In Thuringia the first case of tularemia in a hare was reported in 2006 . In Lower Saxony 2,162 European brown hares and European rabbits (Oryctolagus cuniculus) were screened for tularemia between 2006 and 2009 using cultivation and PCR assays. Francisella specific PCR assays were positive in 23 hares and 1 rabbit which were further confirmed by cultivation of F. tularensis subsp. holarctica in 12 hares . In the present study, cases of tularemia in hares in Germany from 2005 to 2010 were investigated. During this period a total of 52 hares were found positive in PCR assays for F. tularensis subsp. holarctica DNA and from 31 of these cases Francisella strains could be isolated. MALDI-TOF analysis was also used to rapidly identify Francisella to the subspecies level as was previously shown by Seibold et al. .
Several positive specimens were found on the North Sea islands Langeoog and Spiekeroog (LS), around Soest (NR), Darmstadt (H), and Böblingen (BW). These natural foci and also sporadic cases in other regions of Germany were found below 500 m above sea level. In the Czech Republic typical natural foci of tularemia occurred in alluvial forests and field biotopes below 200 m sea level with mean annual air temperature between 8.1-10.0°C and mean annual precipitation of 450–700 mm . In Germany, an outbreak of tularemia in a colony of semi-free living marmosets was located in a region with geographic and ecological conditions similar to the hare habitats in the Czech Republic: field biotopes 175 m above sea level (<200 m) with 9.2°C mean annual air temperature and 642 mm mean annual precipitation . In Germany, tularemia of hares occurs in regions with rather humid soil like in alluvial forests and alongside rivers, but this obviously corresponds with the natural habitat of hares.
Specimens were screened using a PCR assay targeting Ft-M19 described by Johansson et al.  which allows the simultaneous identification of the species F. tularensis and the differentiation of the subspecies holarctica from other (sub-) species. All samples could be attributed to F. tularensis subsp. holarctica.
We found a clear segregation of clade B.I and clade B.IV in Germany, B.I strains dominate in eastern Germany and B.IV within western Germany (Figure 1). Clade B.I is known to dominate in Europe between Scandinavia and the Black Sea [15, 16, 21–23]. The other dominating European clade is B.IV (B.18) which can be found over a large area of western and central Europe, and, based upon this study, western Germany [21, 23–26]. We found only one strain of the B.II clade isolated in Bavaria. Strains of the B.II clade are most frequently isolated in the USA, but are found sporadically in Europe as well [16, 21].
The phylogeographical pattern of clade B.I and B.IV, coincide with the geographical distribution of biovar II and biovar I strains, respectively. Previously, biovar I strains (erythromycin sensitive) have been reported from Western Europe (France, Germany, Spain and Switzerland), North-America, Eastern Siberia and the Far East while biovar II is present in the European part of Russia as well as Northern, Central and Eastern Europe (Austria, Germany, Sweden and Turkey) [27–31]. A mixture of both biotypes has been reported in Sweden, Norway, Bulgaria, Russia and Kazakhstan [27, 28, 32]. Isolation of both biovars from rodents in a single settlement in Moscow as well as from water samples collected in the Novgorod region  indicate coexistence of the biovars in the same epidemiological foci. Taken together, a geographical separation of F. tularensis strains seems to exist in Germany. The phenotypically defined biovar I (erythromycin sensitive) and phylogenetically defined clade B.IV strains are confined in western Germany, whereas biovar II (erythromycin resistance) and clade B.I strains cluster in eastern Germany. This is interesting and may reflect a competition between the two subpopulations or unknown underlying ecological or epidemiological differences.
A deletion in the genome of F. tularensis subsp. holarctica in RD23 is typical for strains of F. tularensis subsp. holarctica in France, the Iberian Peninsula and also Switzerland, where biovar I predominates [24, 25, 27]. However, in one erythromycin susceptible isolate from Bavaria (08T0013), classified in this study as belonging to the B.II clade, RD23 was not deleted, thus showing that deletion of RD23 is not correlated with sensitivity to erythromycin. The molecular mechanisms of resistance to erythromycin have not been functionally established, but mutations identified in domain V of the 23S rRNA of biovar II strains, could provide a likely explanation .
Although 25 VNTR markers have been described for the typing of Francisella, it is pragmatic to investigate only loci of interest depending on the prevalent subspecies of F. tularensis, the efficiency of PCR assays for single loci, and existing data [1, 13, 34]. Sequence analysis of the locus Ft-M3 resulted in two different repeats denominated here as Ft-M3a corresponding with SSTR9E and Ft-M3b corresponding with SSTR9A as described previously by Johansson et al. . Johansson et al. and Byström et al. also found that locus Ft-M3 is the most variable marker [1, 13]. In the Francisella genome variations of DNA sequences in spite of identical repeat length have been described for short-sequence tandem repeats [35, 36]. Locus Ft-M6 showed less variability with only three PCR fragment sizes being observed among the strains. We obtained the same amplicon sizes that were described in previous studies for locus Ft-M3 (Additional file 1: Table S2) [14, 37] and for locus Ft-M6 (Additional file 1: Table S2) [14, 37]. Svensson et al. developed a sophisticated real-time PCR array for hierarchical identification of Francisella isolates . Only three (Ftind33, Ftind38, Ftind49) out of five INDEL loci were discriminatory among our set of F. tularensis subsp. holarctica isolates. Ftind48 is a marker for B.I to B.IV clades (non-japonica/non-california) and is not expected to vary for these isolates, and Ftind50 is targeting a specific deletion that so far only has been found in LVS. It was possible to simplify these assays to conventional PCR assays that allowed a simple read out based on gel electrophoresis. We identified clusters of strains that had the same INDELs and SNPs as strains described by Svensson et al. . In our study the analysis of VNTR and INDELs of two F. tularensis subsp. holarctica strains (06T0001, 10T0191) that were passaged twenty times in Ma-104 cells showed that these genomic elements were stable. Johansson et al. demonstrated for two VNTR loci (SSTR9 and SSTR16) that they were actually stable over 55 passages . The VNTR pattern for strains belonging to clade B.I was more variable compared with the pattern obtained for clade B. IV (Additional file 1: Table S2), as was observed previously [21, 23–25]. This might indicate that clade B.IV is more recently introduced in Germany than clade B.I.
We have applied several typing tools in a polyphasic approach in order to determine their value for identifying groups of Francisella strains in Germany. We found strains belonging to biovars I and II of F. tularensis subsp. holarctica. Although SNP loci are the most informative markers for typing of Francisella this method may have to be adapted to local strains [37, 38].