Modelling the expansion of a grey squirrel population: implications for

squirrel control

Clara Tattoni1, Damiano G. Preatoni1, Peter W.W. Lurz2, Steven P. Rushton2, GuidoTosi3, Sandro Bertolino3, Adriano Martinoli1,* & Lucas A. Wauters11Dipartimento Ambiente-Salute-Sicurezza, Universita degli Studi dell’Insubria, Varese, Via J.H. Dunant 3,I-21100, Varese (VA), Italy; 2IRES, School of Biology, Devonshire Building, University of Newcastle uponTyne, Newcastle, NE1 7RU, UK; 3DI.VA.P.R.A. Entomology and Zoology, University of Turin, I-10095,Grugliasco (TO), Italy; *Author for correspondence (e-mail: adriano.martinoli@unisubria.it)

Received 19 July 2004; accepted in revised form 10 March 2005

Key words: geographic information systems, grey squirrel, interspecific competition, invasive speciescontrol, life history parameters, North Italy, range expansion, red squirrel, spatially explicit populationdynamics models

Abstract

A recently discovered population of the North American grey squirrel (Sciurus carolinensis), introduced toTicino Park, Lombardy (N Italy), is likely to spread into continuous prealpine broadleaf forests ofLombardy and the south of Switzerland. We used Spatially Explicit Population Dynamics Models (SEP-Ms), successfully used to predict the spread of grey squirrels in England and Piedmont, Italy, to examinethe effects of different control scenarios on grey squirrel expansion in a 20000 km2 area around Ticino Park.Without control, grey squirrels will invade Switzerland within the next two decades, and, concomitantly,the size and distribution of local populations of native red squirrels will be reduced. Simulating differentgrey squirrel control or removal scenarios suggests that: (i) efficient control is possible and mainly deter-mined by the spatial distribution and woodland size of the ‘target’ control areas; and (ii) immediate actionsmust be taken, since delay in grey squirrel control will result in the population increasing and spreading,which makes the problems of successful containment more difficult. Model scenarios were based on surveysthat may underestimate the real distribution range and current population size of grey squirrels. Therefore,a combination of hair–tube monitoring and a public participation survey to detect grey squirrel presence,which may also help to increase public awareness, is recommended. Successful containment of further greysquirrel spread will require local co-operation between Italian and Swiss authorities involved in wildlifemanagement.

Abbreviations:GIS – geographic information system; SEPM – spatially explicit population dynamics model

Introduction

In many ecosystems, the introduction of invasive,alien species is a major threat for the mainte-nance of biodiversity (e.g., Pimm et al. 1995). It isclaimed to be the second most important reasonfor loss of biodiversity, after the destruction and

fragmentation of natural habitats, causing theextinction or decline of native species in theregions where they have been introduced(Vitousek et al. 1996; Williamson 1996; Macket al. 2000; IUCN 2000). Alien species interfere withthe native fauna by different ecological processes:predation, interspecific competition or acting as

Biological Invasions (2006) 8:1605–1619 � Springer 2006

DOI 10.1007/s10530-005-3503-z

vector or reservoir of (new) diseases (Gurnelland Pepper 1993; Ragg et al. 1995; Alterio et al.,1998; Sainsbury et al. 2000; Byrom et al.2002; Tompkins et al. 2002; Gurnell et al. 2004).Introduced species not only present a seriousecological problem, but also a socio-economical,and political one (Gosling and Baker 1989; Shineet al. 2000; Barr et al. 2002; Bertolino andGenovesi 2003). Many alien species cause directeconomic damage to human activities (farming,forestry, zoo-technology, disease risk), but never-theless their control often becomes an emotionalissue, in particular with mammal pest species(Lockwood 1997; White and Whiting 2000; Barret al. 2002; Bertolino and Genovesi 2003).

A well documented case of competition by aninvasive alien species is the wide-scale replace-ment of the native Eurasian red squirrel (Sciurusvulgaris) by the introduced eastern grey squirrel(Sciurus carolinensis) on the British Isles and inparts of northern Italy. The rapid increase of thegrey squirrel’s distribution range, coincided witha dramatic decline of the range of the native redsquirrel (Gurnell and Pepper 1993; Wauters et al.1997a), and the grey squirrel has now replacedthe red squirrel over much of its range in GreatBritain, Ireland, and in the fragmented land-scapes it currently occupies in Piedmont, north-ern Italy (Gurnell and Pepper 1993; Wauterset al. 1997a; O Teangana et al. 2000; Lurz et al.2001; Bertolino and Genovesi 2003). An essentialparameter to assess the risk of extinction of redsquirrel populations, and plan effective manage-ment strategies for controlling the invasive spe-cies, is the rate at which replacement will occur,which mainly depends on factors that facilitatethe spread of the alien species (reviewed by Lever1994). For grey squirrels major factors are land-scape structure (connectivity between good habi-tats) and abundant food supplies. Logisticregression models have been useful in the predic-tion of squirrel presence/absence in relation tovariables such as woodland size, quality anddegree of isolation (distance from a large, sourcearea) at a given moment in time (Fitzgibbon1993; Van Apeldoorn et al. 1994; Wauters et al.1997b; Delin and Andren 1999; Rodriguez andAndren 1999; Verbeylen et al. 2003a). Their pre-dictive ability may be further improved by con-sidering the landscape matrix to quantify the

degree of isolation, by applying ‘Cost Distance’models(a combination of distance and resistanceof the landscape) with different resistances fordifferent landscape types (Verbeylen et al.2003b). However, these models are static andcannot be used to predict future spread, distribu-tion and population size of as species that is col-onising new areas, as is the case of introducedinvasive species (Rushton et al. 1997).

One approach to overcoming these problemshas been the use of spatially explicit populationdynamics models (SEPM). These models com-bine spatial variables and species’ life-historytraits. The drawback of these models is that theirperformance is sensitive to the accuracy of theestimates of the life-history parameters input,and, thus, the models can only be used for spe-cies for which detailed data of populationparameters (fecundity, mortality, dispersal dis-tance, density) in different habitat types areavailable (Rushton et al. 1997, 2000). Fortu-nately, many studies on the population dynamicsof red and grey squirrels have produced reliableestimates of these life-history traits under vari-able environmental (habitat) conditions (seeTable 1 for references). Thus, accurate SEPMmodels can be applied to predict current andfuture distribution and population size, theeffects of red and grey squirrel interspecific com-petition and of disease spread within andbetween grey and red squirrel populations(Rushton et al. 1997, 2000; Lurz et al. 2001).

The replacement of red by grey squirrels thatis taking place in Piedmont, northern Italy, doesnot only have serious implications for red squir-rel conservation in Italy, but also for the wholeof Europe (Genovesi 2000; Genovesi andBertolino 2001a, b; Lurz et al. 2001). In effect,political concern about the lack of action inmany countries has been expressed by the Perma-nent Commission of the Bern Convention, whichhas produced several recommendations (n. 57and 77 of 1997) urging countries to eradicatealien invasive species where possible. A specificrecommendation to the Italian authorities (n. 78of 1997) underlines the necessity to eradicategrey squirrels in Italy in order to stop the spe-cies’ further diffusion in Europe, and save the na-tive red squirrel from wide-scale extinction (seeGenovesi and Bertolino 2001a). Moreover, in

1606

Great Britain, the impact of grey squirrels onbiodiversity, not only on red squirrels but alsoon birds and other wildlife, has recently beenconfirmed (Gurnell and Mayle 2003). Until now,no control has taken place, and recently greysquirrel populations have been discovered in

mixed deciduous woodland belts along the Ticinoriver, the natural border between Lombardy(east) and Piedmont (Figure 1) (Fornasari et al.2002; Bertolino and Genovesi 2003). Since thesewoodlands connect with contiguous hardwoodalong the eastern side of Lake Maggiore and into

Table 1. Habitat description and squirrel densities (squirrels/ha) based on field work and observations for the digitised landcover

types (at 250 m resolution). Unsuitable habitats indicated by ‘–’ had a value of zero in the model.

Land cover type Grey squirrel Dmax References Red squirrel Dmax References

Riparian forests 1.0 Bertolino et al. (2003) 0.3 Bertolino et al. (2003)

Oak and oak-hornbeam 5.0 Gurnell (1996a); Kenward

et al. (1998)

0.8 Kenward et al. (1998);

Wauters et al. (2001b)

Hornbeam woodland 2.5 Authors’ estimates 0.4 Authors’ estimates

Black locust-oak woodland 1.0 Authors’ estimates 0.2 Wauters et al. (2001b)

Black locust or other exotics 0.1 Authors’ estimates 0.1 Authors’ estimates

Mixed broadleaf dominated by

chestnut

5.0 Gurnell (1991); Koprowski

(1994); Wauters and

Gurnell (1999)

1.1 Wauters and Lens (1995);

Wauters et al. (2001b)

Mixed deciduous (CORINE) 2.0 Koprowski (1994); Gurnell

et al. (2001b)

0.4 Wauters et al. (2001b);

Authors’ estimates

Beech 2.0 Gurnell (1991) 0.4 Cagnin et al. (2000);

Wauters and Lens (1995)

Maple-ash-lime 5.0 Authors’ estimates 1.1 Authors’ estimates

Conifers and mixed broadleaf 1.5 Gurnell (1991); Gurnell

(1996a); Gurnell etal.

(2001b)

1.3 Wauters and Lens (1995)

Scots pine 0.3 Smith and Gurnell (1997);

Authors’ estimates

0.5 Authors’ estimates

Norway spruce 0.2 Smith and Gurnell (1997);

Authors’ estimates

0.35 Wauters et al. (2000);

Authors’ estimates

Poplar plantations 0.3 Bertolino et al. (2003) 0.1 Wauters et al. (1997a);

Wauters et al. (1997b);

Authors’ estimates

Pastures (CORINE) 0.2 Authors’ estimates 0.1 Authors’ estimates

Alder 0.1 Authors’ estimates 0.1 Authors’ estimates

Cliffs, rock slides, riverbeds – –

Scrub – –

Herbaceous vegetation – –

Mixed agriculture 0.013 Bertolino et al. (2003); Lurz

et al. (2001)

0.005 Bertolino et al. (2003)

Maize, wheat 0.013 Bertolino et al. (2003); Lurz

et al. (2001)

0.005 Bertolino et al. (2003)

Rice – –

Permanent grassland 0.013 Authors’ estimates 0.005 Authors’ estimates

Vineyards, orchards and

hazelnut woods

1.0 Authors’ estimates 0.2 Authors’ estimates

Urban areas 0.013 Authors’ estimates;

Koprowski (1994)

0.013 Wauters and Gurnell (1999)

Urban parks 1.0 Authors’ estimates 0.02 Authors’ estimates

Water bodies – –

Sterile land – –

Scrub, newly planted areas

and uncultivated grassland

– –

Infrastructure – –

1607

Switzerland, immediate monitoring and controlof this population is a priority for the localauthorities responsible for wildlife managementand conservation.

Here, we use a SEPM linked to a GeographicInformation System (GIS) which has been testedon grey squirrels in Britain and Piedmont, Italy(Rushton et al. 1997; Lurz et al. 2001) to: (i)illustrate the potential expansion of this intro-duced species in and beyond the Ticino RegionalPark, Lombardy, and (ii) test the effects of differ-ent control scenarios on the expansion rate,future distribution and population size of thegrey squirrels. We propose effective control mea-sures for the alien species and discuss the impli-cations of our findings in relation to the survivalof the native red squirrel in Europe.

Material and methods

Study species

Recent studies have revealed, at least in part, themechanisms of interspecific competition betweengrey and red squirrels. In broadleaf woodlandswhere they temporarily co-occur, the two species

show a high niche-overlap, basically selecting thesame tree seed species as major food supplies,have similar space use and activity patterns, andgrey squirrels seem to pilfer many of the seedsscatter-hoarded by the congener (Wauters andGurnell 1999; Wauters et al. 2001a, 2002a, b).Moreover, in deciduous woods, grey squirrelsheavily feed on acorns, while red squirrels feedmuch less on this often abundant seed supply,resulting in the introduced species having anadvantage in terms of food-energy available in itshome range (Kenward and Holm 1993; Gurnell1996a; Kenward et al. 1998; Wauters et al.2001a, b, b). In forests dominated by conifers,some degree of niche separation seems to exist,but red squirrels tend to avoid the habitat pat-ches with the highest food availability, occupiedby greys, and are often found at low densities inpoor-quality patches of exotic spruce species(Gurnell 1996b; Wauters et al. 2000; Bryce et al.2002). This interspecific competition for primaryfood resources results in reduced body growth,juvenile recruitment and reproductive success(fewer females with two litters/year) of red squir-rels in woodlands and forests where they co-occur with grey squirrels, compared with red-onlysites of similar habitat composition (Wauters et al.

Figure 1. Study area location (a). Parco Lombardo della Valle del Ticino boundaries (b). Habitat blocks positive to grey squirrel

presence (c).

1608

2001a; Gurnell et al. 2004). It is argued that re-duced breeding output and juvenile recruitmentwill eventually cause density to decline and pushred squirrels to local extinction (Gurnell et al.2004). Finally, at least in the UK, grey squirrelsseem to act as vector for a poxvirus which causesa lethal disease in red squirrels (Sainsbury et al.2000; Tomplins et al. 2002).

Study area and data collection

The Ticino Regional Park in Lombardy, North-Italy, extends for about 908 km2 along the east-side of the Ticino river from where it leaves LakeMaggiore to its confluence with the Po river atPavia (Figure 1). Our study area extended for40 km in all directions from the park boundaries.

The current distribution of grey (and red)squirrels in the park was investigated using hair-tube surveys (Gurnell et al. 2001). Twenty-onetransects of 15 hair-tubes each were placed invarious parts of the park. Hair-tubes were baitedwith sunflower seeds and broken hazelnuts andactivated for 3 months (Fornasari et al. 2002).Every 2 weeks, they were checked for squirrelhairs, and sticky tape containing hairs wasremoved. Hairs were identified as pertaining tored or grey squirrels using a reference collectionand following methods in Gurnell et al. (2001a).

Modelling methodology and approach

The model for simulating the distribution of greysquirrels in the landscape has two main compo-nents. The first is a Geographical InformationSystem (GIS) which stores habitat and animalpopulation information. GRASS (Westerweltet al. 1990) was the GIS used to store and re-trieve habitat information and the model output.This system undertook data manipulation andabstraction and provided input for the secondcomponent, which consisted of programmes sim-ulating the population dynamics of grey squirrelsand their interactions and dispersal within theGIS-held landscape (spatially explicit populationdynamics model, SEPM). The populationdynamics programmes were written in the pro-gramming language ‘C’ and integrated with theGIS component in a UNIX-shell environment. Adetailed description of the original model, used

for investigating the spread of the grey squirreland decline of the red squirrel in East Anglia,England, is given in Rushton et al. (1997). Themodel was applied and tested in Piedmont andthe model is described in Lurz et al. (2001).

The model, as developed by Rushton et al.(1997) has been enhanced adding the capacity toselectively apply species control. The possibilityto supply different control maps to the simula-tion programme allows a more realistic feature:the original SEPM has thus been enabled to testseveral different management scenarios.

Here we extend the same approach to: (i) pre-dict grey squirrel expansion in the Ticino Re-gional Park and surrounding areas, including theItalian boundaries with Switzerland; (ii) predictcontemporary persistence of native red squirrels;and (iii) test the effects of different control strate-gies.

Squirrel habitat suitability

In order to predict grey squirrel range expansion,we obtained land cover data for the regions ofPiedmont and Lombardy in digitised format at250 m resolution (CORINE Land Cover, Com-mission of the European Communities 1993).Habitat types were classified using CORINE-biotopes, and were derived from different sources:the Ticino Regional Park forestry map (10 m res-olution) for areas inside the park that are not inthe Varese province (Lombardy), the vegetationmap of the Varese province (10 m resolution,Tosi and Zilio 2002), and CORINE Land Coverfor the other areas of Lombardy. The Vareseprovince covers the northern part of the parkand surrounding areas, reaching the Swiss bor-der. Land cover and vegetation type data wereedited and a single habitat map with a 250 mresolution was produced. Each 250 m by 250 mcell was characterised by a single habitat type(predominant land cover type, Table 1a). Themajor areas utilised by grey squirrels are broad-leaf and to some extend conifer woodlands andparks, although they will feed on maize in fieldsand on fruit in orchards (Gurnell 1996a, b;Wauters et al. 1997a, b). As a result of theresolution scale of the land cover data, somesmall woodlands (<3.2 ha) suitable for grey squir-rels may not have been recognised as squirrel

1609

habitat, making model predictions of grey squirrelspread more conservative by reducing the totalavailable habitat.

Carrying capacity and life history processes

The population dynamics model is individual-based and predicts the distribution of squirrelsby simulating the life history processes of births,deaths, home range formation and dispersal inyearly time steps. Land cover data (e.g., habitattype, location, area) were classified either asareas of habitat used by the squirrel or areas ofnon-habitat through which animals could movewhen dispersing. Suitable habitats were definedaccording to land cover type and the size, com-position and distribution of habitat blocks wasused to identify blocks in which the two squirrelspecies could be found. The carrying capacity ofeach type of habitat for red and grey squirrelswas calculated on the basis of an estimatedsquirrel density (Table 1). For forested habitatsand general land cover categories we relied onpublished estimates for similar habitats. Whereno published information on squirrel density wasavailable, they were estimated from presence/ab-sence data of the 1996 grey squirrel survey inPiedmont (Wauters et al. 1997a), and from au-thors’ unpublished data.

Each block of suitable habitat within the land-scape was modelled as having separate popula-tions of squirrels. Squirrel population processeswere modelled at the integer level, that is, frac-tional squirrels were not permitted. Each popula-tion had its own dynamics, which interactedthrough the process of dispersal with other habi-tat patches. For each habitat block, in each year,population size was modelled in terms of the netchange due to several factors. These were (i)gains due to recruitment from breeding and

immigration of adults and (ii) losses due to adultand juvenile mortality and emigration. Repro-duction was assumed to occur in two waves, inearly and late spring. Mean litter size was variedin three different fecundity scenarios and mod-elled as a Poisson deviate (Table 2). The percent-age of adult female grey squirrels reproducingonce or twice a year was kept constant at,respectively, 87% and 42%, since we did not findsufficient literature data on variation in the pro-portion of reproducing females in relation tohabitat quality and/or food abundance, while thereproductive rate for red squirrels used in thesimulation were 67% and 26%. Mortality wasassumed to occur after breeding and was mod-elled at the level of the individual as a stochasticprocess and different values were applied to juve-nile and adult life stages (Rushton et al. 1997).Dispersal was modelled as a process occurringonce a year in autumn and the mechanism wasassumed to be saturation dispersal, i.e. occurringas a result of intra-specific competition. Dispersalamongst grey and red squirrels tends to occur inautumn and spring and mainly amongst sub-adult/juveniles, normal dispersal movements areusually within the range of a few kilometres(Thompson 1978; Wauters and Dhondt 1993;Koprowski 1994). Individuals were allowed todisperse to blocks of habitat that were below thecarrying capacity for each species. In addition, ifthere were no suitable habitat blocks availablefor occupation within the maximum dispersaldistance of 10 km, dispersers were assumed todie. As some individuals are likely to move be-yond the 10 km dispersal distance parameterused (Koprowski 1994), model predictions ofgrey squirrel expansion were intentionally conser-vative. A flow diagram illustrating the competi-tion and population dynamics processes in themodel is shown in Figure 2. Squirrel life history

Table 2. Red and grey squirrel life history parameters in relation to annual changes in tree seed crops.

Parameter Species Poor year Good year Mast year

Fecundity Red squirrel 2.00 2.80 3.50

Adult survival Red squirrel 0.55 0.60 0.75

Juvenile survival Red squirrel 0.25 0.30 0.40

Fecundity Grey squirrel 1.50 2.50 3.50

Adult survival Grey squirrel 0.55 0.60 0.65

Juvenile survival Grey squirrel 0.25 0.30 0.40

1610

parameters such as mortality and fecundity varywith annual tree seed crop abundance. We havetherefore simulated a random pattern of 2 poor,2 good and one mast year, and varied the lifehistory parameters accordingly (Table 2).

Running the model

The start distribution of grey squirrels was de-rived from presence/absence data collected fromhair-tube transects. In the survey of 1999, 2000and 2001, 7, 10 and 5 transects respectively hadgrey squirrel hairs (Fornasari et al. 2002) in7–40% of hair-tubes. All habitat blocks contain-ing hair-tubes with signs of grey squirrel presencein at least one of the 3 years were considered tohave a grey squirrel population and were used asthe starting point for the SEPM. Two transectswere omitted (Cossolnovo and BereguardoOvest) since they fell within large habitat blocksthat extended outside the park boundaries andtheir inclusion would have strongly overestimated

the starting distribution and population size.Since grey squirrel density (population size) in2001 was not known, we assumed that, in thisearly colonisation phase, none of the habitatblocks had reached carrying capacity, and thatsquirrels occurred at the same (low) density of0.33/ha in all colonised blocks. This extrapolatedto an initial population size of 150 grey squirrelsin nine habitat blocks (sub-populations, see Fig-ure 1c), this figure is considered realistic becausethe number of squirrels supported by those 9blocks at carrying capacity would be more than1500 individuals.

Model simulations were run for 40 years overa study area of ca. 20000 km2, comprising a40 km ‘buffer zone’ around the park boundaries.The output for each model run in each year wasstored as a presence or absence of grey squirrelpopulations in each habitat block in each year.We then ran the model 10 times over for each setof inputs. The 10 repeat runs provide 10 replicatemaps of the incidence of squirrel populations in

Figure 2. Flow diagrams of the population dynamics (a) and interspecific competition (b) models.

1611

each habitat blocks for each year. For any oneyear a map of the average density can be calcu-lated by dividing the average number of individ-uals by the block size.

Different scenarios of grey squirrel control

The same model was used to predict grey squirreldynamics according to different control schemes.A fixed level of control was introduced as a fur-ther demographic factor inside the model (theothers were birth rate, natural mortality, immi-gration, emigration) affecting the populations insingle habitat blocks. The number of replicatesand the time frame used in the control simula-tions were the same as used in the previous run,i.e., 10 replicates over 40 years, leaving all theother parameters unchanged. We assumed thatthe grey squirrel population would be be man-aged using the so called ‘dispersal control’, i.e. afixed rate of individuals would be removed eachyear while the population was still small andspreading. We tested different schemes in the at-tempt to make a compromise between the effec-tiveness and the (logistic) feasibility of control. Inall the runs, squirrels were removed inside thepark area and only in blocks covered by wood-land, since these were more likely to host a squir-rel population. Moreover, with a total parkwoodland cover of about 20 000 ha and woodlotsize ranging from 0.098 to 865 ha, we judged itimpossible to trap squirrels in all woodlots.Therefore, we created three different control sce-narios selecting different numbers of woodland

blocks in which to trap grey squirrels accordingto different combinations of block size andwoodland type. For each control scenario,removal started in 2005 (no removal in the firstthree years of the simulation) and the effects of aremoval rate of 50% and 80% were tested.

‘Best habitat’ scenarioIn this scenario, grey squirrels were removedonly in good habitat blocks having a carryingcapacity greater than 1/ha and larger than 60 ha,with 18 blocks, covering 2800 ha, occurringacross the whole park’s territory (Figure 3a).

‘Extended control’ scenarioIn this second control scheme we extended thecontrol to habitat blocks with a carrying capacitygreater or equal to 1 grey squirrel/ha (for habitattypes see Table 1a). Control was applied in 17patches, each one larger than 60 ha, for a totaltrapping area of 4600 ha, with larger controlblocks in the northern part of the park comparedto the previous control scheme (Figure 3b).

‘Target-area control’ scenarioTo further test whether the effectiveness of squir-rel management was linked to the spatial patternand size of controlled blocks, this third scenarioused the same woodland habitat categories as inthe ‘extended control’ scenario (medium-highcarrying capacity for grey squirrels), but theblocks selected for control covered a smallertotal area and had an asymmetric distributionwithin the Park (Figure 3c). Grey squirrels were

Figure 3. Habitat blocks (in gray) where the management of grey squirrel take place in each of three simulated scenarios. (a) ‘Best

habitat’ scenario; (b) ‘Extended control’ scenario; (c) ‘Target-area control’ scenario.

1612

trapped on 2600 ha, and most of the 23 habitatblocks where the model simulated grey squirrelremoval were situated in the north, with only afew blocks in the central and southern parts ofthe protected area.

Competition scenarioTo evaluate the effects of grey squirrel expansionon the native red squirrel population, we also rana competition model. This simulation took placein two steps: it calculated the number of squirrelsfor each species in each block according to theappropriate life history parameters (following theflow chart of Figure 2), and then applied compe-tition effects of grey on red squirrels in theblocks where both species co-occured. The num-ber of red squirrels from the competition interac-tion were input in the next model run. (Figure 2).

Because there was no information about thereal density of red squirrel in the study area, weused an average density of 0.16 ind./ha for eachwoodland block as a starting value.

Results

‘No control’ scenario

The landscape in and around Ticino RegionalPark is very suitable for squirrels and in theno-control scenario, our results indicate that greysquirrels are likely to spread over a wide areaover the next 40 years. The model predicts 4103sub-populations totalling approximately 370,000individuals in an area covering about 20,000 km2

(Table 3). The spatial spread of grey squirrel

population over the simulation period at 10 yearintervals is shown in Figure 4. The expandingpopulation first occupies the northern part of thePark and during the first 10 years it occurs onlyin the mixed broad-leaf woodlands dominated bychestnut. During the next 10 years it also spreadsinto the black locust woods (Figure 4a, b). Ourpredictions suggest that it will take more than20 years for grey squirrels to start invading thesouthern part of the Park and to spread outsidethe Park boundaries reaching Switzerland (Fig-ure 4c, d). Squirrels are predicted to show differ-ent patterns of spread: inside the Park dispersiontends to occur along the wooded riverbank ofTicino, while outside the spread has no preferentialdirection. The model simulations indicate that itwill take up to 15 years to reach carrying capacitywithin Ticino Park (Table 3). Dispersal beyondpark boundaries may therefore be slow and therapid implementation of control measures is likelyto be successful in slowing or even preventingfurther spread.

‘Best habitat control’ scenario

The model simulations show that the effect ofthis control effort is a delay of only 2 years inthe spread of the population compared to the nocontrol situation, without stopping populationgrowth either in the 50% or in the 80% removalrate scenario (Table 4). Grey squirrel dynamics arevery similar to the no control scenario in terms ofnumber of individuals and spatial spread. The lackof control success in this scenario is due to thespatial distribution of the control patches as manyof them are actually avoided in the first 15 years ofspread. Consequently, only few individuals areremoved, which represents an insufficient numberof animals trapped with respect to the reproductivesuccess of the entire population.

‘Extended control’ scenario

This strategy is very effective in preventing greysquirrel population growth, and with a 50%removal rate the population remains below 200individuals, while with an 80% removal rate greysquirrels went extinct in all replicates (Table 4).Moreover, only limited numbers of squirrels(<40) need to be trapped each year, suggesting

Table 3. Average predicted grey squirrel population size in

the ‘no control’ scenario. Results are presented at five years

intervals. SD: standard deviation.

Year Individuals SD Populations SD

2001 150 – 9 –

2006 115 16 5 2

2011 247 37 8 4

2016 1639 288 35 8

2021 19,311 3182 461 61

2026 58,485 5530 1160 75

2031 109,720 7084 2009 109

2036 201,521 12,421 2962 88

2041 370,237 12,802 4103 99

1613

Table 4. Average predicted grey squirrel population size in all simulation runs with squirrel removal.

Year Best habitat Extended Target

50% SD 80% SD 50% SD 80% SD 50% SD 80% SD

2001 150 – 150 – 150 – 150 – 150 – 150 –

2006 115 13 134 20 63 11 47 11 60 14 46 6

2011 236 53 378 104 76 10 41 15 73 16 41 11

2016 787 151 2177 538 58 22 15 7 58 9 14 9

2021 8872 2910 20,646 8085 119 69 9 13 125 38 10 13

2026 33,717 9387 51,503 15,344 180 67 8 12 200 72 8 12

2031 97,378 16,083 127,828 21,102 133 101 3 9 145 52 2 5

2036 174,145 22,538 224,277 37,722 178 42 1 3 200 114 0 0

2041 261,034 36,725 317,684 45,965 133 81 0 2 205 171 0 0

Top column stands for the different control pattern used, for more explanations please refer to the Results section.

Results are presented at 5 years intervals.

SD – standard deviation.

Figure 4. Simulation results for ‘No control’ scenario. Habitat blocks occupied by grey squirrels are evidenced in grey. Different

shade of grey represent the different densities (individuals per hectare). (a) year 2011; (b) year 2021; (c) year 2031; (d) year 2041.

1614

that the alien species can be contained with animmediate and well designed control programme.We believe that these results are conservative, aswe assumed grey squirrels were present onlywhere they had been detected, and no reliable(exhaustive) data were available on the currentdistribution of greys in- and outside the Park.An external population can be a source of in-dividuals not included in these runs of the model.

‘Target-area control’ scenario

The previous results suggest that the effectivenessof squirrel management is linked to the spatialpattern and size of controlled blocks at equal re-moval rates. This scenario confirms that an effec-tive control can be achieved on a relatively smallarea (2600 ha with respect to 4600 ha) if propertarget areas are defined (Table 4, Figure 5). Byremoving 50% of the average number of squir-rels, the whole grey squirrel population remainsbelow 300 individuals. A higher trapping successwith 80% removal results in total eradication in30 years. This can be achieved by removingaround 50 individuals/yr for the first 10 years,dropping to less than ten squirrels/yr thereafter.These results show that it is possible to tackle thespread of grey squirrels if policy makers andwildlife managers take immediate actions to con-trol the alien species. They also underline theimportance of a well designed control scheme

and careful choice of target areas to stop thespread, controlling only a small portion of thelandscape. However, for control to be effective,even on a small area, new hair-tube surveys, cov-ering all woodlands inside the park (and thosejust outside the northern border) must be carriedout immediately to reveal the present distribu-tion.

Competition scenario

Red squirrels start to go extinct in the parkwoodlands from 2020, as greys continue toexpand and their numbers increase (Figure 6).During the first 10–15 years the red squirrelpopulation is increasing from the average densityto carrying capacity. This is a model artifact re-lated to initial conditions being set at an averagedensity for the whole landscape. Red squirrelsreach equilibrium (based on carrying capacitiesfor the different habitat types) in the landscapefrom the 10th to the 20th year of simulation. Thepopulation is fluctuating around 280,000 individ-uals and the species is spread in all woodlandblocks (Table 5). After the 20th year of simula-tion greys begin to be an impacting factor andthe predicted 21,000 grey squirrels are likely todrive red squirrel to extinction within the Parkboundaries. Populations of red squirrels seem topersist longer in the northern part of the studyarea where the prealpine mountains are coveredwith conifer forests (Figure 6), and the habitat isslightly more suitable for reds than for greys (seealso Table 1a). However, reds go extinct in thecentral and most of the southern part of thestudy area within the 40 year run simulations.

Discussion

A previous study applying spatially explicit popu-lation models of grey squirrel expansion in North-ern Italy predicted an increasing rate of spread inthe next 20–30 years, with dramatic increase ofpopulation size after colonisation of continuousbroad-leaf forests of the Prealps in Piedmont (Lurzet al. 2001). The SEPM simulations of grey squir-rel expansion along the Ticino river woodlandsconfirm this risk and predict that within the nexttwo decades this population will invade

Figure 5. Simulation results for ‘Target-area control’ scenario,

compared with population increase of ‘No control’ scenario.

Dashed lines indicate different control efficiencies, i.e. 80%

and 50% removal success.

1615

Switzerland. The model forecasts the spread ofgrey squirrels in the whole study area within thenext 40 years and the concomitant reduction of

native red squirrels if no action is taken. The majorfindings of the dynamics model and the variouscontrol scenarios are that: (i) the main direction ofspread of grey squirrels will be north into contigu-ous deciduous forests along the East bank of LakeMaggiore, including southern parts of Switzerland(Canton Ticino) in the near future; (ii) efficientcontrol is possible, thus the alien species can stillbe removed, and the success is mainly determinedby the spatial distribution and total woodland sizeof the ‘target’ trapping areas; and (iii) that immedi-ate actions must be taken, as delay in grey squirrelcontrol will result in the population augmentingand spreading beyond the park borders, whichwould increase the logistic problems of successfulcontainment of grey squirrels. The model scenarioshave indicated that control will significantlyreduce grey squirrel expansion, and that complete

Figure 6. Contraction of red squirrel habitat in the competition scenario. Different shade of grey represent the different densities

(individuals per hectare) as predicted by the model. Pictures show the output at 10 year intervals. (a) year 2011; (b) year 2021; (c)

year 2031; (d) year 2041.

Table 5. Average predicted population size of grey and red

squirrels for the competition model runs.

Year Individuals Populations

Grey SD Red SD Grey SD Red SD

2001 150 – – – –

2006 112 15 206,644 7984 6 2 2728 191

2011 274 87 267,800 4307 9 5 4073 159

2016 1828 646 298,100 512 40 17 4390 22

2021 21,347 6138 284,113 1270 492 116 4403 58

2026 61,839 11,446 274,293 2693 1204 168 4020 98

2031 112,908 13,756 230,043 6621 2047 182 3189 128

2036 208,198 24,588 221,213 6753 3004 176 2805 180

2041 374,636 19,302 179,204 10,200 4120 131 1959 129

Results are presented at 5 years intervals.

SD – standard deviation.

1616

extirpation of the Ticino population is still possibleif the target (trapping) areas are carefully planned(‘target-area control’ scenario), even by removing arelatively small number of animals (ca. 40/yr) over a10-year time span. However, model scenarios werebased on the assumption that hair-tube surveyscarried out in the past reflect the actual grey squir-rel distribution: thus that no sub-populations oc-cur in woodlands within the park that were notcontiguous with the sampled woodlots, and thatno grey squirrels are present outside the parkboundaries. This may not be accurate, and realdistribution range and current population size maybe underestimated. Hence, it is absolutely essentialand urgent to determine the current, exact distri-bution of grey squirrels in and around the park,and to include in the monitoring the western bankof the river, which is covered by mixed woodlands.We suggest using a combination of hair-tube mon-itoring and a public participation survey to detectgrey squirrel presence. This may also help withincreasing public awareness and create a platformon which to explain the dangers of introducedalien species to native fauna and local economy(forestry, crop damage) (e.g., Lever 1994; Barret al. 2002; Bertolino and Genovesi 2003).

A third indication of the control scenarios isthat trapping can already commence within thepark in those woodlands where: (i) grey squirrelpresence has already been confirmed; and (ii)habitat quality is high and carrying capacity ex-ceeds 1 squirrel/ha (‘extended control’ scenario).If the logistics efforts of this scenario are judgedimpractical (total trapping area: 4600 ha), the‘target-area control’ scenario (total trapping area:2600 ha) can also be applied, having a similarprobability of successful containment of the alienspecies. However, the scale and location of trap-ping operations should be modified once reliabledata on the current grey squirrel distributionbecome available.

A fourth model prediction is a drastic declinein the population of native red squirrel concomi-tant with the progressive grey squirrel rangeexpansion. In the study area inter-specific compe-tition will be a strong factor affecting the dynam-ics of native squirrels, who may survive inmountain coniferous woodland while greys in-vade the low elevated forests. These results agreewith the dynamic predicted by the application of

SEPM in other parts of Italy (Lurz et al. 2001;Bertolino and Genovesi 2003) and by field sur-veys in the country (Wauters et al. 1997a). It islikely that Italy will experience a similar declineas that observed in Great Britain where greyshave already replaced red squirrels in most oftheir original habitat (Gurnell and Pepper 1993).

The successful containment of further greysquirrel spread will require local co-operation ofthe Park authorities with the Regional Parks andWildlife Services of the Regions of Lombardyand Piedmont. Furthermore, the Swiss authori-ties of the Ticino District (Ufficio della Caccia edella Pesca, Canton Ticino) must be advised ofthe risk of grey squirrel invasion and the estab-lishment of grey squirrel populations within thenext 20–30 years (potentially sooner, when dis-persing individuals are considered).

General conclusions

In the medium to long-term, preventing furtherinvasions will require a global strategy based onco-operation among nations, and a sensible regu-lation of the massive trade of living animals andplants (Ruesink et al. 1995; Sjoberg and Hokkanen1996). However, in the short-term, singlecountries, or local authorities at a more local le-vel, will have to take political responsibility forthe removal of an alien species that threatensbiodiversity and causes significant damage totimber, based on the international guidelines forthe conservation of nature and biodiversity (e.g.,IUCN 2000; Genovesi and Bertolino 2001a, b).The recently developed European strategy oninvasive alien species (Genovesi and Shine 2003)approved by the Standing Committee of the BernConvention and welcomed by the Council of theEuropean Union, calls for European countries toadopt a regional approach to reduce the adverseimpacts of alien species and urges them to pre-pare and implement national containment/eradi-cation plans.

Acknowledgements

The study was financed by the Parco Regionaledella Valle del Ticino Lombardo to Istituto

1617

Oikos, ONG, Varese. CT holds a fellowship fromthe Fondazione Fratelli Confalonieri, Milano.

References

Alterio N, Moller H and Ratz H (1998) Movements and habitat

use of feral house cats Felis catus, stoats Mustela erminea,

and ferrets Mustela furo, in grassland surrounding yellow-

eyed penguin Megadyptes antipodes breeding areas in spring.

Biological Conservation 83: 187–194

Barr JF, Lurz PWW, Shirley MDF and Rushton SP (2002)

Evaluation of immunocontraception as a publicly acceptable

form of vertebrate pest species control: the introduced grey

squirrel in Britain as an example. Environmental Manage-

ment 30: 342–351

Bertolino S and Genovesi P (2003) Spread and attempted

eradication of the grey squirrel (Sciurus carolinensis) in Italy,

and consequences for the red squirrel (Sciurus vulgaris) in

Eurasia. Biological Conservation 109: 351–358

Bertolino S, Wauters LA, De Bruyn L and Canestri-Trotti G

(2003) Prevalence of coccidia parasites (Protozoa) in red

squirrels (Sciurus vulgaris): effects of host phenotype and

environmental factors. Oecologia 137: 286–295

Bryce J, Johnson PJ and MacDonald DW (2002) Can niche use

in red and grey squirrels offer clues for their apparent

coexistence? Journal of Applied Ecology 39: 875–887

Byrom AE (2002) Dispersal and survival of juvenile ferrets

Mustela furo in New Zealand. Journal of Applied Ecology

39: 67–78

Cagnin M, Aloise G, Fiore F, Oriolo V and Wauters LA (2000)

Habitat use and population density of the red squirrel Sciurus

vulgaris meridionalis in the Sila Grande mountain range

(Calabria, South Italy). Italian Journal of Zoology 67: 81–87

Commission of the European Communities (1993) CORINE

Land cover, Guide technique. Directorate-General Environ-

ment, Nuclear Safety and Civil Protection, Office for Official

Publications of the European Communities, Luxembourg

Delin AE and Andren H (1999) Effects of habitat fragmenta-

tion on Eurasian red squirrel (Sciurus vulgaris) in a forest

landscape. Landscape Ecology 14: 67–72

Fitzgibbon CD (1993) The distribution of grey squirrel dreys in

farm woodland: the influence of wood area, isolation and

management. Journal of Applied Ecology 30: 736–742

Fornasari L, Galbussera R and Sacchi M (2002) Progetto per il

monitoraggio e l’eradicazione dello scoiattolo grigio nel

Parco Regionale della Valle del Ticino Lombardo. Technical

report

Genovesi P (2000) Guidelines for Eradication of Terrestrial

Vertebrates: a European Contribution to the Invasive Alien

Species Issue. Council of Europe, tpvs65e-2000

Genovesi P and Bertolino S (2001a) Linee guida per il controllo

dello scoiattolo grigio (Sciurus carolinensis) in Italia. Qua-

derni di Conservazione della Natura 4: 5–52

Genovesi P and Bertolino S (2001b) Human dimension aspects

in invasive alien species issues: the case of the failure of the

Grey squirrel eradication project in Italy. In: Mcneely JA

(ed) The Great Reshuffling: Human Dimensions of Invasive

Alien Species, pp 113–119. IUCN, Gland, Switzerland and

Cambridge, UK

Genovesi P and Shine C (2003) European Strategy on Invasive

alien species. Council of Europe. T-PVS (2003) 7

Gosling LM and Baker SJ (1989) The eradication of muskrats

and coypus from Britain. Biological Journal of the Linnean

Society 38: 39–51

Gurnell J (1991) The grey squirrel. In: Corbet GB and Harris S

(eds) The Handbook of British Mammals., pp 186–191.

Blackwell Scientific Publications, Oxford. Gurnell 1991

Gurnell J (1996a) The effects of food availability and winter

weather on the dynamics of a grey squirrel population in

southern England. Journal of Applied Ecology 33: 325–338

Gurnell J (1996b) Conserving the red squirrel. In: Ratcliffe P

and Claridge J (eds) Thetford Forest Park the ecology of a

pine forest, pp 132–140. Forestry Commission, Edinburgh

Gurnell J and Pepper H (1993) A critical look at conserving the

British red squirrel Sciurus vulgaris. Mammal Review 23:

125–136

Gurnell B and Mayle J (2003) In: Keeble V (ed) Ecological

impacts of the alien grey squirrel in Britain. Mammaliens

2002 PTES, London.

Gurnell J, Lurz PWW and Pepper H (2001) Practical techniques

for surveying and monitoring squirrels. Forestry Commis-

sion Practical Notes 11: 1–12

Gurnell J, Wauters LA, Lurz PWW and Tosi G (2004) Alien

species and interspecific competition: effects of introduced

eastern grey squirrels on red squirrel population dynamics.

Journal of Animal Ecology 73: 26–35

Gurnell J, Wauters LA, Preatoni DG and Tosi G (2001b)

Spacing behaviour, kinship and dynamics of the grey

squirrel in a newly coolnised deciduous woodland in north

Italy. Canadian Journal of Zoology 79: 1533–1543

IUCN (2000) Guidelines for the Prevention of Biodiversity Loss

caused by Alien Invasive Species. IUCN Gland, Switzerland

Kenward RE and Holm JL (1993) On the replacement of the

red squirrel in Britain: a phytotoxic explanation. Proceedings

of the Royal Society, London B 251: 187–194

Kenward RE, Hodder KH, Rose RJ, Walls CA, Parish T,

Holm JL, Morris PA, Walls SS and Doyle FI (1998)

Comparative demography of red squirrels (Scirurus vulgaris)

and grey squirrels (Sciurus carolinensis) in deciduous and

conifer woodland. Journal of Zoology, London 224: 7–21

Koprowski JL (1994) Sciurus carolinensis. Mammalian Species

480, American Society of Mammalogists.

Lever C (1994) Naturalized Animals: The Ecology of a

Successfully Introduced Species. T & A D Poyser Natural

History, London

Lockwood JA (1997) Competing values and moral imperatives:

An overview of ethical issues in biological control. Agricul-

ture and Human Values 14: 205–210

Lurz PWW, Rushton SP, Wauters LA, Bertolino S, Currado I,

Mazzoglio P and Shirley MDF (2001) Predicting grey

squirrel expansion in North Italy: a spatially explicit

modelling approach. Landscape Ecology 16: 407–420

Mack RN, Simberloff D, Londsdale WM, Evans H, Clout M

and Bazzaz FA (2000) Biotic invasions: causes, epidemiol-

ogy, global consequences, and control. Ecological Applica-

tions 10: 689–710

1618

Pimm SL, Russell GJ, Gittleman JL and Brooks TM (1995)

The future of biodiversity. Science 269: 347–350

Ragg JR, Moller H and Waldrup KA (1995) The prevalence of

bovine tuberculosis (Mycobacterium bovis) infections in feral

populations of cats (Felis catus), ferrets (Mustela furo) and

stoats (Mustela erminea) in Otago and Southland, New

Zealand. New Zealand Veterinary Journal 43: 333–337

Rodriguez A and Andren H (1999) A comparison of Eurasian

red squirrel distribution in different fragmented landscapes.

Journal of Applied Ecology 36: 649–662

Ruesink JL, Parker IM, Groom MJ and Kareiva PM (1995)

Reducing the risk on non-indigenous species introductions.

BioScience 45: 465–477

Rushton SP, Lurz PWW, Fuller R and Garson PJ (1997)

Modelling the distrubution of the red and grey squirrel at the

landscape scale: a combined GIS and population dynamics

approach. Journal of Applied Ecology 34: 1137–1154

Rushton SP, Lurz PWW, Fuller R and Gurnell J (2000)

Modelling the spatial dynamics of parapoxvirus disease in

red and grey squirrels: a possible cause of the decline in the

red squirrel in the UK?. Journal of Applied Ecology 37: 997–

1012

Sainsbury AW, Nettleton P, Gilray J and Gurnell J (2000) Grey

squirrels have high seroprevalence to a parapoxvirus asso-

ciated with deaths in red squirrels. Animal Conservation 3:

229–233

Shine C, Williams N and Gndling L (2000) A guide to designing

legal frameworks on alien invasive species. IUCN Environ-

mental Policy and Law Paper No. 40 IUCN Gland,

Switzerland

Sjoberg G and Hokkanen HMT (1996) Conclusions and

recommendations of the OECD workshop on the ecology

of introduced, exotic wildlife: fundamental and economic

aspecsts. Wildlife Biology 2: 131–133

Smith D and Gurnell J (1997) The ecology of the grey squirrel

in conifer forest. In: Gurnell J and Lurz PWW (eds) The

Conservation of Red Squirrels Sciurus vulgaris L, pp 109–

119. People’s trust for Endangered Species, London

Teangana DO, Reilly S, Montgomery WI and Rotchford J

(2000) The distribution and status of the red squirrel (Sciurus

vulgaris) and Grey squirrel (Sciurus carolinensis) in Ireland.

Mammal Review 30: 45–56

Thompson DC (1978) The social system of the grey squirrel.

Behaviour 64: 305–328

Tompkins DM, Sainsbury AW, Nettleton P, Buxton D and

Gurnell J (2002) Parapoxvirus causes a deleterious disease in

red squirrels associated with UK population declines.

Proceedings of the Royal Society, London B 269: 529–533

Tosi G and Zilio A (2002) Conoscenza delle risorse ambientali

della provincia di Varese – Progetto SIT Fauna. Provincia di

Varese, Settore Politiche per l’Agricoltura e Gestione Faun-

istica, Varese, Italy

Van Apeldoorn RC, Celada C and Nieuwenhuizen W (1994)

Distribution and dynamics of the red squirrel (Sciurus

vulgaris) in a landscape with fragmented habitat. Landscape

Ecology 9: 227–235

Verbeylen G, De Bruyn L and Matthysen E (2003a) Patch

occupancy, population density and dynamics in a frag-

mented red squirrel Sciurus vulgaris population. Ecography

26: 118–128

Verbeylen G, De Bruyn L, Adriaensen F and Matthysen E

(2003b) Does matrix resistance influence Red squirrel

(Sciurus vulgaris L. 1758) distribution in an urban landscape.

Landscape Ecology 18: 791–805

Vitousek PM, D’Antonio CM, Loope LL and Westbrook R

(1996) Biological invasions as global environmental change.

American Scientist 84: 468–478

Wauters LA, Currado I, Mazzoglio PJ and Gurnell J (1997a)

Replacement of red squirrels by introduced grey squirrels in

Italy: evidence from a distribution survey. In: Gurnell J and

Lurz PWW (eds) The Conservation of Red Squirrels, Sciurus

vulgaris L., pp 79–88. People’s Trust for Endangered Species,

London

Wauters L and Dhondt AA (1993) Immigration patterns and

success in red squirrels. Behavioral Ecology and Sociobiol-

ogy 33: 159–167

Wauters LA and Gurnell J (1999) The mechanism of replace-

ment of red by grey squirrels: a test of the interference

competition hypothesis. Ethology 105: 1053–1071

Wauters L, Gurnell J, Currado I and Mazzoglio PJ (1997b)

Grey squirrel Sciurus carolinensis management in Italy –

squirrel distribution in a highly fragmented landscape.

Wildlife Biology 3: 117–124

Wauters LA, Gurnell J, Martinoli A and Tosi G (2001a) Does

interspecific competition with introduced grey squirrels

affect foraging and food choice of Eurasian red squirrels?

Animal Behaviour 61: 1079–1091

Wauters LA, Gurnell J, Preatoni DG and Tosi G (2001b)

Effects of spatial variation in food availability on spacing

behaviour and demography of Eurasian red squirrels.

Ecography 24: 525–538

Wauters LA and Lens L (1995) Effects of food availability and

density on red squirrel Sciuris vulgaris reproduction. Ecology

76: 2460–2469

Wauters LA, Gurnell J, Martinoli A and Tosi G (2002a)

Interspecific competition between native Eurasian red squir-

rels and alien grey squirrels: does resource partitioning

occur? Behavioral Ecology and Sociobiology 52: 332–341

Wauters LA, Lurz PWW and Gurnell J (2000) The interspecific

effects of grey squirrels (Sciurus carolinensis) on the space use

and population demography of red squirrels (S. vulgaris) in

conifer plantations. Ecological Research 15: 271–284

Wauters LA, Tosi G and Gurnell J (2002b) Interspecific

competition in tree squirrels: do introduced grey squirrels

(Sciurus carolinensis) deplete tree seeds hoarded by red

squirrels (S. vulgaris)? Behavioral Ecology and Sociobiology

51: 360–367

Westervelt JM, Shapiro M, Goran W and Gerdes D (1990)

Geographic Resource Analysis Support System, Version 4.0

User’s Reference manual. USACERL ADP Report N-87/22,

Champaign, Illinois.

White PCL and Whiting SJ (2000) Public attitudes towards

badger culling to control bovine tuberculosis in cattle.

Veterinary Record 147: 179–184

Williamson M (1996) Biological Invasion. Chapman and Hall,

London

1619