Integration of community ecology with disease biology is viewed as a
promising avenue for uncovering determinants of pathogen diversity, and
for predicting disease risks. Plant-infecting viruses represent a vastly
underestimated component of biodiversity with potentially important
ecological and evolutionary roles. We performed hierarchal spatial
analysis of wild plant populations to characterise the diversity and
coexistence structure of within-host virus communities, and their
predictors. Our results show that these virus communities are
characterised by single infections of few, dominating virus taxa as well
as diverse, non-random coinfections. Using a novel graphical modelling
framework we demonstrate that after accounting for environmental
heterogeneity at the level of both individual host plants and
populations, most virus co-occurrence patterns can be attributed to
virus-virus associations. Moreover, we show that conditioning variables
changed virus association networks especially through their indirect
effects. This highlights a previously underestimated mechanism of how
human-driven environmental change can influence disease risks.