INTRODUCTION
Parasite ecologists have recognized the fundamental role of habitat, seasonality and climate as well as host age, sex and breeding condition in influencing the exposure and susceptibility of hosts to parasitic infections (Wilson et al. Reference Wilson, Bjørnstad, Dobson, Merler, Poglayen, Randolph, Read, Skorping, Hudson, Rizzoli, Grenfell, Heesterbeek and Dobson2002). There is increasing evidence that interactions between parasites may also play a significant role in affecting susceptibility of hosts to infection (Hochberg and Holt, Reference Hochberg and Holt1990; Sousa, Reference Sousa1994; Petney and Andrew, Reference Petney and Andrew1998; Cox, Reference Cox2001; Lello et al. Reference Lello, Boag, Fenton, Stevenson and Hudson2004; Maizels et al. Reference Maizels, Balic, Gomez-Escobar, Nair, Taylor and Allen2004; Faulkner et al. Reference Faulkner, Turner, Behnke, Kamgno, Rowlinson, Bradley and Boussinesq2005; Hartgers and Yazdanbakhsh, Reference Hartgers and Yazdanbakhsh2006; Cattadori et al. Reference Cattadori, Albert and Boag2007, Reference Cattadori, Boag and Hudson2008; Lello and Hussell, Reference Lello and Hussell2008). However, while these findings indicate that we should consider the whole community of parasites to understand the dynamics of each component species, results from previous studies are not always consistent (Behnke et al. Reference Behnke, Bajer, Sinski and Wakelin2001; Poulin, Reference Poulin2001; Behnke, Reference Behnke2008; Graham, Reference Graham2008; Telfer et al. Reference Telfer, Birtles, Bennett, Lambin, Paterson and Begon2008). Studies in controlled laboratory conditions have not only identified strong parasite interactions but also teased apart some of the molecular mechanisms involved (Curry et al. Reference Curry, Else, Jones, Bancroft, Grencis and Dunne1995; Maizels et al. Reference Maizels, Balic, Gomez-Escobar, Nair, Taylor and Allen2004; Edwards et al. Reference Edwards, Buchatska, Ashton, Montoya, Bickle and Borrow2005; Kamal and El Sayed Khalifa, Reference Kamal and El Sayed Khalifa2006; Graham et al. Reference Graham, Cattadori, Lloyd-Smith, Ferrari and Bjørnstad2007; Bradley and Jackson, Reference Bradley and Jackson2008; Fenton et al. Reference Fenton, Lamb and Graham2008). These experiments are usually undertaken with high parasite doses or parasite and host strains that have been selected for laboratory purposes, and exhibit high responding characteristics. One consequence of these laboratory conditions is that the findings may not be relevant to the epidemiology of free-living natural host-parasite systems. If such interactions are relevant then we would expect to see this reflected in the dynamics of the infra-community of parasites of free-living hosts. Counter to this expectation, comparative studies of field data have found that the variation in the distribution and abundance of parasite infra-communities are mainly the result of variation in host exposure and habitat characteristics (Christensen, Reference Christensen, Nansen, Fagbemi and Monrad1987; Haukisalmi and Henttonen, Reference Haukisalmi and Henttonen1993a, Reference Haukisalmi and Henttonenb; Lotz and Font, Reference Lotz and Font1994; Sousa, Reference Sousa1994; Behnke et al. Reference Behnke, Bajer, Sinski and Wakelin2001, Reference Behnke, Gilbert, Abu-Madi and Lewis2005, Guègan et al. Reference Guègan, Morand, Poulin, Thomas, Renaud and Guégan2005; Behnke, Reference Behnke2008). According to these findings, parasite interactions play a trivial role in free-living systems.
Parasite interactions can either be synergistic (positive) or antagonistic (negative). A synergistic interaction occurs when a parasite species increases its fitness (life-time, reproductive output) as a consequence of the presence of another species, whereas an antagonistic interaction leads to a fitness reduction of one of the parasite species as a response to the presence of the other species. Such interactions may arise from different mechanisms, ranging from direct competition for resources to indirect host-mediated effects, including the active role of the immune system or the passive effect of parasite produced metabolic compounds made available through the host (Sousa, Reference Sousa1992; Rohde, Reference Rohde1994; Behnke et al. Reference Behnke, Bajer, Sinski and Wakelin2001; Cox, Reference Cox2001; Lello et al. Reference Lello, Boag, Fenton, Stevenson and Hudson2004; Maizels et al. Reference Maizels, Balic, Gomez-Escobar, Nair, Taylor and Allen2004; Graham et al. Reference Graham, Cattadori, Lloyd-Smith, Ferrari and Bjørnstad2007; Bradley and Jackson, Reference Bradley and Jackson2008; Fenton et al. Reference Fenton, Lamb and Graham2008).
In this paper we address 2 broad questions (i) are the dynamics of any one parasite species altered by the presence of other parasites in the infra-community and (ii) are the effects of any observed interaction relevant to the epidemiology of free-living host-parasite systems? We examined these questions by studying 2 common parasitic species of the yellow-necked mouse (Apodemus flavicollis): the gastrointestinal nematode Heligmosomoides polygyrus and the ectoparasitic tick, Ixodes ricinus. Previous laboratory studies have shown that the immature infective stage of the closely related species Heligmosomoides bakeri elicited protective immunity, but adult parasites down-modulated the humoral immune response and may establish chronic infections (Monroy and Enriquez, Reference Monroy and Enriquez1992; Behnke et al. Reference Behnke, Wahid, Grencis, Else, Bensmith and Goyal1993, Telford et al. Reference Telford, Wheeler, Appleby, Bowen and Pritchard1998; Maizels et al. Reference Maizels, Balic, Gomez-Escobar, Nair, Taylor and Allen2004, Cable et al. Reference Cable, Harris, Lewis and Behnke2006). This immuno-suppressive action permitted greater parasite survival and reproduction but also facilitated the infection by other parasite species (Colwell and Wescott, Reference Colwell and Wescott1973; Courtney and Forrester, Reference Courtney and Forrester1973; Jenkins, Reference Jenkins1975; Bruna and Xenia, Reference Bruna and Xenia1976; Jenkins and Behnke, Reference Jenkins and Behnke1977; Behnke et al. Reference Behnke, Wakelin and Wilson1978; Behnke and Ali, Reference Behnke and Ali1984; Alghali et al. Reference Alghali, Hagan and Robinson1985). For example, when laboratory mice were concurrently infected with H. bakeri and Trichinella spiralis there was a reduction in the acute response against the second nematode (Behnke et al. Reference Behnke, Wakelin and Wilson1978). Field investigations have identified an association between H. polygyrus and other gastrointestinal parasites, but failed to establish whether H. polygyrus played a role in structuring the parasite community (Behnke et al. Reference Behnke, Gilbert, Abu-Madi and Lewis2005).
To examine the effect of H. polygyrus in natural systems, we monitored the numerical response of I. ricinus in free-living yellow-necked mice. We focused on I. ricinus because tick abundance can be monitored on serially caught mice without invasive techniques. Secondly, artificial infestations of I. ricinus on yellow-necked mice have identified a progressive suppression of the protective response that resulted in increased host susceptibility (Dizij and Kurtenbach, Reference Dizij and Kurtenbach1995). We first conducted an extensive cross-sectional study in different mouse populations and examined how the relationship between tick infestation and H. polygyrus infections were affected by both environmental factors and host characteristics (breeding condition, sex and age). Second, we undertook an experimental manipulation (reduction/increase) of H. polygyrus abundance in mice and monitored changes in tick infestation. The hypothesis we tested was that H. polygyrus will show suppressive effects on host response, similar to the closely related species H. bakeri, and this will result in an increase in I. ricinus infestation per host.
MATERIALS AND METHODS
Species description and monitoring
H. polygyrus inhabits the small intestine of yellow-necked mice and has a direct life cycle with infection occurring after the ingestion of third-stage larvae, either with contaminated food or through grooming (Slater and Keymer, Reference Slater and Keymer1986; Hernandez and Sukhdeo, Reference Hernandez and Sukhdeo1995). Adult males, as opposed to females, were responsible for the majority of the nematode transmission (Ferrari et al. Reference Ferrari, Cattadori, Nespereira, Rizzoli and Hudson2004). H. polygyrus represented the most common helminth with population prevalences reaching 85% (Rosso et al. Reference Rosso, Manfredi, Ferrari, Scalet and Rizzoli2002; Ferrari, Reference Ferrari2005). Six additional helminth species were also identified but occurrence was always very low; the second most common parasite was Hymenolepis fraterna with a mean prevalence of 34·8% (Rosso et al. Reference Rosso, Manfredi, Ferrari, Scalet and Rizzoli2002; Ferrari, Reference Ferrari2005).
I. ricinus is found with a high prevalence in the Trentino Province (Chemini et al. Reference Chemini, Rizzoli, Merler, Furlanello and Genchi1997; Rizzoli et al. Reference Rizzoli, Merler, Furlanello and Genchi2002, Reference Rizzoli, Rosà, Mantelli, Pecchioli, Hauffe, Tagliapietra, Beninati, Neteler and Genchi2004; Carpi et al. Reference Carpi, Cagnacci, Neteler and Rizzoli2008; Rosà et al. Reference Rosà, Pugliese, Ghosh, Perkins and Rizzoli2007) and infests a large range of mammalian species although it is the primary tick species found on yellow-necked mice (Perkins et al. Reference Perkins, Cattadori, Tagliapietra, Rizzoli and Hudson2003, Reference Perkins, Cattadori, Tagliapietra, Rizzoli and Hudson2006). Infestation can be as high as 74 larvae per mouse and the distribution within the host population is aggregated with large body mass breeding males, carrying large infestations (Perkins et al. Reference Perkins, Cattadori, Tagliapietra, Rizzoli and Hudson2003, Reference Perkins, Cattadori, Tagliapietra, Rizzoli and Hudson2006). There are 3 tick stages: larvae, nymphs and adults (Sonenshine, Reference Sonenshine1992). In general the larvae, and a smaller proportion of nymphs, feed on small mammals and a meal lasts for 3–5 days (Sonenshine, Reference Sonenshine1992; Randolph, Reference Randolph1998). Mice reduce tick infestations through self-grooming (Osfeld et al. Reference Osfeld, Miller and Schnurr1993). In yellow-necked mice the immune response toward I. ricinus does not provide a protective defence and multiple tick infections may induce immuno-suppression (Dizij and Kurtenbach, Reference Dizij and Kurtenbach1995).
Yellow-necked mice were sampled in the Dolomitic Alps of the Province of Trentino, Northern Italy, in 2002 (Fig. 1). We used multi-capture live traps (Ugglan type 2, Graham Sweden) located in broadleaf woodlands with mature stands of beech (Fagus sylvatica), some Scots pine (Pinus sylvestris), spruce (Picea abies) and a sparse under-storey (mean altitude±s.e.: 850±22 m a.s.l.), the typical habitat of yellow-necked mice in the Italian Alps (Locatelli and Paolucci, Reference Locatelli and Paolucci1998).
Location of the extensive and intensive study areas: black points identify the 6 sites of the cross-sectional study and white points the area of the intensive experimental study (the individual grids are too close to be appreciated at this scale).
Extensive cross-sectional study
The extensive cross-sectional trapping of yellow-necked mice was performed in July 2002 using multi-capture live traps at 6 study sites for 3 nights, with a total effort of 3456 trap nights (Fig. 1). Each site comprised of 3 trapping grids of 64 multi-capture live traps, each set up following the procedures established by Myllymäki et al. (Reference Myllymäki, Paasikallio, Pankakoski and Kanervo1971) and Henttonen et al. (Reference Henttonen, Oksanen, Jortikka and Haukisalmi1987). For each mouse trapped we recorded sex, body mass, breeding condition and tick infestation. Individuals were classified in breeding condition when we observed descended testes for males and perforated vagina or pregnancy for females (Gurnell and Flowerdew, Reference Gurnell and Flowerdew1990). Mice with body mass above 15 g were euthanized and stored in a single plastic bag, following the animal procedures of the European Commission Directive 86/609/EEC implemented by Italy. Mice below 15 g often carry very low or no infections and since we were interested in identifying a clear I. ricinus-H. polygyrus interaction they were released (Perkins et al. Reference Perkins, Cattadori, Tagliapietra, Rizzoli and Hudson2003; Ferrari et al. Reference Ferrari, Cattadori, Nespereira, Rizzoli and Hudson2004). In the laboratory, mice were carefully inspected: the number of each tick stage recorded, the gastrointestinal tract removed and H. polygyrus extracted, using the filtration-sedimentation technique, and the total number counted (Euzeby, Reference Euzeby1982). Eye lenses were collected and the mass of both eye lenses was used as relative measure of mouse age (Morris, Reference Morris1972; Gregory et al. Reference Gregory, Montgomery and Montgomery1992).
Experimental manipulation of H. polygyrus
The experimental reduction or increase of H. polygyrus abundance in yellow-necked mice was undertaken through an intensive live trapping program over 2 nights, every other week from May to August 2002, for a total of 5376 trap nights. Animals were trapped in 6 grids of 64 multi-capture live traps each (i.e. 8×8 traps at 15 m inter-trap interval covering an area of 1·1 ha, Fig. 1). Grids were located in the same valley and habitat, but more than 500 m apart with natural and artificial barriers (road, open field etc.) between them to minimize movements of individuals between grids. For every trapped individual we recorded sex, body mass, breeding condition and tick number and life-stages. Additionally, each individual was identified with an implantable subcutaneous passive induced transponder tag (PIT tag; Trovan ID 100, Ghislandi & Ghislandi, Italy) that allowed us to monitor the course of the infections in the mouse population using capture-mark-recapture techniques.
H. polygyrus was manipulated in mice weighing above 15 g. The parasite was removed from every other individual caught in 4 grids, through oral treatment with the anthelmintic Pyrantel pamoate (Gellini pharmaceutical; dose: 100 mg/Kg). We selected Pyrantel pamoate since it is active on gastrointestinal nematodes but is not systemic and does not affect the ectoparasite infestation (Wahid and Behnke, Reference Wahid and Behnke1996; Quinnell, Reference Quinnell1992). The mice caught that were not anthelmintic treated, were orally infected with an average number of 30, third-stage infective H. polygyrus larvae to increase nematode abundance (Keymer and Hiorns, Reference Keymer and Hiorns1986; Gregory et al. Reference Gregory, Keymer and Clarke1990). Infective H. polygyrus larvae were obtained from eggs collected from yellow-necked mice faeces from the study area. Eggs were developed to third-stage larvae and passed through 2 yellow-necked mice (trapped in the study area and treated with anthelminthic); the pure worm culture was then used for the field infection (Keymer and Hiorns, Reference Keymer and Hiorns1986).
The implanted ID tag code allowed us to identify and re-treat each individual every 2 weeks and for the duration of the experiment. The pre-patent period of H. polygyrus is 13–15 days (Keymer, Reference Keymer, Rollinson and Anderson1985) so a 15-day period elapsed between treatments to guarantee the success of the manipulation. Individuals caught from the remaining 2 trapping grids were used as controls. Since H. polygyrus manipulation affects nematode transmission and abundance in mice sharing the same habitat (Ferrari et al. Reference Ferrari, Cattadori, Nespereira, Rizzoli and Hudson2004), control mice were located in untouched grids close to the treatment sites. An a priori analysis of tick infestation and nematode abundance data (based on eggs per gram of mouse faeces, EPG) in mice trapped before the onset of the experimental manipulation, confirmed the similarity between the control and treatment sites both for tick infestation (χ22=0·68, P=0·711) and H. polygyrus infection (χ22=1·67, P=0·432). Mouse faeces were removed from the traps of single caught individuals and the EPG of H. polygyrus estimated using the McMaster technique. No faeces were gathered when traps contained more than 1 individual.
Data analysis
To identify the relationship between H. polygyrus abundance and I. ricinus infestation in mice from the cross-sectional trapping survey we used Generalised Linear Models (GLM, with negative binomial errors). The total tick infestation per host was used as a response variable, while H. polygyrus abundance, host characteristics (sex, age and breeding condition), and habitat composition (3 habitats: pure mature beech, mature beech with Scots pine and mature beech with spruce wood) were included as independent factors. The minimal adequate model was selected using stepwise backward deletion from the maximal initial model including all factors and their second order interactions (Crawley, Reference Crawley2002). To test the relative contribution of I. ricinus and host properties on H. polygyrus abundance the analysis was repeated using H. polygyrus abundance as the response variable, and tick infestation and host characteristics as independent factors.
To investigate the response of I. ricinus infestation to the experimental manipulation of H. polygyrus abundance we used Generalised Linear Mixed Models (GLMM-IRREML). Total number of ticks per host, as a response variable, was examined in relation to H. polygyrus treatment (H. polygyrus infection, H. polygyrus removal and control with no host manipulation), host sex, age and breeding condition, as independent factors. This analysis was based on mice recaptured following the initial treatment. Since the intensive longitudinal monitoring of the mouse population was performed every 2 weeks and because the blood meal of a tick lasts 3–5 days, the number of ticks counted on mice trapped between weeks can be considered an independent measure. However, to deal with pseudo-replication due to autocorrelation for capture-recapture of the same individuals in the same trapping week, we removed the weekly recaptures and treated the individuals as a random factor. To account for temporal variation in the longitudinal trapping we entered the trapping week as an additional random factor. An a posteriori multiple comparison Tukey test was conducted on the between-treatment effects computed by the GLMM, to identify which group contributed the most to the pattern observed. To statistically confirm the experimental success of H. polygyrus treatments a GLMM analysis was repeated using EPG as response and treatment as explanatory variable, an a posteriori Tukey test was then performed to highlight which treatment mainly contributed to the between-treatment differences. All statistical analyses were performed using Genstat 6th edition (Lawes Agricultural Trust, 2002) and the cut off for statistical significance was fixed at a probability P<0·05. Different models were tested and the best minimum models are presented and discussed. Data are presented as means±1 standard error.
RESULTS
Extensive cross-sectional sampling
A total of 230 yellow-necked mice were trapped from the 6 sampling sites, the mean capture was 38·3 (95% CL 9–68)15·1 mice per site. Over the 6 sites, the prevalence of H. polygyrus was 66·1% (95% CL 59·9–72·2%), and the geometric mean abundance 10·7 (95% CL 8·9–11·2) worms/host. I. ricinus was found on 49·6% (95% CL 43·1–56·0%) of the individuals, the geometric mean infestation was 8·8 (95% CL 7·9–10·0) and 96·4% of ticks were in the larval stage while the remaining were nymphs with a very occasional adult stage. We found that 67·8% (95% CL 59·4–76·2%) of mice infested with I. ricinus were co-infected with H. polygyrus.
The minimal model that best described the variation in tick infestation across the 6 study sites showed that the total number of ticks per host was negatively related to H. polygyrus abundance, the individual hosts not in breeding condition, the presence of spruce vegetation and the combined effect of H. polygyrus abundance and non-breeding condition of mice (Table 1). In contrast, pure beech woodland habitat and the combined effect of host age and H. polygyrus abundance had a positive effect on the tick number (Table 1). In particular the model showed that host breeding condition and age interacted with H. polygyrus abundance such that non-breeding mice as well as younger hosts exhibited a more pronounced reduction of tick infestation with increasing H. polygyrus abundance compared to older and breeding hosts (Fig. 2). To examine which of the two parasites had a stronger effect on the dynamics of the other species, we repeated this analysis using H. polygyrus as a response variable, and I. ricinus and host characteristics as independent components. The results suggested that host breeding status, sex and pure beech woodland exhibited a positive effect on nematode infection while tick infestation had a negative influence and only when interacting with host characteristics (Table 2).
Model prediction of the relationship between Ixodes ricinus infestation and Heligmosomoides polygyrus abundance in breeding and non-breeding mice of different ages. Predictions are based on mice in the 1st and 3rd quartiles, representing younger and older individuals, respectively.
Extensive, cross-sectional sampling of yellow-necked mice for Ixodes ricinus
(Generalized Linear Model between total tick number per host, as a response, and H. polygyrus abundance, host breeding condition and age and habitat, as explanatory variables.)
Extensive cross-sectional sampling of yellow-necked mice for Heligomosomoides polygyrus
(Generalized Linear Model between H. polygyrus abundance per host, as a response, and tick abundance, host breeding condition, sex and age and habitat, as explanatory variables.)
Experimental manipulation of H. polygyrus
Overall, a total of 87 yellow-necked mice were trapped (45 males and 42 females; 41 in breeding and 46 in non-breeding status) and marked with PIT tags between May and August 2002. Of these individuals, 25% were recaptured twice, 9% 3 times and 17% more than 3 times; the mean time-period between the first and the last capture was 5 weeks.
The level of tick infestation was affected by treatment (GLMM-IRREML, Wald=6·87, d.f.=2, P=0·03) while host characteristics (age, sex and breeding condition) did not significantly contribute to the pattern observed (Fig. 3). The experimental infection/removal of H. polygyrus from mice was successful in that there was a significant change in EPG between treatments (GLMM-IRREML, Wald=9·12, d.f.=2, P=0·01). The a posteriori pairwise comparison between treatments revealed that the anthelmintic treatment caused a significant decrease in H. polygyrus EPG (Table 3B), coupled with a significant increase in tick infestation (Table 3A). However, the infection with H. polygyrus was not sufficiently effective in increasing the nematode abundance compared to the control, which also caused no apparent change in tick infestation (Table 3). In fact, we found a large variation in EPG between mice, suggesting that the infection was successful for some individuals but not others (Fig. 3).
Changes in the geometric mean of Ixodes ricinus infestation (ticks/host) and number of Heligmosomoides polygyrus eggs per gram of host faeces (EPG) in relation to the experimental treatment. The 95% confidence limits are reported.
Average values (±95% confidence limits) for Heligmosomoides polygyrus eggs per gram (EPG) and Ixodes ricinus by treatment and a posteriori pairwise Tukey test based on GLMM estimates
(Comparisons between treatment groups (control, infection and anthelminthic) for: A-I. ricinus infestation and B-H. polygyrus number (EPG).)
DISCUSSION
We undertook an extensive cross-sectional monitoring of populations and an intensive experimental manipulation of individuals of free-living yellow-necked mice and examined the hypothesis that the increase in the abundance of H. polygyrus would increase the infestation of the co-infecting I. ricinus. We selected H. polygyrus because experiments with the laboratory model H. bakeri have identified suppression of immune-mediated mechanisms by the parasite adult stages, which should result in a positive effect on other co-infecting parasites (Colwell and Wescott, Reference Colwell and Wescott1973; Courtney and Forrester, Reference Courtney and Forrester1973; Jenkins, Reference Jenkins1975; Bruna and Xenia, Reference Bruna and Xenia1976; Jenkins and Behnke, Reference Jenkins and Behnke1977; Behnke et al. Reference Behnke, Wakelin and Wilson1978; Behnke and Ali, Reference Behnke and Ali1984; Alghali et al. Reference Alghali, Hagan and Robinson1985; Monroy and Enriquez, Reference Monroy and Enriquez1992; Behnke et al. Reference Behnke, Wahid, Grencis, Else, Bensmith and Goyal1993; Telford et al. Reference Telford, Wheeler, Appleby, Bowen and Pritchard1998; Maizels et al. Reference Maizels, Balic, Gomez-Escobar, Nair, Taylor and Allen2004). We made the assumption that H. polygyrus would behave similarly to H. bakeri. We also selected I. ricinus because previous work identified that infestation levels were negatively related to host susceptibility (Dizij and Kurtenbach, Reference Dizij and Kurtenbach1995) and we could monitor the response of this ectoparasite without invasive techniques. If H. polygyrus influences tick infestation, this can potentially have a significant impact on the dynamics of tick-borne diseases in the mouse populations. For example, I. ricinus is the vector of the zoonotic infection that causes tick-borne encephalitis and changes in the tick dynamics could have a major effect on the dynamics of the viral infection (Labuda et al. Reference Labuda, Kozuch, Zuffova, Eleckova, Hails and Nuttall1997; Randolph, Reference Randolph2000).
In our cross-sectional study we found that there was a negative relationship between the two parasite species and the experimental removal of H. polygyrus resulted in an increase in I. ricinus. These findings were counter to the prediction based on H. bakeri studies; indeed our field manipulation showed that this was a consequence of H. polygyrus infection influencing I. ricinus rather than the reverse. The comparative field study revealed that habitat and host breeding condition explained 24% and 8%, respectively, of the variation in tick infestation; nematode infection alone explained 2% but its effect was further increased to 4% when interacting with host breeding condition and to 3% with age. Despite the low statistical contribution, compared to habitat characteristics, H. polygyrus appeared to have a significant impact on tick infestation.
Previous studies on free-living mouse populations found that H. polygyrus infection is strongly influenced by host age, but habitat can play an important role in causing spatial differences in the intensity of infection among host populations (Gregory et al. Reference Gregory, Montgomery and Montgomery1992; Behnke et al. Reference Behnke, Lewis, Mohd Zain and Gilbert1999; Behnke Reference Behnke2008). Detailed large-scale studies and modelling also showed how weather and vegetation are crucial for the distribution and abundance of questing ticks (Randolph and Storey, Reference Randolph and Storey1999; Randolph, Reference Randolph2000). Our extensive cross-sectional results confirm the importance of environmental characteristics on the distribution of both parasites and also support the hypothesis of a possible effect of the nematode on the spatial distribution and level of tick infestation among different sampling sites of Trentino.
Younger non-breeding individual mice exhibited the strongest H. polygyrus-I. ricinus interaction, I. ricinus decreased with increasing H. polygyrus abundance and this was more apparent in non-breeding than breeding mice, and between young rather than old individuals. Host breeding conditions and age could have altered the nematode-tick interaction and further affected the dynamics of co-infection. Indeed we found that H. polygyrus abundance was mainly driven by host breeding condition, age and habitat, while I. ricinus had a marginal contribution. This finding suggests that H. polygyrus may act as a dominant species in the infra-community of parasites and probably modulates the dynamics of the other parasite species.
The negative relationship between H. polygyrus and I. ricinus was identified in the extensive cross-sectional study and the anthelmintic manipulation experiment confirmed that this relationship occurs as a cause and effect mechanism, i.e. H. polygyrus causes changes in the I. ricinus population. Unfortunately, the design of the experiment did not provide the opportunity to clarify the mechanisms generating the pattern observed. We can exclude the hypothesis that this was caused by direct competition for space since these two parasites use different parts of the host body. We also exclude the possibility of direct competition for resources since ticks and H. polygyrus depend on different trophic elements, the first specialized as bloodsuckers and the second feeding on intestinal tissues (Bansemir and Sukhdeo, Reference Bansemir and Sukhdeo1994). We can also rule out alternative explanations of a potential role of host behaviour on parasite interactions. H. polygyrus can be ingested through fur grooming and grooming is also used to control tick infestation (Hernandez and Sukhdeo, Reference Hernandez and Sukhdeo1995; Osfeld et al. Reference Osfeld, Miller and Schnurr1993). Therefore, mice that groom heavily will have fewer ticks and will potentially ingest more H. polygyrus infective larvae, than less active individuals. It is also possible that differences in ranging behaviour between male and female as well as during their life cycle (breeding vs non-breeding or dispersing young vs territorial adults) may have contributed to changes in exposure (Randolph, Reference Randolph1997; Stradiotto, Reference Stradiotto2008). However, the evidence of a cause and effect mechanism, as by the manipulation of H. polygyrus load on tick infestation, suggests that parasite interaction and host behaviour are independent mechanisms.
Our initial assumption that H. bakeri and H. polygyrus behave in the same manner may not be correct. This difference may be because H. bakeri has been inadvertently selected to be more immuno-suppressive than its natural counterpart or because the larval stages stimulate an immune response that overrides the adult immuno-suppressive abilities. This possible difference warrants further investigation. One hypothesis is that H. polygyrus and I. ricinus interact through the host, either through the immune response or the release of toxic products that may potentially affect the other species (Behnke et al. Reference Behnke, Bajer, Sinski and Wakelin2001; Bradley and Jackson, Reference Bradley and Jackson2008; Fenton et al. Reference Fenton, Lamb and Graham2008). At the moment we do not have evidence to tease apart these hypotheses. We can, however, recognize that the interaction is modulated by the host reproductive status and age; indeed we found that non-breeding mice and young individuals had fewer ticks but more H. polygyrus. Reproduction and age can both be associated with the strength of the host immune response (Schalk and Forbes, Reference Schalk and Forbes1997; Woolhouse, Reference Woolhouse1998). Previous studies on other systems found a trade-off in energy allocation between the immune system and host reproduction, with a deprivation of resources and a more efficient immune response during the breeding season (Apanius, Reference Apanius1991). Age may also affect host susceptibility and time of exposure to infective stages, and contribute to the development of an acquired immune response (Woolhouse, Reference Woolhouse1998). The trickle bi-weekly infections with H. polygyrus larvae may also have enhanced the immune response against this nematode in some individuals, and this may have caused the failure of the worm dosing and the large variability in worm EPG observed among the mice. In this sense, the absence of a significant change in tick infestation in trickle-infected mice may be the result of a large variability in the immune response between individual hosts. This conclusion is supported by previous experimental laboratory infections, where different mouse strains trickle-dosed with H. polygyrus exhibited different responses, with some becoming completely resistant to re-infections as a consequence of a full protective immune response (Brailsford and Behnke, Reference Brailsford and Behnke1992; Behnke et al. Reference Behnke, Lowe, Clifford and Wakelin2003).
In conclusion, both laboratory experiments and field observations leave part of the question on the role of co-infections on the dynamics of a single parasite species unresolved, and are therefore inadequate to provide a full picture of the consequences of parasite interactions in the real world. The approach used in the present study was aimed to overcome some of these limitations, by providing simultaneously a large-scale comparative approach and an experimental manipulation in natural conditions. Despite the fact that the manipulation of H. polygyrus abundance in mice was effective only at reducing H. polygyrus infection, we found clear evidence that changes in I. ricinus infestation are related negatively to changes in H. polygyrus abundance, and that these changes occur as a cause effect process. Different mechanisms, i.e. host immunity, release of parasite toxic products, may have caused this pattern and we do not exclude that their relative role may have changed throughout the time of the infection according to host and environmental condition. The testing of this system in the laboratory will allow us to disentangle the underlying processes affecting the interaction between I. ricinus and H. polygyrus.
We thank Andrea L. Graham for her valuable contributions on a previous version of this manuscript. We are also grateful to the three referees who provided helpful comments to improve the manuscript. This study was funded by the Centro di Ecologia Alpina, and Provincia Autonoma di Trento (Grant number 1060: ECODIS-Ecology and Control of some zoonotic Diseases). P. J. H. is supported by NSF-NIH Ecology of Infectious Disease programme grant number 0520468 and I. M. C. by a Royal Society University Fellowship.
