Despite these limitations, one clear point is that divergence times are three to ten times older for phylogroup 2 Pav than for phylogroup 1 Pav. Indeed, even the most rapid substitution rates result in estimated divergence times for both lineages that

predate the emergence of hazelnut decline by thousands of years. The finding that Pav has been diversifying for a long period of time without being observed in the field is surprising. In Greece, Pav had a particularly heavy impact on the hazelnut cultivar Palaz during the late 1970s [3]. This cultivar was introduced from Turkey in the late Stattic 1960s where there are no records of hazelnut bacterial canker. In contrast, there has been a long history of hazelnut cultivation in Italy, although the Palaz cultivar is not grown. Italian hazelnut cultivation AZD1390 increased rapidly during the decades leading up to the first observed outbreak during the 1970s, going from 3500 hectares in 1945 to almost 20,000 hectares by 1990 in the province of Viterbo [26]. Much of the new cultivation buy BLZ945 in both Greece and Italy

occurred on marginal lands with acidic soils, which are conditions that are likely to make hazelnut more susceptible to Pav infection. How can the long time since Pav divergence be reconciled with the recent occurrence of hazelnut decline? Microbiological surveys of in Italy have found that wild hazelnut trees are often infected by phylogroup 2 Pav[27], suggesting that wild trees might act as a reservoir. It is possible that phylogroup 1 Pav are associated with wild hazelnut in Greece, but similar surveys have not been carried out. Taken together, these data strongly suggest that both Pav lineages have been cryptically infecting hazelnut trees or wild relatives for a long time, and that the emergence of hazelnut decline in the 1970s was most probably due to changes in agricultural practice. While there is

no evidence of horizontal transfer between Pav lineages, we do find a large number of genes that have been horizontally acquired RANTES from other bacteria. Over 250 ORFs from the three Pav genomes lack orthologs in any other sequenced P. syringae strain. This includes over 200 genes that are present in one of the phylogroup 2 Pav strains but not the other, suggesting extensive gene acquisition and loss in this lineage. Over 80% of these genes have homologs in other Proteobacteria. Many of the strain-specific genes are organized into large genomic islands with signatures of mobile elements. Two of these genomic islands are homologous to regions found in other plant-associated bacteria, although the genetic similarity is low. This suggests either that the genetic exchange occurred in the distant past or that the donor strain is only distantly related to the sequenced strains in the database. It would be interesting to sequence other hazelnut-associated bacteria such Xanthomonas arbicola pv.