Temporal and between-site variation in helminth communities of bank voles (Myodes glareolus) from N.E. Poland. 1. Regional fauna and component community levels

SUMMARY Helminth infections were studied in bank voles (Myodes glareolus) from 3 woodland sites in N.E. Poland in the late summers of 1999 and 2002, to assess the temporal stability of derived statistics describing the regional helminth fauna and component community structure, and spatial influence on the latter. Regional helminth fauna changed dramatically between the two years, primarily due to a fall in the abundance of Syphacia petrusewiczi but was partially compensated for by an increase in Mesocestoides lineatus and Cladotaenia globifera. It was dominated by nematodes overall, but more so in 1999 than in 2002 when larval cestodes were more frequent. Most derived parameters for component community structure varied considerably between sites and the two surveys, the hierarchical order for sites not being maintained between surveys. They were susceptible to the disproportionate influence of three relatively rare, unpredictable species with the greatest overall aggregated distribution among hosts. Jaccard's similarity index was less influenced by the rare species, showing greater stability between sites and across years. In conclusion, temporal variation confounded any site-specific characteristics of the summary measures quantified in this study and their usefulness is therefore restricted to the years in which the surveys were conducted.


I N T R O D U C T I O N
The study of helminth parasite communities in small wild mammals has attracted a lot of attention over the years, and at the heart of the on-going debate is the uncertainty about the degree to which these communities are structured, interactive or isolationist, stable or ephemeral (and therefore the extent to which they are predictable) and the relative importance of the factors that contribute to variation in community structure (Montgomery and Montgomery, 1990 ;Poulin, 1997Poulin, , 2007. Typically, hierarchical arrangements characterize helminth communities (Esch et al. 1990 ;Poulin, 2004) and to some extent structure, stability and predictability depend upon the level of organization that is considered. However, the extent to which the derived parameters reflecting the characteristics of each community at each level of organization can vary temporally or spatially has seldom been evaluated comprehensively through empirical studies (Hartvigsen and Kennedy, 1993 ;Timi and Poulin, 2003 ;Calvete et al. 2004). Despite the wealth of publications on helminths of rodents, there are few studies that tackle these issues satisfactorily (Kisielewska, 1970 a, b;Haukisalmi et al. 1987 ;Montgomery, 1989, 1990 ;Tenora and Stanek, 1995).
At the highest level of organization is the total helminth fauna of a given species of host across its entire geographical range (Poulin, 1997). Most often, however, animals are sampled on a smaller geographical scale and, based on pooled data derived from different populations of hosts sampled from a range of sites within the defined region, the regional helminth fauna form the next level of organization. Helminths encountered in populations of hosts within any region constitute component communities, and these in turn are comprised of infracommunities represented by all the parasites within individual hosts. Each level of organization is subject to different evolutionary and ecological pressures, and the degree to which derived parameters used to describe communities vary/show temporal and/or spatial stability will depend on the particular organizational level (Poulin, 1997).
Since infracommunities are the shortest lived, corresponding to a fraction of the life-time of their host, we may expect these to be the most dynamic, showing the greatest spatial and temporal variation. In the case of bank voles, Myodes glareolus, infracommunities may last just a few months in the summer period, although autumn-born voles live longer and each year sufficient individuals survive through the long winter to repopulate the woodland habitats in the following spring (Pucek et al. 1969 ;Alibhai and Gipps, 1985). Component communities are presumed to be longer lived (between several and many host life-times, depending on the stability of the particular population), and therefore intuitively might be expected to be more predictable. Changes at this level depend primarily on colonization events introducing new species to the population of hosts, or extinction through loss of transmission potential. The regional parasite fauna should be even more stable, cushioned against excessive temporal variation by the large number of component communities, temporally and spatially linked to differing extents, which are included within it. Most stable of all should be the total helminth fauna of the host species across its entire geographical range, which can change only through evolutionary events such as the capture of new helminth species through host switching or the allopatric subdivision of parasite lineages through host isolation. These processes commence initially within infracommunities and, if they persist these can spread through component communities to eventually feed over time into regional helminth communities and the total helminth fauna of the host.
Helminth component communities are known to vary in composition, species richness, prevalence etc. between populations of hosts living in sites which may be quite proximal to one another, and in some cases these variations have been linked to differences in habitat quality (Montgomery and Montgomery, 1990 ;Abu-Madi et al. 2000 ;Calvete et al. 2004 ;Behnke et al. 2004 ;Eira et al. 2006). In other cases, however, marked variation in component communities has been found in rodents living in virtually indistinguishable habitats . But, are such differences between sites maintained over years or are they ephemeral, and if so over what time scale do they vary ? Is the rank order of measures of community structure stable over the short/medium/long-term or does it change unpredictably across component communities that are spatially distant ? Are predictability or change properties of regional, component or infracommunities ? There are very few long-term studies in this field that adequately address these issues (Montgomery and Montgomery, 1990).
In our earlier work we found that the helminth infracommunities of bank voles were characterized by medium-term stability in a woodland site in N.E. Poland (Bajer et al. 2005). Based on a study conducted in 1999, we also reported detectable and quantifiable differences in community structure of helminths in voles living in 3 semi-isolated woodland habitats in the same region . We re-examined each of these populations in 2002, sampling at the same time of year as in the original study to avoid seasonal factors. Over evolutionary time these three individual subpopulations of voles constitute part of a continuum across the western Palaearctic region to which bank voles are indigenous (Corbett, 1978 ;Spitzenberger, 1999), and by pooling data on helminth infections from all three sites, the derived summary statistics reflect the local helminth fauna of the Mazury region in Poland. Although bank voles can colonize unoccupied territory at the rate of 2-4 . 5 km year x1 given suitable continuous habitat (Smal and Fairley, 1984), their dispersal distance is generally much shorter (Gliwicz and Ims, 2000) and even closely located populations differ genetically (Aars et al. 1998 ;Gerlach and Musolf, 2000) suggesting that mixing is neither frequent nor extensive in the short term. It is therefore unlikely that there had been any substantial or frequent exchange of individuals between the three woodlands in our study during ecological time because of lack of connectivity between the woods (Van Apeldoorn et al. 1992) : they are too distant from one another and are separated by substantial barriers (lakes, canals and open pasture/cultivated land ; see Behnke et al. 2001) that impede the movement of voles. In ecological time each of these essentially represents a semi-isolated population of hosts, and therefore, summary statistics for helminth infections of voles from each location at this intermediate level reflect helminth component communities of different host populations within the region.
Here we first assess how the regional parasite fauna has changed between the two surveys and then quantify the extent to which between-site differences in component community structure are confounded by temporal effects. Given the above discussion concerning our expectations of stability at the regional and component community levels, our null hypothesis is that the regional helminth fauna and helminth component communities in wild bank voles living in semi-isolated sites within the region, are generally predictable and relatively unchanging in the medium term (2-5 years or so), despite the evident within-year seasonal effects, and the expected age-effects on individual infracommunities. In the accompanying paper we consider effects on infracommunities (Behnke et al. 2008).

Voles
In this paper we refer to bank voles as Myodes glareolus following the recent recommendations that Myodes rather than Clethrionomys is the valid name for this genus of rodents (Wilson and Reeder, 2005 ;Carleton et al. 2003). The methods used for trapping rodents, for sampling and processing trapped animals and establishing age categories have all been fully described in Behnke et al. (2001).

Measures of community structure
Measures of component community structure were identical to those used by Behnke et al. (2001), following Kennedy and Hartvigsen (2000). The total number of helminth species, the Berger-Parker Dominance Index and the unbiased Simpson's Index of Diversity (ID) for each site in each year and on pooled data are given. In some analyses we have excluded S. petrusewiczi, and in others this species as well as the 2 larval cestodes, M. lineatus and C. globifera, are excluded because these three species can cause extremely high infections, numbering several hundred/thousand worms, despite an overall low prevalence in the population. We tested the idea that these species, in particular, exert a disproportionate influence on indicators of community structure, and we compared values with and without these species.
Similarity between sites and in different years was compared by Jaccard's and Sale's Similarity Indices for the latter following Montgomery and Montgomery (1990) using : where C ij is the proportional similarity between surveys i and j, and varies from 0 to 1. p ik xp jk are the proportions of the total helminth fauna at each survey comprising the kth species and s is the number of species. The sign of p ik xp jk is converted to positive before summing.

Statistical analysis
The statistical approach adopted has also been documented comprehensively in the earlier publications Bajer et al. 2005).
Prevalence data (percentage of animals infected with all helminths) are given with 95 % confidence limits, calculated as described by Rohlf and Sokal (1995) employing bespoke software. The distribution of voles, among age and sex classes and in relation to different sites and the years of the surveys, and prevalence of infection with all helminths combined were analysed by maximum likelihood techniques based on log linear analysis of contingency tables using the software package SPSS (version 12.0.1). Full factorial models incorporated the intrinsic factors age (3 age classes) and sex (2 levels, males and females) and the extrinsic factors site (3 levels) and year of survey (2 levels, 1999 and 2002). Infection was considered as a binary factor (present/absent). Beginning with the most complex model, involving all possible main effects and interactions, those combinations that 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 x 2 was not significant, indicating that the model was sufficient in explaining the data. The importance of each term (i.e. interactions involving infection) in the final model was assessed by the probability that its exclusion would alter the model significantly and these values are given in the text.
We used Taylor's Power Law (Taylor, 1961) to identify species of helminths that showed most variation and hence less predictability. The logarithms of mean abundance and variance (of abundance) were calculated separately for each of the 3 sites and in the 2 years of the surveys for each of the 7 most prevalent species of helminths recorded, and when prevalence was zero, those cells were excluded. The line graphs were fitted in Microsoft Office Excel 2003 (by least squares methods) and the statistical analysis of the regression of log transformed variance on abundance was carried out in SPSS version 12.01 for Windows.

Myodes glareolus
In total, 358 bank voles were sampled from the 3 study sites : 139 in 1999 and 219 in 2002. The structure of the sampled host population by year, site and age is summarized in Table 1.
Changes in the regional helminth community structure with time Quantitative changes. In total, combining all 3 sites and both years, 13 501 helminths were recovered, but there was a substantial reduction (71 . 8 %) in the total number of helminths recovered between 1999 (10 529) and 2002 (2972) even though more voles were sampled in 2002 (Table 1). When sample size was taken into consideration the reduction was even more marked (82 . 1 %; mean in 1999=75 . 7 and in 2002=13 . 6). This change between years was predominantly due to S. petrusewiczi which accounted for 70 . 5 % (9516) of all the worms recovered, but which declined in abundance substantially between the 2 surveys (Behnke et al. 2008). Without S. petrusewiczi and with sample size taken into consideration, helminth burdens declined moderately (by 13 . 5 % ; mean number of helminths in 1999=12 . 1 and in 2002=10 . 5).
Two other species had a marked effect on the total number of helminths recovered : Mesocestoides sp. and C. globifera, both of which are relatively infrequent but when present, often number in hundreds. Without these species, and without S. petrusewiczis and with sample size taken into consideration, the remaining total worm burdens changed more markedly (totals of worms recovered=1547 in 1999, and 1388 in 2002 and means=11 . 1 and 6 . 3 respectively). The change represented a reduction of 43 . 0%, mainly attributable to these species increasing abundance in 2002 (when 460 M. lineatus and 450 C. globifera were recovered, compared with 140 and zero, respectively in 1999).
Two hundred and ninety one voles ( Taxa comprising the regional community. Fourteen species of helminths were recorded in total, 8 nematodes and 6 cestodes (Tables 2 and 3 (Table 2 and for further details see Behnke et al. 2008). The regional helminth community, combining all the data, was clearly dominated by nematodes which accounted for 90 . 7 % of all the helminths (68 . 6 % when S. petrusewiczi was excluded). Larval    Kennedy and Hartvigsen (2000). The Berger-Parker dominance index is the proportion of all helminths sampled represented by the dominant species. The dominant species is that showing the highest proportion in each subset of data. The unbiased Simpson's index of diversity was calculated as (1xS i p i 2 )r(n/(nx1)) where n=total number of helminths in data subset and p is the proportion of the i th species (Lande, 1996) and even more so when S. petrusewiczi was excluded (47 . 5 % versus 9 . 5 % of total helminths).

Measures of component community structure
Higher taxa. Eighty nine percent of the helminths recovered were from Urwitałt and Tałty (Table 4), but whereas at Urwitałt the values were comparable in 1999 and 2002, at Tałty the percentage distribution fell by half, and in 2002 a greater percentage of the helminths was recovered from Pilchy (25 . 9% in 2002 versus 6 . 3 % in 1999). Examining the higher taxa more closely, values for intestinal nematodes, adult and larval cestodes were more stable at Urwitałt compared with Tałty and Pilchy. In both years the highest percentages of adult cestodes were observed in voles from Urwitałt, and despite a fall of approximately one third, most of the larval cestodes were recovered in both years from voles from Tałty. When S. petrusewiczi was excluded from the data, the percentage distribution of the remaining helminths was more similar across the 3 sites in the combined data set and in 2002. However, in 1999 the situation was reversed in comparison to the earlier analysis. This time, 46 % of the remaining helminths were encountered at Pilchy (rather than just 6 . 3 % as in the total database). Exclusion of M. lineatus and C. globifera changed the balance even more in favour of Pilchy in 1999 and in the combined database, although in 2002 there was little difference between Pilchy and Urwitałt. Much the same was evident for intestinal nematodes. Exclusion of S. petrusewiczi resulted in most remaining nematodes being attributable to voles from Pilchy.
Total species richness, dominant species and diversity indices by site. Between years, the number of helminth species recorded at Urwitałt and Tałty increased by 3 and 5, respectively and at Pilchy decreased by 1 (Table 3). Berger-Parker dominance indices were lower at all three sites in 2002 and the highest index was obtained for Tałty in 2002, compared with Urwitałt in 1999. In both years the index was lowest at Pilchy.
At both Urwitałt and Pilchy the dominant species remained the same (S. petrusewiczi and A. tetraptera, respectively) but at Tałty Mesocestoides sp. replaced S. petrusewiczi in 2002 (Table 3). Because the communities were less dominated by the oxyuroid nematodes in 2002 compared with 1999, and with the additional species recovered at 2 sites, Simpson's indices of diversity were higher in 2002, overall and at each site.
Similarity indices. Table 5 shows the values calculated for Sale's and Jaccard's similarity indices. Jaccard's index, which is based on presence/absence data, varied less (range=0 . 44-0 . 81) than Sale's index, which takes relative abundance of species into consideration.
The greatest similarity in Jaccard's index was between Urwitałt in both years (1999and 2002) and between Urwitałt 1999and Tałty in both 1999, between Urwitałt and Tałty in 2002, and 2002 and also between these two sites with years combined. Values for Pilchy were generally lower irrespective of the comparison. When S. petrusewiczi, M. lineatus and C. globifera were excluded the index changed very little in most cases, a reflection of the fact that these species were relatively rare in the population.  Sale's index showed markedly greater changes, ranging from 0 . 01 to 0 . 93. Between years, Sale's index was very similar and midrange at Urwitałt and Pilchy, but dropped markedly at Tałty. Urwitałt and Tałty were very similar in 1999, but following the change at Tałty between the years, far less so in 2002, Sale's index dropping from 0 . 93 to 0 . 28. Nevertheless, with years combined Urwitałt and Tałty were very similar but quite different from Pilchy. When S. petrusewiczi, M. lineatus and C. globifera were excluded the index changed quite dramatically in most cases, emphasizing that despite their low prevalence in the population, these helminths had an enormous influence on the index. The greatest similarity was between Tałty and Pilchy, both of which now generated a much higher index for the comparison between years.

Relationship between variance and mean abundance for the most prevalent taxa
The data suggest that S. petrusewiczi and the larval cestodes, M. lineatus and C. globifera, have the greatest influence on the indicators of community structure both at the local regional fauna level and the component community level. Therefore we examined the relationship between the logarithms of the means of abundance and logarithms of the variances of abundance (Taylor, 1961) with these species (C. globifera could not be tested because of insufficient degrees of freedom since it was encountered only in 2 sites in 2002, but not at all in 1999) and compared these to the slopes for this relationship generated for the other prevalent species. We expected S. petrusewiczi and M. lineatus to show the steepest relationships with the highest values for the slope (b). The results are summarized in Fig. 1 and the statistical analysis is given in Table 6, and these show that the slope of the regression of log variance of the mean abundance of M. lineatus on log mean abundance was steeper than that of any of the other species. S. petrusewiczi had a very similar slope, both just less than 2. The shallowest slopes were those of H. mixtum and H. glareoli, and only the former among these two generated a slope less than 1. It is worth noting that a slope of less than 1 for Taylor's Power Law indicates an under-dispersed, uniform distribution, unusual for a parasitic organism.

D I S C U S S I O N
The data and the analysis presented in this paper show that the majority of the indicators of regional helminth fauna and component community structure were remarkably fluid, with substantial variation between the 2 years. Given the evidence that there is some stability of the more common species in the medium-term at the infracommunity level (Bajer et al. 2005 ;Behnke et al. 2008), and our expectations for stability at the higher levels of community organization (regional and component communities ; see Introduction section), this was not what we had expected. The temporal variability of the derived values utilized in this and in other studies (Montgomery and Montgomery, 1990) to summarize regional and component community structure is therefore of fundamental interest, and raises the question of whether they serve any useful purpose. It is pertinent to assess next the species that were primarily responsible for temporal instability of the key measures.  Table 4.
At the core of this between-year variation were 3 species, the oxyuroid S. petrusewiczi and two cestodes, M. lineatus and C. globifera, that exploit bank voles as their intermediate hosts. All 3 species show low to modest prevalence in the bank voles in the region Bajer et al. 2005) and elsewhere (Le Pesteur et al. 1992 ;Haukisalmi and Henttonen, 1993), but when they do occur their numbers can be extremely high. This was reflected in this study in the highest gradients for Taylor's Power Law relationship, and in the case of S. petrusewiczi the low values of k, the aggregation constant, and high indices of dispersion and discrepancy (see accompanying paper, Behnke et al. 2008).
The highest individual worm burden recorded for S. petrusewiczi in this study was 4026 worms (mature male vole from Urwitałt in 1999), this one vole accounting for 29 . 8 % of all the helminths recovered from the 358 voles in the study. Although not common, in our experience such heavily infected animals are not rare either. In another study we reported a bank vole with 6130 worms of this species (Bajer et al. 2005). In fact, in the current study, only 25 voles were infected and of these 8 had worm burdens in the range 181 to 4026. Thus, although prevalence was low (Behnke et al. 2008) S. petrusewiczi nevertheless contributed substantially to total worm burdens and the Berger-Parker index, Sale's similarity index and, as can be seen from Table 3, figured prominently among the dominant species by this criterion. Bajer et al. (2005) discussed some of the underlying factors that are responsible for this aggregation of worms in specific hosts. Syphacia spp. females migrate through the intestine when fully gravid and emerge through the anal sphincter to lay their eggs in large batches on the peri-anal surface. Since the eggs are rapidly infective, transmission most probably relies on close contact between animals, for example when sleeping in family groups in their nests and allogrooming, and if one individual becomes infected through contact with voles from outside social groups, the worms spread rapidly through the remainder of the group, particularly among naïve juveniles (Lewis, 1987). There is also some evidence that prevalence and abundance of Syphacia spp. increase when the population density is high and fall when it is low (Lewis, 1968 ;Montgomery and Montgomery, 1988 ;Haukisalmi and Henttonen, 1990), and the substantial drop in the prevalence of S. petrusewiczi between 1999 (when vole densities were very high ; 18 . 8/1000 trap hours at Urwitałt, see Bajer et al. 2001) and 2002 (when they were lower at 8 . 6, 8 . 1 and 15 . 7/1000 trap hours at Urwitałt, Tałty and Pilchy respectively ; Behnke et al. unpublished data) is consistent with this view. Because S. petrusewiczi were aggregated in just a few hosts in large numbers, they had a major impact on indicators of regional fauna and component community structure. When prevalence and abundance were high, such as in 1999, the influence of S. petrusewiczi on diversity indices was so overwhelming that other species hardly contributed to the calculations, generating Berger-Parker dominance indices above 0 . 9 (i.e. dominant species accounting for more than 90 % of all worms) and correspondingly low Simpson's indices of diversity. Thus the absence of any animals heavily infected with S. petrusewiczi in 2002 (the heaviest infection was 388) was mainly responsible for the substantial drop in mean helminth burdens between the two years of our study, fall in the Berger-Parker dominance index and increase in Simpson's ID, illustrating how derived summary statistics such as those commonly used as measures of component community structure can be easily biased to give an illusion of major change.
The two cestodes exploit bank voles as intermediate hosts, but as far as is known, neither replicates asexually in these hosts  as some other cestodes, including Mesocestoides corti, Table 6. Taylor's power law relationships for the most prevalent species in the study (For each species the log of the variance of the mean abundance was regressed on the log of the mean abundance, with data calculated separately for each of the three sites in each of 2 years, hence maximum n=6. However, because in some subsets of data specific parasites were not found (prevalence=0 %) these subsets were excluded, and for this reason the degrees of freedom fall below 4 in some cases.) are known to do (Specht and Voge, 1965 ;Chernin and McLaren, 1983). Although we have seen many infected animals, we have never encountered dividing forms and this is consistent with the findings of . It is worth also drawing attention here to the controversy about the species of Mesocestoides in bank voles in the region. A second species, M. leptodactylus was described by Loos-Frank (1980 a) and shown to be the species infecting voles in Germany (Loos-Frank, 1980 b). However, it was considered nomen dubium by Premier (1983) and Tenora (2004) and we have chosen to refer to it by its original name M. lineatus. The definitive hosts for M. lineatus are wild carnivores and the most likely host is the fox (Vulpes vulpes), although raccoon dogs (Nyctereutes procyonoides) also inhabit the forests where we trapped. Rodents probably become infected with M. lineatus when they feed on oribatid mites that in turn must get infected by feeding on contaminated food or directly on the faeces of the definitive host, although the role of a first intermediate host in the life-cycle of this species is still not fully resolved (Loos-Frank, 1991). M. lineatus was rare in our voles, affecting just 8 out of the 358 voles, but the heavy infections, 4 of which were in the range 35-393 (and total number of worms re-covered=612), had a disproportionate influence on summary statistics for the subsets of voles to which they belonged. C. globifera matures in owls and raptors such as kestrels, hawks and buzzards (Ferrer et al. 2004), for which bank voles are a major dietary component. Heavy infections in voles (2 infected animals harbouring 200 and 250 larvae respectively) are probably attributable to the consumption of large batches of transmission stages on just one occasion, probably through feeding on bird faeces or vegetation heavily contaminated with eggs from freshly deposited faeces. Parasite burdens with these species were not as numerous as those with S. petrusewiczi (Behnke et al. 2008), and both were more common in 2002, partly compensating for the loss of S. petrusewiczi in that year. Another species that had some effect on summary measures was A. tetraptera. This species is normally considered to be a parasite of Mus domesticus and M. musculus (Behnke, 1975(Behnke, , 1976, but it has also been recorded from bank voles elsewhere (Sharpe, 1964). Experimental data show that M. glareolus cannot be easily infected with eggs derived from M. musculus (Behnke, 1974), and so it is possible that the form in bank voles is a cryptic species not yet clearly differentiated from that infecting Mus spp. However, we are not aware of any molecular genetic studies characterizing Aspiculuris spp. and comparing isolates from M. glareolus and Mus spp. The prevalence of A. tetraptera also increased from 1999 to 2002, but curiously the abundance fell overall (Behnke et al. 2008), and we have data that suggest a long-term trend of rising prevalence of this species across several years (Bajer et al. unpublished). As the accompanying paper shows (Behnke et al. 2008) the prevalence of A. tetraptera remained stable at about 58 % at Pilchy, but increased markedly at both Urwitałt and Tałty. Thus, in addition to the major fluctuations in summary statistics caused by S. petrusewiczi and the two cestodes, there is also a slower long-term trend reflected in the increasing importance of this species in the component community structure of bank voles in our study sites.

Species
Underlying the dramatic changes associated with S. petrusewiczi and the slower long-term change with A. tetraptera, there are also changes in prevalence and abundance of the more stable core species, such as H. mixtum and H. glareoli, and for both of these site effects are more important than temporal ones (see Behnke et al. 2008).
Although most of the summary statistics reflecting regional and component community structure could not be tested statistically, it is evident that, to a large extent, the more marked changes between the years were driven by the disproportionate influence of S. petrusewiczi, M. lineatus and C. globifera. In contrast, the core species provided an element of underlying stability in the summary indices between years and the sporadic occurrence of the other rare species probably made little difference to these calculations other than by contributing to species richness at regional helminth fauna and component community levels. Whilst the range of rarer species in any given location can be eventually assessed comprehensively, it requires large sample sizes since species richness at the community level is dependent on sample size (Walther et al. 1995 : Dove andCribb, 2006). To a large extent these rare species are not specific to bank voles and their transmission is dependent on the presence of other hosts in the vicinity. Taenia martis and T. mustelae are both parasites of mustelids (Verster, 1969 ;Loos-Frank, 2000), and their population sizes and worm burdens determine the risk of infection for voles, which act as intermediate hosts.
Both hosts are highly mobile mammals, and their populations probably show marked fluctuations in specific woodlands, but when they are present they contribute infective stages of the tapeworms they harbour, adding species richness at the component community level for that site.
We conclude that, with the exception of overall prevalence of helminths at the regional faunal level and Jaccard's similarity index at the component community level, other indicators of regional and component community structure in bank voles in our study sites varied considerably between years and the rank order of sites in relation to any specific measure changed unpredictably. Attention has been drawn previously to the variability in measures of helminth species richness between surveys of the same host (Poulin and Mouillot, 2004). To a large extent the variation in the derived measures reported in the current study was accounted for by those species of helminths that have the potential for extreme aggregation, developing exceptionally high worm burdens, or depend on the presence of vagile definitive hosts that seed the environment with transmission stages and have a greater potential than bank voles to move between sites, and hence to crosscontaminate bank vole populations. A further tier of complexity was derived from those helminth species whose prevalence in a specific component community was subject to long-term trends of increasing prevalence and abundance. Clearly the usefulness of the summary measures, reflecting regional helminth fauna and component community structure of helminths in bank voles, employed in this study was restricted to the years in which the surveys were conducted. These indices varied considerably between the two years of the surveys, and hence temporal changes confounded any site-specific characteristics at this level. On these grounds we can now reject our prediction that helminth communities in bank voles should not vary markedly between years at the regional and component community levels.
We are grateful to the M.B. Grabowski Fund for the provision of a grant for this project and the State Committee for Scientific Research (KBN), Poland and the British Council UK (BC) for additional financial support (KBN-BC Grants UM 855 and UM 930). We thank Professor M. Kozakiewicz and Dr Grzegorz Gorecki for use of the field station at Urwitałt (Department of Ecology, Institute of Zoology). We are indebted to D. Bray, T. Race, J. Dinmore, K. Frake and J. Osmond who contributed to both field and laboratory work conducted in 1999. We also thank Dr Tom Reader for his advice with statistical analyses. Finally we wish to acknowledge the support of the forestry departments responsible for the woodland sites utilized in our study (Nadleśnictwa Giżycko, Mikołajki and Orzysz).