Yam (Dioscorea spp.) samples (n = 690) from seven South Pacific Islands were screened for badnavirus infection by ELISA using two antisera to African badnaviruses. Positive readings were obtained for 26.4-34.6% of samples representing both known (D. bulbifera, D. nummularia and D. pentaphylla) and unreported host species (D. alata, D. esculenta, D. rotundata and D. trifida) in this region. Total DNAs were extracted from 25 ELISA-positive plants and 4 ELISA-negative controls and subjected to PCR amplification with badnavirus-specific primers targeting the reverse transcriptase (RT)-RNaseH genes. All 29 samples yielded the expected size PCR-product for badnaviruses, which were cloned and sequenced. Phylogenetic analyses of the resulting 45 partial (500-527 bp) RT-RNaseH sequences revealed 11 new sequence groups with <79% nucleotide identity to each other or any EMBL sequence. Three sequences (two groups) were highly divergent to the other nine new South Pacific yam badnavirus groups (47.9-57.2% identity) and probably represent either new Caulimoviridae genera or endogenous pararetrovirus sequences. Some sequence groups appeared specific to particular Dioscorea host species. Four 99.9% identical RT-RNaseH sequences possessing nine amino acid deletions from D. esculenta from three islands represent a putative integrated sequence group. The distribution of sequence groups across the islands indicates that badnaviruses have spread extensively between islands and continents through infected germplasm.

The discovery of endogenous pararetroviral sequences (EPRVs) has had a deep impact on the approaches needed for diagnosis, taxonomy, safe movement of germplasm and management of diseases caused by pararetroviruses. In this article, we illustrate this through the example of yam (Dioscorea spp.) badnaviruses. To enable progress, it is first necessary to clarify the taxonomical status of yam badnavirus sequences. Phylogeny and pairwise sequence comparison of 121 yam partial reverse transcriptase sequences provided strong support for the identification of 12 yam badnavirus species, of which ten have not been previously named. Virus prevalence data were obtained, and they support the presence of EPRVs in D. rotundata, but not in D. praehensilis, D. abyssinica, D. alata or D. trifida. Five yam badnavirus species characterised by a wide host range seem to be of African origin. Seven other yam badnavirus species with a limited host range are probably of Asian-Pacific origin. Recombination under natural circumstances appears to be rare. Average values of nucleotide intra-species genetic distances are comparable to data obtained for other RNA and DNA virus families. The dispersion scenarios proposed here, combined with the fact that host-switching events appear common for some yam badnaviruses, suggest that the risks linked to introduction via international plant material exchanges are high.