red fox is smaller in northern areas thanin southern habitats (Englund 1965), but it isstill about 60% heavier than the arctic fox atcomparable latitudes. It also has morepointed ears and a relatively longer tail andneck (Hersteinsson and Macdonald 1982).

The colour of red foxes is very variable,ranging from light greyish to bright orange.There are also silver (black) and “cross”colour morphs.

Food and Population DynamicsArctic and red foxes are opportunistic feed-ers, but the arctic fox generally has a lessvaried diet. This is probably related to thelow diversity of prey in tundra areas (Her-steinsson and Macdonald 1982). However,arctic fox populations can be divided intocoastal and inland (“lemming”) foxes, as theecology of the species differs between theseenvironments (Braestrup 1941). Most inlandareas are barren and small rodents with cyclicpopulation fluctuations dominate the preybase. In years of low food availability, foxreproduction typically fails. When food avail-ability is high, an inland vixen may carry upto 21 foetuses and wean at least 16 cubs, oneof the largest litter sizes among Canids(Macpherson 1969, Ewer 1973, Tannerfeldtand Angerbjörn 1998). As a consequence, thepopulation dynamics of inland foxes ischaracterised by large fluctuations, as thefoxes respond numerically to the availabilityof their cyclic prey (e.g. Elton 1924,Macpherson 1969, Angerbjörn et al. 1995,Angerbjörn et al. 1999, Strand et al. 1999).

Arctic foxes in coastal areas have accessto a greater variety of resources than inlandfoxes, such as sea birds, seal carcasses, fishand marine invertebrates. Prey availability istherefore relatively stable between years. Thedifference between inland and coastal areashas led to the evolution of differentreproductive strategies. Coastal foxes pro-duce relatively small litters of similar sizeevery year and they appear to have lost thecapacity to have very large litters. Inlandfoxes cannot reproduce successfully all years,but have larger mean and maximum littersizes (Frafjord 1993a, Tannerfeldt andAngerbjörn 1998).

Red foxes can have a maximum of 12cubs, but most litters consist of 3-6 cubs(Larivière and Pasitschniak-Arts 1996). Red

7

fox populations are relatively dense andstable in productive habitats and thepopulations may be socially regulated. Onthe contrary, populations in less productivehabitats may be food regulated (Lindström1989b). For example, small rodents are cyclicin boreal Fennoscandia and in other Eurasianareas with similar snow conditions, such asPoland and Russia (Hansson and Henttonen1985). Both the proportion of breeding redfox vixens and litter sizes are related to voleavailability (Englund 1970, Lindström 1988).Consequently, the size of red fox populationsin areas with cyclic voles fluctuates inrelation to vole abundance (Englund 1970,Angelstam et al. 1985, Goszczynski 1989a).

Social organisationHome ranges of adult arctic foxes are 4-60km2 (e.g. Eberhardt 1982, Birks and Penford1990, Prestrud 1992, Angerbjörn et al. 1997,Eide et al. 2002a). The smallest home rangesare found in productive coastal habitats, butaverage home range size increases as habitatsbecome less productive in inland areas (Eideet al. 2002a). Comparing red fox homeranges in different habitats, home range sizealso increases as productivity decreases(Lindström 1982, Jones and Theberge 1982,Goszczynski 1989b, Meia and Weber 1995).

The basic social unit in arctic and redfoxes is a breeding pair where both the maleand the female participate in cub rearing(Kleiman 1981, Hersteinsson and Macdonald1982, Angerbjörn et al 2003a). Withinhabitats, the pair establishes a home range ofthe size necessary to sustain it when foodavailability is low (Lindström 1989b, Eide etal. 2002a). When food availability in thehome range increases, arctic and red foxesincrease group size rather than diminishhome range size, as older offspring remain intheir natal home ranges (Lindström 1989b,Angerbjörn et al. 2003a). In some cases,resident older offspring reproduce and thefoxes practice communal breeding with morethan one litter in a den (Macdonald 1979,Hersteinsson and Macdonald 1982, Baker et

al 1998, Strand et al. 2000, Angerbjörn et al.2003a). The resource dispersion hypothesissuggests that the dispersion of food patchesand group size determine home range size bythe richness of those patches. This mightexplain the social organisation of both arcticand red foxes (Hersteinsson and Macdonald1982, Macdonald 1983).

Interspecific relationsArctic and red foxes are sympatric in arelatively narrow overlap zone in the lowArctic and the similarities between them indi-cate that they should be competitors (Her-steinsson and Macdonald 1982). Due to itslarger size, a red fox needs more food and alarger home range than an arctic fox tosustain its higher energy demands. Thus, thenorthern limit of the red fox may bedetermined by resource availability which inturn depends on climate (Hersteinsson andMacdonald 1992). Red foxes are dominantover arctic foxes in direct interactions(Rudzinski et al 1982, Schamel and Tracy1986, Frafjord et al. 1989, Korhonen et al.1997). The southern distribution limit ofarctic foxes may therefore be determined byinterspecific competition with red foxes. Ifthat is the case, an amelioration of theclimate would increase productivity andallow red fox expansion, while arctic foxpopulations would retreat due to increasedcompetition from red foxes.

As a hypothetical example, northernFennoscandia experienced a 2°C increase intemperature in 1910-1935. This would havepushed isotherms and the distribution of redfoxes up c. 330 m in mountain tundra areas.If the lowest altitude of arctic foxes in 1900would have been 600 m a.s.l, the newdistribution limit in 1935 would be 930 ma.s.l, and the arctic fox population wouldhave lost more than half of its former habitat(Hersteinsson and Macdonald 1992).Archaeological finds indicate that majorreductions in the arctic fox distribution rangehave occurred before, as all Norwegian foxbones from the warm period 9000-5000 B.P

8

Fig. 1 Estimated number of breeding arctic foxes inSweden in 1974-2002 (from Angerbjörn et al. 2002).

were from red foxes, while arctic fox boneswere either older or younger (Frafjord andHufthammer 1994).

Arctic and Red foxes in Sweden

The Fennoscandian mountain range stretchesfrom the Kola Peninsula in the north-east tosouth-western Norway, encompassing areasin Russia, Finland, Sweden and Norway. Dueto a combination of relatively high latitudesand altitudes, the area has tundra climate andvegetation, while the rest of Fennoscandia isforested. The habitat can be described asmountain tundra with arctic species like thearctic fox, Norwegian lemming Lemmuslemmus and reindeer Rangifer tarandus.

Assuming that present densities ofbreeding arctic foxes in Siberia are com-parable to former densities in Fennoscandia,the Swedish population before 1880 con-sisted of c. 4000 breeding adults (range 730-9430 in 33 000 km2 of available habitat; datarecalculated from Angerbjörn et al. 1999 andTannerfeldt pers. comm.). In the late 19th

century, increased prices of arctic fox peltsled to excessive hunting and a consequentpopulation decline (Lönnberg 1927). Ahealthy arctic fox population was regarded asa valuable asset to people in the mountainregions. Hence, the arctic fox was protectedin 1928, and protection in Norway (1930)and Finland (1940) soon followed. Despite

this, the population did not recover and peakpopulation sizes in the 1970’s were about180 breeding adults. A further decreasefollowed from 1980, probably due to foodshortage during an absence of lemming peaks(Fig. 1, Angerbjörn et al. 1995). Populationestimates for 2002 indicate that there areabout 70 adults in Sweden and 10 in Finland.Population densities in Norway and Kola aresimilar to those in Sweden (Angerbjörn et al.2002). Low lemming abundances, decreasedavailability of reindeer carcasses in winter,increased competition with red foxes,inbreeding and Allee effects are factorswhich may have hampered a recovery(Hersteinsson et al. 1989, Linnell et al. 1999b).

Hunting bags indicate that the Fenno-scandian red fox population about tripled insize between 1930 and 1960 (Lindström1989a, Selås et al. unpublished). The increasecoincided with a northward spread of redfoxes in North America and Russia (Marsh1938, Macpherson 1964, Chirkova 1968).There are several factors which may havecontributed to the increase. A general amelio-ration of the climate, decreased interferenceand exploitative competition with large pre-dators, increased access to food left-overs inhuman garbage, and changes in forestrypractices including large clear-cut areaswhich favour field voles Microtus agrestis,the main prey of boreal red foxes (Hersteins-son and Macdonald 1992, Lindström 1989a).

The red fox population was severely hitby an epidemic of sarcoptic mange in the1980’s (Lindström and Mörner 1985, Danelland Hörnfeldt 1987). Although the epidemicbegan in northern Sweden, reductions appearto have been largest in the dense populationsin southern Sweden. Three Swedish countieshave mountain tundra habitat with arcticfoxes, Norrbotten, Västerbotten and Jämt-land. There was no reduction in red foxhunting bags in Norrbotten and a relativelyslight reduction in Västerbotten (Fig. 2; Swe-dish Sportsmen’s Association 1960-2001).Hence, the mange epidemic probably hadlittle effect on the degree of interspecific

Year

No

. o

f re

pro

du

cin

g a

du

lts

0

40

80

120

160

200

74 78 82 86 90 94 98 02

9

competition between arctic and red foxes inthese counties. There is only one observationin Fennoscandia of arctic foxes infested bymange and those foxes were captured, treatedand thereafter released (Mörner 1988).

When the red fox population increasedafter 1930, the number of red foxes alsoincreased in Fennoscandian mountain tundrahabitat (Østbye et al. 1978). There isevidence for a parallel redistribution of arcticfoxes which retreated to peripheral areas anddens at higher altitudes (Østbye et al. 1978,Linnell et al. 1999a). These findings areconsistent with Hersteinsson and Macdon-ald’s (1992) theory on interspecific compe-tition. However, if red foxes have different

habitat or prey preferences than arctic foxes,the retreat could also result from otherchanges in the environment (Chirkova 1968,Linnell et al. 1999a).

Questions

The aim of this thesis is to investigate therelationship between arctic and red foxesthrough studies of den use, food niches andinterference competition, and to estimate thepossible effects of red fox competition on anarctic fox population. The specific questionsare:

Fig. 2 Biogeographical zones in Sweden. Arctic and red foxes are sympatric only in mountain tundra habitat in thealpine zone of Norrbotten (BD), Västerbotten (AC) and Jämtland (Z) counties. Red foxes are found in all of Sweden,but the population was severely hit by a mange epidemic which started to spread in 1975 (see text). However,estimated numbers of shot red foxes indicate that the epidemic had less serious effects on the northernmost red foxpopulations compared to the rest of Sweden (figures to the right; data from the Swedish Sportsmen’s Association(1960-2001).

Red foxes shot in Norrbotten (BD)

0

2000

4000

6000

8000

1960 1970 1980 1990 2000

Red foxes shot in Västerbotten (AC)

0

2000

4000

6000

8000

1960 1970 1980 1990 2000

Red foxes shot in Jämtland (Z)

0

2000

4000

6000

8000

1960 1970 1980 1990 2000

Total no. of red foxes shot in Sweden

0

20000

40000

60000

80000

1960 1970 1980 1990 2000

10

Paper I Do arctic and red foxes differ intheir use of dens in mountaintundra habitat?

Paper II The prey base in mountain tundrahabitat is restricted. To what extentis the arctic fox dependent onlemmings for food and successfulreproduction?

Paper III Do arctic and red foxes have dif-ferent virtual food niches?

Paper IV Is there interference competitioncausing arctic foxes to avoid redfoxes?

Paper V Using a spatially explicit model tosimulate population effects, what isthe effect of red fox competitionon an arctic fox population?

METHODS

Study area

The study area was located in the nature re-serve of Vindelfjällen in Västerbotten county,Swedish Lapland. In total, thestudy area covered 1300 km2 ofmountain tundra, but there was asmaller core area of 450 km2

(Fig. 3). The major part of thearea is situated within the lowalpine vegetation zone and themain vegetation types are dryheath, grass heath and dry fen.Most geomorphological struc-tures are of glacial origin andsubglacially engorged eskers,kames and other hummockyglacifluvial accumulations arecommon (Ulfstedt 1977). Sum-mers are short with snowmelt inJune and first snow in Septem-ber, but there is no permafrost.

The tree line is situated at about 750 m a.s.land valleys with birch and coniferous forestsintersect the mountain tundra (Fig. 3). Arcticfoxes bred only above the tree line. Thelarger part of the red fox population wasfound below the tree line, but we have onlycollected data on red foxes which inhabiteddens on the mountain tundra, i.e. those thatwere most likely to interact with arctic foxes.

All population peaks are by definition“peaks”. However, Fennoscandian lemmingand vole cycles generally have amplitudes inthe order of 1:200 (Hörnfeldt 1994, unpub-lished data). As a consequence, there is apopular notion of “peak years” specificallymeaning peaks reaching approximately thesedensities. For lemmings in particular, suchpeaks are easily distinguished in the field aslemmings suddenly are seen everywhere onthe tundra, while they rarely are seenotherwise. In this thesis, I use this qualitativeterm to distinguish such lemming and volepopulation peaks from other, relatively small,increases.

Four species of microtine prey werepresent in the study area, Norwegianlemming, field vole, bank vole Clethriono-mys glareolus and grey-sided vole Clethrio-nomys rufocanus. Lemming and vole popula-

Fig. 3 The study area in Vindelfjällen. Only areas above the tree linewhere arctic fox dens were surveyed regularily are shown.

11

tions are generally cyclic in Fennoscandiantundra and boreal forests, but this pattern hasbeen interrupted during the last 20 years(Hansson and Henttonen 1985, Angerbjörn etal. 2001, Hanski et al 1993, Hörnfeldt 1994).The last lemming peak in Västerbotten beforethe start of this study occurred in 1982, butfluctuations have continued with a loweramplitude (Angerbjörn et al. 2001). Like-wise, densities of grey-sided voles in theforests of Västerbotten have declined drasti-cally and the last population peak occurred in1985. Forest populations of field and bankvoles have continued to peak but there hasbeen a declining trend in peak densities(Hörnfeldt 1994). The declining trends infield and bank vole populations were, at leasttemporarily, broken in 1998 when there wasa large population peak (Hörnfeldt 1998). In2001, the lemming population finally peakedin all Swedish mountain tundra habitat(unpublished data).

Other species in the study area whicharctic and red foxes may prey upon werereindeer Rangifer tarandus, mountain hareLepus timidus and shrews Sorex spp. Therewere also rock ptarmigan Lagopus mutus,willow ptarmigan Lagopus lagopus,passerines Passeriformes, ducks Anserii-formes, gulls and waders Charadriiformes.

Other predators which may interact withthe foxes were wolverine Gulo gulo andgolden eagle Aquila chrysaëtos.

General methods

The arctic fox population in the study areahas been surveyed by the Swedish Arctic FoxProject since 1985. The aim of the projecthas been to study the ecology and behaviourof the arctic fox (e.g. Tannerfeldt 1997).

Dens originally made by arctic foxes aretypically large structures that have been inuse for hundreds of years. They are coveredwith lush vegetation which makes themconspicuous and easy to find (Zetterberg1945, Macpherson 1969). Den surveys dur-

ing the breeding season is therefore the mostcommon method for monitoring of arctic foxpopulations (Eberhardt et al. 1983, Anger-björn et al. 1995). Dens originally dug by redfoxes are smaller than arctic fox dens. Thevegetation on a red fox den does not differmarkedly from the surroundings and fewdens have more than 5 openings (Østbye etal. 1978). In the early years of the Arctic FoxProject, den surveys mainly covered the corearea, but more areas were added as time wenton. Most surveyed dens were arctic fox dens,but some red fox dens have also beenchecked on a regular basis. However,reproducing red foxes were only found inarctic fox dens. Therefore, all dens men-tioned in the following are dens originallyconstructed and used by arctic foxes, but Irefer to them as arctic or red fox densdepending on which species that occupiedthem.

We visited dens in July, when cubsemerge, and searched for foxes or signs ofactivity such as scats, prey remains wear andtear in the vegetation, fur and newly ex-cavated openings. When we suspected thatarctic foxes were using the den, we observedit for at least 24 hours to decide if there werecubs or not and minimum litter sizes wererecorded. Within the core area we taggedvirtually all cubs and two thirds of thereproducing adults. Red foxes in the areawere very shy compared to the arctic foxes,and they often moved even after a short visitat their dens. Hence, we rarely had more thanone chance to see them and we only recordedpresence or absence of cubs (either visualobservation of cubs, or small scats and wearand tear in the grass which indicate that cubswere present). On a few occasions, we didnot see any foxes although the dens wereclearly inhabited. We then relied on tracksand the size of newly excavated den openingsto determine which fox species used the den.

Lemmings build nests under the snow inwinter and lemming densities can beestimated by nest counts (Sittler 1995). From1990 and onwards, we counted the number of

12

lemming nests per 100 km walked on themountain tundra and used this as an index forlemming winter population densities. Thedirection of change in the index betweenyears was used to determine the status of thelemming population in the summer (increaseor decrease/low phase).

Fox den use (Paper I)

Most arctic fox dens in the study area werealready known by local rangers in the early1980’s, while others have been found duringlater surveys. In 1999-2000 we gatheredmorphological data on all dens. We countedthe number of openings that were in goodcondition (“non-collapsed”) and measuredthe distance between the most outspacedopenings in two directions (den “length” and“width”). We used the resulting rectangle asan approximation of den area. Altitude anddistance from the tree line were estimatedfrom maps.

Survey data showed when dens had beeninhabited by foxes. For each den, we used thefrequency of litters in relation to surveyedyears to analyse and compare the denpreferences of arctic and red foxes.

Fox diets and food niches (Paper II, III)

We needed data on fox diets to investigate towhat extent arctic foxes depend on lemmingavailability and to compare the food nichesof arctic and red foxes. We collected freshscats at inhabited den and analysed thecontents. Unfortunately, all scats collected in1991-1992 and some collected in 1993-1995were lost during storage.

First, we studied arctic fox diets in 1990and 1994-1997 and tested if there was anumerical response to lemming availability,i.e. if the number of foxes which establishedand tried to reproduce depended on lemmingavailability in winter (Paper II).

Second, we compared arctic and red foxdiets in 1993 and 1996-1998 (Paper III). Thebest way to determine if two species competeis to compare their virtual and actual nichesby studying resource use under allopatric andsympatric conditions (Colwell and Futuyma1970, Clode and Macdonald 1995). Arcticand red foxes are sympatric in Vindelfjällenand we have no allopatric populations tocompare with in Sweden. However, aslemming and vole populations fluctuate, thecomposition of the prey base changesbetween summers. If arctic and red foxesrespond similarly to those changes, theyshould have the same virtual food niches. Inthat case, percent overlap (Schoener 1970)between their diets should be constant. Wetested if this was the case by comparing thedegree of overlap between arctic and redfoxes diets over four summers.

Interference competition (Paper IV)

If there is interference competition betweenarctic and red foxes and the arctic fox is theweaker competitor, arctic foxes should avoidcontact and not breed close to dens inhabitedby red foxes. Although defended territoriesmay be smaller, arctic foxes are often seen upto 6 km from their dens (Angerbjörn et al.1997, Tannerfeldt pers. obs.). The larger redfox should have a 90% larger home range(Harestad and Bunnell 1979, Hersteinssonand Macdonald 1992). The 6 km radius forarctic foxes should thus be equivalent to an 8km radius for red foxes. We used den surveydata from the core area in 1987-2000 andtested how often high quality dens wereinhabited by breeding arctic foxes whenbreeding red foxes were absent or presentwithin an 8 km distance.

Population effects (Paper V)

Spatially explicit population models can beused to simulate population effects of behav-

13

iours like interference competition (Rushtonet al. 1997, 2000). Hence, we developed aspatially explicit population model to studylong-term effects of interspecific competitionon the arctic fox population in Vindelfjällen.The model had a GIS component whichstored information on den and fox locations.Thereby, we could simulate arctic fox avoid-ance of red foxes by not allowing them toestablish in dens situated within 8 km of densoccupied by red foxes, as described in PaperIV.

The modelled area consisted of the studyarea in Vindelfjällen and surrounding forests.The forests contained a source population ofred foxes from which dispersal to mountaintundra dens occurred. We parameterised themodel using available data on arctic and redfox reproduction and mortalities duringdifferent lemming and vole populationphases (low, increase, peak and crash). Weused the observed sequences of lemming andvole phases in the period 1985-2000 as amodel input and simulated the correspondingfox population dynamics during this period.Initial fox populations in the simulationswere those of Vindelfjällen and surroundingforests in 1985.

We compared model outputs with theobserved population dynamics of arctic andred foxes in the core area. We then tested theeffect of red fox competition on the arctic foxpopulation by running the model with andwithout red foxes, comparing populationtrends.

RESULTS AND DISCUSSION

Fox den use

There were 77 arctic fox dens in the studyarea, all excavated in glacifluvial eskers orterrace formations. The lowest lying den wassituated 760 m a.s.l, which is only a fewmeters above the treeline, while the highestden was situated 1160 m a.s.l (Fig. 4, Paper

I). The highest mountain peaks in the studyarea rise above 1600 m a.s.l, but areas above1100 m generally consist of boulder fields.Rock dens under boulders are used by arcticfoxes in high-arctic Svalbard (Prestrud 1992).We have observed adult arctic foxes use suchdens as temporary shelters, but there is noindication that they use them for breeding inSweden at present low population densities.The average den in Vindelfjällen was situatedin an esker on a south-facing slope 915 ma.s.l, 3.5 km from the treeline. It had an areaof 277 m2 and 44 openings (Paper I). Thus,the dens were among the largest reported forarctic foxes in tundra habitats (e.g.Chesemore 1969, Østbye 1978, Garrott et al.1983, Anthony 1996, Frafjord 2003).

Unlike many other tundra areas, there isno permafrost in Vindelfjällen and there aremany glacifluvial sand formations. It istherefore possible to dig dens of com-paratively high quality. Arctic foxes may digseveral new openings in one summer.However, construction of dens with morethan 100 openings, as the largest dens inVindelfjällen, and fertilisation of the den areato an extent where the vegetation differscompletely from the surroundings (Österdahl2003), must be the work of generations offoxes. The dens were constructed before thearctic fox population started to decline in theearly 1900:s (Zetterberg 1945) and thedistribution of dens should therefore mirrorthe former distribution of arctic foxes.

Arctic foxes showed a preference fordens with many openings, while red foxespreferred dens which covered large areas.These two measurements were positivelyrelated to each other and both speciesconsequently preferred large dens (Paper I).Large dens are equally available at allaltitudes (Spearman’s r = - 0.054, p = 0.97),which indicates that the former arctic foxdistribution reached down to the presenttreeline (Fig. 4). Zetterberg (1945) worked asa ranger in the area 1904-1929 and alsogathered information on the arctic fox fromold Sami and settlers. He did not find arctic

14

Fig. 4 Distribution of arctic fox dens in Vindelfjällen(n=77) to the left. Median and mean altitudes for thedens were 905 and 915 m a.s.l. respectively. Onaverage, dens used by red foxes were situated 860 ma.s.l. (n=11) and dens used by arctic foxes 940 m a.s.l.(n=30). The right figure show means (small boxes),SE:s (large boxes) and SD:s (whiskers). In theanalyses for paper I, we accidentally labelled a denwith a red fox litter as having an arctic fox litter, thusn-values differs from the ones presented here. Themistake should not alter the overall conclusions ofPaper I as the den was situated 810 m a.s.l, anddifferences between arctic and red fox den useincrease when it is labelled correctly.

foxes breeding below the tree line, butmentions that arctic foxes hunted in the birchforests, which indicates that dens at relativelylow altitudes were used. Arctic foxes werealso trapped in low altitude dens. Despitethis, only 12% of the arctic fox litters born in1985-2001 were found in dens below themean altitude of all dens (915 m a.s.l.) andthe mean altitude of arctic fox breeding denswas 940 m a.s.l (Fig 4).

A positive relationship between denaltitude and density of den openings indicatesthat dens at lower altitudes disrepair(Spearman’s r = 0.64, p < 0.001). Forexample, one of the largest dens inVindelfjällen (den area 800 m2) is situated900 m a.s.l. It had at least 172 openings inthe early 1900:s, but only 40 in 2000. It wasinhabited by arctic foxes in 1911 (Zetterberg1945), but it was only inhabited once (by redfoxes) in 1985-2002. Red foxes used dens atlower altitudes, closer to the tree line, thanarctic foxes (Paper I). Although they prefer

large dens, they therefore use dens withcomparatively less openings than those usedby arctic foxes at higher altitudes, whichexplain why their den use is more related toden area than den openings.

Thus, low altitude dens are rarely usedby arctic foxes and they are collapsing, whichindicates that the habitat preferences of arcticfoxes have changed after the populationcrashed. Similar patterns of den use havebeen found in Norway, where radio trackingalso has shown that arctic foxes use lowaltitude tundra less than expected in relationto availability, and that they avoid forestedhabitats (Østbye 1978, Landa et al. 1998,Linnell et al. 1999a). The retreat of arcticfoxes to higher altitudes, and the fact that redfoxes use former arctic fox dens at loweraltitudes, support the hypothesis thatinterspecific competition with red foxesdetermines the distribution of arctic foxes.

Fox diets and virtual food niches

Spatial segregation between arctic and redfoxes may be a consequence of differentvirtual niches rather than interferencecompetition (Chirkova 1968, Linnell et al.1999a). For example, Chirkova (1968) sug-gested that parallel changes in arctic and redfox distributions could be caused by changesin prey distributions, as arctic foxes had astrong preference for lemmings, while redfoxes preferred voles.

Lemmings, field voles, birds andreindeer were the main constituents of botharctic and red fox diets in Vindelfjällen(Paper II, III). Lemmings were the domina-ting prey of arctic foxes, while field volesdominated red fox diets (Fig 5). Generally,lemmings occurred significantly more oftenin arctic fox scats than red fox scats, whilefield voles and birds occurred significantlymore often in red fox scats.

Dens inhabited by arctic foxes weresituated at higher altitudes than densinhabited by red foxes (Paper I, III). For bothspecies and for all years combined, occur-

No of dens

Alti

tud

e (

m a

.s.l.

)

760

840

920

1000

1080

1160

0 2 4 6 8 10 12

Red foxes

Arctic foxes

15

rence of field voles and birds decreasedsignificantly at higher altitudes, while theoccurrence of lemmings tended to increase(Paper III). Productivity generally decreaseswith increasing altitude and vole densities arehigher in productive habitats than inunproductive habitats (Oksanen and Oksanen1992, Oksanen et al. 1999). There is also anegative relationship between bird abundanceand altitude in our study area (LindaAxelsson, unpublished data). Hence,lemmings should typically make up a largerpart of the prey base around arctic fox denscompared to red fox dens, while the relativeavailabilities of field voles and birds should

Fig. 7 A comparison between the o

quartiles.

be higher around red fox dens. Frafjord(1995, 2000) found that lemmings occurredmore frequently in arctic fox scats than in redfox scats at comparable altitudes. Hesuggested that arctic foxes might be lessopportunistic than red foxes are, and that theytherefore would be more dependent onlemmings also in the same habitat as redfoxes. However, the large fluctuations inlemming and vole populations between yearsimply substantial changes in the prey base.Hence, comparisons between fox diets atsimilar altitudes should been done in thesame year to be completely comparable,

Pre

y re

main

s in

sca

ts (

% v

olu

me)

������������������

��������������

�������

��������

���������

��������

���������������������

������������ ������ ����� �����

����������������

0

20

40

60

80

100

1990 1993 1994 1995 1996 1997 1998

Lemming

��Voles and shrews��������

Birds

Reindeer

Pre

y re

main

s in

sca

ts (

% v

olu

me)

����������

��������

��������

��������

��������������

����������

����������

������������������������

0

20

40

60

80

100

1993 1996 1997 1998

b)

a)

Fig. 5 Summer diets of arctic and red foxes in Vindelfjällen. Food niche overlap between arctic and red foxes wasrelatively constant in 1993, 1996 and 1998. In 1996, both arctic and red fox diets were dominated by lemmings,while they switched to voles (mainly field voles), birds and reindeer the other years (from Paper II and III).

16

which was not always the case in Frafjord’sstudies.

We took the effect of temporal changesin the prey base into account by comparingniche overlap between arctic and red foxes indifferent years. Despite large changes inarctic and red fox diets, food-niche overlapremained almost constant in 3 out of 4 years(75-80%; Fig. 5, Paper III). This indicatedthat arctic and red foxes switched to the sameprey species as the composition of the preybase changed between years, and conse-quently, that they have the same searchimages, foraging habitats and diet pre-ferences (Taylor 1984). One year (1997)deviated from this pattern with a relativelylow overlap of 44%. Two red fox dens wereincluded in the sample this year and theirdiets were very different (overlap 29%).Individual variation in red fox diets wasconsequently larger than that between arcticand red foxes in 1997.

Hence, we conclude that arctic and redfoxes generally respond in a similar way totemporal changes in the prey base, andsuggest that arctic and red foxes have thesame virtual food niches (Paper III).

Specialists, generalists & numerical responses

Arctic foxes are often described asopportunists which readily change their dietsto exploit the most profitable food resources(e.g. Garrott et al. 1983, Birks and Penford1990, Stickney 1991, Frafjord 1993b, Bantleand Alisauskas 1998, Kapel 1999, Anthonyet al. 2000, Dalerum and Angerbjörn 2000,Samelius and Alisauskas 2000, Eide et al.2002b). This was supported by our studies(Paper II, III). Nevertheless, arctic foxes ininland habitats are often strongly dependenton rodent availability for successfulreproduction. The number of arctic foxes thatoccupied dens in our study area was relatedto lemming availability in spring, i.e. duringmating and gestation (Fig. 6) and there was asignificant relationship between lemming

Fig 6 The relationship between lemming densities inthe winter and the number of occupied dens in ourstudy area in July (p = 0,019, r2 = 0.51; from Paper I).

phase and number of arctic fox litters (Table1). Other studies have also shown that thenumber of arctic fox litters and litter sizes arerelated to rodent abundance (Macpherson1969, Angerbjörn et al. 1995, Kaikusalo andAngerbjörn 1995, Angerbjörn et al. 1999,Strand et al. 1999).

The boreal red fox population in Swedenpreys heavily on field voles and respondsnumerically to fluctuations in the volepopulation (Englund 1965, Englund 1970,Angelstam et al. 1985). However, when redfoxes establish in mountain tundra areas theyreadily prey on lemmings. In fact, both arcticand red foxes appeared to prefer lemmings tofield voles, unless field voles were far moreabundant than lemmings (Paper III). Thereason may be that it is more profitable tohunt for lemmings, which are larger and usemore open microhabitats than field voles(Heske and Steen 1993). Similarly, red foxesin boreal forests consume more field volesthan Clethrionomys spp, although the latterare more abundant. It may be easier and moreprofitable to hunt field voles thanClethrionomys spp. as the latter live inprotective shrubs, while field voles moverelatively slow and live in more open habitats(Englund 1965).

As red foxes in mountain tundra habitatprey on lemmings, what determines red fox

Lemming nests/100 km

No. of occ

upie

d d

ens

0

2

4

6

8

0 20 40 60 80

1990

1991

19921993

1994

1996

1997

1998

1995

17

numbers? A comparison of red fox abun-dance with (mountain tundra) lemming phaseand (boreal) field vole abundance indicatedno relationship between lemming phase andthe number of red foxes breeding inmountain tundra habitat. Instead, red foxnumbers both in mountain tundra and borealhabitats were significantly related to fieldvole abundance in boreal forests (Table 1,Fig. 7). Thus, different factors determinearctic and red fox abundance in mountaintundra habitat, but the species use the preybase in similar opportunistic manners.

The specialist-generalist concepts are us-ed with two different meanings. Sometimes,they are regarded as immutable, internalcharacteristic of a species, which implies thata specialist does not switch to other types ofprey even when they are readily available. Atother times, they are used to describe thefunctions of predators in specific com-munities (Reid et al. 1997, Wiklund et al.1999). Either way, specialists have morepronounced numerical responses to oneparticular prey species than generalists do, asthey cannot fully compensate for declines inthat prey by switching to other types of prey(Andersson and Erlinge 1977). Species thatare immutably specialised could be describedas ”virtual specialists”, which function asspecialists in all environments. An ”actualspecialist” could also be a “virtual gene-ralist”, an opportunist forced into a specialiststrategy by a limited prey base. Such specieswould change strategy and become “actualgeneralists” in environments with higher

biodiversity.The red fox is regarded as a generalist

(e.g. Englund 1965, Ferrari and Weber 1995,O’Mahony et al. 1999). However, red foxesin southern Sweden have more diversifieddiets than those in northern Sweden (Englund1965, Lindström 1982). Red fox populationsare also relatively stable in the south andthere is no direct relationship betweenchanges in the abundance of any particularprey species and reproduction. On thecontrary, northern red fox populations are

Fig. 7 Correlograms between boreal field volepopulations and red fox populations in differenthabitats in 1987-2001. Both fox populations weresignificantly correlated with the field vole populations,with a time lag of one year (see Table 1). Data on fieldvole abundance (no trapped voles/100 trapnights inspring in Västerbotten) was available from theSwedish Environmental Protection Agency(http://www.eg.umu.se/personal/ hornfeldt_birger/).Estimated number of red foxes shot in Västerbotten(Swedish Sportsmen’s Association 1960-2001) wasused as an index of boreal red fox population size, andnumber of red fox litters in the core area as an index ofthe number of red foxes in tundra habitat.

Table 1 Degree of crosscorrelation between boreal red foxes, tundra red foxes, arctic foxes and field voles/lemmings. The highest correlation coefficient (r) and the corresponding timelag of the foxes are shown for each fox-rodent pair. Significant r-values are bold.

Boreal field vole(1972–1986)

Boreal field vole(1987-2001)

Tundra lemming(1990-2001)

Boreal red fox rtimelagn

0.67015

-0.73114

0.45411

Tundra red fox rtimelagn

-0.68115

0.52412

Arctic fox rtimelagn

0.66413

-0.44115

0.78012

Time lag (years)

Cro

ssco

rrela

tion c

oeffic

ient

-1.0

-0.5

0.0

0.5

1.0

-8 -6 -4 -2 0 2 4 6 8

Boreal red foxes

Tundra red foxes

18

food regulated and fluctuate in relation to thevole cycle (Englund 1970). In a transitionzone between northern and southern popula-tions, red foxes are food regulated in years oflow food availability and socially regulatedwhen vole densities are increasing or high(Lindström 1989). Thus, there appears to be asouth-north gradient in the feeding strategiesof red foxes. Continuing along the gradient ofdecreasing productivity from boreal forests tolow-production mountain tundra, red foxesshould have progressively less varied diets.Thus, red foxes in mountain tundra andboreal habitat can be regarded as actual fieldvole specialists relative to red fox popula-tions in southern Sweden.

The arctic fox uses a specialist strategyfor feeding and reproduction in inland areasand a generalist strategy in coastal areaswhere there are more types of prey andavailability is more stable (e.g. Hersteinssonand Macdonald 1996, Tannerfeldt andAngerbjörn 1998, Dalerum and Angerbjörn2000). Inland arctic foxes, including thepopulation in Vindelfjällen, function asactual specialists on lemmings as reproduc-tion suffers when lemming availability de-creases. Their opportunistic behaviour indi-cates that they exploit all available resources,but alternative prey is insufficient and doesnot allow for successful reproduction.Nevertheless, using these resources mayfacilitate survival of adult foxes duringlemming lows, and when lemming popula-

tions start to increase, they respond by prod-ucing large numbers of cubs. Some cubs willsurvive long enough to exploit the nextlemming peak and the population will beviable despite drastic fluctuations inpopulation size (Tannerfeldt and Angerbjörn1996). Thus, the arctic fox is an opportunisticspecialist stuck in a world of lemmings(Paper II).

Interference competition

Prey is generally more abundant at lowaltitudes in mountain tundra habitat.However, both radio-tracking of arctic foxesand patterns of den use indicate that arcticfoxes rarely use these areas (Paper I, Landaet al. 1998, Linnell et al. 1999a). Thisbehaviour differs from that in otherenvironments, such as Svalbard, where arcticfoxes prefer productive habitats to barrenareas (Jepsen et al. 2002). It also differs fromprevious behaviour in Sweden, as arcticfoxes used to hunt also in the birch forestsbefore the population decline (Zetterberg1945). Arctic and red foxes have the samevirtual food niches and they shouldconsequently compete for territories in lowaltitude areas where prey is most abundant.Observations of encounters between wildarctic and red foxes show that arctic foxes areinferior to red foxes and they avoid closecontact when possible (Schamel and Tracy1986, Frafjord et al. 1989).

We tested if patterns of den use by arcticfoxes indicated small-scale patterns of avoid-ance of red foxes. We found that arctic foxesused dens for breeding more often in theabsence of red foxes, than when red foxeswere breeding within an 8 km distance.Actually, only three arctic fox litters wereborn within this distance from red foxes. Intwo of these cases, arctic fox cubs werekilled by red foxes, whereupon the litterswere moved to secondary dens, farther fromthe dens inhabited by red foxes (Fig. 8, PaperIV).

Red fox < 8 km No red fox < 8 km

Na

tal a

rctic

fo

x d

en

s (%

)

���������������������������������������������������������������������������������������

������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������

0

10

20

30

40

50

60

Fig. 8 Arctic fox breeding rate in relation to whetherthere were reproducing red foxes in the vicinity(within 8 km) or not. Black denotes that there was redfox predation on arctic fox cubs (from Paper IV).

19

The fact that arctic foxes directly avoidbreeding close to red foxes implies that redfoxes can affect arctic fox den preferencesand distribution patterns. Increased inter-specific competition with an expanding redfox population, through territorial andencounter competition (Schoener 1983), maytherefore explain why the distribution of thearctic fox in Fennoscandia has shifted and thearctic fox has retreated to higher altitudes.This is similar to the relation between redfoxes and coyotes in North America, whereinterspecific territoriality and predationsuppress red fox populations and force redfoxes to retreat to coyote territory borders(Sargeant et al. 1987, Harrison et al. 1989,Sargeant and Allen 1989).

It has been suggested that sizedifferences between similar carnivores couldlead to coexistence (Rosenzweig 1966).However, this probably requires largerdifferences in size and food niche than thosebetween arctic and red foxes. For example,despite high mortalities from coyote preda-tion, kit and swift foxes seem able to coexistwith coyotes by having a somewhat differentfood niche and by using dens for protection(White et al. 1994, Kitchen et al. 1999).Interactions between predators of verydifferent sizes and/or niche requirements, asin the kit/swift fox-coyote case, may besimilar to those between predators and herbi-vore prey. Predators would then be able toco-exist as long as there is no hyperpredationon the “predator prey” species. The lattermay occur if alternative prey sustain suchlarge top-predator populations that predationpressure on the “predator prey” becomesextremely high (Roemer et al. 2001). Aspredator species become more similar in sizeand niche requirements, interactions betweenthem will also involve an increasing amountof competition. In the end, interferencecompetition over territories could have alarge impact on the inferior species, whichwould have little or no habitat left at itsdisposal.

Population effects

Chirkova (1970) and Linnell et al. (1999)argued that the presence of red foxes couldnot have a large impact on arctic fox popula-tions, since red fox densities are low in themountain tundra and there are many vacantdens. However, we have shown that onereproducing red fox can exclude arctic foxesfrom a relatively large area containing manydens (Paper IV). Thus, red fox reproductionin mountain tundra dens decrease the amountof habitat available to arctic foxes. Suchreductions in potential habitat may havelong-term consequences for the arctic foxpopulation.

We simulated arctic and red fox popula-tion dynamics in the Vindelfjällen area for 15years, using a spatially explicit populationmodel which incorporated arctic fox avoid-ance of dens located less than 8 km fromdens inhabited by red foxes (Paper V).Lemming and vole population phases duringthe simulated time period corresponded tothose observed in Vindelfjällen in 1986-2000. There was an increasing trend inmodelled arctic fox population under allo-patric conditions, but the population decreas-ed when interference competition with redfoxes was included in the model (Fig. 9). Inthe latter case, the model recreated theobserved population dynamics of arctic foxesin Vindelfjällen relatively well (Fig. 10,Paper V). Thus, these results indicate that arelatively small number of red foxes couldhave a large impact on the arctic foxpopulation.

Modelled numbers of red foxesreproducing in dens above the tree line wassimilar to that in the core area in Vindel-fjällen, i.e. not exceeding two occupied densper year (Paper V). It may seem strange thata small numbers of red foxes could have suchdrastic effect on the arctic fox population inan area encompassing 450 km2 and almost 30dens. However, if two red fox dens would be“strategically” located in the core area, in thecentral parts of the eastern and western core

20

area respectively, only two dens would besituated farther than 8 km from either of thered fox dens. Even limited red fox presencemay therefore exclude arctic foxes from largeparts of their habitat and in that case, arcticfoxes would no longer be able to respond toincreased lemming availability by increasedestablishment and reproduction.

Sterilised red foxes were successfullyused as biological control agents for arcticfox populations on two islands (Bailey 1992).Arctic foxes were still present on the islandssix months after red foxes were introduced,indicating that direct killings of adult arcticfoxes were rare, but all arctic foxesdisappeared in the following years while redfoxes remained. Possibly, the arctic foxeswere excluded from the best dens andfeeding beaches. In combination withpredation on arctic fox cubs, this may haveled to their extinction (Bailey 1992). InNorway, the mountain area with the largestand most stable arctic fox population alsoappears to have the least red fox activity(Frafjord 2003). These examples demonstratethe importance of “competition refuges” forinferior competitors (Durant 1998). Then,could high altitude dens function ascompetition refuges for arctic foxes inFennoscandia?

The Swedish arctic fox populationdecreased from an estimated 180 breedingadults (90 litters) during the lemming peak in1982, to a total of 18 breeding adults (9litters) during the lemming peak in 2001(Angerbjörn et al. 2002). The absence oflemming peaks has therefore been consideredas the most acute problem for the arctic foxpopulation. Food shortage has a negativeeffect on the number of foxes that establishand try to reproduce, litter sizes and juvenilesurvival in summer (Angerbjörn et al. 1991,Tannerfeldt et al. 1994, Angerbjörn et al.1995, Paper II). The declining populationtrend and our model results (Paper V)indicate that high altitude areas were of toolow quality to function as competitionrefuges for Swedish arctic foxes after 1982.

0

4

8

12

16

20

1 3 5 7 9 11 13 15

Year

Occ

upie

d d

ens

Arctic foxes without red foxes

Arctic foxes with red foxes

Maximum number of red foxes

Fig. 9 The simulated effect of red fox competition on arctic fox den occupancy (from Paper V).

0

4

8

12

16

20

1985 1987 1989 1991 1993 1995 1997 1999

Occ

up

ied

de

ns

model median

field data

Fig. 10 A comparison between den occupancy in theobserved arctic fox population and the predictions ofthe model. The dotted lines indicate the lower andupper quartiles (from Paper V).

21

On the contrary, the arctic foxpopulation was relatively small during theperiod with lemming peaks in 1960-1982, butpeak population densities appear to havebeen relatively stable (Angerbjörn et al.1995). The higher amplitude of the lemmingcycle should have had limited effect on thered fox population as red fox presence inmountain tundra habitat primarily depends onthe status of boreal vole populations. Hence,provided that red fox presence in mountaintundra habitat has been relatively constantsince 1960, the impact of red fox competitionin 1960-1982 may have been ameliorated bytwo factors:

1) A combination of low lemmingavailability and red fox presence mightrefrain arctic foxes from establishment inotherwise suitable habitat, while highlemming abundances may increaseoverall habitat quality to acceptablelevels. Arctic foxes have overlappinghome ranges in habitats of highproductivity, but are strictly territorial inhabitats of low quality (Eide et al.2002a). Interspecific overlap might beexplained by arctic foxes paying the costof predation by using dens close to redfoxes during lemming peak years, sincethe alternative may be not to breed at all.

2) A lemming cycle of normal amplitudemay increase habitat quality at highaltitudes. Hence, arctic foxes in highaltitude dens might have had more andlarger litters before 1982.

Thus, it is possible that the total reproductiveoutput of arctic foxes, despite red fox com-petition, was higher when lemming popula-tions peaked regularly. This might have keptthe arctic fox population from decliningbefore 1982. Under these circumstances, highaltitude areas might have functioned ascompetition refuges for a small arctic foxpopulation.

CONSERVATION IMPLICATIONS

The arctic fox in Fennoscandia is endangeredand the remaining population is fragmented(Angerbjörn et al. 2003b). The low popula-tion size is in itself a threat since small,isolated populations are more likely to goextinct (Loison et al. 2001). There has prob-ably been some gene flow between theFennoscandian population and the largeSiberian population in the 1900’s, but there isa risk of inbreeding in the small Fenno-scandian population (Dalén et al. 2002).Conservation biology is divided into twomajor fields of research, the small-populationparadigm and the declining-populationparadigm. The former deals with the persis-tence of small populations, such as problemsrelated to inbreeding, and the latter withreasons why populations decline (Caughley1994). However, the ultimate cause of anextinction is usually that the habitat is lost oraltered, and problems concerning smallpopulations come into play only during thelast generations when a population already isdoomed (Thomas 1994). To turn a decliningpopulation trend, conservation actions todecrease inbreeding may be necessary, but itis also vital to look beyond these and takemeasures to restore habitat.

The results of this thesis indicate thatinterference competition with red foxes hashampered a recovery of the arctic foxpopulation after the initial population decline.Competitive exclusion by red foxes hascaused a substantial reduction in arctic foxhabitat. Further, red foxes have taken overthe most productive areas and remainingarctic fox habitat is of such low quality that itis uncertain whether it can maintain even asmall arctic fox population.

In 1998-2002, there was a joint Swedish-Finnish conservation project (SEFALO) topreserve the arctic fox. The aim was toincrease population size through supple-mental feeding and red fox control. Red foxcontrol, i.e. culling of red foxes in importantarctic fox habitats, was intended to facilitate

22

establishment of arctic foxes and decreasepredation on arctic fox cubs. Supplementalfeeding aimed to increase establishment andlead to larger litter sizes and higher survivalof cubs (Angerbjörn et al. 2002).

In Vindelfjällen, we tried to perform redfox control by hunting at carrion, but themethod did not succeed (Paper IV). Instead,rangers began hunting from snow mobiles,which is more efficient, in the county ofJämtland in 2001. They selected a relativelysmall area of mountain tundra and aimed tokeep it free from red foxes. Snow-tracking inspring 2002 indicated that there were fewerred foxes within the hunting area than outsideit. There were also fewer red foxes than inVindelfjällen. There were only four arctic foxlitters in Sweden the following summer, butall of them were found within the red foxcontrol area in Jämtland (Angerbjörn et al.2002). Comparisons with previous den pre-ferences of arctic foxes in Jämtland indicatethat the arctic foxes selected dens within thered fox control area rather than previouslyhigh-ranked dens (Sverker Dalén, unpublish-ed data). The lemming population peak in2001 extended through winter 2002, butcrashed in late spring. Food availability wasconsequently low in summer and the arcticfox cubs appeared to be highly dependent onsupplemental feeding. However, establish-ment seemed to have been favoured by redfox control, as feeding did not begin until thedens had become inhabited.

Six dens in Vindelfjällen were inhabitedby arctic foxes in spring 2002. In summer, allbut one den was abandoned and no arcticfoxes were born. Red foxes were observedvisiting two of the abandoned dens and athird den was situated close to a den with ared fox litter. We found red fox litters in sixarctic fox dens in our study area. Thus, redfox competition probably contributed to thecomplete reproductive failure of the arcticfoxes in Vindelfjällen (Angerbjörn et al.2002).

The arctic fox was one of the earliestpost-glacial colonisers of Fennoscandia. The

increase in the red fox population is likely tobe, by large, an effect of human inter-ventions; the green house effect might havecaused a warming of the climate, hunting haslead to a decrease in populations of largepredators, and clear-cutting of forest favour-ed the field vole and thus indirectly the redfox. In the long run, the future of theFennoscandian arctic fox population maydepend on human activities also outside themountain tundra.

In a shorter perspective, the relativesuccess of the conservation efforts in Jämt-land indicates that it is possible to reverse thedeclining trend in the arctic fox population.An extensive conservation programme cantherefore be used as a temporary measure toovercome the most acute situation. If thelemming peak in 2001 was the first sign thatwe are entering a new period of regularlemming peaks, we may have a fair chanceto, at least, get back to the relatively stablesituation we had in 1960-1980.

Perhaps we owe the arctic fox that chance.

CONCLUSIONS

The distribution of the Fennoscandian arcticfox population changed in the 20th century,parallel to an increase in the red foxpopulation. Arctic and red foxes have similarvirtual food niches and they should thereforecompete for the same territories. Arctic foxesare inferior to red foxes in direct conflicts,and they are consequently ousted from therelatively productive low altitudes areaspreferred by red foxes. There were regularlemming peaks in 1960-1982 and highaltitude habitats may then have functioned ascompetition refuges for a small arctic foxpopulation. However, high altitude habitat isnot productive enough to support a viablearctic fox population in the absence oflemming peaks, such as the period between1983 and 2000. Under such circumstances,

23

the arctic fox might not have a realised nichein Sweden and the population was very closeto extinction in spring 2001. In the summerof 2001, the mountain tundra experienced itsfirst lemming peak in almost 20 years and 91arctic fox cubs were born in Sweden. Thishas given the arctic fox population somebreathing space, but the situation is still graveand the Fennoscandian arctic fox might goextinct within the near future. However,experience from SEFALO, a Swedish-Finnish project to preserve the arctic fox,indicates that a combination of red foxcontrol and feeding of arctic foxes mayimprove the situation.

REFERENCES

Abrams P. 1980. Some comments on measuring nicheoverlap. Ecology 61: 44-49.

Andersson M, Erlinge S. 1977. Influence of predationon rodent populations. Oikos 29: 591-597.

Angelstam P, Lindström E, Widén P. 1985.Synchronous short-term population fluctuations ofsome birds and mammals in Fennoscandia –occurrence and distribution. Holarctic Ecol. 8: 285-298.

Angerbjörn A, ArvidSon B, Norén E, Strömgren L.1991. The effect of winter food on reproduction inthe arctic fox, Alopex lagopus: A field experiment.J. Anim. Ecol. 60: 705-714.

Angerbjörn A, Tannerfeldt M, Bjärvall A, Ericson M,From J, Norén E. 1995. Dynamics of the arctic foxpopulation in Sweden. Ann. Zool. Fenn. 32: 55-68.

Angerbjörn A, Ströman J, Becker D. 1997. Homerange patterns in arctic foxes in Sweden. J. Wildl.Res. 2: 9-14.

Angerbjörn A, Tannerfeldt M, Erlinge S. 1999.Predator - prey relations: arctic foxes and lem-mings. J. Anim. Ecol. 68: 34-49.

Angerbjörn A, Tannerfeldt M, Lundberg H. 2001.Geographical and temporal patterns of lemmingpopulation dynamics in Fennoscandia. Ecography24: 298-308

Angerbjörn A, Tannerfeldt M, Henttonen H, ElmhagenB, Dalén L. 2002. Bevarande av fjällräv Alopexlagopus i Sverige och Finland [Conservation of thearctic fox Alopex lagopus in Sweden and Finland.In Swedish]. Final Report. EU LIFE-Nature.

Angerbjörn A, Hersteinsson P, Tannerfeldt M. 2003a.Consequences of resource predictability in thearctic fox two life history strategies. In: DWMacdonald & C Sillero-Zubiri (eds.). The Biology& Conservation of Wild Canids. Oxford UniversityPress. In press.

Angerbjörn A, Hersteinsson P, Tannerfeldt, M. 2003b.Arctic fox. In: D W Macdonald & C Sillero-Zubiri(eds.). Canid Action Plan. IUCN. In press.

Anthony RM. 1996. Den use by arctic foxes (Alopexlagopus) in a subarctic region of western Alaska.Can. J. Zool. 74: 627-631.

Anthony RM, Barten NL, Seiser PE. 2000. Foods ofarctic foxes (Alopex lagopus) during winter andspring in western Alaska. J. Mamm. 81: 820-828.

Audet AM, Robbins CB, Larivière S. 2002. Alopexlagopus. Mamm. Species 713: 1-10.

Bailey EP. 1992. Red foxes, Vulpes vulpes, asbiological control agents for introduced arcticfoxes, Alopex lagopus, on Alaskan islands. Can.Field-Nat. 106: 200-205.

24

Baker PJ, Robertson CPJ, Funk SM, Harris S. 1998.Potential fitness benefits of group living in the redfox, Vulpes vulpes. Anim. Behaviour 56: 1411-1424.

Bantle JL, Alisauskas RT. 1998. Spatial and temporalpatterns in arctic fox diets at a large goose colony.Arctic 51: 231-236.

Birks JDS, Penford N. 1990. Observations on theecology of arctic foxes Alopex lagopus in Eqalum-miut Nunaat, West Greenland. Medd. Grønl. Biosci.32: 1-2.

Braestrup FW. 1941. A study of the arctic fox inGreenland. Immigrations, fluctuations in numbersbased mainly on trading statistics. Medd. Grønl.Biosci. 131: 1-101.

Brown JS, Laundré JW, Gurung M. 1999. The ecologyof fear: optimal foraging, game theory, and trophicinteractions. J. Mamm. 80: 385-399.

Caughley G. 1994. Directions in conservation biology.J. Anim. Ecol. 63: 215-244.

Chesemore DL. 1968. Notes on the food habits ofarctic foxes in northern Alaska. Can. J. Zool. 46:1227-1130.

Chirkova AF. 1968. The relationship between arcticfox and red fox in the far north. Problems of theNorth 11: 129-131.

Clode D, Macdonald DW. 1995. Evidence for foodcompetition between mink (Mustela vision) andotter (Lutra lutra) on Scottish islands. J. Zool. 237:435-444.

Colwell RK, Futuyma DJ. 1970. On the measurementof niche breadth and overlap. Ecology 52: 567-576.

Connell JH. 1983. On the prevalence and relativeimportance of interspecific competition: evidencefrom field experiments. Am. Nat. 122: 661-696.

Dalén L, Götherström A, Tannerfeldt M, AngerbjörnA. 2002. Is the endangered Fennoscandian arcticfox (Alopex lagopus) population geneticallyisolated. Biol. Cons. 105: 171-178.

Dalerum F, Angerbjörn A. 2000. Arctic fox (Alopexlagopus) diet in Karupelv valley, East Greenland,during a summer with low lemming diversity.Arctic 53: 1-8.

Danell K, Hörnfeldt B. 1987. Numerical responses bypopulations of red fox and mountain hare during anoutbreak of sarcoptic mange. Oecologia 73: 533-536.

Durant SM. 1998. Competition refugees andcoexistence: an example from Serengeti carnivores.J. Anim. Ecol. 67: 370-386.

Eberhardt LE, Garrott RA, Hanson WC. 1983. Den useby arctic foxes in northern Alaska. J. Mamm. 64:97-102.

Eide NE, Jepsen JU, Prestrud P. 2002a. Spatialorganisation of reproductive arctic foxes (Alopex

lagopus) – responses to changes in spatial andtemporal distributions of prey. In: Eide, N.E. 2002.Spatial ecology of arctic foxes: Relations toresource distribution, and spatiotemporal dynamicsin prey abundance. DPhil. thesis, AgriculturalUniversity of Norway. Norway.

Eide NE, Eid PM, Prestrud P, Swenson JE. 2002b.Summer diets of arctic foc (Alopex lagopus) acrossan Arctic landscape. In: Eide, N.E. 2002. Spatialecology of arctic foxes: Relations to resourcedistribution, and spatiotemporal dynamics in preyabundance. DPhil. thesis, Agricultural University ofNorway. Norway.

Elton CS. 1924. Periodic fluctuations in the numbers ofanimals: their causes and effects. Brit. Journ. Exper.Biol. Edinburgh 2: 119-163.

Englund J. 1965. Studies on food ecology of the redfox (Vulpes vulpes) in Sweden. [In Swedish withEnglish summary] Swedish Wildlife (Viltrevy) 3:377-485.

Englund J. 1970. Some aspects of reproduction andmortality rates in Swedish foxes (Vulpes vulpes)1961-63 and 1966-69. [In Swedish with Englishsummary] Swedish Wildlife (Viltrevy) 8: 1-82.

Ewer RF. 1973. The carnivores. Cornell Univ Press,New York.

Fedriani JM, Palomares F, Delibes M. 1999. Nicherelations among three sympatric Mediterraneancarnivores. Oecologia 121: 138-148.

Ferrari N, Weber JM. 1995. Influence of the abundanceof food resources on the feeding habits of the redfox, Vulpes vulpes, in western Switzerland. J. Zool.236: 117-129.

Frafjord K. 1993a. Reproductive effort in the arctic foxAlopex lagopus: A review. Norwegian Journal ofAgricultural Sciences 7: 301-309.

Frafjord K. 1993b. Food habits of arctic foxes (Alopexlagopus) on the western coast of Svalbard. Arctic46: 49-54.

Frafjord K. 1995. Summer food habits of arctic foxesin the alpine region of southern Scandinavia, with anote on sympatric red foxes. Ann. Zool. Fenn. 32:111-116.

Frafjord K. 2000. Do arctic and red foxes compete forfood? Z. Säugetierk. 65: 350-359.

Frafjord K. 2003. Ecology and use of arctic fox Alopexlagopus dens in Norway: tradition overtaken byinterspecific competition? Biol. Cons. 111: 445-453.

Frafjord K, Becker D, Angerbjörn A. 1989.Interactions between arctic and red foxes inScandinavia - predation and aggression. Arctic 42:354-356.

25

Frafjord K, Hufthammer AK. 1994. Subfossil recordsof the arctic fox (Alopex lagopus) compared to itspresent distribution in Norway. Arctic 47: 65-68.

Fuglei E, Øritsland NA. 1999. Seasonal trends in bodymass, food intake and resting metabolic rate, andinduction of metabolic depression in arctic foxes(Alopex lagopus) at Svalbard. J. Comp. Physiol. B.169: 361-369.

Fuller TK, Keith LB. 1981. Non-overlapping ranges ofcoyotes and wolves in northeastern Alberta. J.Mamm. 62: 403-405.

Garrott RA, Eberhardt LE, Hanson WC. 1983. Summerfood habits of juvenile arctic foxes in northernAlaska. J. Wildl. Manage. 47: 540-545.

Geffen E, Mercure A, Girman DJ, Macdonald DW,Wayne RK. 1992. Phylogenetic relationships of thefox-like canids: mitochondrial DNA restrictionfragment, site and cytochrome b sequence analyses.J. Zool. 228: 27-39.

Goszczynski J. 1989a. Population dynamics of the redfox in Central Poland. Acta Theriol. 34:141-154.

Goszczynski J. 1989b. Spatial distribution of red foxesVulpes vulpes in winter. Acta Theriol. 34: 361-372.

Hanski I, Turchin P, Korpimäki E, Henttonen H. 1993.Population oscillations of boreal rodents: regulationby mustelid predators leads to chaos. Nature 364:232-235.

Hansson L, Henttonen H. 1985. Gradients in densityvariations of small rodents: the importance oflatitude and snow cover. Oecologia 67: 394-402.

Harestad AS, Bunnell FL. 1979. Home range and bodyweight – a reevaluation. Ecology 60: 389-402.

Harrison DJ, Bissonette JA, Sherburne JA. 1989.Spatial relationships between coyotes and red foxesin eastern Maine. J. Wildl. Manage. 53: 181-185.

Hersteinsson P. 1989. Population genetics and ecologyof different colour morphs of arctic foxes Alopexlagopus in Iceland. Finnish Game Res. 46: 64-78.

Hersteinsson P, Macdonald DW. 1982. Somecomparisons between red and arctic foxes, Vulpesvulpes and Alopex lagopus, as revealed by radiotracking. Symp. Zool. Soc. Lond. 49: 259-289.

Hersteinsson P, Angerbjörn A, Frafjord K, KaikusaloA. 1989. The arctic fox in Fennoscandia andIceland: management problems. Biol. Cons. 49: 67-81.

Hersteinsson P, Macdonald DW. 1992. Interspecificcompetition and the geographical distribution of redand arctic foxes Vulpes vulpes and Alopex lagopus.Oikos 64: 505-515.

Hersteinsson P, Macdonald DW. 1996. Diet of arcticfoxes (Alopex lagopus) in Iceland. J. Zool. 240:457-474.

Heske EJ, Steen H. 1993. Interspecific interactions andmicrohabitat use in a Norwegian low alpine rodent

assemblage. In: Stenseth, N. C. and Ims, R. A.(eds.). Biology of lemmings. Academic Press,London.

Hutchinson GE. 1957. Concluding remarks. – Coldspring Harbor Symposium on Quantitative Biology22: 415-427. Cited in: Begon, M., Harper, J. L. andTownsend, C. R. 1990. Ecology. Individuals,populations and communities, 2nd edition.Blackwell Scientific Publications. Massachusetts.

Hörnfeldt B. 1994. Delayed density dependence as adeterminant of vole cycles. Ecology 75: 791-806.

Hörnfeldt B. 1998. Miljöövervakningen visar påminskande sorkstammar [Environmental super-vision indicates decreasing vole populations. InSwedish]. Fauna och Flora 93: 137-144.

Jepsen JU, Eide NE, Prestrud P, Jacobsen L. B. 2002.The importance of prey distribution in habitat useby arctic foxes (Alopex lagopus). Can. J. Zool. 80:418-429.

Jones DM, Theberge JB. 1982. Summer home rangeand habitat utilisation of the red fox (Vulpes vulpes)in a tundra habitat, northwest British Columbia.Can. J. Zool. 60: 807-812.

Kaikusalo A, Angerbjörn A. 1995. The arctic foxpopulation in Finnish Lapland during 30 years,1964-93. Ann. Zool. Fenn. 32: 69-77.

Kapel CMO. 1999. Diet of foxes (Alopex lagopus) inGreenland. Arctic 52: 289-293.

Kitchen AM, Gese EM, Schauster ER. 1999. Resourcepartitioning between coyotes and swift foxes:space, time and diet. Can. J. Zool. 77: 1645-1656.

Kleiman DG, Malcolm JR. 1981. The evolution ofmale parental investment in mammals. In:Gubernick DJ, Klopfer PH (eds). Parental care inmammals. Plenum Publishing Corporation.

Korhonen H, Alasuutari S, Mäkinen A, Niemelä P.1997. Inter- and intraspecific competition betweenthe fox species Alopex lagopus and Vulpes vulpes:an evaluation trial under penned conditions. PolarBiology 17: 330-336.

Landa A, Strand O, Linnell JDC, Skogland T. 1998.Home-range sizes and altitude selection for arcticfoxes and wolverines in an alpine environment.Can. J. Zool. 76: 448-457.

Larivière S, Pasitschniak-Arts M. 1996. Vulpes vulpes.Mamm. Species 537: 1-11.

Lindström E. 1982. Population ecology of the red fox(Vulpes vulpes) in relation to food supply. DPhil.thesis, Stockholm University. Sweden.

Lindström E. 1988. Reproductive effort in the red fox,Vulpes vulpes, and future supply of a fluctuatingprey. Oikos 52: 115-119.

Lindström E. 1989a. The role of medium-sizedcarnivores in the Nordic boreal forest. Finn. GameRes. 46: 53-63.

26

Lindström E. 1989b. Food limitation and socialregulation in a red fox population. Holarctic Ecol.12: 70-79.

Lindström E, Mörner T. 1985. The spreading ofsarcoptic mange among Swedish red foxes (Vulpesvulpes L) in relation to fox population dynamics.Rev. Ecol. (Terre Vie) 40: 211-216.

Lindström ER, Andrén H, Angelstam P, Cederlund G,Hörnfeldt B, Jäderberg L, Lemnell P-A, MartinssonB, Sköld K, Swenson JE. 1994. Disease reveals thepredator: Sarcoptic mange, red fox predation, andprey populations. Ecology 75: 1042-1049.

Linhart SB, Robinson WBJ. 1972. Some relativecarnivore densities in areas under sustained coyotecontrol. J. Mamm. 53: 880-884.

Linnell JDC, Strand O, Landa A. 1999a. Use of densby red Vulpes vulpes and arctic Alopex lagopusfoxes in alpine environments: Can interspecificcompetition explain the non-recovery of Norwegianarctic fox populations? Wildl. Biol. 5: 167-176.

Linnell JDC, Strand O, Loison A, Solberg EJ, JordhøyP. 1999b. A future plan for arctic foxes in Norway?A status report and action plan. [In Norwegian withEnglish summary] NINA Oppdragsmelding 575: 1-37.

Loison A, Strand O, Linnell J. 2001. Effect of temporalvariation in reproduction on models of populationviability: a case study for remnant arctic fox (Alo-pex lagopus) populations in Scandinavia. Biol.Cons. 97: 347-359.

Lönnberg E. 1927. Fjällrävsstammen i Sverige 1926.[The arctic fox population in Sweden 1926. InSwedish.] K. Sv. Vetenskapsakad. Skr.Naturskyddsärenden No 7. Royal SwedishAcademy of Sciences, Uppsala, Sweden.

Macdonald DW. 1979. ‘Helpers’ in fox society. Nature282: 69-71.

Macdonald DW. 1983. The ecology of carnivore socialbehaviour. Nature 301: 379-384.

Macpherson AH. 1964. A northward range extensionof the red fox in the eastern Canadian arctic. J.Mamm. 45: 138-140.

Macpherson AH. 1969. The dynamics of Canadianarctic fox populations. Can. Wildl. Serv. Rep. Ser.8: 1-49.

Major JT, Sherburne JA. 1987. Interspecificrelationships of coyotes, bobcats, and red foxes inwestern Maine. J. Wildl. Manage. 51: 606-616.

Marsh DB. 1938. The influx of the red fox and itscolour phases into the Barren Lands. Can. Field.Nat. 52: 60-61.

Meia JS, Weber JM. 1995. Home ranges andmovements of red foxes in central Europe: stabilitydespite environmental changes. Can. J. Zool. 73:1960-1966.

Mercure A, Ralls K, Koepfli KP, Wayne RK. 1993.Genetic subdivisions among small canids: mito-chondrial DNA differentiation of swift, kit andarctic foxes. Evolution 47: 1313-1328.

Mörner T. 1988. Botade från skabb. [Cured frommange. In Swedish] Svensk Jakt 126: 1086-1087.

Oksanen L, Oksanen T. 1992. Long-term microtinedynamics in north Fennoscandian tundra: the volecycle and the lemming chaos. Ecography 15: 226-236.

Oksanen T, Schneider M, Rammul Ü, Hambäck P,Aunapuu M. 1999. Population fluctuations of volesin North Fennoscandian tundra: contrastingdynamics in adjacent areas with different habitatcomposition. Oikos 86: 463-478.

O’Mahony DO, Lambin X, MacKinnon JL, Coles CF.1999. Fox predation on cyclic field volepopulations in Britain. Ecography 22: 575-581.

Palomares F, Ferreras P, Fedriani JM, Delibes M.1996. Spatial relationships between Iberian lynxand other carnivores in an area of south-westernSpain. J. Appl. Ecol. 33: 5-13.

Palomares, E, Caro, T. M. 1999. Interspecific killingamong mammalian carnivores. Am. Nat. 153: 492-508.

Park T. 1962. Beetles, competition and populations.Science 138: 1369-1375.

Polis GA, Meyers CA, Holt RD. 1989. The ecologyand evolution of intraguild predation: potentialcompetitors that eat each other. Ann. Rev. Ecol.Syst. 20: 297-330.

Prestrud P. 1991. Adaptations by the arctic fox (Alopexlagopus) to the polar winter. Arctic 44: 132-138.

Prestrud P. 1992. Denning and home-range charac-teristics of breeding arctic foxes in Svalbard. Can.J. Zool. 70: 1276-1283.

Ralls K, White PJ. 1995. Predation on San Joaquin kitfoxes by larger canids. J. Mamm. 76: 723-729.

Reid DG, Krebs CJ, Kenney AJ. 1997. Patterns ofpredation on noncyclic lemmings. Ecol. Monogr.67: 89-108.

Roemer GW, Coonan TJ, Garcelon DK, Bascompte J,Laughrin L. 2001. Feral pigs facilitate hyper-predation by golden eagles and indirectly cause thedecline of the island fox. Animal Cons. 4: 307-318.

Rosenzweig ML. 1966. Community structure insympatric carnivora.