INTRODUCTION
Bartonella spp. are vector-borne bacteria associated with numerous emerging infections in humans and animals (Breitschwerdt and Kordick, Reference Breitschwerdt and Kordick2000). Bartonella spp. typically parasitize the erythrocytes of mammalian hosts, resulting in long-lasting infections (Seubert et al. Reference Seubert, Schulein and Dehio2002). Diverse Bartonella species/strains have been isolated recently from a wide range of wild mammals, including rodents (Kosoy et al. Reference Kosoy, Regnery, Tzianabos, Marston, Jones, Green, Maupin, Olson and Childs1997; Heller et al. Reference Heller, Riegel, Hansmann, Delacour, Bermond, Dehio, Lamarque, Monteil, Chomel and Piémont1998; Hofmeister et al. Reference Hofmeister, Kolbert, Abdulkarim, Magera, Hopkins, Uhl, Ambyaye, Telford, Cockerill and Persing1998; Birtles et al. Reference Birtles, Hazel, Bennett, Bown, Raoult and Begon2001; Holmberg et al. Reference Holmberg, Mills, McGill, Benjamin and Ellis2003; Tea et al. Reference Tea, Alexiou-Daniel, Papoutsi, Papa and Antoniadis2004; Jardine et al. Reference Jardine, Appleyard, Kosoy, McColl, Chirino-Trejo, Wobeser and Leighton2005; Knap et al. Reference Knap, Duh, Birtles, Trilar, Petrovec and Avsic-Zupanc2007). In Europe, the most common small mammal species in rural environments are bank voles (Myodes glareolus), field voles (Microtus agrestis), common voles (Mi. arvalis), wood mice (Apodemus sylvaticus), yellow-necked mice (A. flavicollis) and common shrews (Sorex araneus) and these have been identified as reservoirs of many microparasites, including Bartonella spp. but also Anaplasma phagocytophilum, Borrelia burgdorferi and Babesia microti, all of which are also important pathogens of humans and domesticated animals (Healing, Reference Healing1981; Birtles et al. Reference Birtles, Harrison and Molyneux1994; Bown et al. Reference Bown, Bennet and Begon2004; Pawełczyk et al. Reference Pawełczyk, Bajer, Behnke, Gilbert and Siński2004; Siński et al. Reference Siński, Bajer, Welc, Pawełczyk, Ogrzewalska and Behnke2006; Bray et al. Reference Bray, Bown, Stockley, Hurst, Bennett and Birtles2007; Welc-Falęciak et al. Reference Welc-Falęciak, Bajer, Behnke and Siński2008b). Different groups of arthropods (fleas, lice, sandflies) act as vectors for Bartonella spp. Among rodent Bartonella spp., fleas are well-recognized vectors (Bown et al. Reference Bown, Bennet and Begon2004) but transmission by the tick Ixodes ricinus has also been suggested recently (Cotté et al. Reference Cotté, Bonnet, Le Rhun, Le Naour, Chauvin, Boulouis, Lecuelle, Lilin and Vayssier-Taussat2008).
There are relatively few studies on the ecology of Bartonella spp. in naturally infected hosts (Birtles et al. Reference Birtles, Hazel, Bennett, Bown, Raoult and Begon2001; Kosoy et al. Reference Kosoy, Mandel, Green, Marston and Childs2004; Telfer et al. Reference Telfer, Begon, Bennett, Bown, Burthe, Lambin, Telford and Birtles2007a). Rodents constitute very good models for such studies because of the high heterogeneity and dynamics of their populations, facilitating investigations on the contribution of a range of quantifiable intrinsic and extrinsic factors to the cause of the patterns of variation observed in the field. Each rodent community is subdivided into different functional subgroups including for example settled, territorial adults of both sexes and mobile juveniles. The role of these subgroups as hosts for Bartonella spp. is not known, nor has the precise contribution of unpredictable external factors (i.e. temperature, humidity) that create complex unique temporary combinations of environmental effects been quantified comprehensively. Little is currently known about the long-term dynamics of Bartonella in host populations/communities, because of the limited number of systematic long-term ecological studies in naturally infected hosts (Kosoy et al. Reference Kosoy, Mandel, Green, Marston and Childs2004). In our preliminary studies in N.E. Poland, we demonstrated that here wild rodents harbour 2 Bartonella species, including B. grahamii – a species associated with human illness (Welc-Falęciak et al. Reference Welc-Falęciak, Paziewska, Bajer, Behnke and Siński2008a), and were generally heavily infested with parasitic arthropods, especially fleas and immature stages of the tick I. ricinus (Pawełczyk, Reference Pawełczyk2003; Welc-Falęciak et al. Reference Welc-Falęciak, Bajer, Behnke and Siński2008b).
The aims of the present study were (1) to evaluate the stability of Bartonella infections in 2 rodent communities of 4 species by monitoring infections over a 3-year time period and (2) to evaluate the influence of selected ecological factors (host sex and age; season and year of study) on the infection levels in these rodent populations. Improved knowledge of the ecology of these parasites in naturally infected rodent populations will result in a better understanding of their epidemiology and transmission opportunities in nature.
MATERIALS AND METHODS
Field studies
Small rodents were live-trapped during, at minimum, 5 consecutive nights in the breeding season (May–September) in the years 2004–2006 in 2 habitats: bank voles (Myodes glareolus) and yellow-necked mice (Apodemus flavicollis) in forest sites, common voles (Microtus arvalis) and root voles (Microtus oeconomus) in abandoned fallow agricultural land. Our 3 forest sites were located in Mazury in the North-Eastern corner of Poland, close to the towns of Mikołajki, Ryn and Pisz – named ‘Urwitałt’, ‘Tałty’ and ‘Pilchy’, respectively. The study sites, including a map, were fully described by Behnke et al. (Reference Behnke, Barnard, Bajer, Bray, Dinmore, Frake, Osmond, Race and Siński2001). These 3 sites were chosen on the basis of their similar habitat quality (managed mixed forests). As no significant differences in prevalence/abundance of Bartonella were found among the sites, all the sampled animals were treated as a ‘woodland community’. Fallow land was located close to the field station at Urwitałt, on the east side of Lake Łuknajno and near one of our forest sites. This grassland habitat consists of small dry hillocks (up to 5 m of elevation) and wet lower terrain with small temporary ponds, creating 2 kinds of microhabitats. The dry territory is inhabited mainly by common voles and the damp surroundings of the ponds, by the root voles. Forest species were sampled during 3 seasons (spring [May and June], summer [July/August] and autumn [September]) in the year 2004 and during 2 seasons (spring [May and June] and autumn [September]) in the years 2005–2006. Microtus spp. were caught only during the summer period (July/August) in each year. Rodents were live trapped in wooden traps that were inspected twice daily, and processed according to the procedures described in detail by Bajer et al. (Reference Bajer, Pawełczyk, Behnke, Gilbert and Sinski2001).
Sampling of hosts
At the field station in Urwitałt, all the animals were inspected, identified, sexed, relevant morphometric data were recorded and they were weighed (to the nearest 0·5 g). Ectoparasites (ticks) were removed mostly from ears and limbs and the fur was inspected carefully for fleas. After inspection most animals were live-sampled, marked and released as near as possible to the original site of capture, whilst others (approx. 35%) were bled by cardiac puncture under terminal (ether) anaesthesia.
Ageing of rodents
Three age classes were established on the basis of body weight (Morris, Reference Morris1972) and sexual development, supported with dry lens weight (when available), corresponding to immature juveniles (age class 1), young mature animals (age class 2) and adults (age class 3) (see Behnke et al. Reference Behnke, Barnard, Bajer, Bray, Dinmore, Frake, Osmond, Race and Siński2001; Bajer et al. Reference Bajer, Bednarska, Pawełczyk, Behnke, Gilbert and Siński2002).
Blood collection and DNA extraction
Thin blood smears were prepared from drops of blood taken from the tail vein or heart. Blood smears were air-dried, fixed in absolute methanol and stained for 1 h in Giemsa stain in buffer at pH 7·2. Each smear was examined under oil immersion. From the sacrificed animals, 200 μl of whole blood were collected into 0·001 m EDTA and frozen at a temperature of −20°C. From the live-sampled individuals a few drops of blood were collected from the tail vein into 200 μl of 0·001 m EDTA and frozen. From individuals that were found dead in the trap, the whole heart was isolated and homogenized in 400 μl of 0·001 m EDTA. For this group of animals (12%), blood smears were not obtained and diagnosis for Bartonella infection was done exclusively on the basis of PCR. For the remaining 88% of samples both microscopical analysis of stained blood smears and DNA amplification were used. A strong correlation between the results of these two methods was obtained by the Fisher exact test (only 3% of discrepancy, in favour of PCR). Genomic DNA was extracted from whole blood or heart homogenates using AxyPrep MiniPrep Blood kit (AxyGen, USA) and stored at a temperature of −20°C. The extracted DNA was subjected to a PCR with the primers CS140f (Birtles and Raoult, Reference Birtles and Raoult1996) and BhCS1137n (Norman et al. Reference Norman, Regnery, Jameson, Greene and Krause1995), targeting a specific fragment of the gene encoding the enzyme citrate synthase (gltA) (Birtles and Raoult, Reference Birtles and Raoult1996). The PCR was incubated at 95°C for 2 min to denature genomic DNA and the thermal cycle reaction programmed for 40 cycles of 1 min at 95°C, 1 min at 54°C and 2 min at 72°C, with a final extension step of 7 min at 72°C. PCR products were subjected to electrophoresis on a 1·5% agarose gel and stained with ethidium bromide.
Prevalence (percentage of animals infected) was estimated based on microscopical observations and the PCR, and values are reported with the 95% confidence limits (given in brackets in the text), calculated by bespoke software based on the tables of Rohlf and Sokal (Reference Rohlf and Sokal1995). Abundance of infection was estimated as the number of Bartonella spp. infected erythrocytes/200 fields of vision at×1000 magnification. In cases when the samples were only positive by PCR (7% of all cases), an intensity of 0·001 infected erythrocytes/200 fields of vision was implemented into quantitative statistical analysis. Quantitative data are reported as geometric means (GM means) of all animal in a given subset (infected and non-infected, as defined Margolis et al. Reference Margolis, Esch, Holmes, Kuris and Schad1982), with 95% confidence limits calculated as described by Elliott (Reference Elliott1977).
Statistical analysis
Prevalence of infection was analysed by maximum likelihood techniques based on log linear analysis of contingency tables, implemented by the software package, SPSS v. 14. Prevalence of infection as a binary factor (infected=1, not infected=0) and then habitat (2 levels), host species (4 levels, My. glareolus, A. flavicollis, Mi. arvalis and Mi. oeconomus), year (3 levels), host age (3 levels), host sex (2 levels), season (3 levels, only for forest species) were entered as factors. Beginning with the most complex model, involving all possible main effects and interactions, those combinations not contributing significantly to explaining variation in the data were eliminated stepwise, beginning with the highest-level interaction.
A minimum sufficient model was then obtained, for which the likelihood ratio of χ2 was not significant, indicating that the model was sufficient in explaining the data. The analysis of prevalence of infection was first conducted for all 4 host species together, and when host species appeared as a significant main effect, the analysis was repeated for each host species separately.
In an additional set of analyses, flea and tick prevalence (as a binary factor: infested=1, not infested=0) were entered as factors. However, as no significant correlations resulted from that analysis, these data are not presented in this paper.
Quantitative data reflecting parasite abundance within hosts were expressed as geometric means (GM) because the data were highly overdispersed (Elliott, Reference Elliott1977; Dash et al. Reference Dash, Hall and Barger1988). These means reflect the abundance of infection as defined by Margolis et al. (Reference Margolis, Esch, Holmes, Kuris and Schad1982) and include all subjects within the specified group, infected and not infected, for which relevant data were available. Parasite abundance was analysed by multifactorial GLM using models with normal errors after normalization of the data by log10 (x+1) transformation (Crawley, Reference Crawley1993; Wilson and Grenfell, Reference Wilson and Grenfell1997). The same factors as those used for analysis of prevalence were employed for analysis of abundance, implementing the approach of step-wise backward simplification of models by removal of non-significant terms.
RESULTS
Rodent communities
Overall a total of 1100 rodents were sampled over the 3-year period and of these 74·5% were woodland rodents (of which 80·2% were bank voles, 19·8% were yellow-necked mice) and 25·5% were grassland rodents (of which 53·2% were common voles, 46·8% were root voles). The structure of the sampled host communities/population by year, season, host species and sex is summarized in Table 1. The dynamics of relative densities of hosts are presented in Table 2. During the period of study, prevalence of bacterial infections was similar in the two rodent communities (woodland and grassland rodents – 32·5% and 26·1%, respectively) with 1 leading host species in each habitat (46·3% in A. flavicollis versus 29·1% in My. glareolus, or 36·9% in Mi. arvalis versus 13·7% in Mi. oeconomus) (Table 3).
Table 1. The structure of the sampled host populations by year, season, host species and sex in the 3-year period of study
nd, Not done.
M – males; F – females.
Σ – total.
Table 2. Dynamics of relative densities of hosts
nd, Not done.
Table 3. Overall prevalence and abundance of Bartonella spp. in rodent host (all host species and all years combined)
N, number of animals examined.
Prevalence and abundance of Bartonella spp. in four host species
Overall prevalence and abundance values, across the 3 years of the study by host species and by host sex, are presented in Table 3. The effect of host species, year, sex and age was analysed in a dataset comprising 1085 animals. In the resulting minimal sufficient model, the influence of year of study on Bartonella spp. prevalence was evident (year×Bartonella prevalence: χ22=10·2, P=0·006): 2006 was the year of highest prevalence (33·6% [27·6%–40·3%]) and the lowest number of infected rodents was observed in 2005 (27·3% [22·5%–32·7%]).
Also the influence of interaction of host species and sex on prevalence was marked (Table 3) (host species×sex×Bartonella prevalence: χ32=9·3, P=0·025; goodness of fit for model: χ922=110·9, P=0·088). Prevalence of Bartonella spp. was highest in A. flavicollis (46·3% [37·7%–55·2%) and lowest in Mi. oeconomus (13·7% [9·0%–20·1%]). The multifactorial GLM revealed a strong main effect of host species on the abundance of Bartonella spp. (F 3, 951=11·5, P<0·001). As earlier in the analysis of prevalence, abundance was much higher in mice (geometric mean [GM]=4·66 [3·33–6·41]) in comparison to other host species (Table 3) and lowest in root voles (GM=0·37 [0·00–0·87]). Because of the strong influence of host species on both infection parameters, the analysis was repeated for each host species separately, beginning with full factorial models incorporating year, and season of study (for forest species), host sex and age as factors.
Dynamics of Bartonella spp. infection in My. glareolus
This dataset consisted of 643 bank voles sampled during the 3-year period. There was only 1 significant term in the minimal sufficient model (goodness of fit: χ772=58·7, P=0·928). Infection rates with Bartonella spp. varied markedly in relation to year (year×Bartonella prevalence: χ22=20·2, P<0·001) with the highest noted prevalence in 2006 (36·4% [31·2%–42·0%]) compared to the previous years of study (17·7% [12·2%–24·9%] and 24·1% [17·1%–32·7%] in 2004 and 2005, respectively).
Season was not a significant component of the minimum sufficient model but nevertheless some variation was noted. Bartonella spp. infections were identified in 21·3% [13·7%–31·7%] voles in spring; in 15·1% [7·8%–25·5%] voles in summer and in 32% [29·2%–35·5%] of individuals in autumn. Likewise, host age did not achieve statistical significance and was not a component of the minimum sufficient model, but some pertinent trends were evident. The highest prevalence of Bartonella spp. was noted in juveniles (35·8% [27·7%–44·6%] in age class 1 voles) in comparison to 2 older age classes (29·7% [25·7%–33·9%] and 24·6% [20·6%–29·2%] in age class 2 and 3, respectively).
Multifactorial GLM generated a simple model for the abundance of Bartonella spp. with, again, the only significant factor being the main effect of year (F 2, 551=5·87, P=0·003). The highest GM was noted in 2006 (GM=1·38 [0·77–2·22]) compared to the previous years of study (2004 – GM=0·60 [0·07–1·38]; 2005 – 0·80 [0·26–1·58]).
Dynamics of Bartonella spp. infection in A. flavicollis
This dataset consisted of 162 yellow-necked mice sampled over the 3-year period. In the analysis of the prevalence of Bartonella spp., sex contributed to 2 significant terms in the minimal sufficient model (goodness of fit: χ742=41·0, P=0·999). A higher infection rate was found in male mice compared with females during each season but the extent of the difference between the sexes varied from season to season (Fig. 1A; season×sex×Bartonella prevalence: χ22=6·26, P=0·044). Overall, prevalence of Bartonella spp. was 5 times higher in autumn in comparison to spring (54·5% [47·4%–61·2%] vs 11·8% [2·1%–35.0%]). In each of the 3 years prevalence was higher in male mice (Fig. 1B) and also in the pooled dataset (54·5% [40·6%–67·7%] vs 33·3% [23·7%–44·3%] but the discrepancy between the sexes varied and was least in 2006 (year×sex×Bartonella prevalence: χ22=6·77, P=0·034).
Fig. 1. (A) The effect of host sex and season of study on Bartonella spp. prevalence in yellow-necked mouse (Apodemus flavicollis). (B) The effect of host sex and year of study on Bartonella spp. prevalence in yellow-necked mouse (A. flavicollis).
Multifactorial GLM generated a simpler model for the abundance of Bartonella spp. with 2 main effects (year of study and host sex). The geometric mean number of Bartonella-infected red blood cells was higher in 2006 (GM=47·2 [12·8–175·60] vs 2·90 [0·53–8·95] and 1·33 [0·09–3·97] than in 2004 and 2005, respectively, as for prevalence (main effect of year on Bartonella abundance: F 2, 152=8·64, P<0·001). Markedly higher abundance was found in males (GM=5·98 [3·51–17·11]) compared to females (GM=2·30 [0·00–3·68]) (main effect of sex on Bartonella abundance: F 1, 152=8·87, P=0·003). Host age did not affect Bartonella infection in the yellow-necked mouse population. Prevalence of Bartonella spp. infection varied in the range 27·3% [12·6%–50·0%] – 50% [37·4%–62·6%] among the 3 age classes.
Dynamics of Bartonella spp. infection in Mi. arvalis
This dataset consisted of 149 common voles trapped during the 3-year period. In this host species, neither the analysis of prevalence nor the analysis of abundance revealed any significant terms. The prevalence remained similar in consecutive years of study (2004 – 41·5% [31·7%–51·8%]; 2005 – 37·0% [27·7%–47·3%]; 2006 – 31·0% [17·0%–47·5%]). There was little difference between the two sexes (Table 3) and 3 age classes (age class1 – 25·0% [9·0%–50·0%]; age class 2 – 34·4% [21·0%–52·8%]; age class 3 – 35·2% [23·8%–48·3%]). However, a reversed trend was observed with respect to abundance. The overall highest GM number of Bartonella-infected erythrocytes was noted in age class 1 during the 3 years of study but this was not significant.
Dynamics of Bartonella spp. infection in Mi. oeconomus
This dataset consisted of 129 root voles trapped during the 2-year period (only 2 years of study were analysed because only 2 root voles were trapped in year 2004; Table 1). In the analysis of prevalence of Bartonella spp., none of the initially fitted extrinsic or intrinsic factors significantly affected this infection parameter. Prevalence across age classes varied slightly with the prevalence being higher in the youngest animals (age class 1, 2 and 3 – 27·3% [7·9%–59·6%], 13·0% [6·4%–21·6%] and 12·5% [7·1%–21·2%], respectively) and between sexes (10·3% [4·7%–19·6%] in males, 18·0% [10·9%–27·9%] in females).
Multifactorial GLM generated a simple model for the abundance of Bartonella spp. with 2 main effects (year of study and host age) and an interaction of year of study with host age. A higher geometric mean number of Bartonella infected erythrocytes was noted in 2006 (1·41 [0·57–2·71]) than in 2005 (0·36 [0·03–0·82]) (main effect of year on Bartonella abundance: F 1, 108=4·75, P=0·032). Abundance of infection in juvenile root voles was nearly 8 times higher than in the 2 older age classes (age class 1 – 2·55 [0·77–6·10]; age class 2 – 0·33 [0·06–0·68] and age class 3 – 0·26 [0·00–0·63]) (main effect of age on Bartonella abundance: F 2, 108=3·94, P=0·023). The same pattern of age-related variation was found during the 2 years of study (year×host age×Bartonella abundance: F 2, 108=3·36, P=0·039) although in 2006 in age class 1 mean abundance was up to 9 times higher in comparison to similar values in age class 2 and 3 (Fig. 2).
Fig. 2. The effect of host age and year of study on Bartonella spp. prevalence in the root vole (Microtus oeconomus).
Host sex did not affect Bartonella infection in root voles (Table 3).
DISCUSSION
The results reported in this paper, based on a 3-year ecological study in the Mazury Lake District region of Poland, are consistent with earlier work (Bajer et al. Reference Bajer, Pawełczyk, Behnke, Gilbert and Sinski2001) and extend that study by establishing the relative importance of the 4 host species as reservoirs of Bartonella spp. The hosts were considered as falling into 2 ecological communities, one associated with the forests and one with open fallow grassland and, in each of these 2 communities, the prevalence and abundance of Bartonella infection was high in relation to many other published reports (Engbaek and Lawson, Reference Engbaek and Lawson2004; Tea et al. Reference Tea, Alexiou-Daniel, Papoutsi, Papa and Antoniadis2004). Moreover, in each habitat one host species dominated as the more heavily infected; yellow-necked mice in the forest habitat and common voles on fallow land. Both parameters (prevalence and abundance) were the lowest in root voles, and to our knowledge this is a new host record for Bartonella.
Overall, prevalence of Bartonella spp. in the Mazury Lake District estimated both on the basis of blood-smear examination or DNA amplification was remarkably high (45%). Prevalence in each rodent community was higher than reported from central Sweden for the same and related rodent species (17% in A. flavicollis; 24% in A. sylvaticus; 15% in My. glareolus and 1 infected of 3 tested for Mi. agrestis;Holmberg et al. Reference Holmberg, Mills, McGill, Benjamin and Ellis2003). Prevalence in our study was also higher than reported from small mammal communities in Denmark (28% in Mi. agrestis, A. flavicollis, A. sylvaticus and Sorex vulgaris; Engbaek and Lawson, Reference Engbaek and Lawson2004), Greece (31% in 7 species, including A. flavicollis; Tea et al. Reference Tea, Alexiou-Daniel, Papoutsi, Papa and Antoniadis2004), and comparable to that reported in Slovenia (40% in A. flavicollis, A. sylvaticus, A. agrarius and My. glareolus; Knap et al. Reference Knap, Duh, Birtles, Trilar, Petrovec and Avsic-Zupanc2007) and the UK (64%; Birtles et al. Reference Birtles, Hazel, Bennett, Bown, Raoult and Begon2001), where molecular methods were used to assess the presence of infection. High prevalence of Bartonella infection in naturally infected rodent populations is caused most likely by persistent or long-term bacteraemia, and/or by high incidence rates of infection (Kosoy et al. Reference Kosoy, Regnery, Tzianabos, Marston, Jones, Green, Maupin, Olson and Childs1997; Birtles et al. Reference Birtles, Hazel, Bennett, Bown, Raoult and Begon2001). Our microscopy-based epizootiological studies in rodents revealed that the vast majority of Bartonella infections occur with low bacteraemia, thus suggesting that these infections are in the chronic and persistent phase. In an experimental study, injection of B. birtlesii into BALB/c mice induced a long-lasting bacteraemia of 5 to 8 weeks duration p.i. (Boulouis et al. Reference Boulouis, Barrat, Bermond, Bernex, Thibault, Heller, Fontaine, Piémont and Chomel2001). Given the results presented in this report and the fact that the average life span of wild rodents is about 3–4 months in nature (Pucek et al. Reference Pucek, Ryszkowski, Zejda, Petrusewicz and Ryszkowski1970), it is likely that once acquired, infections persist until the end of the rodent's life. However, there are also strong indications that re-infection can take place under natural conditions and there is evidence for exchange of different species/genotypes in a single individual (Birtles et al. Reference Birtles, Hazel, Bennett, Bown, Raoult and Begon2001).
Between-year variation was well marked for Bartonella spp. in bank and root voles as well as in yellow-necked mice but was less evident in common voles. Between-year dynamics of infection had been observed in bank voles during our earlier study (Bajer et al. Reference Bajer, Pawełczyk, Behnke, Gilbert and Sinski2001). In contrast, in a recent study in the UK no effect of year on Bartonella spp. prevalence was found in common shrews (Bray et al. Reference Bray, Bown, Stockley, Hurst, Bennett and Birtles2007). These contrasting patterns are probably caused by fluctuations in relative population densities of both hosts and vectors (fleas) and resulting alterations in transmission efficiency. The relatively high prevalence of Bartonella infections in bank and common voles and in yellow-necked mice in 2006 may be linked to the higher flea infestations of rodents in 2006 (84%) compared with 2004 (34%) (own data, not published). On the other hand, the positive association between host densities and parasite distribution has been noted in other ecological studies on rodent microparasites (Begon et al. Reference Begon, Hazel, Baxby, Bown, Cavanagh, Chantrey, Jones and Bennett1999; Bajer, Reference Bajer2008).
In our study the prevalence of Bartonella infections in bank voles and yellow-necked mice varied among seasons, increasing from spring to peak in autumn. Under the climatic conditions in N.E. Poland, only a small proportion of animals survive the winter period, and these serve as a source of infection for newly-born cohorts later in the season, among which infections spread in the naïve young animals. A similar pattern was found by Turner (Reference Turner1986), Bajer et al. (Reference Bajer, Pawełczyk, Behnke, Gilbert and Sinski2001) and Telfer (Reference Telfer, Begon, Bennett, Bown, Burthe, Lambin, Telford and Birtles2007a, Reference Telfer, Clough, Birtles, Bennett, Carslake, Helyar and Begonb) and most likely explained by the increase in flea densities from winter to summer (Bown et al. Reference Bown, Bennet and Begon2004). Here, the spring to autumn increase in prevalence of Bartonella spp. was paralleled by flea infestations with the highest level of infestation noted in autumn (68%) in comparison to spring (56%) and summer (33%) (Bajer et al., unpublished data).
Host age played a more important role in the ecology of Bartonella infections than host sex. There was an indication of an age effect on prevalence and abundance of Bartonella spp. in bank voles and a similar tendency for such an effect in root voles. In both cases, the infections were more common and more intense in the 2 youngest age classes (1 and 2) compared with age class 3 suggesting that age-dependent immunity may be a feature of this host-parasite relationship. A similar age-dependent pattern was found by Healing (Reference Healing1981), Pawełczyk et al. (Reference Pawełczyk, Bajer, Behnke, Gilbert and Siński2004) and Holmberg et al. (Reference Holmberg, Mills, McGill, Benjamin and Ellis2003) but not by Young (Reference Young1970), Bajer et al. (Reference Bajer, Pawełczyk, Behnke, Gilbert and Sinski2001) and Bray et al. (Reference Bray, Bown, Stockley, Hurst, Bennett and Birtles2007). As juvenile rodents are more mobile than territorial adults, they may spread the infections between different populations in the local ecosystem.
Fichet-Calvet et al. (Reference Fichet-Calvet, Jomâa, Ben Ismail and Ashford2000) suggested that Bartonella infection may be self-limiting and immunizing, such that older animals that have cleared their earlier infections, remain immune to further infections. On the other hand there is evidence for re-infection in individual rodents (Birtles et al. Reference Birtles, Hazel, Bennett, Bown, Raoult and Begon2001), and this may be a consequence of infection by different species of Bartonella in sequence. The high Bartonella prevalence in young rodents may be a result of more efficient horizontal transmission among naive youngsters or it may be a result of vertical transmission between females and their litters. Isolations of these bacteria from the fetuses of rodents (white-footed mouse and cotton rats) suggest that transmission of Bartonella spp. in utero may occur among natural hosts (Kosoy et al. Reference Kosoy, Regnery, Kosaya, Jones, Marston and Childs1998).
The significance of rodents as a reservoir of pathogens of public health interest depends on the parasite genotypes that actually occur in naturally infected populations. Based on earlier comparative sequence analysis of the citrate synthase gene (gltA) fragment, which is commonly used for the genotyping of Bartonella, a considerable heterogeneity exists among Bartonella isolates derived from these rodent populations (Welc-Falçciak et al. 2008 a). During our preliminary genotyping studies 2 Bartonella species were found – B. taylorii and B. grahamii. B. grahamii has been reported earlier from the ocular fluids of a patient with neuroretinitis, suggesting that these bacteria can be a cause of pathological changes in humans (Kerkohoff et al. Reference Kerkhoff, Bergmans, van Der Zee and Rothova1999) but the infectivity and pathogenic potential of B. taylorii in humans remains unknown.
This study represents the first comprehensive analysis of the Bartonella infection in 2 rodent communities comprising 4 species of hosts in N.E. Europe, and for the first time including Mi. oeconomus. To our knowledge, this is also the first ecological study on the dynamics of natural infections in wildlife that took account of the range of intrinsic and extrinsic factors known to influence Bartonella infection, and the dynamics of relative host densities. This 3-year field study conducted in 2 contrasting habitats confirmed that, overall, Bartonella spp. are prevalent in rodent hosts in N.E. Poland, despite between-year, spatial and seasonal variation. In consequence, there is likely to be widespread dissemination of these bacteria in the local ecosystem, where human encroachment is currently increasing rapidly because of the expanding local populations and increasing numbers of summer visitors that are attracted to this part of Europe by the opportunities for aquatic and outdoor leisure activities.
ACKNOWLEDGMENTS
The study was supported by the Ministry of Science and Higher Education (previous State Committee for Scientific Research [KBN]) Grant no. N303 093134. The study was conducted in accordance with the guidelines of the Local Ethical Committee (No. 576/2006).
