The data concerning the occurrence of BGC outbreaks in Germany was obtained from the Animal Disease Information System (https://www.tsis.fli.de/) hosted by the Friedrich-Loeffler-Institut, the German Federal Research Institute for Animal Health. This data was curated from reports submitted by competent veterinary authorities of the federal states to the Federal Ministry of Food and Agriculture following the German Animal Health Act on notifiable animal diseases.

Sixty-three Cfv strains were provided from the German NRL for Bovine Genital Campylobacteriosis (Figure 1, Supplementary Table S1). Out of 63 strains, 57 and six were collected between 2001 and 2015 and between 1985 and 1993 in Thuringia, respectively. Among them, five Cfv in the period 1985 to 1988 were isolated in the GDR (East Germany) and were sent from Thuringian laboratories to the German reference laboratory only after the re-unification of both German states (17).

The genomic DNA was extracted using DNeasy Blood and Tissue kit (QIAGEN^®, Hilden, Germany) as per the manufacturer's instructions. The DNA was eluted in 200 μl elution buffer. DNA was quantified spectrophotometrically using a Nanodrop^® ND-1000 (Fisher Scientific GmbH, Schwerte, Germany). The quality of the DNA was determined using the Qubit dsDNA BR Assay Kit (Invitrogen, Carlsbad CA, USA). The extracted DNA was sequenced using an Illumina MiSeq2000 platform with Nextera XT sequencing library kits to generate 300 bp paired-end reads (Illumina, https://www.illumina.com) and an average sequencing depth of 108-fold (range, 38–226-fold) for all genomes. In addition, six genomes previously sequenced using the Illumina Hi-Seq platform and having an average sequencing depth of 328-fold (range, 88–507) were included (18) (Supplementary Table S1). For all genomes, fastp (v0.23.2) (19) was used to report sequencing metrics followed by taxonomic classification of sequence reads using the Kraken2 database (20). Genome assembly was performed using shovill (v1.0.4) (21) and information about the assembled genomes was collected with seqkit (v2.3.1) (22) (Supplementary Table S3). We deposited the genome sequence data at the National Center for Biotechnology Information (NCBI) under the BioProject accession number PRJNA891116.

For molecular genotyping, all genomes were screened with cfvCatch (no version; https://gitlab.com/FLI_Bioinfo/cfvcatch; accessed on 01.2022) (7). cfvCatch reports the phylogenetic position of C. fetus genomes, as well as 11 nucleotide markers specific for clade 1 genomes (Supplementary Table S1). In addition, it predicts C. fetus-specific PCR amplicons, the insertion element ISCfe1, and MLST sequence types (Supplementary Table S1).

The raw read sequence data obtained from 63 Cfv in this study was compared with sequence data from NCBI (Supplementary Table S2). The software snippy (v4.6, https://github.com/tseemann/snippy) was used in default mode to identify sequence variants with the Cfv genome 97/608 (accession number CP008810) as a reference. The resulting alignment file was filtered using Gubbins (v2.4.1) (23) and HomoplasyFinder (no version) (24) applying default settings to mask putative recombinant regions and homoplastic SNPs, respectively. Based on a stripped alignment of 1,619 variant sites, a maximum likelihood (ML) tree was built with RAxML-NG (v1.1) (25) using the GTR + Γ nucleotide substitution, Lewis' correction of ascertainment bias and 650 bootstrap replicates (the bootstrapping test estimated applying “majority rules extended” (MRE) was converged after 650 trees). The Cfv tree was mid-point rooted based on the phylogenetic structure of the C. fetus species (7, 18). The ML tree was visualized using iTOL (v5) (26).

To estimate the divergence time of Cfv lineages, BactDating (v1.1; https://github.com/xavierdidelot/BactDating) was applied to the recombination-filtered phylogeny using the additive relaxed clock (ARC) model (27) and 100 million Markov chain Monte Carlo (MCMC) iterations sampled every 10⁴ steps (28). The effective sample sizes (ESS) were computed for the parameters α, μ, and σ >200. The MCMC convergence was verified by visualizing traces of the model parameters. The estimated emergence time of Cfv lineages were transposed onto the ML phylogeny produced from SNP analysis.

The WOAH defines an outbreak as one or more animals infected with a pathogen, with or without clinical signs, in an epidemiological unit (29). According to this definition, there have been a total of 68 BGC outbreaks in Germany since 1994, as recorded in the Animal Disease Information System reports (https://www.tsis.fli.de/; accessed on 01.03.2022). BGC has been reported in nine federal states, with a range of zero to nine outbreaks per year. The spatial and temporal distribution of reported BGC outbreaks is depicted (Figure 1) and shows that most outbreaks occurred in Southern Germany: Bavaria (n = 42) and Baden-Wuerttemberg (n = 14), followed by Lower Saxony (n = 4), North Rhine-Westphalia (n = 3), and Saxony (n = 2). One BGC outbreak each was reported in the federal states of Thuringia, Brandenburg, Rhineland-Palatinate, and Saxony-Anhalt, while no reports exist for the other federal states (Figure 1B). From 2001 to 2009, 45 outbreaks, and since 2010, 12 outbreaks were reported.

Whole-genome sequencing was used to characterize 63 C. fetus strains from German BGC cases. These strains were collected from seven out of nine BGC-affected federal states. Out of them, nine strains had no known date of isolation (three from Baden-Wuerttemberg, one from Bavaria and five from Lower Saxony). No strains were available from North Rhine-Westphalia, Rhineland-Palatinate, Saxony and Saxony-Anhalt (Figure 1, Supplementary Table S1).

All German strains were assigned to clade 1, the bovine-specific clade of C. fetus (9, 18), recently proposed as Cfv (7). These strains carry ISCfe1 and parA genes [characteristic markers of Cfv subspecies (4)], belong to MLST sequence type (ST) 4 except for one strain of ST75, and show positive alignment to Cfv-specific PCR primers with the successful prediction of Cfv PCR amplicons (Supplementary Table S1).

For a comparative phylogeographic analysis of the 63 German Cfv strains, sequences of an additional 91 geographically-diverse strains collected in 11 countries and previously assigned to clade 1 (Cfv clade) were downloaded from the NCBI database (Supplementary Tables S1, S2). The mutation rate for the Cfv clade was estimated to be 1.47 × 10⁻⁷ s/s/y (substitutions per site per year; 95% CI = 0.64 × 10⁻⁷ – 2.2 × 10⁻⁷), i.e., an average of ~0.3 substitutions in the 2 Mb Cfv genome occurs every year. This is 150-fold lower than previous estimates reported for C. fetus species (2.9 × 10⁻⁵ s/s/y), reflecting the conserved genetic diversity possibly due to the clade's restrictive adaptation to cattle (18). Similarly, van der Graaf–van Bloois et al. (30) reported a molecular clock rate of 1.5–3.5 × 10⁻² substitutions per kb per year for mammalian-associated C. fetus, with the Cfv genome-cluster having a high dN/dS ratio (0.4) suggesting recent diversification (30). The use of genetic data to estimate evolutionary rates and timescales of pathogens provide valuable insights into their biological processes. These estimates, however, can be influenced by access to informative sequence data and careful model selection (31).

The phylogeny of global Cfv reveals two main subclades (1.1 and 1.2) and nine descending genetic lineages named hierarchically (Figure 2A). The average number of SNP differences over sequence pairs within the identified lineages was 46 (13–88), between the lineages was 154 (60–245) and between the two main subclades was 215 SNPs. Importantly, some of the identified Cfv genetic lineages display distinct geographic features with strains from one country grouped into a single lineage (e.g., Germany and Canada) or in multiple lineages (e.g., Argentina and Spain; Figure 2A). The absence of geographic clustering of some other lineages was mainly confined to regions (and strains) currently underrepresented in our data set.

The so-called South American strains constitute the ancestral population of subclade 1.1 and might be the origin of this subclade. This was extrapolated by first confirming the existence of a significant temporal signal across this clade evidenced by a positive correlation (R² = 0.04) between the root-to-tip genetic distance and the sampling times (p = 6.51 × 10⁻³; Supplementary Figure 1). The temporal estimates support the phylogenetic branching order from the recombination-corrected SNP analysis. The most recent common ancestor of the South American strains corresponds to the branching of subclade 1.1 into seven lineages which are dated to 1595 (95% CI = 1109–1778), indicating earlier divergence than the other strains (estimated branching from 1660 to 1839; Figure 2A). This was also reflected in the pairwise nucleotide diversity of the South American strains that was higher (52–88 SNP differences) compared to strains from other countries in subclade 1.1 (6–67 SNPs).

Phylogenetic analysis of Cfv strains clusters all German strains (1985–2015) into one genetic lineage (1.1.1) within the subclade 1.1 (Figure 2A) with an average pairwise genetic distance of 43 SNPs. Interestingly, this lineage included strains recovered at the time of the GDR (East Germany, TH-1) before re-unification. This suggests that the strains of the GDR and Germany have a common ancestor and that the date of Cfv introduction into Germany predates the division of the country after World War II (WWII). We estimated that the most recent common ancestor of the German strains circulated since the nineteenth century (1839; CI: 1645–1912; Figure 2A). The grouping of Cfv strains from both German states into one lineage is striking. This lineage might have persisted under the tight physical borders and the different farming and breeding systems present for nearly 45 years in both German states during the Cold War times. For the Thuringian strains isolated in the 1980's, no metadata are available. Thus, it is not clear if they were isolated from endemic cattle as no official records were made. Notification of BCG outbreaks by official state veterinarians can be taken for granted, considering that BCG was a notifiable disease in the GDR. Also, for the Thuringian strain of 1993, no metadata are available and no BGC notification can be identified either. One can speculate that it originated from animal trade as a consequence of the replacement of the population after re-unification which leaves the question unanswered why this sublineage (SL) was found 1985–1988 already in the GDR. These events stay cryptic. Further socio-economic research is also required to clarify why other strains were not introduced into Germany and whether the German lineage was not exported. It is plausible that the spread of Cfv across German borders is limited today due to strict eradication programs in Germany and tight international trade controls. In addition, the fact that the host range is limited to cattle restricts the dynamics of the pathogen to the movement of animals or the transport of animal products, such as infected semen, thus promoting control of the spread. A similar observation can be made in Spain (Figure 2A). Those strains (2002–2014) are distributed in two genetically distinct lineages that include only Spanish strains, with a mean pairwise genetic distance of 9 and 13, respectively (Figure 2A).

At the country level, the German lineage has diversified into two major lineages [German Lineage 1 (G-1) and G-2] and five sublineages (BW, BA, TH-1, LS, TH-2; Figure 2B). The first lineage (G-1) comprises 44 strains isolated in Southern and Middle Germany. This lineage has a common ancestor dated earlier [1892 (CI: 1790–1943)] than the second lineage [1943 (CI: 1860–1977)] suggesting a later diversification of the second lineage (Figure 2B). The second lineage (n = 19 strains) occurs mainly in Northern Germany, but was also occasionally detected in recent years in Southern Germany, Thuringia, and Brandenburg (Figure 2B). Interestingly, the distribution of German Cfv lineages is in line with the distribution of German cattle breeds. For example, specialized dairy cattle (Niederungsvieh, lowland cattle) is bred in Northern Germany, while mixed-purpose cattle (Höhenvieh, highland cattle) is bred in Southern Germany (32). This North-South division is further supported by genetic studies of bovine Y-chromosomes. Unfortunately, no data was available on the cattle breed from which Cfv was recovered.

The five German sublineages (SL) show the distinct geographic distribution in federal states. Hence, these sublineages were labeled according to the major representative state (Figure 2B). The first SL (BW SL) includes 18 strains from Baden-Wuerttemberg (2001–2006) and three strains from Bavaria (2005, 2006, and 2015; Figure 2B). The clustering of strains from Baden-Wuerttemberg and Bavaria can presumably be explained by the geographic proximity of these two states or a sporadic transmission from Baden-Wuerttemberg to Bavaria. However, the detailed epidemiological information of the three Bavarian strains was unavailable to confirm this.

The second SL (BA SL) was only confined to Bavaria and comprises 17 strains (2004–2008; Figure 2B). Thus, the co-circulation of two SLs (BA SL and BW SL) in Bavaria is likely as Bavaria has the largest cattle population and bull semen production centers nationwide.

The third SL (TH-1 SL) includes six strains labeled and cataloged in Thuringia between 1985 and 1993; five strains date back to the time of the GDR (1985–1987). Interestingly, the GDR strains cluster separately from the other German strains, reflecting the restriction of cattle trade between both states. The fourth SL (TH-2 SL) was also detected in Thuringia, including four identical strains (zero SNPs). Three strains were obtained from prepuce and semen of two bulls in March 2004 in Thuringia and the same strain was obtained from an aborted fetus in Baden-Wuerttemberg one month later, suggesting an epidemiological link.

The fifth SL (LS SL) contains 11 strains from Lower Saxony and two strains isolated from Baden-Wuerttemberg (2004) and Bavaria (2007). It also includes the only two strains isolated in Brandenburg in 2013. The Brandenburg strains are genetically close to Lower Saxony strains (pairwise SNPs, 0–7). The TH-2 SL and LS SL prove that with the replacement of GDR stocks with Western German cattle also Cfv strains were introduced to Eastern German federal states, Thuringia and Brandenburg.