Variation in the helminth community structure in bank voles (Clethrionomys glareolus) from three comparable localities in the Mazury Lake District region of Poland

We tested the null hypothesis that populations of hosts trapped in isolated neighbouring locations showing comparable habitat quality, should support similar helminth parasite communities. The study was undertaken in a 2-week period in late summer in NE Poland in a single year, thereby eliminating seasonal and between-year variation in parasite burdens. A total of 139 Clethrionomys glareolus (bank vole) were sampled from 3 forest sites of similar habitat quality. Total species richness was 11 (6 nematodes and 5 cestodes) with 85·6% of the voles carrying at least 1 species and an overall mean species richness of 1·4. At the component community level, the fewest species of helminths were recorded from site 2 (n = 6, compared with 9 at each of the other sites), but site 3 had the lowest Berger-Parker Dominance Index and the highest Simpson's Index of Diversity. At the infracommunity level, site 3 had the highest mean no. of helminth species/vole, the highest mean Brillouin's Index of Diversity but the lowest mean no. of helminths/vole. Voles from sites 1 and 3 differed in the nematodes that were most common (site 1, Heligmosomum mixtum – 95%; site 3, Heligmosomoides glareoli – 79·3%). At site 2 no species exceeded 50% but prevalence of Syphacia petrusewiczi was higher than at the other sites. The prevalence of cestodes was too low to test reliably (12·9%), but the highest prevalence of adult cestodes was recorded at site 1 (22·5% compared with 4·9 and 1·7% for sites 2 and 3 respectively). Host sex did not influence infection, but mean species richness increased with age. The different sites were responsible for most of the variation in our data, and the intrinsic factors (sex and age) were less important in shaping the component community structure of helminths. We conclude that even locations in relative close proximity to one another (13–25 km), selected on the basis of similar habitat quality, have rodent populations that differ in their helminth parasite communities, although for reasons other than the factors quantified in the present study.



Parasitic infections are being increasingly recognized as a major factor influencing host behaviour, lifehistory strategies and population dynamics (Sheldon & Verhulst, 1996 ;Hillgarth & Wingfield, 1997 ;Hudson & Dobson, 1997 ;Barnard & Behnke, 2001). Fragmentation of the environment by natural barriers (e.g. lakes, mountain ranges) and\or human activities (e.g. towns, major roads, agriculture) can lead to isolated subpopulations of hosts (Hanski, 1997). Such subpopulations may be subject to different selection pressures created by local conditions, with profound implications for the dynamics of the overall population. Local differences in parasite communities infecting subpopulations are likely to create one important source of variation in selection pressure.
The structure of helminth component communities in wild rodents in Europe has been well documented and is known to be affected by both extrinsic (year, season etc.) and intrinsic (host age, sex) factors (Montgomery & Montgomery, 1990 ;Behnke et al. 1999 ;Abu-Madi et al. 2000). Perhaps surprisingly intrinsic factors have generally been found to play a lesser role in shaping these communities than extrinsic factors : host sex has rarely been found to contribute significantly (Kisielewska, 1970c ;O'Sullivan, Small & Fairley, 1984 ;Abu-Madi et al. 2000), although there are some reports of particular species showing a significant sex bias (Lewis, 1968 ;Behnke et al. 1999). Accurate estimation of the age of wild rodents is notoriously difficult (Morris, 1972) but nevertheless, employing relatively crude measures of age e.g. body weight or eye lens weight, a significant role for host age in structuring component communities has been identified. Young animals will have had less opportunity to encounter infective stages than adults, and since many rodent helminths cause chronic infections, worm burdens and mean species richness accumulate with time and hence age of host (Haukisalmi, Henttonen & Tenora, 1988 ;Montgomery & Montgomery, 1989 ;Gregory, 1992 ;Behnke et al. 1999).
Extrinsic factors exert the most important influence on helminth communities with strong evidence in support of temporal variation arising through seasonal and annual changes in mean species richness (Langley & Fairley, 1982 ;Haukisalmi et al. 1988 ;O'Sullivan et al. 1984 ;Montgomery & Montgomery, 1989). Marked differences have also been reported in the structure of helminth component communities from rodents captured in different sites within similar climatic zones (Kisielewska, 1970c ;Montgomery & Montgomery, 1990 ;Abu-Madi et al. 2000), although to date few studies have systematically examined and compared helminth component communities from relatively isolated populations living in comparable habitats within close proximity to each other. Rather, the emphasis has been to select habitats of contrasting habitat quality (Montgomery & Montgomery, 1990 ;Abu-Madi et al. 2000), despite the inevitable difficulties of interpretation in the absence of clear baseline values on the extent of natural variation in helminth communities when habitat quality remains constant (Kennedy, Hartvigsen & Halvorsen, 1991 ;Hartvigsen & Kennedy, 1993).
The study reported in this paper was based in Mazury in north-eastern Poland, a region rich in managed forests, interspersed with numerous lakes (in excess of 1000), rivers and canals that impede free movement of rodents between forest patches, and in some cases may create impassable barriers. Numerically the dominant woodland rodent is the bank vole, Clethrionomys glareolus, the ecology and population dynamics of which have been well documented over the years in Poland and elsewhere (Kozakiewicz, 1976a ;Alibhai & Gipps, 1985). Moreover, the east European parasite fauna of bank voles is also well known as a result of the extensive studies of Kisielewska (1970a, b, c) and others (Zarnowski, 1955), although advances in our understanding of the taxonomy of bank vole helminths have resulted in changes to the nomenclature of several of the species reported by these earlier workers (Haukisalimi & Tenora, 1993 ;Tenora & Meszaros, 1975 ;Wiger, Barus & Tenora, 1978 ;Genov & Yanchev, 1981).
We selected 3 study sites, within an extensive area of fragmented forest. The sites were chosen on the basis of the similarity of their habitat structure and type, and isolation from one another. Trapping was restricted to a single season in one year to eliminate other quantifiable extrinsic factors and hence the only known extrinsic factor in our study, at this time of year, was site of capture. We tested the null hypothesis that the composition of helminth component communities of the bank vole C. glareolus should not differ significantly between host populations living in habitats of similar quality.

Study sites
Our study sites were located in Mazury in the north eastern corner of Poland, in the vicinity of the largest lake in the region, Jezioro S; niardwy and the towns of Mikołajki, Orzysz and Pisz (Fig. 1). Site 1 was located east of the nature reserve surrounding Lake Łuknajno, and north of Lake S; niardwy (long. 21m 39n6h, lat 53m 47n7h, Les! nistwo Łuknajno). Site 2 was located 13n5 km north-east of site 1, on the eastern side of Lake Tałty (long. 21m 32n4h, lat 53m 53n5h, Les! nistwo Skorupki) and site 3 was located 11n5 km southeast of site 1, on the eastern side of Lake S; niardwy (long. 21m 48n8h, lat 53m 42n3h, Les! nistwo Lisiej Jamy). All 3 sites were chosen on the basis of their similar habitat quality. All were mature woodlands with Scots pine (Pinus sylvestris) and silver birch (Betula verucosa) as the dominant trees but contained patches of dense Norway spruce (Picea abies), occasional oaks (Quercus robur), and common alders (Alnus glutinosa). The trees formed a high canopy over most of the area on each site and the shrub layer comprised mostly hazel (Corylus avellana). Ground cover comprised Oxalis acetosela, Convalaria mayalis and Stelaria holostea and in places dense patches of moss. There were several small woodland glades dominated by various grasses. The area occupied by the sites ranged over approximately 900-1200 hectares and each had one boundary comprising a lake shore but was otherwise surrounded by open agricultural land.
Comparability of the habitat structure was tested by taking line transects along each trapline and scoring the vegetation every 5 m. At each sample point, the nearest item of vegetation (a) vertically above the observer, (b) 1 m to the left and 1 m to the right at a height of 1n2 m from the ground and (c) on the ground itself was recorded as : (i) high canopy (tree tops constituting the highest vegetation layer at the site), (ii) mid-canopy (tree tops below the high canopy), (iii) shrub (shrubs, bushes and immature trees up to 5 m in height), (iv) the trunk or branch of a high or mid-canopy tree), (v) tall herb (dicotyledonous plants over 1 m in height), (vi) herb (herb layer monocotyledonous and dicotyledonous plants), (vii) leaf litter and shed conifer needles and (viii) bare soil. The scores were then used to calculate the percentage occurrence of each vegetation category (0-100 % in each case) at the sample points along each trapline.

Collection of voles
At each site, rodents were caught live in Longworth, Sherman or locally constructed wooden traps, with a small metal platform internally which when triggered released a metal door. Approximately 100 traps were used in total. These were set out at 20 m intervals in parallel lines, 10 m either side of tracks running through the sites and 2 traps were placed within 2-3 m of one another at each point. Trapping sessions comprised 4 days at each site with traps being inspected in the early morning and just before dusk. Any traps containing animals were replaced by fresh traps and the animals were brought to the University of Warszawa 's field stations at Urwitalt (sites 1 and 2) or Pilchy (site 3).

Sampling of hosts
At the field stations, all animals were identified, culled, weighed (to the nearest 0n1 g), sexed and relevant morphometric data were recorded. Finally, the entire alimentary tract was removed, placed in plastic universals containing 10 % formalin and the livers, lungs and body cavity were inspected for helminths.
Three age classes were established primarily on the basis of the combined dry weight of each vole 's pair of eye lenses (Morris, 1971) as follows : juveniles (lens weight l 3-5 mg), young adult (5n1-5n9 mg), mature adults ( 6n0mg). These categories correspond approximately to voles 63, 63-100 and 100 days old (Kozakiewicz, 1976b). However, adjustments were made, particularly at the limits of each category, using other morphometric measurements which included the following parameters : visual assessment of maturity at capture (juvenile\ adult), externally apparent testes (scrotal\non-scrotal), pregnancy and lactation, head length, head width, nose-anal length, tail length and foot length. To confirm that age-related factors showed the predicted increase from age class 1 through to age class 3 we carried out 2-way analyses of variance with each of these measurements in turn as the dependent variable, and the 3 age classes and sex as the factors. All tests confirmed highly significant effects of age class (Fig. 2), but in the case of body weight, nose-anal and total bodyjtail length there were also significant differences between the sexes and significant interactions between sex and age class.

Measures of community structure
Following the report by Kennedy & Hartvigsen (2000), measures of component community structure were total number of helminth species, the Berger-Parker Dominance Index and Simpson 's Index of Diversity (unbiased as (1kD)i(n\nk1) where n l total number of helminths in the sample and D l Simpson 's Index ; Washington, 1984) for each site in turn. Infracommunity structure was assessed by mean number of helminth species per vole, maximum number of helminths and species density distribution across the sample, mean number of helminth individuals per vole, mean Brillouin 's Index per vole (infected and uninfected), maximum Brillouin 's Index, mean abundance and prevalence of individual species.

Statistical analysis
The frequency distribution of infracommunity species richness was tested for goodness of fit to the positive binomial distribution (assumption of the null model is a regular distribution), the Poisson distribution (assumption of the null model is a random distribution), the negative binomial model (assumption of the null model is an aggregated distribution) and the null model of Janovy et al. (1995) (assumption of the null model is that, in the absence of associations and interactions between species, the frequency distribution of infracommunity species richness is predicted by prevalence values of all the species comprising the component community). All distributions were tested for goodness of fit by χ#.
Prevalence data (percentage of animals infected) are shown with 95 % confidence limits, calculated as described by Rohlf & Sokal (1995). Prevalence was analysed by maximum likelihood techniques based on log linear analysis of contingency tables using the software package Statgraphics Version 7. Beginning with the most complex model, involving all possible main effects and interactions, those combinations which did not contribute significantly to explaining variation in the data were eliminated in a stepwise fashion beginning with the highest-level interaction. A minimum sufficient model was then obtained, for which the likelihood ratio of χ# was not significant, indicating that the model was sufficient in explaining the data.
Summary figures for parasite abundance are expressed as arithmetic means p..., but where distributions were aggregated, and where data required transformation for analysis, geometric means (GM) are given (Elliott, 1977 ;Dash, Hall & Barger, 1988) together with 95 % confidence limits (95 % CL). These means reflect the abundance of infection as defined by Margolis et al. (1982) and Bush et al. (1997) and include all subjects within the specified group, infected and not infected, for which relevant data were available. The degree of aggregation in the data was calculated by the Index of Discrepancy (D) as described by Poulin (1993) (a value of 0 indicates an even distribution of counts across all hosts and a value of 1 indicates all parasites aggregated in a single host) and the Index of Dispersion (I, variance to mean ratio, where values l indicate overdispersed data). Frequency distributions of individual species were also tested for goodness of fit to negative binomial, positive binomial and Poisson models by χ# as described by Elliott (1977) and the negative binomial exponent k is given as appropriate.
Parasite abundance was analysed by GLIM (A statistical system for generalized linear interactive modelling ; GLIM 4, PC version, Royal Statistical Society, 1993) as described previously, using models with normal errors after normalization of the data by log "! (xj1) transformation (Crawley, 1993 ;Wilson & Grenfell, 1997 ;Behnke et al. 1999). Site (3 levels, 1, 2 and 3), host age (3 levels) and host sex (2 levels) were entered as factors. We began in all cases with the full factorial models, including all main effects and interactions, and then progressively simplified them by deletion of terms, beginning with the l 50n8, P 0n0001, but there is also a significant effect of sex F ","$$ l 17n7, P 0n0001, and a significant interaction between sex and age F #,"$$ l 6n0, P l 0n003). (B) Nose-anal length (main effect of age F #,"$$ l 42n0, P 0n0001, but there is also a significant effect of sex F ","$$ l 13n8, P l 0n0003, and a significant interaction between sex and age F #,"$$ l 5n8, P l 0n004). (C) Total body and tail length (main effect of age F #,"$$ l 56n2, P 0n0001, but there is also a significant effect of sex F ","$$ l 21n3, P 0n0001, and a significant interaction between sex and age F #,"$$ l 6n6, P l 0n002). (D) Average hind foot length (only the main effect of age is significant F #,"$$ l 15n8, P 0n0001). (E) Skull length (only the main effect of age is significant F #,"$$ l 20n7, P 0n0001). (F) Dry lens weight (only the main effect of age is significant F #,"$$ l 120n6, P 0n0001). highest order interactions, and progressing to the main effects. Two-way interactions were first deleted to register the change in deviance and then reinstated in turn until all 3 had been evaluated. All 2-way interactions were then removed, and the procedure was repeated with the main effects. For models with normal errors the change in deviance is divided by the scale parameter and the result divided by the change in degrees of freedom (..) following each deletion, to give a variance ratio, F. Finally, minimum sufficient models were fitted, entering only the significant terms, and the residuals from these were checked for approximately normal distribution.

Habitat structure
Comparability of habitat structure at the 3 sites was confirmed quantitatively by analysis of variance (ANOVA) of the percentage scores at the sampling points for the traplines in each of the 3 sites. Scores were aggregated by vegetation level as (a) mean percentage high plus mid-canopy (canopy layer), (b) mean percentage shrub plus tall herbs and tree branch\trunk (shrub layer), and (c) the ratio of percentage ground-layer herbaceous material to percentage bare soil and litter (ground layer). In all 3 cases there was no significant difference between sites (canopy layer  Clethrionomys glareolus A total of 139 bank voles was sampled from the 3 study sites. The structure of the host population by site, host sex and age is summarized in Table 1. Although samples from the 3 sites did not differ significantly in respect of the relative numbers of the two sexes, there was a significant difference in the numbers of voles among the 3 age classes (χ# l l3n9, .. l 4, P l 0n0075). Voles in age class 2 formed the largest proportion of the sampled animals from all 3 sites (57n5, 51n2 and 44n8 % for sites 1, 2 and 3 respectively). However, voles from the youngest age class comprised a larger proportion of the sample from site 3 and were least well represented in site 1 (7n5, 26n8 and 36n9 % for sites 1, 2 and 3 respectively).

Measures of component community structure
Total species richness and component species. Eleven species of helminths were recorded in total, 6 nematodes and 5 cestodes (Table 2) and 119 voles (85n6 %) carried at least 1 of these species. The 2 heligmosomid species, H. mixtum and H. glareoli, were the most prevalent helminths among all the voles combined (40n3 % and 36n0 % respectively), but there were marked differences in their prevalence, as well as that of other helminths, between sites.
Total species richness, dominant species and diversity indices by site. The number of helminth species recorded at sites 1 and 3 were higher than at site 2 (Table 3). However, sites 1 and 2 had the highest Berger-Parker dominance indices and the lowest Simpson 's diversity indices, because of the very high worm burdens with S. petrusewiczi in a small proportion of voles from these sites (  1n44p0n076 (variance to mean ratio l 0n55). Mean species richness was marginally higher at site 3 compared with site 1, but relatively low at site 2 (Table 5 ; 3-way ANOVA in GLIM with normal errors and site, host age and sex as factors, main effect of site F #,"$& l 9n134, P 0n001). There was also a significant main effect of host age (Fig. 3A, F #,"$& l 4n611, P l 0n01) with mean species richness increasing from the youngest voles (age class 1) through to the oldest (age class 3) but no effect of host sex (mean species richness in males l 1n45p 0n109, females l 1n43p0n107). There were no significant interactions.
Measures of infracommunity diversity. The maximum number of helminth species per vole ranged from 2 in site 2 to 4 in site 3. The mean number of helminths harboured per vole appeared to differ with sites 1 and 2 showing considerably higher mean values than site 3, principally because of the few animals from the former sites carrying very heavy S. petrusewiczi burdens. However, statistical analysis (3-way ANOVA in GLIM with site, sex and age as factors on log "! (xj1) transformed data) revealed only a significant 3-way interaction (F $,"#& l 3n212, 0n025 P 0n01).
There was a significant difference in the value of Brillouin 's diversity index across the 3 sites (Table  5 ; 3-way ANOVA in GLIM with site, sex and age as factors, F #,"$& l 10n431, P 0n001), the highest mean value being generated by voles from site 3 and the lowest by those from site 1. There were no other significant terms in this analysis but the main effect of age was only just outside significance (F #,"$& l 2n383, 0n1 P 0n05), and the mean values show an increase in diversity with increasing age (Fig. 3B).
Species density distributions. The species density distributions at each site are shown in Fig. 4. As can be seen the majority of animals harboured 1 or 2 species. At sites 1 and 3 some animals harboured 3 species but only 1 vole, from site 3, harboured the maximum number of 4 species. These distributions were adequately described by both positive and negative binomial models but only that for site 2 was compatible with the Poisson distribution. The data from each site were combined (Fig. 4D) and compared to the distribution predicted by the null model for interactions of parasite species in an assemblage (Janovy et al. 1995). No significant difference was found (χ# l 2n9, .. l 4, P 0n05).
Prevalence of species and higher taxa. The prevalence data are summarized by site in Table 2. At the highest taxonomic level (all helminths combined), voles from site 2 had lower prevalence of helminths than those from sites 1 and 3. The minimum sufficient model (analysis by log linear maximum likelihood) included only 2 terms, siteiage and siteiinfection (χ# l 26n6, 24 .., P l 0n323), indicating that the number of animals in respective age classes and prevalence of infection with helminths varied between sites. An identical picture emerged for nematodes, since all cestodes were encountered only in animals that also carried nematodes and no other higher helminth taxa were recorded in this study.
The prevalence of H. mixtum varied considerably between sites (minimum sufficient model l sitei infection and siteiage, χ# l 28n0, .. l 24, P l 0n260) with the majority of voles in site 1 carrying this species, compared with none of the 58 voles sampled from site 3 ( Table 2).
Prevalence of infection with H. glareoli showed an opposite pattern with 79n3 % of voles from site 3 infected, and only 7n5 % at site 1. The minimum sufficient model for this species was more complex than that for H. mixtum (χ# l 11n4, .. l l6, P l 0n783). The siteiage interaction term was expected but there were also 2 additional terms : a significant effect of site on infection (Table 2) but also an independent effect of sex and age on infection (Fig.  5). The oldest animals showed the lowest prevalence with this species, but whereas in males peak prevalence was in the youngest animals, among females it was in age class 2.
Prevalence of infection with the 2 oxyuroids varied only in relation to site, but in different combinations. Voles from site 2 had the highest prevalence of S. petrusewiczi (the minimum sufficient model incorporated interaction terms : siteiage and sitei infection ; χ# l 27n6, .. l 24, P l 0n277) whereas those from site 3 the lowest (Table 2). However, A. tetraptera showed the highest prevalence in site 3 and the lowest in site 1 (similar model but χ# l 26n1, .. l 24, P l 0n348).
M. muris was encountered too infrequently to permit analysis of prevalence but Table 2 shows that this species was only recovered from sites 1 and 3. A single specimen of T. muris was recovered from a female vole from site 3.
None of the cestodes showed prevalence in excess of 10 %. Among adult cestodes, C. henttoneni was the most common, 8 of the 11 animals harbouring this species originating from site 1. In addition Paranoplocephala gracilis was only found at site 1, and consequently when the prevalence of adult cestodes (all species pooled) was analysed, the minimum sufficient model confirmed that a siteiinfection term was essential (χ# l 22n2, .. l 24, P l 0n567).
Larval cestodes were rarer still and hence no further analysis was undertaken. Three animals (1 from each site) harboured M. lineatus, the single specimen of T. martis came from site 3, and 2 of the 3 T. mustelae were from site 3 with the remaining worm being from a vole from site 1.

Frequency distributions and measures of aggregation.
The frequency distributions of all species for which quantitative analysis was possible were examined and the summary statistics are given in Table 6. It was not possible to test all for significance because of insufficient degrees of freedom arising from too few animals carrying infection in some cases, particularly in low worm burden classes. Among those which could be tested, none were compatible with positive binomial or Poisson distributions. In site 1, H. mixtum and S. petrusewiczi could not be tested for goodness of fit to the negative binomial but the values of I indicated overdispersion, and both distributions differed significantly from those expected of the positive binomial and Poisson distribu-tions. In site 2 the distributions of H. mixtum and S. petrusewiczi were well described by the negative binomial, and the values of D suggested that H. glareoli and A. tetraptera were also aggregated. In site 3 H. glareoli and A. tetraptera conformed to the negative binomial and the value of D suggested also that S. petrusewiczi was aggregated.
Abundance of infection. Analysis of H. mixtum by 3way ANOVA in GLIM (after normalization by log "! [xj1] transformation) revealed that there was a highly significant main effect of site (F #,"$& l 43n0, P 0n001), but also a significant interaction between site and sex (F #,"#( l 5n88, P 0n005). As Fig. 6A shows, abundance was highest in voles from site 1, but higher in females compared with males. In site 2 males carried marginally higher infections than females and in site 3 none of the voles were infected with H. mixtum.
In both oxyuroid species, following analysis of log(xj1) transformed data, the only significant main effect was that of site (S. petrusewiczi, F #,"$& l 43n0, P 0n001 ; A. tetraptera, F #,l$& l 12n38, P 0n001). However, in the case of S. petrusewiczi there was a significant 2-way interaction between site and sex (F #,"#( l 5n98, P 0n005) and this is illustrated in Fig. 6C where it can be seen that voles from site 3 had very low mean abundance and those from site 2 the highest, but this was confounded by a sex difference with a higher mean abundance in female voles. In site 1 S. petrusewiczi showed an intermediate abundance but was higher in male compared with female voles. In contrast to S. petrusewiczi, voles from site 3 had the highest mean abundance of A. tetraptera (Fig. 6D). However, this was complicated by a 3-way interaction with sex and age (F $,"#& l 3n46, 0n025 P 0n01). None of the remaining parasites were present in sufficient animals to merit analysis.


The data presented in this paper clearly show that differences in the helminth communities harboured by bank voles from the 3 study sites existed at both component and infracommunity levels. Although 11 species of helminths were identified in total, the species list did not exceed 9 for any of the 3 sites. At the component community level, voles from site 2 had fewer species of helminths but it was site 3 that differed with respect to dominance and diversity indices. These indices were very similar at sites 1 and 2 (high for the Berger-Parker dominance index, and low for the Simpson 's index) and were primarily generated through the voles that were very heavily infected with S. petrusewiczi. Nematodes of the genus Syphacia can accumulate to a high abundance through autoinfection, facilitated by eggs that are infective within hours of being released by female worms onto the perianal surface of the host (Lewis, 1968). At site 3 A. tetraptera was the dominant species, representing 68n4 % of all recovered helminths, but, as with S. petrusewiczi, this nematode was not the most prevalent at the site.
Differences between the sites were also evident at the infracommunity levels in respect of mean species richness, mean Brillouin 's index and in the prevalence and quantitative worm burdens of particular species. The lowest mean species richness, lowest prevalence of both nematodes and cestodes was recorded for voles from site 2 and none of the 6 species of helminths from this site infected more than 43n9 % of the voles. In fact the prevalence of the most common species, the heligmosomids combined, did not exceed 50 % in voles from this site. However, voles from site 2 showed the highest prevalence of S. petrusewiczi with 5 animals harbouring in excess of 100 worms. It was these few heavily infected animals that generated the high value for the mean number of helminths\vole from this site. In contrast, voles from sites 1 and 3 both had a higher but similar mean species richness but in each site a different species of nematode showed the highest prevalence. In site 1 it was H. mixtum that infected 95 % of the animals and in site 3 H. glareoli showed the highest prevalence, with 79n3 % of the voles carrying this species. Each of these 2 closely related nematodes can be regarded as a core species in its respective site. Other species were rare in site 1 (although the prevalence of adult tapeworms was the highest) whereas site 3 was also characterized by the highest prevalence with A. tetraptera and M. muris. When the quantitative data were analysed at higher taxonomic levels, no clear distinctions were found between the sites, because the individual species compensated for lack of others.
As expected from other studies of helminths in wood mice and voles, there was no evidence in our data for sex bias in infections (Kisielewska, 1970c ;O'Sullivan et al. 1984 ;Abu-Madi et al. 2000). Agerelated differences in parasite burdens are more difficult to assess because of the problem of accurate ageing (Morris, 1972). Most workers have employed host weight to estimate age but, at best, this can be regarded as only an approximate indicator since older animals can lose condition for a variety of reasons, not the least from heavy parasite infections and seasonal changes in availability of food. A more reliable indicator is the dry weight of lenses, but few studies have exploited this as an indicator of age or combined lens weight with other morphometric measurements. However, we paid special attention to age estimation, using eye lens weight and a range of other morphometric measurements to allocate voles to 3 age classes. Hence it was surprising to find that there were few age effects in our data. Interestingly mean species richness increased significantly from age class 1 through to class 3, and the mean Brillouin 's index showed a corresponding increase although it was just outside significance. Both trends are consistent with the idea that older animals will have had a longer opportunity to acquire infections and that most of the helminth parasites are relatively long-lived in relation to the natural life-span of the host. At the individual species level, the only significant age-related effect was on the prevalence of H. glareoli. The oldest age-class showed the lowest prevalence which, although intuitively unexpected, might indicate that animals acquire resistance with duration of exposure. Gregory (1992) also identified a trend of decreasing worm burdens among the heaviest wood mice (Apodemus sylvaticus) infected with Heligmosomoides polygyrus. Thus the peak prevalence of infection in females in age class 2 is consistent with this idea, but perhaps surprisingly it was the youngest males that showed the highest prevalence. Analysis of the quantitative data did not confirm a significant effect of age, even when confined to voles from site 3, but a trend of falling worm counts from the youngest to the oldest voles was evident.
Our study adds to those of Kisielewska et al.  & Henttonen (1999) which have drawn attention to spatial heterogeneity, at both local and regional levels, as a factor in shaping the helminth component communities in wild rodents. Montgomery & Montgomery (1990) studied a range of sites which encompassed deciduous, coniferous and grassland sites in Northern Ireland and concluded that neither proximity nor the gross habitat characteristics were key determinants of helminth communities. Moreover, some helminth communities of Apodemus sylvaticus changed quite radically whilst others remained unchanged over a 5 year period. Abu-Madi et al. (2000) studied 3 sites which differed quite markedly in their habitat structure, and associated high worm burdens and mean species richness with mice from semi-isolated populations living in hedgerows on agricultural land. Haukisalmi & Henttonen (1999) concluded that aggregation of helminth parasites was primarily determined within sites rather than between them, and hence local factors played a key role in promoting dissimilarity between sites (see also Hartvigsen & Kennedy, 1993).
In our study, voles from 3 woodland habitats, which were selected because of their gross similarity to each other in respect of habitat quality, were shown to carry quite different combinations and intensities of infection. Our study sites were sufficiently distant from one another to exclude the possibility that voles interchanged between them regularly, and each was surrounded by a combination of open ground and water, making them essentially island habitats (Gerlach & Musolf, 2000). Our data add weight to the idea that spatial factors contribute significantly to shaping helminth component communities in rodents even at a local level ( 30 km) but the principal conclusion from our study is that even sites intentionally selected on the basis of the similarity of their habitat quality may differ substantially in this respect.
The question then arises as to why this should be the case. If the voles are indeed isolated in each of our 3 study sites (allopatric host-parasite combination), then genetic differences may play a role (Lively & Dybdahl, 2000 ;Gerlach & Musolf, 2000). In this context it is relevant that we were able to detect significant phenotypic variation between the vole populations from the 3 sites, as reflected in differences in morphometric measurements, including body size and limbs (after controlling for age and sex ; Barnard et al. manuscript in preparation). Moreover, recent studies have shown that rivers and even highways can act as isolating mechanisms for C. glareolus, with detectable differences at the genetic level between populations separated by such barriers (Gerlach & Musolf, 2000). It is equally possible that local variation in host population structure combined with subtle differences in life-history strategies as reflected in behaviour, reproduction and investment in immune function, may generate sufficient dif-ferences between the populations to affect risk of exposure, susceptibility and resistance to infection (sympatric host-parasite combination).
On the other hand, attention has been drawn to the stochastic nature of the processes that govern helminth communities in other animals, for example fish (Hartvigsen & Kennedy, 1993). Thus local factors have been suggested to play an important role in promoting dissimilarity in helminth community structure in brown trout from reservoirs of comparable habitat quality. Similarly, differences between our sites may have been based to some extent on chance, and the likelihood that transmission stages of some parasites became extinct (extinction events) or were introduced to the local vole population (colonization events) either by migrating or dispersing voles (helminths with direct cycles) or prey\predators (helminths with indirect cycles). With respect to cestodes, which only accounted for 1n5 % of all the helminths recovered (total l 10 530 individual worms), the availability of insect intermediate and mammalian definitive hosts may have been a factor. However, all the 6 nematode species have direct life-cycles, 3 are specific to bank voles, 2 (A. tetraptera and T. muris) are specific to rodents and only M. muris may have a wider range of hosts. At this stage, there are no indications as to the identity of the factors responsible for the contrasting helminth component communities in rodents from our 3 sites. However, their identification, and that of the processes involved, are the long-term goals in our project.