Methodology

The Swanson ABC method: pathogen A — shared vector B — pathogen C. The A↔B link and the B↔C link are each taken from EPPO; the inferred A↔C is a hypothesis about a possible shared vulnerability through the vector.

A is drawn from regulated quarantine lists (Russia, the United Kingdom, Türkiye — the only ones with non-empty lists in the EPPO API; EU lists are centralised and not exposed through the API). C is every pathogen associated with the same vector in EPPO.

Scope: 99 regulated pathogens in role A, 264 chains in the main catalog.

The single data source is the EPPO Global Database. Every chain rests on two EPPO bibliographic references: one for the A↔vector edge and one for the vector↔C edge.

Pathogen, vector and crop names on the site are clickable and lead to EPPO taxon pages; some sources link straight to the publication via DOI.

• Vector (the primary bridge). Vector specificity = 1 / the number of pathogens it carries. The narrower the range carried, the more meaningful the link.
• Shared host-crop range. The share of common crops via the Jaccard coefficient (intersection over union of the host sets). Major host and generic Host classes are used; laboratory and weed hosts are excluded.
• Shared geography. The overlap of the distribution countries of A and C per EPPO.

Doubly confirmed = vector plus a shared host range. Triply = plus a country overlap.

The main catalog is ranked by non-obviousness (vector specificity multiplied by novelty from host-family distance): links between botanically distant crops, which a specialist would not keep in mind by default, rise to the top.

The confirmed section is ranked differently — by strength of confirmation, not by non-obviousness. The "strong" mark = a specific vector (no more than 6 pathogens carried) AND at least 2 shared crops AND a shared-host volume between 5 and 150 AND a Jaccard no lower than 0.1.

For confirmation the novelty metric is unsuitable: here shared hosts are a sign of link strength, not a penalty (in novelty, shared hosts instead lower the score).

• EPPO does not record vectors for many pathogens, especially new ones — vector coverage is incomplete.
• Most host records are generic Host with no class specified; this lowers the completeness of the host bridge.
• Seeds come only from the three countries with non-empty EPPO lists — the real regulated picture is broader.
• The method does not determine the direction of transmission.
• This is association mining from the EPPO literature, not a biological experiment; every hypothesis requires expert review.
• Crop convergence is an operational signal of shared crop risk, not Swanson novelty.
• Temporal dynamics (when the vector and both pathogens actually coincided) are not taken into account.