Over 4,200 grapevines showing yellows symptoms were geolocalised and sampled (fresh leaves) from 2015 to 2022 during official surveillance campaigns. Vegetal material was stored at 4 °C until total nucleic acids (TNA) extraction within three days of receipt.

Data were provided by the phytosanitary official services of the cantons. The geolocalisation of each plot containing positive samples was projected onto a map of the pertinent cantonal vineyard surfaces using the open-source software QGIS 3.34.2-Prizren. The data were organised and projected according to the year of detection to trace the progression of the epidemic.

The spatial distribution of malG genotypes was visualised using QGIS 3.34.2-Prizren. To improve map readability and prevent overlap between closely located samples, genotypes were aggregated within 3 × 3 km grid cells, and their relative frequencies were displayed as pie charts to summarise local diversity.

Approximately 1 g of leaf mid-vein tissue was used to extract total nucleic acids as previously described (Debonneville et al., 2022). FDp and Bois noir phytoplasma presences were tested by triplex quantitative PCR (qPCR) following a method adapted from (Pelletier et al., 2009).

TNA from seventeen plants were extracted and subsequently enriched with microbial DNA using NEBNext^® Microbiome DNA Enrichment Kit as described in (Debonneville et al., 2022). Illumina sequencing was performed by Macrogen (South Korea).

Raw Illumina reads were quality trimmed with the software Trimmomatic and mapped to FDp reference genome Bowtie 2 to exclude all reads not corresponding to FDp. Pre-processed Illumina reads were mapped to the reference genome with Bowtie2 and SNPs were detected using Geneious 2023.0.4. Parameters for variant calling were a minimum coverage of 10 and a minimum variant frequency of 80%. SNPs found in repeated regions or representing deletions and insertions in a base-repetition context were discarded due to low reliability.

Genetic characterization was done by multilocus sequence typing. The number of samples successfully analysed varied among loci, ranging from 48 to 675, and is detailed for each locus in Table 1. Loci reported in the literature were amplified by PCR or nested PCR using published primers and amplification programs. Primers for newly identified SNPs were designed using Geneious 2023.0.4. All primers and references are listed in Table 1. PCR conditions for primers ftsH5, ftsH7, ftsH17, rpoC, pcrA, helicase, intergenic 38.423 and dnaX were 2 min at 94 °C, 30 s at 94 °C, 35 cycles with one cycle consisting of 30 s at 55 °C, 1 min 30 s at 72 °C and final extension of 10 min at 72 °C. All amplification reactions were performed using GoTaq ^® G2 Flexi polymerase (Promega), except for malG reactions, for which a Q5 High-Fidelity DNA Polymerase (New England Biolabs) was used.

After quality control by electrophoresis on 1% agarose gel, PCR products were purified by ultrafiltration with NucleoFast 96 PCR plates (Macherey-Nagel) and vacuum processing. Products were sent to Fasteris (Switzerland) for forward and reverse sequencing using Sanger technology. Raw sequences were trimmed and de novo assembled. Resulting consensus were multi-aligned using MUSCLE algorithm and compared to references.

Expasy online tool¹ was used for nucleotide sequence translation into protein to define whether SNPs produce synonymous or non-synonymous translations.

Between 2015 and 2022, a total of 4,212 grapevines exhibiting symptoms characteristic of grapevine yellows, including leaf discoloration and downward rolling, withering of inflorescences and berries, and lack of shoot maturation, were geolocated and sampled as part of the national surveillance program conducted by the phytosanitary services of the Cantons of Vaud, Valais, and Geneva (Switzerland). Samples were analysed by quantitative PCR (qPCR) for the detection of Grapevine flavescence dorée phytoplasma (FDp) and Candidatus Phytoplasma solani (Grapevine bois noir phytoplasma), two causal agents of grapevine yellows that produce identical symptoms and can only be differentiated by molecular assays. Of the tested samples, 26.26% were positive for FDp, 49.76% for Bois noir (BN) phytoplasma, 0.17% tested positive for both pathogens, and 23.81% were negative for both. These results highlight the predominance of BN among symptomatic grapevines in the surveyed regions between 2015 and 2022 and indicate that dual infections with both phytoplasmas are exceptionally rare.

For analytical purposes, the study area was divided into five zones: Lavaux, Chablais, Central Valais, La Côte, and Geneva (Figure 1).

The two earliest FD outbreaks north of the Alps were recorded in the Lavaux region in 2015 and exhibited a typical epidemic expansion in subsequent years, characterized by concentric spread. Over time, additional clusters emerged, culminating in a peak of positive cases between 2018 and 2019. Although control measures led to an important reduction in incidence, new positive cases continued to be detected thereafter.

In the Chablais region, FDp was first identified in 2020, followed by a sharp increase in cases and the emergence of multiple new outbreak sites in subsequent years. In Central Valais, an isolated case was detected in 2016, but prompt management prevented further spread, and no cases were recorded in 2017. Nevertheless, an outbreak was confirmed in 2020, with another reported the following year approximately 10 km to the south-west.

Only isolated cases have been detected to date in Geneva (2021) and La Côte (2019 and 2022), with no evidence of sustained or onward transmission in these zones. The quality of the DNA extractions allowed us to use only samples from the outbreak of 2022 in La Côte for the MLST.

To better understand the introduction and spread of FDp in Swiss vineyards north of the Alps, a representative selection of over 700 qPCR-positive samples was analysed using a multilocus sequence typing (MLST) approach. Samples were selected based on their geographic origin and year of collection, with the aim of maximizing spatial and temporal coverage. The MLST analyses were conducted in two phases: the first phase targeted established genetic loci (map, dnaK, vmpA and malG) previously shown to exhibit sequence variability in other geographic regions. The second phase employed newly identified loci that demonstrated informative genetic variability specific to the Swiss FDp context.

A total of nine distinct genotypes were identified based on all observed combinations of variable loci present within individual samples (Table 3). Genotypic differentiation was achieved using the previously described malG locus (Rossi et al., 2019), in combination with the eight newly detected SNPs. Two genotypes were predominant: CH_A (corresponding to the malG G3/G3 matching the published FDp whole genome sequence; (Debonneville et al., 2022)) and CH_D. This was followed by CH_I, the sole genotype detected in Central Valais region, where all eight samples from different locations exhibited this genotype (Figure 4). Genotype CH_F was primarily found in the eastern area of Lavaux, whereas the remaining genotypes appeared to be rare.

Regional genotypic analyses showed contrasting levels of genetic diversity and spatial structure among FDp populations across Swiss vineyards north of the Alps (Figure 4). Chablais showed the highest diversity, with six multilocus genotypes detected. In contrast, only three genotypes (CH_A, CH_E, and CH_F) were found in Lavaux. SNP analysis within the malG G3/G3 group revealed that the FDp population in La Côte differed from that in Lavaux by one base. In Geneva, the quality of the Illumina sequencing for the malG G3/G3 sample was insufficient to allow SNP-based genotyping, but the absence of the SNP identified in the 2020 outbreak of La Côte was confirmed. Finally, the population structure in Central Valais remained unchanged, as no SNPs were detected within the malG G1/G1 group. Interestingly, CH_E was detected in only three samples within a predominantly CH_A area in Lavaux. Both genotypes share the same combination of newly detected SNPs and differ solely at the malG locus. A similar pattern was observed with genotype CH_B, found in two samples within a CH_D-dominated region in Chablais, also sharing SNP combinations and differing only by the malG locus.

The map locus proved highly informative for investigating the origin of FDp in the studied region, where a single map genotype, M54, was consistently detected across all samples spanning more than 100 km. This genotype is almost exclusively associated with the Vitis vinifera–Scaphoideus titanus pathosystem (Malembic-Maher et al., 2020) and has rarely been reported in alternative perennial hosts (Casati et al., 2017; Moussa et al., 2023; Rigamonti et al., 2023). Moreover, the initial outbreaks of FD occurred in the middle of continuous vineyard areas, far from potential alternative hosts and geographically isolated from the closest infested zones. Taking into account all this information, we hypothesize that FDp was most probably introduced primarily through infected grapevine planting material. Nonetheless, further investigations are ongoing to clarify the potential role of alternative insect vectors and plant hosts in local FDp persistence and spread.

This low diversity in the map locus contrasts sharply with observations in other European regions, where much higher diversity has been documented over relatively short distances. For instance, four map genotypes were identified within 240 km in Croatian grapevines (Plavec et al., 2019), six within 50 km in Slovenia (Kogej Zwitter et al., 2023), and four within 140 km in France’s Aquitaine region (Malembic-Maher et al., 2020). Elevated genotype richness, particularly in the Balkans, likely reflects the presence of FDp in alternative plant hosts, which have been hypothesized to be the origin of the epidemics in Serbia and Montenegro (Krstić et al., 2022 and Radonjić et al., 2023, respectively). In the studied region, this low diversity might be due to scarce introductory events together with the recent introduction of the phytoplasma.

MLST analysis using the additional known loci vmpA, dnaK and malG confirmed very low genotypic diversity in the FDp populations studied: no variation was detected in vmpA, or dnaK, and only limited variation was observed at the malG locus, identifying three genotypes (G1/G1, G1/G3, and G3/G3). This contrasts with findings from other regions: for instance, Rossi et al. (2019) reported up to 11 malG types in grapevines in the Piedmont area.

To better resolve local epidemic dynamics and assess potential links between geographically close outbreaks, we sought to increase the genotyping resolution beyond what was possible with malG alone. Indeed, the three malG genotypes (G1/G1, G1/G3, and G3/G3) suggested only a limited number of populations, which did not fully account for the observed spatial complexity and temporal progression of the epidemic. We therefore explored genome-wide SNP variation within these groups; an approach that, to our knowledge, is novel in FDp population studies. This enabled the identification of eight additional variable loci, primarily within the G1/G3 group, increasing resolution from three to nine multilocus genotypes.

Using the refined genotyping approach, only one genotype was detected from geographically distant samples in Central Valais, consistent with a single introduction followed by local spread, likely mediated by infected S. titanus. The initially scattered distribution observed until 2022 may be attributable to delayed symptom expression caused by the latency period of phytoplasma development within the host or by incomplete surveillance. By 2023, vineyards previously free of infection between positive plots had become affected (data not shown).

In Lavaux, the two dominant genotypes showed clear geographic segregation, likely influenced by a motorway acting as a physical barrier to insect movement (Figure 2b). CH_E was detected in only three samples from spatially scattered locations within an area otherwise dominated by CH_A. Notably, these CH_E samples carried a malG G1/G3 genotype, contrasting with the surrounding population, which was overwhelmingly characterised by malG G3/G3. Interestingly, CH_E shares the same SNP profile as both CH_A (from Lavaux) and CH_I (from Valais), differing only at the malG locus. This close genetic similarity suggests a recent divergence or a shared ancestral origin.

The rarity and scattered presence of CH_E on three different locations in Lavaux region raise questions about its origin. Given its extremely low frequency and its occurrence within a largely homogeneous area, the most parsimonious explanation is the introduction of this genotype through planting material, either via a shared source used for vine replacement or through a single introduction event followed by local dissemination. Its low frequency might reflect limited dispersal or establishment capacity. Alternative explanations cannot be excluded, and CH_E could have emerged locally from a CH_A background through changes affecting one of the two malG copies, converting from malG3 to malG1, potentially within the insect vector. Rossi et al. (2019) reported that approximately 50% of the 183 malG types detected in their study were found exclusively in insects, and suggested a high variation rate of FDp populations within the vector as one possible explanation. Nevertheless, the plausibility of a direct malG3-to-malG1 shift is highly questionable, as it would require at least four nucleotide changes. Homologous recombination occurring in the insect hosting different variants of malG could also theoretically be envisaged, but phytoplasmas generally show poor recombination capacity (Bai et al., 2006), and the FDp genome lacks potential mobile units (PMUs) typically involved in such processes (Debonneville et al., 2022). In the absence of experimental evidence, the origin of these strains, which differ exclusively at the malG locus, remains unresolved. Intriguingly, a similar situation was observed in the Chablais region, where the rare genotype CH_B coexisted with the predominant CH_D, sharing identical SNP profiles and differing only at the malG locus.

The Chablais region showed a more complex, mixed distribution of six genotypes across disconnected vineyard patches separated by distances exceeding 3 km, which surpasses the currently known active flight range of S. titanus. Studies have demonstrated that S. titanus typically flies within 30 m for 80% of its movements, with occasional flights up to 330 m (Lessio et al., 2014), and recent evidence supports active and common movement up to at least 210 m (Petit, 2023). Moreover, human-assisted movement via agricultural machinery has been shown to transport S. titanus over distances up to 2.5 km (Chavarri-Padilla, 2023), and wind can also act as a passive dispersal mechanism (Petit, 2023). Vine growers may manage several spatially dispersed plots across different villages, use shared agricultural machinery (including service providers operating the same equipment across multiple vineyards), and exchange planting material for dead vine replacements. However, detailed plot-level information on these practices is generally not recorded and was not available for the vineyards where positive samples analysed in this study were identified. The causes of the complex epidemiological dynamics observed in the Chablais region remain uncertain. Nevertheless, the hypothesis of a more pronounced influence of region-specific passive diffusion processes points to the possible role of local socio-agricultural determinants.

Overall, these spatial and genetic patterns are consistent with a scenario in which the spread of FDp in Swiss vineyards may be shaped by two main forces: first, the natural epidemiological expansion of the phytoplasma via its insect vector in regions where grapevine cultivation is spatially continuous, such as Lavaux, and second, a combination of anthropogenic and ecological drivers of dispersal. Anthropogenic factors may include the movement of infected planting material, vineyard management practices, and physical barriers such as roads or motorways, while ecological drivers include wind-mediated vector dispersal and the fragmented structure of certain viticultural landscapes, as observed in Chablais. In this context, the presence of rare genotypes in spatially distant or disconnected plots suggests that vector movement alone may not fully account for the observed patterns. Rather, the interplay between local environmental conditions and human-mediated practices could have contributed to shaping the epidemic landscape. Together, these observations highlight the value of combining molecular surveillance with landscape-level epidemiological monitoring to improve our understanding of FDp spread and inform disease management strategies.