Mete-data of biocrust cover in global drylands
We compiled data of biocrust cover in drylands from the published references in Web of Science during January 1980 and July 2019. Meanwhile, we also combined the unpublished materials, and recent reviews and books. References without reporting biocrust cover were excluded. We confined our research scope to drylands (including hyper-arid land, arid land, semiarid land and dry subhumid land) according to Aridity Index of around 0-0.65. However, cryptogam in dryland forest ecosystems were excluded in this study because they significantly differed from what we discussed here in terms of structures and functions. Data points without location information were also removed because location information was necessary to extract remote sensing data. Meanwhile, natural steady pattern of biocrust cover was of more relevance than the disturbed sites or sites that in the transient states to avoid possible interference to stable states (e.g., grazing, fire, trampling). Therefore, data points coming from the sites where were disturbed or were in transient phases were not included in this study. However, studies that reported sites which experienced long-term restoration >30 years (generally longer than at least one life history of vascular plants, and also estimated years to recovery of biocrusts) were generally kept because the restored ecosystems largely reached the final steady states. Biocrust cover was directly extracted from references which reported vegetation cover in ecosystem scale or indirectly calculated by us by averaging over multiple points in an ecosystem. If microhabitat (or microclimate) was reported, only data points in non-sunny slope were used to estimate biocrust cover. To avoid over-representation, data points among which the distances were smaller than 5 km were integrated and were regarded as a single point, assigned an averaged value.