Descriptive analyses
The virus communities exhibit a significantly nested structure: the mean C-score over the viruses was 290.52 at the individual plant level (with the maximum of 87155 possible checkerboard units); and 2.97 at the population level (with the maximum of 890 possible checkerboard units) (Figure 2B). Almost all viruses (24 OTUs) were present in the most virus-rich host population (population 861; Figure 2B and 2D), while other populations hosted smaller subset of the viruses, with the simplest metacommunity consisting of six viruses (population 3222; Figure 2B and 2C).
All the viruses co-occurred with another virus at least once (Figure 3A). The most abundant viruses, Closteroviridae and Caulimoviridae, occurred as single infections in 39 and 33 host plants, respectively, which made up nearly one third of their total occurrences (133 and 122, respectively). The number of unique virus co-occurrence combinations (of two viruses or more) in the whole data set was 111. Only Avsunoviroidae and Alphasatellitidae never co-occurred (of which the latter was so rare that it was not included in the network models, see below).
The species-area curve shows that all 25 viruses have been encountered when on average 264 plants have been sampled (Figure 3B). The difference between the species-area and coexistence-area curves shows how the cumulative sum of co-occurring viruses is lower than the cumulative viral richness would predict, until the coexistence curve reaches its maximum at 387 sampled host plants. For example, when 100 host plants have been sampled, the number of viruses observed was 20, which would enable 190 co-occurring pairs of viruses while the average number of observed co-occurring virus pairs at that point was 168.