Discussion
This study describes the combined use of population genetics and serological analyses to investigate the infection dynamics of a rabies-related lyssavirus (RRLV) in the greater mouse-eared bat in South Tyrol, the northernmost region in Italy. In particular, serological tests were specifically directed towards EBLV-1, even if the circulation of a highly related virus cannot be completely ruled out due to the lack of virus identification in this study and, more in general, in this species.
In this context, our aim was to define the spread of EBLV-1 (or its related RRLV) in the area and to interpret serological data from different locations using the connectivity between roosts derived from genetic analyses as a proximate measure for the likelihood of viral transmission. In addition, phylogenetic analyses were performed to address the possible movement of individuals across the Alps, which could be related with the dispersal of lyssaviruses that, despite being widely circulating in Europe, have never been described in Italy (Leopardi et al., 2018; Shipley et al., 2019). Our study has focused on maternity roosts, which are easier to locate compared to groups of adult males, and provide higher chances for virus amplification as a consequence of a bigger colony size, more frequent and closer interactions among individuals and due to the yearly replenish of susceptible individuals after the birth pulse (Drexler et al., 2011).
The detection of antibodies against EBLV-1 in South Tyrolean maternity colonies of greater mouse-eared bats dates back to 2012, despite no mortality has ever been related to LYSV circulation so far (Leopardi et al., 2018). The presence of antibodies in healthy bat colonies is peculiar in the ecology of LYSVs, and has been extensively described elsewhere (Amengual, Bourhy, López-Roíg, & Serra-Cobo, 2007; David T S Hayman et al., 2010; Robardet et al., 2017; Suu-Ire et al., 2017). Indeed, LYSVs are widely known for their pathogenicity in non-flying mammals, invariably lethal when symptoms develop (Fisher et al., 2018). In this study, we confirmed that all the investigated colonies are exposed to the virus, as they show highly specific neutralizing antibodies. Antibody titer were lower compared to other mammals; similar pattern has been already described in bats for other pathogens, both in the field and after experimental infection, and might be related with the high efficiency of their innate, non-antibody-mediated pathway for viral control (Baker, Schountz, & Wang, 2013). In addition, we detected a significant variation of sero-positivity across a single reproductive season, supporting seasonal variation in the circulation of LYSVs in all the locations investigated and, overall, an increasing sero-prevalence in late summer compared to spring. These data are consistent with similar findings in Myotis and serotine bats (Amengual et al., 2007; Robardet et al., 2017; Serra-Cobo et al., 2013), and could relate to an increased virus transmission after the birth pulse, as suggested for other viruses as a consequence of the large uptake of susceptible individuals (Drexler et al., 2011). Indeed, higher antibody titers were found during late summer, which could be a sign of seroconversion due to viral exposure. This result differs from recent evidences determined for EBLV-2 in Myotis daubentonii, for which it was suggested a peak in virus transmission later during the autumnal swarming, when males and females aggregate for mating (Horton et al., 2020). However, this is rather predictable, considering that bats are highly variable in their ecology, so that differences in life history traits including hibernation, colony formation, parturition and mating are likely to determine peculiarities in virus dynamics as well (D. T S Hayman et al., 2013). Despite peaks in transmission might depend upon host ecology, most species follow strict seasonal patterns which might affect prevalence of viruses (D. T S Hayman et al., 2013); thus, we suggest that comparison between data of sero-prevalence detected from different years or locations should always account for differences in the sampling period. In this context, 100% peak of sero-prevalence recorded in this study is exceptional compared to previous evidences in the same host species (Amengual et al., 2007; Schatz et al., 2014; Serra-Cobo et al., 2002; Šimić et al., 2018) and is likely related to a very late sampling, when adults were already dispersing for mating. In addition, it is wise to consider to what extent the strong effect of seasonality might hide the influence of other factors, including the geographical, demographical or climatic data. Indeed, although preliminary analyses suggested that the youngest animals were more likely to be serologically positive compared to adults, the effect of age lost significance when accounting for the dependence on seasonality. This is likely related to non-homogeneous sampling across age classes of different sampling campaigns.
None of the five roosts investigated was more likely to be infected, suggesting that EBLV-1 or a related cross-reacting virus is circulating across the area with no geographical restriction. Indeed, genetic data derived from South Tyrolean Myotis myotis are consistent with a single meta-colony, with movement of individuals between colonies likely favoring viral transmission. Despite several studies already documented regional panmixia of M. myotis at the nuclear level, we also found lack of structuring based on mitochondrial DNA, strengthening the hypothesis that population across the whole South Tyrol are highly connected. In addition, FST values inferred from nuclear and mitochondrial markers were similar, suggesting a much lower philopatric behaviour than previously described in Germany, France and Switzerland despite a similar spatial scale (Castella et al., 2001; Petri, Pääbo, Von Haeseler, & Tautz, 1997), and supporting regular movements of individuals between colonies, similarly to what observed in the serotine bat (Eptesicus serotinus ) (Moussy et al., 2015). The potential use of several roosts in the area was also supported by occasional recapture in the big colony two of adult females which were tagged elsewhere, despite a structured capture-recapture study has still to be performed to quantify inter-colonial movements. Interestingly, colony two showed the highest diversity in the area, also supporting the mixing of bats in this spot, characterized by the biggest population size and by a central position within the study area. However, such a higher diversity might also be explained by the historical aggregation of bats in this territory from different colonies already existing in neighbouring zones; indeed, the church was completed only in the early 19th century, well after the remaining roosts included in the study.
A meta-population structure for Myotis myotis is described here for the first time, providing critical information to unravel dynamics of LYSVs in this bat species. Indeed, previous mathematical models suggested this bat was unable to sustain the circulation of EBLV-1 alone, supporting that sero-positivity in the Balearic Islands was likely associated with inter-species transmission from the highly vagile species Miniopterus scheibersii (Colombi et al., 2019; Pons-Salort et al., 2014). In South Tyrol, the greater mouse-eared bats form very large maternity colonies within churches, where the interaction with other species is strongly limited compared to the cave dwelling populations typical of the Mediterranean area. In the area target of our study, colonies mostly consist of M. myotis,with M. blythii being sporadically found throughout the season and a small population of M. emarginatus recorded in a single roost from May to July. No co-roosting with Miniopterus schreibersii have ever been recorded in the studied colonies (Drescher, 2004). However, the yearly detection of seropositive individuals highly supports maintenance of the virus within the population (Leopardi et al., 2018). In this context, the meta-population structure seen in our sampling area as opposed to the close population model in the Balearic Islands could explain virus maintenance in M. myotis even in the absence of M. schreibersii . Indeed, several studies suggest that the exchange of individuals between roosts may be one of the main factors favoring the persistence of viruses across the whole population, sometimes allowing for local fade out of infection with subsequent reintroduction (Blackwood, Streicker, Altizer, & Rohani, 2013; Colombi et al., 2019; Horton et al., 2020; Pons-Salort et al., 2014).
Despite being largely described across Europe, EBLV-1 has never been detected in Italy (Leopardi et al., 2018; Shipley et al., 2019). Indeed, the large majority of individuals included in our sample carried a mitochondrial signature typical of the Italian peninsula, confirming the Alps as major barrier for bats, including M. myotis (Ruedi & Castella, 2003; Ruedi et al., 2008). However, our data also confirmed that individuals belonging to the major European clade A and to the clade C, previously reported only across the border between Switzerland and North-Western Italy, are present in South Tyrol, suggesting that transboundary admixture of animals is possible. Indeed, haplotypes from all three clades were found in the small colony three, located just on the border with Austria, which also showed a weak but significant sign of nuclear fixation. While it is possible that these results are due to continental individuals reaching this location during dispersal, coexistence of several divergent matrilines in this population may also result from the persistence of ancestral polymorphism during coalescence. As colony three is located in the oldest church in the area, dating back to 1337, it is likely that the current genetic pattern is shaped by both present- time and recent- past demographic events. Either way, these data support that South Tyrolean valleys might act as narrow corridors across the Alps, allowing for the movement of individuals from the north. In particular, the dispersal of bats across the Alps could either provide or have provided in the recent past a corridor for the introduction of LYSVs from continental Europe, where they are widespread. Regarding EBLV-1, the origin of the current lineage has been dated back to 1400 (Troupin et al., 2017), so that introduction in Italy could be related to both ongoing or recent admixture of animals from South Tyrol and other northern areas, where the virus has been confirmed virologically. The spread of viruses across the home range of bat populations has already been described for other pathogens, including lyssaviruses, henipaviruses, coronaviruses and astroviruses (Halczok et al., 2017; Leopardi et al., 2016; Olival et al., 2020; Peel, Sargan, et al., 2013). Unfortunately, the lack of viral characterization in this bat species prevented us to compare the geographical structuring of LYSVs and Myotis myotis , which could help to explain the role of this bat in the dispersal and evolution of EBLV-1 (Carver & Lunn, 2020; Olival et al., 2020). Challenges in the detection of bat viruses are not peculiar to our study or to LYSVs, but are likely related with low prevalence and short shedding period of viruses in these animals and, more generally, in wildlife (Wilkinson & Hayman, 2017). In this context, antibodies are easier to detect and persist longer, so that serology is widely used as a proxy for virus circulation in wildlife to model mechanisms of virus persistence within bat population even prior to virus detection (Gilbert et al., 2013; Peel, McKinley, et al., 2013).
Regarding the genetic approach used in this study, we were able to amplify 12 microsatellites described by Castella et al., (2001), and developed a new protocol for the amplification of the complete control region of the mitochondrion. This included sequencing of two hypervariable regions (HVI and HVII), which were successfully amplified and concatenated from most samples. Compared to previous studies which had used the sole HVII, the additional presence of the HVI fragment allowed to identify a higher variability, suggesting that our approach might have a slightly better resolution for genetic analyses. South Tyrolean Myotis bats showed variability in all but one microsatellite (C113), confirming results already obtained by Berthier et al (Berthier, Excoffier, & Ruedi, 2006). In general, mean diversity within colonies was similar to what has already been described in other studies, both at the nuclear and mitochondrial levels (Castella et al., 2001, 2000; Ruedi & Castella, 2003; Ruedi et al., 2008). Our analyses showed more heterogeneity for the parameter “nucleotide diversity” compared to the “gene diversity” calculated within different colonies, ranging between 0.001 to 0.013 and from 0.431 to 0.783, respectively (Table 2). Indeed, gene diversity only considers the number of detected haplotypes while the nucleotide diversity also takes into account the degree of their divergence. Thus, the ten-fold higher nucleotide diversity in colonies two and three is likely related to the admixture of haplotypes that belong to different phylogenetic clades, as shown in other studies (Castella et al., 2001; Ruedi & Castella, 2003; Ruedi et al., 2008). Similarly to other studies, we determined genetic structuring of M. myotis based on both nuclear and mitochondrial markers, in order to test for female philopatry, which is generally supported for this species (Castella et al., 2001). However, we found comparable results from microsatellites and mitochondrial sequences, thus disproving the expected pattern in our sample. These unexpected results could be either related to peculiar ecology of South Tyrolean populations or to lower spatial scale compared to other studies on the same species.
In conclusion, we provide serological evidence for the circulation of a RRLV in the South Tyrolean population of M. myotis, antigenically related to EBLV-1. Unlike the Mediterranean populations, M. myotis in northern Italy is a strictly house dwelling species, where adult females aggregate in church roofs between April and September to form large maternity colonies reaching over 2000 individuals, with a remarkable chance for human encroachment (Zahn, 1999). As all LYSVs can cause clinical rabies in humans, this study has a strong impact on public health, regardless of the viral species actually circulating in the study area. Fortunately, we founded a strong seasonal pattern, suggesting the highest risk for human exposure might be limited in time, easily predictable and likely associated with animal birth. In this context, preventive measures to limit human exposure by limiting access to the colony during most critical times could be beneficial to both humans and animals, which is imperative considering the ecological importance of this endangered species.
Genetic data supported animal movements between colonies, potentially favoring viral dispersal across the host population home range and likely allowing for viral maintenance in the area. In addition, genetic data from this study provide important information on the ecology of the species, suggesting that females might be more faithful to the natal area rather than to the natal roost. Besides being crucial to inform epidemiological modeling for disease dynamics, these results have relevant implications for the conservation of this endangered species, suggesting for example that monitoring and protection should be performed at the regional rather than local level, as animal movements between roosts might confound fluctuations in colony sizes.