Abstract Fungi are major drivers of ecosystem functions. Increases in aridity are known to negatively impact fungal communities in dryland ecosystems globally, however, much less is known on the potential influence of other environmental drivers. To fill this knowledge gap, we reanalyzed fungal data from 912 soil samples, providing the largest and most complete fungal community dataset from global drylands. We used machine learning tools to examine geographical patterns in community composition and spatial, edaphic, and climatic factors driving them. Further, we determined critical thresholds of community turnover along those gradients. Our analysis identifies UV index, climate seasonality, and sand content as the most important environmental predictors of community shifts, harbouring greatest association with the richness of putative plant pathogens and saprobes. Important nonlinear relationships existed with each of these fungal guilds, with increases in UV and temperature seasonality above 7.5 and 900 SD, respectively, being associated with an increased probability of plant pathogens and unspecified saprotrophs occurrence. Conversely, these environmental parameters had a negative relationship with litter and soil saprotrophs richness. Consequently, these functional groups might be differentially sensitive to environmental changes, which might result in an inevitable disturbance of current plant-soil dynamics in drylands.