• emigration;
  • habitat corridor;
  • habitat fragmentation;
  • Monte Carlo simulation;
  • random walk


  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • 1
    We report a field experiment designed to explore the responses by hedgehogs to novel and unfavourable terrain, which they encounter when dispersing between fragmented local populations in farmland. We identified four replicate sites as unused by hedgehogs, but lying between existing populations and containing locally abundant food. At each site, we released six hedgehogs from distant sources at 2-day intervals and monitored their subsequent dispersals over 20 days by radio-tracking. We compared movement trajectories under this treatment to those of 29 hedgehogs released at two sites known to support abundant hedgehogs. A third treatment comprised 20 unmanipulated hedgehogs at three sites. We estimated habitat use and proximity to habitats by comparing each observed trajectory to simulated random walks of equal length. We sought differences between treatments in movement and habitat use from nested analysis of variance.
  • 2
    No two trajectories of any translocated hedgehogs followed the same route; most involved little change in body weight, and took the animal into or through existing populations. Hedgehogs moved substantially further and faster on average from the unfavourable than from the favourable sites. They showed a significantly stronger attraction to habitat edges, which therefore acted as corridors, and a significant proportion of them stayed closer to roads and to urban habitat than random expectation. Habitat preferences also shifted significantly towards urban and away from arable areas. Six were predated, and two killed by road traffic. At least three returned to their release point, with P < 0·05 of this occurring by chance.
  • 3
    In this region, natural dispersals between populations up to 4 km apart are rare events. We have shown, however, that hedgehogs are capable of travelling distances of up to 3·8 km from a release point, and up to 9·9 km in total, compared to an average home range span of 0·8 km, and it appears that none of the local populations in the study area are out of reach of neighbours.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

All living organisms possess the ability to disperse; indeed, self-replication would be pointless without it. Dispersal has long been known to contribute to the structure and dynamics of populations, and the now extensive theories of source-sink dynamics and metapopulations have been developed to model the processes that govern population persistence in heterogeneous environments (Pulliam 1996; Hanski 1999). Dispersal has recently become a focus of attention for terrestrial ecologists in particular, wherever habitat fragmentation at the hands of man poses questions about the ability of wildlife to access more or less small and isolated patches of suitable habitat (e.g. Beier & Noss 1998). Ecologists ask questions about how population persistence is affected by population size and proximity to neighbours (e.g. Andreassen & Ims 1998), why some areas of suitable habitat are empty (Lawton & Woodroffe 1991), which habitats are barriers to dispersal between populations (e.g. roads for wildlife: Forman & Alexander 1998) and which are dispersal corridors (e.g. woodland strips for small mammals: Ruefenacht & Knight 1995).

Direct answers to these questions require monitoring dispersals, but these are usually once-in-a-lifetime events for individuals in undisturbed populations and present formidable problems to observe in practice (Ims & Yoccoz 1997). In consequence, understanding of dispersal ecology has tended to lag behind other disciplines of population ecology (Stenseth & Lidicker 1992; Turchin 1998). Much of the empirical data on dispersal have been collected at the population level, on patch occupancy (e.g. Thomas & Harrison 1992) and traffic across barriers or through corridors (e.g. Desrochers & Hannon 1997). Population-level estimates of migration and mixing between populations are increasingly being obtained indirectly from molecular analysis of genetic mixing (e.g. Real 1994). Although sophisticated techniques are available for studying dispersal behaviour, such as radio-telemetry to mark and follow animals, these are not cost-effective for many species and environments (McShea & Madison 1992). Study at the individual level requires tagging appropriate subjects prior to dispersal, and then either relocating them after dispersal from within an area that increases with the square of the dispersal distance (e.g. Sutcliffe, Thomas & Peggie 1997), or being present for the dispersal events if and when they happen and following their often brief trajectories (e.g. Beier 1995). In this paper we describe an experimental approach to generating and following dispersal trajectories, applied to an insectivorous mammal.

An extensive literature now exists on dispersal in mammals, particularly among microtines where it is implicated in the cyclical dynamics of many Northern Hemisphere populations (Stenseth & Lidicker 1992). The distances over which individuals are capable of dispersing is known for many mammals (Macdonald 1984), but few studies have set out to test hypotheses about the behavioural responses of individuals to their environment during dispersal. Our review of the literature has located only five publications in research journals over the last 3 years (1997–99) using an experimental design to test hypotheses about mammalian dispersal (Jones, Mathews & Porter 1997; Blackburn, Wilson & Krebs 1998; Gundersen & Andreassen 1998; Gillis & Krebs 1999; Nunes et al. 1999; identified by the ISI bibliographic service from among 64 papers concerned explicitly with mammalian dispersal, within a total of 402 on animals using ‘dispersal’ in the title). Although the lack of individual-level study and hypothesis testing reflects the practical difficulties involved, it impedes our understanding of the role of dispersal in population persistence for mammals.

Here we report an experimental field study of dispersal in the European hedgehog (Erinaceus europaeus L.). These insectivores make interesting subjects for studying movement because they are relatively sedentary but do not defend a feeding area, and they have no clearly defined dispersal phase in the life-history (Reeve 1994). Our study was carried out in fragmented farmland and urban habitats of Oxfordshire, where the population dynamics of hedgehogs have already been studied extensively in the field (Dickman 1988; Doncaster 1992, 1993, 1994; Micol, Doncaster & Mackinlay 1994; Ward, Macdonald & Doncaster 1997). Populations here have a discontinuous distribution, which is influenced by the distribution of predators and food resources. Natural dispersal between patches of suitable habitat has not been observed, but is known to take place as some local populations are small and absent in some years (CPD, personal observation). It is unclear how much genetic mixing takes place between local populations, but an analysis of microsatellite loci has indicated restricted gene flow within this patch network (Becher & Griffiths 1998).

In order to observe migration behaviours under controlled conditions, we created replicate dispersal opportunities for hedgehogs at several rural sites and one urban site. We did this by releasing individual hedgehogs at sites lying within their regional distribution, at points which contained locally rich food resources and through which hedgehogs may disperse, but where our surveys had revealed that they were not present as residents. Released hedgehogs were followed by radio-telemetry throughout any subsequent dispersal, and their trajectories were analysed with respect to the environment through which they passed. These data were compared to similar data collected on individuals that were released into sites that did support resident hedgehogs, and to control data collected on unmanipulated individuals.

The general aims of the field experiment were to test the hypotheses (i) that hedgehogs respond to their local environment during dispersal in ways that are distinct from their responses within a home range; and (ii) that the direction and distance of dispersal trajectories is influenced by corridors to movement (such as hedges), and by barriers (such as roads). With observations from previous studies on local population structure (Micol et al. 1994), these experimental data could also answer subsidiary questions about the probable connectivity or isolation of local populations. Direct estimates of dispersal distances and orientation abilities obtained from this study are intended to complement molecular data from the same area on population levels of migration and mixing (Becher & Griffiths 1997, 1998; ongoing).

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Study areas

Four release sites in ‘unfavourable’ habitat were located in spring 1994 in mixed pasture and arable farmland within 30 km of Oxford City, UK (Sandford, Rousham, Wilcote, Great Tew). The sites were separated from each other by 3–15 km, and so could be considered independent with respect to the usual home ranges of hedgehogs of generally less than 40 ha (this study; and Reeve 1994). A fifth release site was located in unfavourable habitat within Oxford City. These five sites were all deemed to be unoccupied by hedgehogs after nocturnal surveys by torchlight on at least three nights revealed no foraging hedgehogs at the release point or in any of the adjacent habitat patches, in accordance with the protocol of Micol et al. (1994). They all lay within habitat that was known to sustain hedgehogs, and in a region that supported discrete local populations. The five sites had nearest neighbouring hedgehog populations between 0·5 and 2·0 km from their release points (identified from field surveys and records of road casualties). The release point itself was chosen on the basis of having a plentiful and available food resource in the immediate vicinity, usually in the form of an abundant earthworm population on a mown playing field or grazed pasture. Hedgehogs released at these sites would therefore be entering an area that could support their short-term food needs, but which lay between existing local populations of hedgehogs.

Comparative data were obtained from two release sites in ‘favourable’ habitat previously studied in similar farmland to the ‘unfavourable’ rural sites, and also within 30 km of Oxford City (Eynsham in 1991 and Kirtlington in 1993, 15 km apart: as described in Doncaster 1992, 1994). Both these sites were known to support substantial hedgehog populations. The releases were timed to replace a similar number just removed from each site, in order to avoid inflating the local population density.

The seven release sites were compared to three ‘control’ sites, which were located in similar and neighbouring rural habitat (Rousham-II, lying 1 km from the unfavourable release site at Rousham, Eynsham-II and Kirtlington-II, abutting the two favourable release sites). These controls were used in summer 1993 to track movements of unmanipulated hedgehogs.

Field tests

During April and May 1994, 30 adult hedgehogs were located in a variety of rural and suburban sites around Oxfordshire, and tagged with radio-transmitters (Biotrack, Dorset, UK: 7 g on acrylic mount). Field trials were carried out on these animals at the five unfavourable sites between June and September 1994. Six adult hedgehogs were released per site, including examples of both sexes, but not noticeably gestating females. Each was translocated directly from a home site that was at least 8 km distant from the release point. Sequential releases of hedgehogs were made at intervals of at least 2 days, to reduce the chance of them trailing each other. Each released hedgehog was located during its dispersal once each day, and located and weighed once each night. This procedure ceased ~20 days after release, if the individual had by that time occupied the same area for at least 5 days, otherwise it continued for longer. Transmitters were removed from all animals at the end of their test.

Translocations to the two favourable sites were made in 1991 (19 to Eynsham) and 1993 (10 to Kirtlington), as part of the replicate perturbation experiment described in Doncaster (1992, 1994). Procedures were the same as at the unfavourable sites, except that all individuals were released simultaneously.

Hedgehogs at the three control sites were tagged with radio-transmitters in summer 1993 (nine at Eynsham-II, six at Kirtlington-II, five at Rousham-II), and their undisturbed use of home ranges was monitored over a period of up to 2 months.

All fieldwork was performed under UK Home Office Licence PPL: 30/00043.


The unfavourable release site lying within Oxford City was analysed separately from the other sites, because the highly divided habitat of urban gardens and streets differed categorically from the semirural habitats of all the other release and control areas.

Our analyses concerned principally the responses by hedgehogs to their local environment, compared between nine sites nested in three treatments. These treatments are henceforth identified as ‘TU’ = translocation to unfavourable rural habitat, ‘TF’ = translocation to favourable rural habitat, and ‘C’ = unmanipulated rural controls.

Dispersal and exploration

Five dispersal parameters were computed for each animal from its radio-locations:

  • 1. ‘MCP area’ = the area of the minimum convex polygon encompassing all its locations.

  • 2. ‘Distance moved’ = the sum of linear distances between consecutive locations.

  • 3. ‘MCP span’ = the maximum width of its MCP area.

  • 4. ‘Speed’= distance moved divided by the total duration of post-release monitoring.

  • 5. ‘Start–end distance’ = the distance between the release point and the last radio-location.

Differences between treatments in these responses were tested with analysis of variance (detailed below).

Dispersal and exploration by released hedgehogs were defined in relation to the mean MCP span for unmanipulated hedgehogs at the control site. The sexes were pooled for these definitions, since MCP span did not vary by sex for any of the treatments. A translocated animal was considered to have dispersed away from the release point if the centroid of its radio-locations lay outside a circle drawn around the release point of radius equal to the upper 99% confidence limit of half the MCP span for control animals (0·56 km). The individual was defined as exploring if its centroid of locations lay within this circle, despite some radio-locations falling outside it. Differences between treatments in the proportion of dispersing and exploring individuals were analysed with a chi-square test.

Use of linear features

All linear features at the sites were digitized from maps produced by the Ordnance Survey (OS, scale 1 : 25 000) with a GIS package (vector digitizing software CartaLinx, raster software IDRISI: Clark Labs, Mass., USA). Each discrete patch in the resulting networks was then assigned a habitat classification, according to ground data collected at the time of the study and supplemented with information from aerial photographs taken by CPD in summer 1994. The four recognized habitat types were: urban buildings and dwellings, pasture (for cattle, horses and sheep), woodland (mainly deciduous), and arable (mainly wheat and barley, also some ley and 3-year rotational set-aside). These habitat data were unavailable for the Kirtlington and Kirtlington-II study sites, which were therefore excluded from the habitat preference analysis.

We computed the distances of each post-release radio-location to four target features:

  • 1. Nearest ‘edge’, including all borders separating polygons of different habitat, and hedgerows identified in the aerial photographs as separating polygons of the same habitat.

  • 2. Nearest ‘road’, including all unpaved and asphalt roads, and railroads.

  • 3. Nearest ‘large road’, as above but excluding unpaved roads and asphalt roads classified on OS maps as generally less than 4 m wide.

  • 4. Nearest ‘urban’ area, including all patches of built-upon land (mainly villages and farms).

The hypothesis against which these distances were tested was that hedgehogs tended to move or stay closer to the features than would be expected if they moved randomly with respect to them. This meant comparing the observed distribution of distances on each individual to a set of distributions given by a simulated hedgehog on replicate random walks at the same speed and through the same area. The comparison would then control for differences between sites in the distribution of possible distances to features. For example, hedgehogs in highly fragmented habitat would have less opportunity to distance themselves from habitat edges than those in uniform habitat.

Monte Carlo simulations were used to generate the random distributions, following Besag & Clifford (1989) and Manly (1997). For each observed trajectory, we simulated 99 random walks from the release point, each with the same number of radio-locations and the same frequency distribution of distances between consecutive points as the observed animal, and each constrained to lie within the minimum convex polygon of locations for all animals at that site. The component of randomness was only in the direction taken from one location to the next and in the order of consecutive distances. The observed median distance was ranked with the 99 simulated medians, and a hedgehog was considered to be significantly closer to the target feature of concern if its observed rank lay within the five smallest medians (P < 0·05). The difference between treatments in proportion of individuals significantly closer to target features was analysed with a chi-square test. A difference between treatments in strength of attraction to target features was tested with analysis of variance (detailed below).

Test of orientation by translocated hedgehogs

For each translocated hedgehog, its minimum distance to the release point at any time after the first three locations was computed for the observed trajectory and each of the 99 simulated trajectories. Any individual with a minimum observed distance < 100 m was considered to have returned to the release point. The return was deemed non-random if this minimum distance ranked among the five smallest values expected for the individual.

Analyses of variance

Nested parametric analyses of variance were employed to seek differences between treatments in dispersal parameters. A log transformation was applied to MCP area, distance moved, MCP span, and start–end distance; an inverse log transformation was applied to speed.

The same method was also used to seek differences between treatments in the proportion of observed radio-locations lying closer to target features than the simulated medians. The median of simulated locations defines the threshold of equal chance that a randomly moving individual will be nearer or further than this distance from the feature. The response features ‘edge’ and ‘urban’ required no prior transformation, while roads and large roads were transformed according to y′= arcsine(y3), and y′ = y2, respectively.

Transformed variables had errors that neither departed significantly from a normal distribution on an Anderson–Darling test, nor had significantly heterogeneous variances on a Levene’s test. Our results tables were compiled with the back-transformed means and (asymmetric) limits of the confidence interval (following Sokal & Rohlf 1995).

All response parameters were analysed in a statistical package (Minitab, version 12·1) with the general linear model:

Response = Treatment ? Site′(Treatment) ? Sex

The three treatment levels were: TU in four replicate sites, TF in two replicate sites, and C in three replicate sites. The main effect of treatment was tested against the error mean square MS[Site′(Treatment)]; Sex and Sex × Treatment were tested against error MS [Sex × Site′(Treatment)]; Site′(Treatment) and Sex × Site′(Treatment) were tested against error MS[Subject′{Sex × Site′(Treatment)}]. The nesting was unbalanced, since the number of individuals differed between sites, and this prevented computation of an exact F-value. We therefore used Satterthwaite’s approximate F-value (following Sokal & Rohlf 1995), and also repeated the analyses in a balanced design of four hedgehogs at each site. This meant ignoring sex effects, and dropping one of the TU sites from the analysis (Wilcote) where data were available from only three individuals. An unbiased choice of subjects for this analysis was ensured by applying two selection criteria. These were to accept the individuals (i) with the most locations, or if there were no differences in the number of locations (ii) those with locations collected over the shortest time-span. Post-hoc tests were computed for balanced designs, with error MS[Site′(Treatment)], following Underwood (1997).

For target features showing a significant treatment effect, we further explored the distribution of distances, measured as a difference between the proportion of observed and simulated locations for each individual at each of 10 distance classes. The general linear model was:

Response = Treatment ? Site′(Treatment) ? Distance

The Distance × Treatment interaction was tested against error MS[Distance × Site′(Treatment)].

Use of habitat

The proportions of observed and simulated radio-locations in each habitat type were computed for pooled subjects and sites within treatments. Any overall departure from randomness was analysed with a chi-square goodness-of-fit test. The compositional analysis of Aebischer, Robertson & Kenward (1993) was then used for post-hoc comparisons between habitat types, in the observed use relative to availability. This required a pooling of individuals within sites, since too few locations were collected on each individual for proportional habitat use by subject. The test sought to identify which habitats were significantly preferred over others in one-tailed multiple t-tests with degrees of freedom given by one less than the number of sites. The TF treatment could not be tested in this way because habitat data were only available at one TF site.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

No overall change in body weight was observed for translocated hedgehogs, with weight gains recorded for 18 of the 24 TU hedgehogs and 12 of the 29 TF hedgehogs between the release day and the last radio-tracking day. Mean ± SE post-release weights were 820 ± 9 g in TU, 838 ± 7 g in TF, and 880 ± 7 g in control sites. Of the 30 hedgehogs released at unfavourable sites, two were killed by road traffic (at Sandford and Great Tew), and six were predated (three at Wilcote, one at each of Great Tew, Rousham and Oxford City). The uneaten remains, consisting solely of the dorsal coat of spines with transmitter attached, suggested the predators were probably badgers, Meles meles L., which are locally abundant and known to predate hedgehogs (Doncaster 1992, 1994). Three at Wilcote, two at Great Tew and one at Oxford City died within 5 days of release, and were removed from further analyses for lack of sufficient data on their movement parameters and habitat use. Of the remaining 73 hedgehogs, 68 were assigned to the three treatments TU, TF, C, and a further five were analysed separately from the Oxford City group (two males and three females).

Movement parameters

Analyses of frequencies

Trajectories of TU hedgehogs are illustrated in Fig. 1 with respect to major habitat categories. No two hedgehogs followed the same route (true also of TF hedgehogs), suggesting that movements were exploratory rather than prescribed by a fixed set of criteria. Our impression from Fig. 1 that many of the hedgehogs took an anticlockwise trajectory was not confirmed by any statistical difference from zero in the angle of turns. The two individuals moving the fastest also had among the largest spans. These were female no. 25 and male no. 26 at Wilcote, with maximum speeds of 2·40 and 3·49 km per day, respectively, and MCP spans of 3·96 km and 3·70 km, respectively. Only female no. 4 at Sandford had a larger span, of 4·21 km. The trajectory of no. 4 took a wide arc with a total distance of 9·90 km. This individual moved up to 2·30 km per day, passing through two hedgehog-populated areas of urban habitat before finishing in urban habitat 1·95 km from the release point. The maximum distance travelled from any release point was 3·78 km, by two females at different TU sites (no. 4 at Sandford and no. 24 at Rousham in Fig. 1).


Figure 1. Trajectories of hedgehogs transplanted to unfavourable sites, in relation to principal habitat categories (TU hedgehogs, Oxford City site not shown). Release points at each site are identified by arrowed dot; animal identification numbers and small arrows show individual trajectories of far-moving individuals. Urban habitat is indicated in dark grey, pasture and woodland habitats in light grey, arable land unshaded.

Download figure to PowerPoint

The frequency distribution of maximum distances from release points was non-normal for the sample translocated to unfavourable sites (Fig. 2a, Anderson–Darling test, A2 = 0·88, P = 0·02), and consistent with normal for the sample translocated to favourable sites (Fig. 2b, Anderson–Darling test, A2 = 0·37, P = 0·4). This means that the sample in unfavourable sites deviated from the predictions of a random walk model, with fewer individuals displacing < 0·5 km, and more > 3 km, than expected by a normal distribution (since data lie below the line at both tails, making the lower tail lighter and the upper tail heavier than expected).


Figure 2. Normal probability plots for maximum distances from release points attained by (a) TU hedgehogs including Oxford City, with mean = 1·48 km, standard deviation = 1·08, N = 30, and (b) TF hedgehogs with mean = 0·88 km, standard deviation = 0·40, N = 29. Plots show probabilities of occurrence assuming a normal distribution. The least-squares line fitted to the points estimates the cumulative distribution function for the population from which data are drawn.

Download figure to PowerPoint

Similar proportions of TU and TF hedgehogs explored away from their release points (13 of 19 and 17 of 29, respectively, not significant by chi-square test). Likewise, similar proportions dispersed (10 of 19 and 13 of 29, respectively, not significant). While a significantly larger number of males than females explored away from their release points (19 of 25 and 11 of 23, respectively, for TU and TF combined, χ12 = 4·06, P = 0·04), the proportions dispersing were similar between the sexes (13 of 25 and 10 of 23, respectively, not significant). Of the five hedgehogs analysed at Oxford, only one male explored, and then dispersed from the release site. Although many released hedgehogs at TU and TF sites apparently settled either in the release area or elsewhere, none of them showed a sufficiently stable site fidelity to permit dividing the tracking period into pre and post-settlement periods.

A capacity to keep track of their dispersal route was demonstrated by at least three TU individuals, which moved away from the release point > 0·9 km and subsequently returned to within 3% of that maximum distance. The minimum distance following a maximum distance, and the probability of this occurring in a random walk, were: to 10 m from 2·72 km at P = 0·02 for male no. 5 at Sandford; to 30 m from 0·93 km at P = 0·03 for male no. 6 at Sandford; to 10 m from 1·32 km at P < 0·01 for female no. 12 at Great Tew. Trajectories are shown in Fig. 1, and they indicate return routes that do not follow the outward path.

Analyses of variance

Table 1a shows that TU hedgehogs gave the highest means for all five movement parameters, and confidence limits that did not overlap those of TF or C individuals for all but start–end distance when sites were pooled within treatments. TU hedgehogs on average had larger MCPs (F2,6 ≈ 13·38, P = 0·02) and moved at faster speeds (F2,6 ≈ 17·76, P = 0·006) than TF and/or C individuals. The overall differences between treatments were not significant for distance moved (F2,6≈5·71, P = 0·06), MCP span (F2,6 ≈ 5·80, P = 0·06) and start–end distance (F2,6≈6·96, P = 0·06). The balanced analysis of variance gave exact F-values that corroborated these results, although distance moved was also found to differ significantly between treatments (F2,5 = 27·01, P = 0·002). Tukey’s multiple comparison test indicated a higher MCP span at TU than at C sites.

Table 1.  Dispersal parameters by treatment group and sex. Treatment groups are: TU = transplanted to unfavourable environment (mean ± SE of 24 ± 2 locations per individual over 20 ± 2 days), TF = transplanted to favourable environment (12 ± 0·6 locations per individual over 54 ± 2 days), C = controls (16 ± 0·4 locations per individual over 53 ± 2 days). Responses are means and 95% confidence limits for N individuals, of: MCP = area of minimum convex polygon, dist mov = total distance moved, span = maximum distance between two radio-locations, speed = average speed over tracking period, start–end = distance between first and last radio-location Thumbnail image of

Despite male hedgehogs producing higher means than females on all these movement parameters (Table 1b), the differences were only significant in an interaction with the random Site′(Treatment), for MCP (F6,50 ≈ 4·78, P < 0·001), MCP span (F6,50 ≈ 2·93, P = 0·02), and start–end distance (F6,50 ≈ 4·76, P < 0·001). There were no significant interactions of treatment with sex for any of the parameters. Differences between sex means did not correlate with time of year when sites were sampled (regressions, F1,7 < 0·93, P > 0·37).

Hedgehogs released into the unfavourable site in Oxford City had mean values of movement parameters lying between those of controls and TF hedgehogs, except for speed, which was close to that of TU hedgehogs. Their mean MCP area was 24·8 ha (0·0–63·5 ha at 95% confidence limits), distance moved was 3·40 km (0·82–6·00 km), MCP span was 0·72 km (0·250–2·07 km), speed was 222 m/day (148–295 m/day), start–end distance was 0·51 km (0·00–0·94 km).

Use of edges, roads and urban areas

Analyses of frequencies

Figure 3 shows the distribution of locations on TU hedgehogs with respect to linear features. Apart from the two hedgehogs known to have been hit by road traffic, others safely crossed trunk roads and small rivers. Although no major barriers to dispersal were detected, no hedgehog crossed the main railway line running through the Oxford City site, and three individuals at Rousham probably used a road bridge to cross the river and adjacent railway passing through that site, on the evidence of locations taken either side of the bridge. A visual comparison of the patterns in Figs 1 and 3 with the distribution of locations on control hedgehogs, shown in Fig. 4, suggests that the TU hedgehogs made more use of urban habitat and avoided arable land, and that they were attracted to linear features. These hypotheses are analysed in detail below.


Figure 3. Locations of hedgehogs transplanted to unfavourable sites, in relation to all linear features (four individuals at Great Tew, six at Rousham, six at Sandford, three at Wilcote, Oxford City site not shown). Roads shown as thick black lines, rivers as thick grey lines, railways as cross-hatched lines.

Download figure to PowerPoint


Figure 4. Locations on hedgehogs at the two control sites, in relation to linear features and also showing principal habitat types (nine individuals at Eynsham-II, 5 at Rousham-II). Shading and lines as in Figs 1 and 3. Habitat patch data are unavailable for the third control site, which is not shown.

Download figure to PowerPoint

Table 2 shows that a low proportion of hedgehogs stayed significantly closer to edges than random expectation, with no differences between treatments. A higher proportion of TU hedgehogs than TF or C hedgehogs were significantly attracted to roads (χ22 = 10·06, P = 0·007, pooling sites within treatments), and to urban habitat (χ22 = 10·86, P = 0·004). We tested the possibility that a preference for urban areas shown by some TU hedgehogs might reflect an attraction to habitat they were familiar with prior to translocation. Four of six TU hedgehogs with urban origins subsequently headed for urban areas, but this proportion was not significantly different to the five of 13 with rural origins that also went to urban areas (Fisher P = 0·35). No sex differences were obtained in the proportions of hedgehogs showing significant attraction to any of these features. Of the five Oxford City hedgehogs, one stayed significantly closer than random expectation to edges and urban, one to all roads, one to large roads only, and one to urban habitat.

Table 2.  Number of hedgehogs staying significantly close to habitat features. Treatment groups TU, TF and C as for Table 1. An individual is included in the category of being close to the feature if the median distance from its observed trajectory ranked among the top five with 99 simulated random trajectories. * Indicates a significant dependency between treatment and response at P < 0·05 by chi-square test
TreatmentClose to edgesClose to all roads*Close to large roadsClose to urban*
Analyses of variance

Hedgehogs differed between treatments in the average proportion of observed locations lying closer to edges than the median random expectation (F2,6 ≈ 10·51, P = 0·02). Figure 5 shows the trend in strength of attraction from highest at TU to lowest at C. Roads and urban habitat, also shown in Fig. 5, did not differ significantly on the unbalanced analysis (all roads: F2,6 ≈ 6·31, P = 0·08; large roads: F2,6 ≈ 0·83, P = 0·49; urban: F2,6 ≈ 0·47, P = 0·65). These results were corroborated by the balanced design (edge: F2,5 = 36·20, P < 0·001, other responses not significant). None of the target features showed significant sex effects in the unbalanced design.


Figure 5. Proportion of observed radio-locations closer to target features than the median simulated distance. Graphs show transformed mean (SE for 19 individuals at TU sites, 29 at TF, and 20 at C). Asterisk indicates a significant treatment effect at P < 0·05 by nested analysis of variance for habitat edges. For this factor, treatment means all differed with Tukey’s pairwise comparisons on the balanced design at P > 0·05; for the other factors, no treatment means differed significantly with this test.

Download figure to PowerPoint

Figure 6 shows how treatment influences the distribution of distances from edge habitats (Distance × Treatment F18,54≈3·30, P < 0·001). A clearly defined pattern is observed only for TU hedgehogs, where the closest distances tend to be the most preferred, and the furthest the least. All three groups have a lower proportion of radio-locations than random expectation at > 90 m, which suggests a general avoidance of open areas.


Figure 6. Distribution of radio-locations from habitat edges, measured as a difference between the proportion of observed and simulated locations for each individual at each distance class. The response is shown as a mean difference (SE for the 20 C, 29 TU and 19 TF individuals).

Habitat use

All treatments yielded a non-random distribution of observed locations with respect to the four habitat categories (Table 3, χ32 > 20·00, P < 0·001). The post-hoc multiple t-tests indicated a significant preference in TU sites for urban over the other three habitat types (Table 3: t3 > 2·7, P < 0·05). The two control sites for which we had habitat data did not reveal a significant preference for one habitat over another in relation to expected use from the simulated trajectories. However, the apparent indifference to arable habitat was explained at least partially by their home ranges being situated within an area of little arable land (see Fig. 4). That this might be a form of avoidance on a larger scale was analysed by applying the post-hoc tests to the area of ‘available’ habitat within a 1·7-km radius around the sites. This broader scale was equivalent to the average area covered by hedgehogs released at each TU site, and it confirmed that control and TU sites had similar overall habitat composition (Table 3, compare values of available habitat at C and TU) and that the control populations did avoid arable land in favour of urban habitat (Table 3: t1 = 6·84, P < 0·05). It is noteworthy that simulated hedgehogs at TU and TF sites used arable land somewhat less, and woodland and pasture more, than they had available from aerial coverage. This is because the fixed release point from which all simulations started had been sited in an area with locally abundant food, and such areas generally lay close to woodland and pasture and far from arable land.

Table 3.  Proportions of observed locations in each habitat category, compared to expected proportions from simulations of random walks, and to aerial coverage of available habitat. For TU and TF sites, available habitat was measured from coverage within the polygon surrounding all locations; for C sites it was measured from within a 1·7-km radius around each site (equivalent to the average area of TU polygons). All fractions sum to one over the four habitat types. * Indicate significant dependence of frequency of use on habitat type, at P < 0·05 by chi-square goodness-of-fit to expected proportions. For TU and control treatments only, vertical lines join habitats with non-significantly different usage according to the post-hoc multiple Student’s t-tests of Aebischer et al. (1993), at P > 0·05


  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The ecology of dispersal concerns population-level expressions of mechanisms operating at both evolutionary and behavioural levels. There is a recognized gap between theory and data on how dispersal evolves (Dieckmann, O’Hara & Weisser 1999), as well as on how vertebrates orientate themselves in unfamiliar terrain (Sherry 1996). Ecological studies can offer direct and indirect evidence to narrow these gaps, since dispersal has measurable effects on the distribution and abundance of populations. Our data on dispersal in hedgehogs have revealed non-random movement patterns in unfamiliar terrain, and differences between favourable and unfavourable habitat in the type and amount of movement behaviour.

Our comparison of TU to TF individuals allowed us to control for the artificial act of translocating hedgehogs into unfamiliar areas. Although the TF subjects were released simultaneously in contrast to the staggered releases of TU subjects, no hedgehog was seen to follow the trail of another. The translocated hedgehogs were all ≥1 year old, whereas anecdotal evidence suggests that dispersal is also, and perhaps predominantly, undertaken by juveniles before their first winter hibernation (Berthoud 1978; Reeve 1994). However, hedgehogs do not have a fixed natal territory from which to disperse, nor a clearly defined dispersal stage. Their farmland habitat is subject to unpredictable and dramatic changes in resource quality as crops are harvested and livestock rotated, to which juveniles and adults alike respond with shifts in their home ranges (Doncaster 1993). On this basis, and from evidence of previous translocations (Doncaster 1992, 1994), we predicted that adult hedgehogs would respond to translocation by seeking improved conditions, as opposed to simply fleeing unfavourable conditions. This was borne out by the general absence of weight loss following release, evidence of orientation over distances > 1 km, and non-random patterns of movement. The total of 59 translocated hedgehogs showed movement behaviours ranging from true dispersal, involving abandonment of one range and travel to a disjunct range, through forms of quasi-dispersal (sensuLidicker & Stenseth 1992) such as excursions out of a stationary range, and nomadic floating.

The experimental protocol had some limitations common to many field trials at this scale. Although the replicate sites were independent in terms of the identities of the hedgehogs using them, we could not avoid some spatial overlap. For example, hedgehog no. 4 released at the Sandford site dispersed far enough to pass through the Great Tew site (Fig. 1). Also all three control sites abutted release sites (cf. Rousham and Rousham-II in Figs 3 and 4), although subjects were followed in different years, and different habitat availabilities were obtained from the smaller scale of movement of unmanipulated hedgehogs. The comparison between treatments assumed all years were similar in terms of dispersal pressure. These design limitations arose from attempting to keep the experiment within a single region of broadly similar habitat. Against this, the significant advantage of the experimental design over descriptive observation was its capacity to generate controlled and replicate trajectories. Prior to this translocation experiment, descriptive data from radio-tracking studies (ours: Doncaster 1993; others: in Reeve 1994) had given no indication of the substantial ranging capacity of this normally sedentary species, with a home range of < 40 ha. Our previous translocations had shown hedgehogs were capable of dispersing 3·0–3·6 km from a release point (Doncaster 1992, 1994). One other study has recorded a dispersal of 5·2 km, by a rehabilitated hedgehog released on Jersey (Morris 1997). We now evaluate the experimental results with respect to three general questions about dispersal behaviour.

Why do species occupy some sites and not others in similar habitat? Animals may have discontinuous ranges because certain habitats are too isolated to be reached or because some are too inhospitable to be colonized. For example, water voles Arvicola terrestris are absent from stretches of river bank that have been colonized by American mink Mustela vison which can consume all water voles while still subsisting on alternative prey (Lawton & Woodroffe 1991). Similarly for hedgehogs, their feeding niche of invertebrates overlaps with that of badgers which also consume hedgehogs, thus rendering some habitat inhospitable and potentially creating barriers to colonization of badger-free fragments (Doncaster 1992, 1994; Micol et al. 1994; Ward et al. 1996, 1997). Badgers were present throughout this study area, except in the immediate neighbourhood of urban habitat. Local variations in their distribution are likely to contribute to the fragmentation of hedgehog populations, in combination with variations in food availability, and to the unfavourability of TU sites. Released hedgehogs tended to disperse into urban areas (Table 2). A probable explanation is that urban habitat tended to be avoided by badgers and often supported existing hedgehog populations. Some released hedgehogs settled in urban areas that did not support existing populations (e.g. nos 5, 6 and 10 at Sandford, Fig. 1), indicating that the attraction to urban habitat may also reflect the overall attraction to roads (Table 2) and edges (Fig. 5), which converge there.

Favourable habitat for dispersing individuals may depend not only on abundant and safe resources, but also on how much they are being depleted by other conspecifics. Two of the TU release points, Rousham and Great Tew, lay within 1 km of existing hedgehog populations, yet not all releases there finished in the nearest population. Some individuals passed through existing hedgehog populations without stopping (e.g. no. 4 from Sandford). This suggests a form of local density-dependence in the arrival phase of dispersal, which would be consistent with the more commonly recorded density-dependence in the leaving phase (e.g. examples and models in Turchin 1998), if dispersal obeys a rule of last in, first out. Non-territorial animals with a nomadic response to crowding of this sort will tend towards an ideal free distribution (sensuFretwell & Lucas 1970; extended to populations and metapopulations in Doncaster 2000), provided individuals are free to roam between populations. Our experimental results allow us to test this provision for hedgehogs by analysing trajectories in relation to habitat structure.

Does habitat structure influence dispersal route? Analysis of dispersal routes should reveal which habitat structures act as corridors for a species increasing connectivity between its populations, and which act as barriers decreasing connectivity. Such structures have potential impacts for conservation biology that have long been recognized, although no general answers have emerged about their influences on dispersal and population persistence (Simberloff et al. 1992). We found that hedgehogs in unfavourable habitat moved further and faster than those in favourable habitat (Table 1a), and were attracted to linear features (Figs 3–5). Edge habitats can therefore be said to have represented movement corridors. A substantial proportion of TU hedgehogs followed road networks (Table 2), although two were killed by road traffic, and movement across railways may have been restricted to bridges. In a previous translocation experiment, Doncaster (1992) monitored one female that crossed a 12-m wide branch of the River Thames and a dual-carriageway trunk road, before eventually being killed by traffic on that same road. Vehicles are well known to kill large numbers of hedgehogs (reviewed in Reeve 1994), and in this sense they are a barrier to movements across roads. Our data indicate, however, that the roads themselves – or most probably their verges – can represent movement corridors, as they do for many species of invertebrates and small mammals (Forman & Alexander 1998). Apart from these general patterns, the trajectories of released hedgehogs were surprisingly variable in both length and direction (Fig. 1), indicative of them exploring the unfamiliar habitat rather than following a fixed set of dispersal rules.

Hedgehogs showed a disassociation from arable land in unfavourable habitat only (Table 3). Those in favourable habitat positioned their home ranges in local areas that contained little arable land. This is an example of a scale-dependent response typical of many species in heterogeneous habitat, but not often recorded (cf. Johnson 1980). Populations tend to locate in areas with favoured habitat characteristics (in this case more urban, pasture and woodland habitat than arable), and individuals dispersing more widely actively seek routes that avoid unfavoured habitats (in this case arable).

Does distance regulate dispersal between neighbouring populations? The simplest models of movement dynamics assume that animals follow a random walk, yielding a normal distribution of distances around the release point (Hanski 1999). Empirical studies, however, often yield distributions with heavier tails than the normal (e.g. Johnson & Gaines 1990). We found that more hedgehogs in unfavourable habitat moved longer distances than is predicted by a normal distribution (Fig. 2). More than one hedgehog from each of the unfavourable release sites moved to, or passed through, an existing local population of hedgehogs even when the nearest population was 2 km distant (Sandford). In this area of Oxfordshire, our surveys revealed gaps between populations that were never > 4 km, which is less than the maximum span of movements by TU hedgehogs. On this evidence, no population was completely isolated from its neighbours. We also found that females were capable of moving just as far as males, despite a general tendency for females to have smaller home ranges than males. This is a significant observation because for any mammal the colonization of unoccupied suitable habitat, or recolonization of abandoned habitat, requires breeding females if a viable population is to be established, yet male-biased dispersal is the norm among mammals (Greenwood 1980). Previous surveys of local distribution pattern have estimated that hedgehogs in our study area use > 90% of suitable habitat, and are thus robust to loss of habitat on the metapopulation scale (Micol et al. 1994; Doncaster, Micol & Plesner Jensen 1996). The present study indicates that such high coverage is facilitated by their dispersal capability.

Currently the most promising way forward for understanding dispersal is to combine ecological with evolutionary perspectives by collecting both field and molecular data (Koenig, Van Vuren & Hooge 1996; Thompson & Goodman 1997). In a concurrent study of genetic mixing between local populations across an area overlapping our study area, Becher & Griffiths (1998) found no correlation of genetic to geographical distance, and concluded that factors other than distance affected dispersal. Our own continuing molecular studies do show a positive regression among local populations from a larger sample size in Hampshire (500 individuals, 80 km south of the Oxfordshire study site). Despite our observations of long dispersals in relation to distances between nearest neighbouring populations, we recorded no displacement > 10 km. We therefore expect local populations to differ in the number of source populations from which they receive immigrants, depending on how far they lie from all other populations in the network. The frequency of dispersal events between local populations will depend on the type of matrix, and also on the availability of – and direction taken by – linear features such as habitat edges and roads.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

We thank Susanna d’Antoni and Susanne Plesner Jensen for assistance with radio-tracking, and Peter Atkinson for advice on simulations. Numerous errors were brought to our attention by two anonymous referees. This study was supported by the BBSRC, by NERC grant GR803680 to CPD, and by a grant to CR from the Università degli Studi di Roma ‘La Sapienza’.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • Aebischer, N.J., Robertson, P.A. & Kenward, R.E. (1993) Compositional analysis of habitat use from animal radio-tracking data. Ecology, 74, 13131325.
  • Andreassen, H.P. & Ims, R.A. (1998) The effects of experimental habitat destruction and patch isolation on space use and fitness parameters in female root vole Microtus oeconomus. Journal of Animal Ecology, 67, 941952.
  • Becher, S.A. & Griffiths, R. (1997) Isolation and characterization of six polymorphic microsatellite loci in the European hedgehog Erinaceus europaeus. Molecular Ecology, 6, 8990.
  • Becher, S.A. & Griffiths, R. (1998) Genetic differentiation among local populations of the European hedgehog Erinaceus europaeus in mosaic habitats. Molecular Ecology, 7, 15991604.
  • Beier, P. (1995) Dispersal of juvenile cougars in fragmented habitat. Journal of Wildlife Management, 59, 228237.
  • Beier, P. & Noss, R.F. (1998) Do habitat corridors provide connectivity? Conservation Biology, 12, 12411252.
  • Berthoud, G. (1978) Note préliminaire sur les déplacements du hérisson européen (Erinaceus europaeus L.). Terre Vie, 32, 7382.
  • Besag, J. & Clifford, P. (1989) Generalised Monte Carlo significance tests. Biometrika, 76, 633642.
  • Blackburn, G.S., Wilson, D.J. & Krebs, C.J. (1998) Dispersal of juvenile collared lemmings (Dicrostonyx groenlandicus) in a high-density population. Canadian Journal of Zoology – Revue Canadienne de Zoologie, 76, 22552261.
  • Desrochers, A. & Hannon, S.J. (1997) Gap crossing decisions by forest songbirds during the post-fledging period. Conservation Biology, 11, 12041210.
  • Dickman, C.R. (1988) Age related dietary change in the European hedgehog, Erinaceus europaeus. Journal of Zoology, London, 215, 114.
  • Dieckmann, U., O'Hara, R. & Weisser, W. (1999) The evolutionary ecology of dispersal. Trends in Ecology and Evolution, 14, 9092.
  • Doncaster, C.P. (1992) Testing the role of intraguild predation in regulating hedgehog populations. Proceedings of the Royal Society Series B, 249, 113117.
  • Doncaster, C.P. (1993) Influence of predation threat on foraging pattern: the hedgehog’s gambit. Revue d’Ecologie (Terre Vie), 48, 207213.
  • Doncaster, C.P. (1994) Factors regulating local variations in abundance: field tests on hedgehogs (Erinaceus europaeus). Oikos, 69, 182192.
  • Doncaster, C.P. (2000) Extension of ideal free resource use to breeding populations and metapopulations. Oikos, 89, 2436.
  • Doncaster, C.P., Micol, T. & Plesner Jensen, S. (1996) Determining minimum habitat requirements in theory and practice. Oikos, 75, 335339.
  • Forman, R.T.T. & Alexander, L.E. (1998) Roads and their major ecological effects. Annual Review of Ecology and Systematics, 29, 207231.
  • Fretwell, S.D. & Lucas, H.L. Jr (1970) On territorial behaviour and other factors influencing habitat distribution in birds. Acta Biotheoretica, 19, 1636.
  • Gillis, E.A. & Krebs, C.J. (1999) Natal dispersal of snowshoe hares during a cyclic population increase. Journal of Mammalogy, 80, 933939.
  • Greenwood, P.J. (1980) Mating systems, philopatry and dispersal in birds and mammals. Animal Behaviour, 28, 11401162.
  • Gundersen, G. & Andreassen, H.P. (1998) Causes and consequences of natal dispersal in root voles, Microtus oeconomus. Animal Behaviour, 56, 13551366.
  • Hanski, I. (1999) Metapopulation Ecology. Oxford University Press, Oxford, UK.
  • Ims, R.A. & Yoccoz, N.G. (1997) Studying transfer processes in metapopulations: emigration, migration, and colonization. Metapopulation Biology: Ecology, Genetics, and Evolution (eds I.A.Hanski & M.E.Gilpin), pp. 247265 Academic Press, London.
  • Johnson, D.H. (1980) The comparison of usage and availability measurements for evaluating resource preference. Ecology, 61, 6571.
  • Johnson, M.L. & Gaines, M.S. (1990) Evolution of dispersal: theoretical models and empirical tests using birds and mammals. Annual Review of Ecology and Systematics, 21, 449480.
  • Jones, M.L., Mathews, N.E. & Porter, W.F. (1997) Influence of social organization on dispersal and survival of translocated female white-tailed deer. Wildlife Society Bulletin, 25, 272278.
  • Koenig, W.D., Van Vuren, D. & Hooge, P.N. (1996) Detectability, philopatry, and the distribution of dispersal distances in vertebrates. Trends in Ecology & Evolution, 11, 514517.
  • Lawton, J.H. & Woodroffe, G.L. (1991) Habitat and the distribution of water voles – why are there gaps in a species’ range? Journal of Animal Ecology, 60, 7991.
  • Lidicker, W.Z. Jr & Stenseth, N.C. (1992) To disperse or not to disperse: who does it and why? Animal Dispersal: Small Mammals as a Model (eds N.C.Stenseth & W.Z.LidickerJr), pp. 2136. Chapman & Hall, London.
  • Macdonald, D.W., ed. (1984) The Encyclopaedia of Mammals. Unwin, London.
  • Manly, B.F.J. (1997) Randomization, Bootstrap and Monte Carlo Methods in Biology. Chapman & Hall, London.
  • McShea, W.J. & Madison, D.M. (1992) Alternative approaches to the study of small mammal dispersal: insights from radiotelemetry. Animal Dispersal: Small Mammals as a Model (eds N. C.Stenseth & W. Z.LidickerJr), pp. 319332. Chapman & Hall, London.
  • Micol, T., Doncaster, C.P. & Mackinlay, L.A. (1994) Correlates of local variation in the abundance of hedgehogs Erinaceus europaeus. Journal of Animal Ecology, 63, 851860.
  • Morris, P.A. (1997) Released, rehabilitated hedgehogs: a follow-up study in Jersey. Animal Welfare, 6, 317327.
  • Nunes, S., Duniec, T.R., Schweppe, S.A. & Holekamp, K.E. (1999) Energetic and endocrine mediation of natal dispersal behavior in Belding’s ground squirrels. Hormones and Behavior, 35, 113124.
  • Pulliam, H.R. (1996) Sources and sinks: empirical evidence and population consequences. Population Dynamics in Ecological Space and Time (eds O.E.Rhodes, R.K.Chesser & M.H.Smith), pp. 4569. University of Chicago Press, Chicago.
  • Real, L.A., ed. (1994) Ecological Genetics. Princeton University Press, Princeton, NJ.
  • Reeve, N.J. (1994) Hedgehogs. Poyser, London.
  • Ruefenacht, B. & Knight, R. (1995) Influences of corridor continuity and width on survival and movement of deermice. Biological Conservation, 71, 269274.
  • Sherry, D.F. (1996) Middle-scale navigation: the vertebrate case. Journal of Experimental Biology, 199, 163164.
  • Simberloff, D., Farr, J.A., Cox, J. & Mehlman, D.W. (1992) Movement corridors: conservation bargains or poor investments? Conservation Biology, 6, 493504.
  • Sokal, R.R. & Rohlf, F.J. (1995) Biometry, 3rd edn. W. H. Freeman, New York.
  • Stenseth, N.C. & Lidicker, W.Z. Jr (1992) The study of dispersal: a conceptual guide. Animal Dispersal: Small Mammals as a Model (eds N.C.Stenseth & W.Z.LidickerJr), pp. 520. Chapman & Hall, London.
  • Sutcliffe, O.L., Thomas, C.D. & Peggie, D. (1997) Area-dependent migration by ringlet butterflies generates a mixture of patchy population and metapopulation attributes. Oecologia, 109, 229234.
  • Thomas, C.D. & Harrison, S. (1992) Spatial dynamics of a patchily distributed butterfly species. Journal of Animal Ecology, 61, 437446.
  • Thompson, P.M. & Goodman, S. (1997) Direct and indirect estimates of dispersal distances. Trends in Ecology & Evolution, 12, 195196.
  • Turchin, P. (1998) Quantitative analysis of movement. Measuring and Modeling Population Redistribution in Animals and Plants. Sinauer, Sunderland, Mass.
  • Underwood, A.J. (1997) Experiments in Ecology. Cambridge University Press, Cambridge.
  • Ward, J.F., Macdonald, D.W. & Doncaster, C.P. (1997) Responses of foraging hedgehogs to badger odour. Animal Behaviour, 53, 709720.
  • Ward, J.F., Macdonald, D.W., Doncaster, C.P. & Mauget, C. (1996) Physiological responses of the European hedgehog, Erinaceus europaeus, to predator and non-predator odour. Physiology and Behavior, 60, 14691472.

Received 29 September 1999; revised version accepted 13 May 2000