From June 1 to August 31, 2007, bats were caught and classified according to species, sex, age category, gravidity, and lactation status. Bat species were typed by morphologic criteria by experienced bat biologists who had worked in the habitat for several years. A total of 315 bats were sampled (Table 1). Overall prevalence of CoV in all bats was 9.8%.

Of the 7 species studied, 5 (M. dasycneme, M. daubentonii, M. bechsteinii, Pipistrellus nathusii, P. pygmaeus) yielded coronavirus, with detection rates of 5.2% to 25.4% per species. Detection rates varied significantly between bat species, with M. dasycneme showing significantly higher rates than any other species (analysis of variance [ANOVA], p<0.0002). Among the M. dasycneme bats, detection rates were not equally distributed but correlated significantly with the location in which bats were caught (one-way ANOVA, p<0.013). Similar observations were made for 3 other virus-positive species, where detection was achieved in 2 of 6 (M. daubentonii), 1 of 3 (P. nathusii), and 1 of 4 (P. pygmaeus) sampling locations.

Spillover of bats between colonies was not seen, with one interesting exception. In location 2, one of 7 examined M. dasycneme bats was already banded. Its ring number showed that it had spilled over from another area. This bat was the only one yielding coronavirus in location 2. By tagging with a radio transmitter, its maternity roost could be traced to location 3, ≈10 km away. When location 3 was sampled, it yielded 45% positive bats (15 of 38 M. dasycneme tested), the highest rate of all locations sampled in the study period.

Factors correlating with coronavirus infection were determined. Analysis was conducted on major physical properties. Approximately half of the bats (56%) were females, 30% of them lactating. Twenty-seven percent were juvenile, 12% subadult, and 61% adult. As shown in Table 2, ANOVA analysis identified that young age and ongoing lactation, but not a particular sex or existing gravidity, correlated significantly positively with coronavirus detection. Among female bats, detection rates were significantly higher in those bats associated with maternity colonies than in those caught in foraging habitats (2-tailed t test, p = 0.026) or swarming sites (2-tailed t test, p = 0.037). Differences between foraging and swarming sites were not significant (p = 0.609).

Sequences of PCR products from all coronaviruses were determined. As shown in Figure 2, the northern German viruses clustered in 1 large monophyletic clade containing no other previously known virus. In a sister relationship was a clade of viruses from Chinese bats with prototype strains A701/2005, HKU6 and A821/2005 (18,34). These viruses were all detected in M. ricketti, which belongs to the same subgenus (Leuconoe Boie) as M. dasycneme and uses a similar ecologic niche (35–37).

Within the Northern German bat-CoV clade, 4 different lineages appeared to be monophyletically associated with certain bat species. Lineage 1 was associated with M. dasycneme, lineage 2 with P. nathusii, lineage 3 with P. pygmaeus, and lineage 4 with M. daubentonii. As expected, the coronavirus from the stray bat from location 2 and the viruses from location 3 clustered closely together in lineage 1. Within the same lineage was 1 coronavirus from an M. dasycneme bat that had been sampled at a different site and at a different time (location 5), along with Pipistrellus bats that carried clearly distinct virus of lineages 2 and 3. Viruses therefore seemed to be more closely associated with bat species than with sampling locations. On the contrary, virus from the only virus-positive M. bechsteinii bat, a very rare and almost extinct species, clustered with lineage 1. Virus detection was successfully repeated from the same sample. These findings suggest that spillover of virus from M. dasycneme into M. bechsteinii might have occurred. In addition, nonrecent host transition of a common ancestor of Myotis-associated CoV into Pipistrellus is suggested by virus phylogeny for lineage 2 and 3 viruses (Figure 2).

To appreciate the diversity of northern German bat-CoV, a nucleotide distance matrix of 30 major taxa of coronaviruses was set up, including established coronavirus species and novel bat-CoV taxa, as recently defined (Appendix Figure). Lineages 1, 2, and 3 had mutual nucleic acid distances between 6% and 8%. Distance of lineage 4 from the aforementioned was 12%–13%. Distance of northern German lineages 1–4 from the sister clade of Chinese M. ricketti CoV was 15%–17%. The Chinese and northern German Myotis-associated CoV and their common sister clade, represented by strain A512/2005, were 20%–22% distant. For comparison, among the established species of CoV the lowest degrees of nucleic acid distances were observed between mouse hepatitis virus, human CoV HKU1, and the hCoV-OC43/bovine CoV pair, at 16%–18%. The 2 established lineages within CoV group 2b (SARS-like CoV, bat-SARS CoV HKU3) were 10% distant.

Similar to our previous studies on anti-coronavirus antibodies in African bats and recent findings of bat CoV in North America (26,27), this study shows that the presence of coronaviruses in bats is not a unique phenomenon in Asia and seems to extend worldwide. The prevalence of coronaviruses in bats in northern Germany was 9.8%, which is in the same range as in studies of similar size from China: Lau et al. found 66 (16%) of 412 bats positive for coronaviruses (38), Chu et al. found a prevalence of 15.8% (43/272 bats) (25), Woo et al. found 4.2% (13/309 bats) (34), and Tang et al. found 6.5% (64/985 bats) (18).

To explain how coronaviruses might be transmitted and maintained in bat populations, we have statistically determined factors that influence virus detection. Young bats of both sexes, as well as lactating bats, but not gravid bats, were significantly more likely to carry coronaviruses. The virus could be transmitted between young bats and mothers in maternity colonies, rather than circulating year-round at equal levels in the population. Indeed, female bats captured near maternity colonies showed significantly higher virus detection rates than those captured in foraging or swarming sites. Our bats reside in maternity colonies during early summer months; after young bats are born, male bats avoid maternity colonies. One could propose a scenario in which the young provide a susceptible population, amplifying the virus and transmitting it to adult females in maternity colonies. Comparable to many respiratory and enteric virus infections in humans, adults would replicate virus less efficiently than the young because they have at least partial immune protection because of infection earlier in life. This would explain lower detection rates in adult bats. For confirmation, the immune status of young and adult bats against homologous coronaviruses would need to be studied. However, we cannot take blood from our bats without harming them because of their small size; bats in Germany are strictly protected.

Similar to studies conducted in China (18,25), viruses might be associated with bat species. The same virus was never detected in different species that occurred simultaneously in the same location. For example, M. daubentonii and M. dasycneme both occurred in locations 2 and 8, but only 1 species per location yielded (different) virus; location 5 harbored all 4 virus-carrying species in 1 place. Three of 4 Myotis species yielded viruses (Table 1), and these belonged to 3 different lineages. On the other hand, the same virus lineage was found in remote colonies of the same bat species (compare M. dasycneme and M. daubentonii in Table 1).

On the local scale of our study, it was difficult to determine whether strict species association or limited local transmission may be responsible for the observed associations is difficult to determine. An influence of local transmission cannot be excluded, considering the hypothesis that virus is likely transmitted in maternity colonies. Groups of bats from the same maternity colony stay together throughout life, and bats of the investigated species never mix in such colonies.

On a larger scale, however, a group I coronavirus hosted by palaearctic (or Old World) Myotis bats, including M. dasycneme and daubentonii in Germany and M. ricketti in China (18,34), might exist. Earlier studies (18) used a threshold of 20% nucleic acid distance in our target gene to define a new species of bat coronavirus. By using these criteria, the palearctic Myotis virus would form a distinct coronavirus species with German and Chinese subspecies. Even though such a classification is preliminary and does not take other aspects of coronavirus classification into account, it would be supported by host biology. Myotis bats do not migrate, but habitats of different Myotis species continuously overlap from China throughout Asia into Europe. M. ricketti, which harbors the Chinese sister clade of our coronaviruses, belongs to the same subgenus (Leuconoe Boie) as M. dasycneme and uses a similar ecologic niche (35–37). A continuous virus population might thus coexist with a continuous palearctic Myotis population. As a conclusive extension of this hypothesis, the recently described North American bat coronaviruses RM11 and RM48 from M. occultus were more closely related to our viruses than the RM65 strain from an unrelated Eptesicus fuscus bat (26).

Finally, the virus observed in 2 different Pipistrellus species would likely have resulted from a host switch of Myotis virus; Pipistrellus is not closely related to Myotis spp. (28). As predicted very recently, coronaviruses may not only be prone to accidental, infrequent host switch between mammals, but may jump from 1 host species to another within the bat reservoir (39). Our study supports this notion and suggests that host transition within geographically closely associated, but genetically distinct, bats may have occurred. Because all related viruses in Europe and China were associated with Myotis, the direction of transition was probably from Myotis spp. into Pipistrellus spp., where virus then would have diversified further. This hypothesis is also suggested by relatively long internal branch lengths on the third level of bifurcation (counting from the basal node of the German Myotis clade) that separates both Pipistrellus virus lineages. This in turn could be a correlate of independent adaptations to P. nathusii and P. pygmaeus, respectively, after host transition.

Should these initial observations be confirmed in future studies, implications on infection control and prevention of zoonotic outbreaks would be considerable. Targeted eradication of bats is technically impossible and ecologically detrimental. Systematic intervention in the ability of bats to carry coronaviruses might be a realistic, but remote scenario. Further research into the association of coronaviruses with natural hosts is necessary to understand their maintenance patterns and zoonotic potential.