We sequenced 12 bat brains positive for Lyssavirus antigen detected by immunofluorescence and reverse transcription–PCR (RT-PCR) as described (10). All viruses were identified as EBLV-1. For phylogenetic analyses, the 400-bp 5′ variable extreme of the nucleoprotein gene of these EBLV-1 strains was amplified by specific EBLV-1 nested RT-PCR and sequenced by using the following primers: SEQVAR1F 5′-₁ACGCTTAACAACCAGATCAAAG₂₂-3′, SEQVAR2F 5′-₅₁AAAAATGTAACACYYCTACA₇₀-3′, EBLVSEQVAR1R 5′-₅₉₆CAGTCTCAAAGATCTGTTCCAT₅₇₅-3′, and EBLVSEQVAR2R 5′-₅₅₂TAGTTCCCAGTATTCTGTCC₅₃₃-3′.

Figure 1

Figure 1. European bat lyssavirus 1 (EBLV-1) phylogenetic reconstruction based on the first 400 bp of the nucleoprotein gene. The tree was obtained by Bayesian inference run for 10⁷ generations; trees were sampled...

All rabies-positive serotine bats came from southern Spain (Huelva, Seville, Murcia, and Badajoz) and were molecularly identified as E. isabellinus (8). An alignment was performed by using ClustalX (www.clustal.org) to combine the obtained sequences and other available EBLV-1 sequences from GenBank, including a Duvenhage virus used as the outgroup (online Appendix Table, www.cdc.gov/EID/content/17/3/520-appT.htm). Before conducting further analyses, we used jModelTest (http://darwin.uvigo.es/software/jmodeltest.html) to select the best fitting substitution model for our sequences according to the corrected Akaike information criterion. Maximum-likelihood phylogenies were reconstructed by using PHYML (http://atgc.lirmm.fr/phyml) software and by using a generalized time-reversible model and the γ parameter estimated in the analyses. Maximum-parsimony analyses were conducted by using PAUP* 4.0b10 (http://paup.csit.fsu.edu/) weighting transversions 15× according to the transitions/transversion ratio estimated in the jModelTest analyses. Confidence in the topologies for the maximum-likelihood and the maximum-parsimony analyses was established with 1,000 bootstrap replicates. A Bayesian phylogenetic inference was obtained by using MrBayes version 3.1 (http://mrbayes.csit/fsu.edu/) with random starting trees without constraints. Two simultaneous runs of 10⁷ generations were conducted, each with 4 Markov chains, and the trees were sampled every 100 generations. Net p-distances between groups were calculated by using MEGA4 (www.megasoftware.net) (Figure 1).

The genetic structure and relationships between haplotypes were examined within the main lineages through a parsimony-based network built with a median-joining algorithm implemented in the Network 4.5.1 program (11). To evaluate and compare genetic variability and polymorphism among lineages, we estimated the number of haplotypes, mutations, and segregating sites as well as haplotype diversity and nucleotide diversity by using DNAsp version 4.5 (12) for the major clades (Table). Finally, to investigate population dynamics across lineages, the Fu Fs and Tajima D statistics were calculated (Table). These 2 statistics are considered to be the most powerful tests for detecting expansion events (13).

Figure 2

Figure 2. Geographic distribution of Eptesicus serotinus bats (red), E. isabellinus bats (blue), and cases of rabies in bats (green dots), Europe, 1990–2009. Obtained from Rabies Bulletin Europe (www.who-rabies-bulletin.org/).

All phylogenetic analyses, regardless of the reconstruction criterion used, formed a monophyletic cluster of the EBLV-1 strains from Spain (only the Bayesian inference reconstruction is shown). The Bayesian inference, maximum-likelihood, and maximum-parsimony analyses identified the cluster from Spain and EBLV-1a and EBLV-1b as being monophyletic (Figure 1), although only maximum-likelihood and maximum-parsimony analyses suggested a closer relationship between EBLV-1a and the cluster from Spain. The genetic differentiation of the EBLV-1 strains from the Iberian Peninsula matches their association with another bat species (Figure 2), which suggests that the host bat’s evolutionary history plays a major role in EBLV-1 molecular epidemiology, as has been proposed for rabies virus in bats in North America (14).

The low genetic diversity and the Fu Fs and Tajima D statistics (Table) all suggest rapid population expansion of EBLV-1a, which is consistent with the star-like structure of the network for this lineage (Figure 1). Conversely, haplotype and nucleotide diversity descriptors (Table) have the highest values for EBLV-1b and a complex network structure with differentiated subnetworks. All these elements indicate that this lineage has a complex evolutionary history. The lineage from Spain also has low diversity and a star-shaped network, but neutral evolution cannot be rejected on the basis of the Fs and D statistics. Net distances are similar within and between lineages, except for EBLV-1a, which is slightly more differentiated (Figure 1). Consequently, the suggested EBLV-1 expansion from Spain into Europe (15) is not supported by our results, which record the highest variability and most complex phylogenetic structure for France and the Netherlands (Figure 1). This complex structure suggests either a longer evolutionary history in these areas or a recent contact of distinct bat lineages in this zone.

The results of this study show that the strains from Spain do not belong to subtype 1b because of their association with a different reservoir (E. isabellinus bats). Moreover, what is currently considered to be EBLV-1b seems to include at least 4 lineages that are more genetically diverse and have a complex history. EBLV-1a, however, has low genetic diversity despite its extensive geographic distribution, suggesting a relatively recent and successful expansion of this lineage. These results call into question the current classification of EBLV-1 into 2 single subtypes. To provide a better understanding of EBLV-1 molecular epidemiology in Europe, additional studies that consider different genes should be conducted and the current classification should be revised accordingly.