Data for: Metagenomics show high spatiotemporal virus diversity and ecological compartmentalisation: virus infections of melon, Cucumis melo, crops and adjacent wild communities

Emergence of viral diseases results from novel transmission dynamics between wild and crop plant communities. The bias of studies towards pathogenic viruses of crops has distracted from knowledge of non-antagonistic symbioses in wild plants. Here we implemented a high throughput approach to compare the viromes of melon (Cucumis melo), and wild plants of crop (Crop) and adjacent boundaries (Edge). Each of the 41-plant species examined was infected by at least one virus. The interactions of 104 virus operational taxonomic units (OTUs) with these hosts occurred largely within ecological compartments of either Crop or Edge, Edge having traits of a reservoir community. The positive correlation of virus and plant richness at each site, the tendency for increased specialist host use through seasons, and specialist host use by OTUs observed only in Melon, characterised local-scale patterns of infection. In this study of systematically sampled viromes of crop and adjacent wild communities most hosts showed no disease symptoms, suggesting non-antagonistic symbioses are common. The coexistence of viruses within species-rich ecological compartments of agro-systems might promote the evolution of a diversity of virus strategies for survival and transmission. These communities, including those suspected as reservoirs, are subject to sporadic changes in assemblages, and so too are the conditions that favour the emergence of disease.

Methods
Four sites where Cucumis melo var. Piel de Sapo was grown (Crop habitat sites C1, C2, C3, and C4, from here, Crop) were compared to relatively permanent communities that form the narrow borders (Edge habitat sites E1 and E3, here from Edge) that separate crops. At each site, 50 plant samples from a 25 m x 2 m area were collected systematically at each resampling according to fixed itineraries, regardless of plants showing symptoms of virus infection or not. Individual RNA extracts from the same plant species and collection (i.e. same time, same site) were pooled to obtain a single HTS library. Paired-end reads of 125 or 150 nt. were sequenced on Illumina HiSeq platforms. Approximately 8.0 x 10^6 reads per library were sequenced. Local BLAST queries against reference genomes of plant viruses. The following validation criteria were used to subset BLAST query output: 1) query coverage of 100%; 2) query length greater or equal to 125 nt.; 3) paired-read matches only; 4) unique virus matches within each library; and 5) the difference between maximum and minimum query start positions in the reference genome had to span more than 1% of the virus genome length. In this way, at least two pairs of reads were required to match the reference. The first two steps of the validation criteria were performed to provide a less-restrictive indication of virus presence, then the last three steps applied to reduce the potential for the presence of false positive detections. Validated OTUs were used to generate plant-virus interaction networks.