Echinococcus multilocularis is a wide-spread cestode causing human alveolar echinococcosis (AE), a severe disease, with canids (mainly red foxes, Vulpes vulpes) acting as final host (1, 2). A wide variety of small mammals are described as intermediate hosts (1, 3). Individuals of some murid species were occasionally detected with E. multilocularis infections [e.g., one Mus musculus (4), one Rattus norvegicus (5)], but their role as intermediate hosts can be neglected (1). Regular records of rodent populations with relevant prevalences are only reported from cricetid species, e.g., Arvicola scherman (formerly A. terrestris) and Microtus arvalis in Europe (6, 7); Myodes rufocanus in Japan (8, 9); Ellobius tancrei and Lasiopodomys brandtii (Brandt’s vole, formerly Microtus brandtii) in Central Asia (10); Peromyscus maniculatus and Microtus pennsylvanicus in north central of the United States (11, 12).

Due to successful rabies vaccination, increased supply of anthropogenic food resources, and changing human–wildlife interactions [e.g., urban tameness (13)], fox populations have substantially increased, especially in the densely populated and urbanized areas in many European regions, e.g., France (14), Switzerland (15), and Germany (16). There is strong evidence that these changes in the population dynamic of foxes led to a marked increase of the incidence of human AE in different regions of continental Europe during the last two decades (17, 18).

In parallel to this development, the parasite spread from historically known endemic regions in Central Europe like northern Switzerland, eastern France, southern Germany, and western Austria, over large distances toward the Baltic States (19, 20), Scandinavian countries (21, 22), and to the west of France (14, 23). In Switzerland, the southern border of the parasite distribution corresponds fairly well to the course of the Alpine crest (24). However, some case records of rodent infections south of the Alpine crest in Italy (25) and Switzerland (24, 26) demonstrate that the border of the distribution area is not just a result of the harsh climatic conditions in the high Alps. In the Swiss Canton Grison, a correlation between the prevalence in foxes and the predation of foxes on cricetid but not on murid species (26) gave evidence that, beside climatic factors (27), the distribution of suitable intermediate hosts is crucial for the distribution pattern of the parasite. In contrast to many other studies that suggest a geographical spread of E. multilocularis, recent investigations in Ticino, the most southern Canton of Switzerland which is located south of the Alpine crest, revealed a stable border of the distribution of the parasite over the last 20 years. Interestingly, its local distribution and its border of distribution matched the restricted areas where the vole M. arvalis was present (24).

Throughout Switzerland, nine different cricetid species occur, which could potentially act as intermediate hosts. However, only four of these species are wide spread and occur in higher densities (28). Two of the species, i.e., A. scherman and M. arvalis, live in open fields and have been described by different authors as important intermediate hosts in Central Europe (1). The prevalence of the widespread vole M. agrestis, a species living mainly in wetland, meadows, and young forests (29, 30), has less been studied. However, a recent investigation demonstrated its high susceptibility to experimental oral inoculations (31). Although no protoscoleces were found 6 weeks p.i. in this study, a recent field study in Sweden confirmed that this species can develop fertile infections (32). Myodes glareolus, the fourth widespread vole in Switzerland, lives mainly in bush lands and forests. It has been regularly described as potential intermediate host in Central Europe, but its epidemiological role for supporting or maintaining the parasite cycle remains unclear (1).

In order to effectively transmit the parasite in the European endemic area, the intermediate hosts have to develop fertile metacestodes with infective protoscoleces (33). In addition to being susceptible to the parasite, only species which share their habitat with foxes and are regularly predated by them can ensure the maintenance of the life cycle (34, 35). Therefore, prevalence studies do not suffice to compare the significance of different rodent species for the maintenance of the parasite life cycle. In addition, the direct comparison between different species is hampered by the fact that transmission intensity can greatly vary in space and time (35). Correspondingly, prevalence rates in rodent populations are strongly affected by changing environmental conditions over time and the infective state of the fox individuals in local fox territories.

With our study, we wanted to elucidate the relative importance of the vole species A. scherman, M. arvalis, M. agrestis, and M. glareolus in selected study plots where several of these vole species co-occur simultaneously. This should ensure (A) that the investigated rodents were most likely exposed to the feces of the same fox family groups (and therewith to the same overall level of environmental egg contamination) and (B) that the foxes of one family group could select between the different rodent prey species. To compare the relative importance of the investigated rodent species, we estimated for each species (1) the prevalence of E. multilocularis, (2) the proportion of fertile infections, (3) the asexual parasite reproduction (number of protoscoleces), and (4) the predation frequency by foxes.

The emergence of AE across large regions of Europe has been associated with the increasing fox populations after the successful control of rabies in many European countries and an increased supply of anthropogenic food resources (1, 15, 18). However, the role and importance of different intermediate host species is under debate (1, 35, 46).

Investigations aiming to identify the key intermediate host species rely mostly on prevalence studies, and only a part of them also considers the parasite burden. In Europe, high prevalences have been found on a regular base in the four cricetid species M. arvalis [e.g., 18.6 (47), 3.0 (48), and 8.6% (49)], A. scherman [e.g., 13.6 (47), 3.6 (48), 6.5 (49), and 14.0% (50)], M. glareolus [e.g., 5.2 (5), 18.2 (48), 10.3 (49), and 4.4% (51)] and the muskrat Ondatra zibethicus [e.g., 22.1 (52), 1.6–62.5 (53), and 0.7% (54)]. Whereas the muskrat is rare in Zurich and, therefore, could not be analyzed in our study, our results confirm the relevance of the other three rodent species which all had relevant prevalence rates and a significant amount of fertile infections.

However, also other European cricetid species have been occasionally found infected like the sibling vole Microtus levis formerly M. rossiaemeridionalis in Svalbard [e.g., 18.9% (55)], Chionomys nivalis [syn. Microtus nivalis] in Romania (e.g., 77), Microtus subterraneus in France [1 infection among 169 individuals (56)], Microtus agrestis in France [1 infection among 16 individuals (5)] and in Sweden [1 infection among 187 individuals (32)], and Arvicola amphibius in Sweden [1.8% (32)]. Indeed, the example of Svalbard shows that M. levis can maintain the parasite life cycle and, therefore, this species possibly could play a significant role for the occurrence of the parasite in Eastern Europe. However, this species does not occur in the historically recognized endemic areas of France, Germany, Austria, Switzerland, and other countries of Western Europe. Considering that only in France one infected M. subterraneus (56) and one infected M. agrestis (5) and in Sweden few infected individuals of M. agrestis and A. amphibius (32) were found, it is very unlikely that these two species play a significant role in highly endemic regions. The same applies for C. nivalis as this species has a much more restricted distribution in Western Europe than E. multilocularis. However, it is known that M. agrestis is susceptible to experimental infections (31), and it replaces M. arvalis in the Scandinavian countries where the species expands more to open habitats than in regions where it co-occurs with M. arvalis (57). Therefore, it is likely that this species together with A. amphibius maintains the life cycle at least at a low level in Scandinavian countries where surprisingly so far no infected M. glareolus have been detected (58).

Although many studies have investigated different murid species for E. multilocularis infections (5, 40, 49, 51), to our knowledge there are so far only two confirmed cases of infected murid rodents in Western Europe, namely a M. musculus which was trapped in the cellar of an inhabited house in a small village in the French Auvergne (59) and a R. norvegicus with small, non-fertile lesions (5). Experimental studies have confirmed the very high resistance of laboratory rats (60) to inoculations with high numbers of E. multilocularis eggs, but elucidated that this resistance can be reduced with immunosuppressive interventions resulting in active infections (61). Therefore, this single case and few cases from Japan cannot be regarded as indicators for the intermediate host competence of R. norvegicus. Also other non-cricetid rodent species have been reported in Europe, as the introduced Nutria Myocastor coypus (53, 54) and the Eurasian beaver Castor fiber (62, 63), which both can harbor fertile infections (53, 62). Considering their potential to disperse over large distances and their longevity (64–66), these two species could occasionally be of some importance in the spread and the persistence of the parasite. However, both species live in low densities and are only occasionally reported in the fox diet [e.g., Ref. (67, 68)].

Based on a systematic review of epidemiological studies, Oksanen et al. (46) confirmed arvicolids (including the genera Microtus, Arvicola, and Myodes) and muskrats as important intermediate hosts for E. multilocularis in Europe. However, most of the studies included in this review did not consider to which extent the different rodent species were preyed on by final hosts. As muskrats are not a frequent prey of foxes, this species is regarded by other authors more as a bioindicator for the presence of the parasite rather than a key intermediate host (51, 54). Only in special cases, e.g., if trapped animals of control programs are left on river banks, the infective cadavers would be available in large numbers to foxes and boost the infection pressure (51).

By reviewing the existing literature, the three cricetid species M. arvalis, A. scherman, and M. glareolus can be regarded as the most important candidates for maintaining the parasite life cycle over large parts of its distribution area in Western Europe. All three species are widespread, can reach high population densities (28, 69), frequently co-occur in high endemic regions of Western Europe and have been regularly reported with fertile infections in the wild. In this study, we ensured by trapping different species in the same plots and during the same periods that the same fox families had access to the different rodent species. Thus, we can assume a similar overall exposition of the rodents to E. multilocularis eggs. Furthermore, foxes had the choice on which of these species the preferably prey. We also included M. agrestis in this comparative study, as this species is another common cricetid species in the Swiss midlands. However, despite of the huge trapping efforts we could catch only 29 individuals, and that none of these animals were infected. This gives evidence that this species can be neglected for maintaining the parasite life cycle in the high endemic region of the Swiss Midlands.

Comparing M. arvalis, A. scherman, and M. glareolus we found significant differences among the species on several levels. Interestingly, M. arvalis had—to our knowledge—the highest prevalence ever recorded which was considerably higher than the prevalence rates for A. scherman and M. glareolus. For example, during spring 2014, 28.6% of all trapped M. arvalis (95% CI: 22.4–35.4) and only 6.5% (3.6–10.7) of A. scherman and 2.6% (0.5–7.5) of M. glareolus were infected. The difference was even more pronounced when comparing only the fertile infections. In total, in 16.8% (95% CI: 11.9–22.8) of all trapped M. arvalis during the same time period were protoscoleces found, whereas the corresponding values were only 0.9% (0.1–3.3) for A. scherman and 0.9% (0.0–4.8) for M. glareolus. The eminent role of this species is also underlined by a recent experimental study which demonstrated the high susceptibility of M. arvalis for fertile E. multilocularis infections (70). Thereby, it has to be considered that M. arvalis, which is known for its short generation time, can reach very high population densities of more than 2,500 individuals per hectare (28), which is higher than the peak values for A. scherman and Myodes [A. scherman: >1,000 ind./ha; M. glareolus: 50–100 ind./ha (28, 69)]. Furthermore, M. arvalis are much smaller (20–35 g) than A. scherman [65–130 g (28)]. Therefore, a fox has to feed on several M. arvalis to have an equivalent of nutrition as from one A. scherman. It can also be assumed that M. arvalis is a much easier prey as it uses more superficial channels than A. scherman, which rarely leaves the tunnel system. This assumption is substantiated by the observation that M. arvalis is a preferred prey compared to other rodent species (71). Nevertheless, in our study both species have been detected in the fox feces in similar frequency.

Interestingly, our data indicate that Myodes is the best intermediate host in terms of fertility of the parasite. In experimental studies, its susceptibility to oral experimental infections was lower as compared with Microtus spp., but infected animals developed fertile infections (72). However, although M. glareolus is a widespread and common species, we recorded no predation on it by foxes in our study. This is in contrast to other studies in which foxes were shown to prey on M. glareolus (73, 74). The species lives—like M. agrestis—more in covered habitats and thus probably is less susceptible to fox predation (75). Taken together, we conclude that this species plays a minor role in the perpetuation of the life cycle in our study region. However, highly infected Myodes could occasionally be eaten by domestic dogs and thus contribute to the transmission of E. multilocularis to human.

Our study supports the evidence for a high relevance of M. arvalis [in accordance with Ref. (24); see Table 2]. However, it cannot be excluded that other European species can replace M. arvalis. Especially A. scherman has shown similar high prevalences in previous studies in Zurich. Thus, in one trapping plot an extraordinary high prevalence of 61% [95% CI (41–78)] was recorded for this species (41). It is possible that foxes prefer Arvicola when M. arvalis is not available or in very low densities. This is supported by a study of Weber and Aubry (78) in the Swiss Jura mountains where A. scherman was the main prey and recorded for 54.5% of the investigated prey items. Indeed, a sigmoid-like functional response to A. scherman density has already been described for the predation of foxes and the predation rate decreased when the density of M. arvalis increased (35, 79). On the other hand, M. arvalis was consumed at a high level even when its density was very low. However, to clarify to which extent such replacement processes buffer the life cycle would need further studies.

In conclusion, our study highlights how differences between rodent species in their susceptibility, exposition to infective eggs, parasite fertility, and predation by foxes affect their relevance for the life cycle of E. multilocularis. Our results provide evidence that M. arvalis and probably to a lesser extent A. scherman distribution models could be good predictors for the distribution and abundance of E. multilocularis in Western Europe. Models on the distribution and abundance of these species, therefore, could allow to model parasite occurrence on a more detailed spatial scale than models on fox distribution, as foxes have a very broad and much more uniform distribution than the different rodent species.

Trapping of animals was performed under the direct supervision of a veterinary specialist, and according to the Swiss law, the guidelines on Animal Welfare and the specific regulations of the Canton of Zurich (permit number 17/2013) by the Veterinary Office and the Ethics Committee of the Canton of Zurich (Kantonales Veterinäramt Zürich, Zollstrasse 20, 8090 Zürich, Switzerland).

All authors listed have made substantial, direct, and intellectual contributions to the work, and all approved its content for publication.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.